annotate src/plugins/fts-squat/squat-trie.c @ 4887:463a4ebba685 HEAD

Another try at memory reduction / leak fixes
author Timo Sirainen <tss@iki.fi>
date Sun, 10 Dec 2006 01:18:58 +0200
parents ae2114f11a0e
children cb3b4153136c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1 /* Copyright (C) 2006 Timo Sirainen */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
2
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
3 #include "lib.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
4 #include "array.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
5 #include "bsearch-insert-pos.h"
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
6 #include "file-cache.h"
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
7 #include "file-lock.h"
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
8 #include "istream.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
9 #include "ostream.h"
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
10 #include "read-full.h"
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
11 #include "write-full.h"
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
12 #include "mmap-util.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
13 #include "squat-uidlist.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
14 #include "squat-trie.h"
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
15
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
16 #include <stdio.h>
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
17 #include <stdlib.h>
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
18 #include <unistd.h>
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
19 #include <fcntl.h>
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
20 #include <ctype.h>
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
21
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
22 #define TRIE_COMPRESS_PERCENTAGE 30
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
23 #define TRIE_COMPRESS_MIN_SIZE (1024*50)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
24
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
25 #define SQUAT_TRIE_VERSION 1
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
26 #define SQUAT_TRIE_LOCK_TIMEOUT 60
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
27
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
28 /* for non-x86 use memcpy() when accessing unaligned int* addresses */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
29 #if defined(__i386__) || defined(__x86_64__)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
30 # define ALLOW_UNALIGNED_ACCESS
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
31 #endif
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
32
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
33 #define BLOCK_SIZE 4
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
34
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
35 #define ALIGN(size) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
36 (((size) + sizeof(void *)-1) & ~((unsigned int) sizeof(void *)-1))
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
37
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
38 struct squat_trie {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
39 char *filepath;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
40 int fd;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
41 dev_t dev;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
42 ino_t ino;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
43
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
44 enum file_lock_method lock_method;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
45 struct file_lock *file_lock;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
46 int lock_count;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
47 int lock_type; /* F_RDLCK / F_WRLCK */
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
48
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
49 struct file_cache *file_cache;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
50 uint32_t file_cache_modify_counter;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
51
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
52 void *mmap_base; /* NULL with mmap_disable=yes */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
53 const uint8_t *const_mmap_base;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
54 size_t mmap_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
55
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
56 const struct squat_trie_header *hdr;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
57 uint32_t uidvalidity;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
58
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
59 char *uidlist_filepath;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
60 struct squat_uidlist *uidlist;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
61 struct trie_node *root;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
62 buffer_t *buf;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
63
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
64 unsigned int corrupted:1;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
65 unsigned int mmap_disable:1;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
66 };
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
67
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
68 struct squat_trie_build_context {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
69 struct squat_trie *trie;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
70
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
71 struct ostream *output;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
72
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
73 uint32_t prev_uid;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
74 unsigned int prev_added_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
75 uint16_t prev_added[BLOCK_SIZE-1];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
76
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
77 unsigned int node_count;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
78 unsigned int deleted_space;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
79
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
80 unsigned int modified:1;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
81 unsigned int failed:1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
82 unsigned int locked:1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
83 };
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
84
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
85 struct squat_trie_compress_context {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
86 struct squat_trie *trie;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
87
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
88 const char *tmp_path;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
89 struct ostream *output;
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
90 int fd;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
91
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
92 struct squat_uidlist_compress_ctx *uidlist_ctx;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
93
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
94 unsigned int node_count;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
95 };
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
96
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
97 struct squat_trie_header {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
98 uint8_t version;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
99 uint8_t unused[3];
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
100
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
101 uint32_t uidvalidity;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
102 uint32_t used_file_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
103 uint32_t deleted_space;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
104 uint32_t node_count;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
105 uint32_t modify_counter;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
106
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
107 uint32_t root_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
108 };
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
109
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
110 /*
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
111 packed_node {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
112 packed ((8bit_chars_count << 1) | have_16bit_chars);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
113 uint8_t 8bit_chars[8bit_chars_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
114 uint32_t idx[8bit_chars_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
115 if (have_16bit_chars) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
116 packed 16bit_chars_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
117 uint16_t 16bit_chars[16bit_chars_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
118 uint32_t idx[16bit_chars_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
119 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
120 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
121 */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
122
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
123 struct trie_node {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
124 /* new characters have been added to this node */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
125 uint8_t resized:1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
126 /* idx pointers have been updated */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
127 uint8_t modified:1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
128 uint8_t chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
129 uint16_t chars_16bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
130
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
131 uint32_t file_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
132 uint32_t orig_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
133
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
134 /* the node pointers are valid as long as their lowest bit is 0,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
135 otherwise they're offsets to the trie file (>> 1).
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
136
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
137 in leaf nodes the children pointers are uint32_t uid_list_idx[]; */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
138 /* uint8_t 8bit_chars[chars_8bit_count]; */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
139 /* struct trie_node *children[chars_8bit_count]; */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
140 /* uint16_t 16bit_chars[chars_16bit_count]; */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
141 /* struct trie_node *children[chars_16bit_count]; */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
142 };
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
143 #define NODE_CHARS8(node) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
144 (uint8_t *)(node + 1)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
145 #define NODE_CHILDREN8(node) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
146 (struct trie_node **) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
147 ((char *)((node) + 1) + \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
148 ALIGN(sizeof(uint8_t) * ((node)->chars_8bit_count)))
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
149 #define NODE_CHARS16(node) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
150 (uint16_t *)((char *)NODE_CHILDREN8(node) + \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
151 sizeof(struct trie_node *) * ((node)->chars_8bit_count))
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
152 #define NODE_CHILDREN16(node) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
153 (struct trie_node **) \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
154 ((char *)NODE_CHARS16(node) + \
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
155 ALIGN(sizeof(uint16_t) * ((node)->chars_16bit_count)))
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
156
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
157 static void free_node(struct trie_node *node, unsigned int level);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
158 static int
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
159 squat_trie_compress_node(struct squat_trie_compress_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
160 struct trie_node *node, unsigned int level);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
161 static int trie_write_node(struct squat_trie_build_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
162 unsigned int level, struct trie_node *node);
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
163 static int
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
164 squat_trie_build_flush(struct squat_trie_build_context *ctx, bool finish);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
165
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
166 static int chr_8bit_cmp(const void *_key, const void *_chr)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
167 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
168 const uint8_t *key = _key, *chr = _chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
169
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
170 return *key - *chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
171 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
172
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
173 static int chr_16bit_cmp(const void *_key, const void *_chr)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
174 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
175 const uint16_t *key = _key, *chr = _chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
176
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
177 return *key - *chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
178 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
179
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
180 void _squat_trie_pack_num(buffer_t *buffer, uint32_t num)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
181 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
182 uint8_t c;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
183
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
184 /* number continues as long as the highest bit is set */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
185 while (num >= 0x80) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
186 c = (num & 0x7f) | 0x80;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
187 num >>= 7;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
188
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
189 buffer_append(buffer, &c, 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
190 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
191
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
192 c = num;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
193 buffer_append(buffer, &c, 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
194 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
195
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
196 uint32_t _squat_trie_unpack_num(const uint8_t **p, const uint8_t *end)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
197 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
198 const uint8_t *c = *p;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
199 uint32_t value = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
200 unsigned int bits = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
201
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
202 while (c != end && *c >= 0x80) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
203 value |= (*c & 0x7f) << bits;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
204 bits += 7;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
205 c++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
206 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
207
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
208 if (c == end) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
209 /* last number shouldn't end with high bit */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
210 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
211 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
212 if (bits > 32-7) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
213 /* we have only 32bit numbers */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
214 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
215 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
216
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
217 value |= (*c & 0x7f) << bits;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
218 *p = c + 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
219 return value;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
220 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
221
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
222 static const void *data_normalize(const void *data, size_t size, buffer_t *dest)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
223 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
224 const uint8_t *src = data;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
225 size_t i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
226
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
227 buffer_set_used_size(dest, 0);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
228 for (i = 0; i < size; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
229 uint16_t chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
230
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
231 if (src[i] <= 32)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
232 chr = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
233 else if (src[i] > 'z')
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
234 chr = src[i] - 32 - 26;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
235 else
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
236 chr = i_toupper(src[i]) - 32;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
237 buffer_append(dest, &chr, sizeof(chr));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
238 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
239 return dest->data;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
240 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
241
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
242 static void
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
243 squat_trie_set_syscall_error(struct squat_trie *trie, const char *function)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
244 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
245 i_error("%s failed with index search file %s: %m",
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
246 function, trie->filepath);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
247 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
248
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
249 void squat_trie_set_corrupted(struct squat_trie *trie, const char *reason)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
250 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
251 i_error("Corrupted index search file %s: %s", trie->filepath, reason);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
252
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
253 (void)unlink(trie->filepath);
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
254 (void)unlink(trie->uidlist_filepath);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
255 trie->corrupted = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
256 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
257
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
258 static void
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
259 trie_map_node_save_children(unsigned int level, const uint32_t *src_idx,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
260 unsigned int count, struct trie_node **children)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
261 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
262 unsigned int i, file_bit;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
263
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
264 file_bit = level == BLOCK_SIZE ? 0 : 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
265
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
266 #ifndef ALLOW_UNALIGNED_ACCESS
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
267 if ((POINTER_CAST_TO(src_idx, size_t) & (sizeof(uint32_t)-1)) == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
268 #endif
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
269 for (i = 0; i < count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
270 children[i] = src_idx[i] == 0 ? NULL :
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
271 POINTER_CAST(src_idx[i] | file_bit);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
272 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
273 #ifndef ALLOW_UNALIGNED_ACCESS
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
274 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
275 /* unaligned access */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
276 uint32_t idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
277
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
278 for (i = 0; i < count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
279 memcpy(&idx, &src_idx[i], sizeof(idx));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
280 children[i] = idx == 0 ? NULL :
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
281 POINTER_CAST(idx | file_bit);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
282 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
283 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
284 #endif
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
285 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
286
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
287 static int trie_map_area(struct squat_trie *trie, uoff_t offset, size_t len)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
288 {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
289 ssize_t ret;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
290
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
291 if (trie->file_cache == NULL)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
292 return 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
293
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
294 ret = file_cache_read(trie->file_cache, offset, len);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
295 if (ret < 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
296 squat_trie_set_syscall_error(trie, "file_cache_read()");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
297 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
298 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
299 trie->const_mmap_base =
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
300 file_cache_get_map(trie->file_cache, &trie->mmap_size);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
301 trie->hdr = (const void *)trie->const_mmap_base;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
302 return 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
303 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
304
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
305 static int
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
306 trie_map_node(struct squat_trie *trie, uint32_t offset, unsigned int level,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
307 struct trie_node **node_r)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
308 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
309 struct trie_node *node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
310 const uint8_t *p, *end, *chars8_src, *chars16_src;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
311 uint32_t num, chars8_count, chars16_count;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
312 unsigned int chars8_offset, chars8_size, chars8_memsize;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
313 unsigned int chars16_offset, chars16_size, chars16_memsize;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
314
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
315 i_assert(trie->fd != -1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
316
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
317 if (trie_map_area(trie, offset, 2+256) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
318 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
319
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
320 if (offset >= trie->mmap_size) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
321 squat_trie_set_corrupted(trie, "trie offset too large");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
322 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
323 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
324
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
325 p = trie->const_mmap_base + offset;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
326 end = trie->const_mmap_base + trie->mmap_size;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
327
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
328 /* get 8bit char count and check that it's valid */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
329 num = _squat_trie_unpack_num(&p, end);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
330 chars8_count = num >> 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
331
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
332 chars8_offset = p - trie->const_mmap_base;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
333 chars8_size = chars8_count * (sizeof(uint8_t) + sizeof(uint32_t));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
334
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
335 if (chars8_count > 256 ||
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
336 chars8_offset + chars8_size > trie->mmap_size) {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
337 squat_trie_set_corrupted(trie, "trie offset broken");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
338 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
339 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
340
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
341 chars8_memsize = ALIGN(chars8_count * sizeof(uint8_t)) +
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
342 chars8_count * sizeof(struct trie_node *);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
343
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
344 if (trie_map_area(trie, chars8_offset, chars8_size + 8) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
345 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
346
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
347 if ((num & 1) == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
348 /* no 16bit chars */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
349 chars16_count = 0;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
350 chars16_memsize = 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
351 chars16_offset = 0;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
352 } else {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
353 /* get the 16bit char count */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
354 p = trie->const_mmap_base + chars8_offset + chars8_size;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
355 end = trie->const_mmap_base + trie->mmap_size;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
356
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
357 chars16_count = _squat_trie_unpack_num(&p, end);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
358 if (chars16_count > 65536) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
359 squat_trie_set_corrupted(trie, "trie offset broken");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
360 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
361 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
362 chars16_offset = p - trie->const_mmap_base;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
363
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
364 /* map the required area size and make sure it exists */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
365 chars16_size = chars16_count *
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
366 (sizeof(uint16_t) + sizeof(uint32_t));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
367 if (trie_map_area(trie, chars16_offset, chars16_size) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
368 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
369
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
370 if (chars16_offset + chars16_size > trie->mmap_size) {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
371 squat_trie_set_corrupted(trie, "trie offset broken");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
372 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
373 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
374
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
375 chars16_memsize = ALIGN(chars16_count * sizeof(uint16_t)) +
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
376 chars16_count * sizeof(struct trie_node *);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
377 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
378
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
379 node = i_malloc(sizeof(*node) + chars8_memsize + chars16_memsize);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
380 node->chars_8bit_count = chars8_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
381 node->chars_16bit_count = chars16_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
382 node->file_offset = offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
383
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
384 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
385 uint8_t *chars8 = NODE_CHARS8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
386 uint16_t *chars16 = NODE_CHARS16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
387 struct trie_node **children8 = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
388 struct trie_node **children16 = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
389 const uint32_t *src_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
390 const void *end_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
391
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
392 chars8_src = trie->const_mmap_base + chars8_offset;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
393 chars16_src = trie->const_mmap_base + chars16_offset;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
394
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
395 memcpy(chars8, chars8_src, sizeof(uint8_t) * chars8_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
396 memcpy(chars16, chars16_src, sizeof(uint16_t) * chars16_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
397
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
398 src_idx = CONST_PTR_OFFSET(chars8_src, chars8_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
399 trie_map_node_save_children(level, src_idx, chars8_count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
400 children8);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
401 if (chars16_count == 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
402 end_offset = &src_idx[chars8_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
403 else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
404 src_idx = CONST_PTR_OFFSET(chars16_src,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
405 chars16_count *
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
406 sizeof(uint16_t));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
407 trie_map_node_save_children(level, src_idx,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
408 chars16_count, children16);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
409 end_offset = &src_idx[chars16_count];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
410 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
411
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
412 node->orig_size = ((const uint8_t *)end_offset -
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
413 trie->const_mmap_base) - offset;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
414 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
415
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
416 *node_r = node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
417 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
418 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
419
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
420 static void free_children(unsigned int level, struct trie_node **children,
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
421 unsigned int count)
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
422 {
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
423 unsigned int i;
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
424 uint32_t child_idx;
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
425
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
426 for (i = 0; i < count; i++) {
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
427 child_idx = POINTER_CAST_TO(children[i], size_t);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
428 if ((child_idx & 1) == 0)
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
429 free_node(children[i], level);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
430 }
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
431 }
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
432
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
433 static void free_node(struct trie_node *node, unsigned int level)
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
434 {
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
435 if (level < BLOCK_SIZE) {
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
436 struct trie_node **children8 = NODE_CHILDREN8(node);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
437 struct trie_node **children16 = NODE_CHILDREN16(node);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
438
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
439 free_children(level + 1, children8, node->chars_8bit_count);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
440 free_children(level + 1, children16, node->chars_16bit_count);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
441 }
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
442 i_free(node);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
443 }
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
444
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
445 static void squat_trie_unmap(struct squat_trie *trie)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
446 {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
447 if (trie->file_cache != NULL)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
448 file_cache_invalidate(trie->file_cache, 0, (uoff_t)-1);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
449
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
450 if (trie->mmap_base != NULL) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
451 if (munmap(trie->mmap_base, trie->mmap_size) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
452 squat_trie_set_syscall_error(trie, "munmap()");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
453 trie->mmap_base = NULL;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
454 }
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
455
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
456 trie->mmap_size = 0;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
457 trie->hdr = NULL;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
458 trie->const_mmap_base = NULL;
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
459
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
460 if (trie->root != NULL) {
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
461 free_node(trie->root, 1);
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
462 trie->root = NULL;
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
463 }
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
464 }
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
465
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
466 static void trie_file_close(struct squat_trie *trie)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
467 {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
468 if (trie->file_cache != NULL)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
469 file_cache_free(&trie->file_cache);
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
470 if (trie->file_lock != NULL)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
471 file_lock_free(&trie->file_lock);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
472
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
473 squat_trie_unmap(trie);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
474 if (trie->fd != -1) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
475 if (close(trie->fd) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
476 squat_trie_set_syscall_error(trie, "close()");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
477 trie->fd = -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
478 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
479
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
480 trie->hdr = NULL;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
481 trie->corrupted = FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
482 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
483
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
484 static int
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
485 trie_map_check_header(struct squat_trie *trie,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
486 const struct squat_trie_header *hdr, uoff_t file_size)
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
487 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
488 if (hdr->version != SQUAT_TRIE_VERSION)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
489 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
490
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
491 if (hdr->used_file_size > file_size) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
492 squat_trie_set_corrupted(trie, "used_file_size too large");
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
493 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
494 }
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
495 if (hdr->root_offset != 0 &&
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
496 (hdr->root_offset > file_size ||
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
497 hdr->root_offset < sizeof(*hdr))) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
498 squat_trie_set_corrupted(trie, "invalid root_offset");
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
499 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
500 }
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
501 if (hdr->uidvalidity != trie->uidvalidity) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
502 squat_trie_set_corrupted(trie, "uidvalidity changed");
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
503 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
504 }
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
505
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
506 return 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
507 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
508
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
509 static int squat_trie_file_was_modified(struct squat_trie *trie)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
510 {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
511 struct squat_trie_header hdr;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
512 int ret;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
513
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
514 ret = pread_full(trie->fd, &hdr.modify_counter,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
515 sizeof(hdr.modify_counter),
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
516 offsetof(struct squat_trie_header, modify_counter));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
517 if (ret < 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
518 squat_trie_set_syscall_error(trie, "pread_full()");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
519 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
520 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
521 if (ret == 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
522 /* broken file, treat as modified */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
523 return 1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
524 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
525 return hdr.modify_counter == trie->file_cache_modify_counter ? 0 : 1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
526 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
527
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
528 static int squat_trie_map(struct squat_trie *trie)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
529 {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
530 const struct squat_trie_header *hdr;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
531 struct stat st;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
532 ssize_t ret;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
533
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
534 if (trie->hdr != NULL) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
535 if (!trie->mmap_disable) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
536 if (trie->hdr->used_file_size <= trie->mmap_size) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
537 /* everything is already mapped */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
538 return 1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
539 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
540 } else {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
541 ret = squat_trie_file_was_modified(trie);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
542 if (ret <= 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
543 return ret < 0 ? -1 : 1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
544 }
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
545 }
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
546
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
547 if (fstat(trie->fd, &st) < 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
548 squat_trie_set_syscall_error(trie, "fstat()");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
549 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
550 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
551 trie->dev = st.st_dev;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
552 trie->ino = st.st_ino;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
553
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
554 squat_trie_unmap(trie);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
555
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
556 if (!trie->mmap_disable) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
557 trie->mmap_size = st.st_size;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
558 trie->mmap_base = mmap(NULL, trie->mmap_size,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
559 PROT_READ | PROT_WRITE,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
560 MAP_SHARED, trie->fd, 0);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
561 if (trie->mmap_base == MAP_FAILED) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
562 trie->mmap_size = 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
563 trie->mmap_base = NULL;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
564 squat_trie_set_syscall_error(trie, "mmap()");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
565 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
566 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
567 trie->const_mmap_base = trie->mmap_base;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
568 } else {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
569 ret = file_cache_read(trie->file_cache, 0, sizeof(*trie->hdr));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
570 if (ret < 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
571 squat_trie_set_syscall_error(trie, "file_cache_read()");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
572 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
573 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
574 if ((size_t)ret < sizeof(*trie->hdr)) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
575 squat_trie_set_corrupted(trie, "file too small");
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
576 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
577 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
578 trie->const_mmap_base =
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
579 file_cache_get_map(trie->file_cache, &trie->mmap_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
580 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
581
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
582 hdr = (const void *)trie->const_mmap_base;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
583 if (trie_map_check_header(trie, hdr, st.st_size) < 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
584 return -1;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
585 trie->hdr = hdr;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
586 trie->file_cache_modify_counter = trie->hdr->modify_counter;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
587
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
588 if (trie_map_node(trie, trie->hdr->root_offset, 1, &trie->root) < 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
589 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
590 return 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
591 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
592
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
593 static void trie_file_open_fd(struct squat_trie *trie, int fd)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
594 {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
595 struct stat st;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
596
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
597 if (fstat(fd, &st) < 0) {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
598 /* don't bother adding complexity by trying to handle this
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
599 error here. we'll break later anyway in easier error
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
600 handling paths. */
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
601 squat_trie_set_syscall_error(trie, "fstat()");
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
602 trie->ino = 0;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
603 } else {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
604 trie->dev = st.st_dev;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
605 trie->ino = st.st_ino;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
606 }
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
607 trie->fd = fd;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
608
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
609 if (trie->mmap_disable)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
610 trie->file_cache = file_cache_new(trie->fd);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
611 }
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
612
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
613 static int trie_file_open(struct squat_trie *trie, bool create)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
614 {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
615 int fd;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
616
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
617 i_assert(trie->fd == -1);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
618
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
619 fd = open(trie->filepath, O_RDWR | (create ? O_CREAT : 0), 0660);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
620 if (fd == -1) {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
621 if (errno == ENOENT)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
622 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
623
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
624 squat_trie_set_syscall_error(trie, "open()");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
625 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
626 }
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
627 trie_file_open_fd(trie, fd);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
628 return 1;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
629 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
630
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
631 static int trie_file_create_finish(struct squat_trie *trie)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
632 {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
633 struct squat_trie_header hdr;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
634 struct stat st;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
635
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
636 if (fstat(trie->fd, &st) < 0) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
637 squat_trie_set_syscall_error(trie, "fstat()");
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
638 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
639 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
640
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
641 if (st.st_size <= sizeof(hdr)) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
642 memset(&hdr, 0, sizeof(hdr));
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
643 hdr.version = SQUAT_TRIE_VERSION;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
644 hdr.uidvalidity = trie->uidvalidity;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
645 hdr.used_file_size = sizeof(hdr);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
646
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
647 if (pwrite_full(trie->fd, &hdr, sizeof(hdr), 0) < 0) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
648 squat_trie_set_syscall_error(trie, "pwrite_full()");
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
649 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
650 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
651 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
652
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
653 return 0;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
654 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
655
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
656 struct squat_trie *
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
657 squat_trie_open(const char *path, uint32_t uidvalidity,
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
658 enum file_lock_method lock_method, bool mmap_disable)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
659 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
660 struct squat_trie *trie;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
661
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
662 trie = i_new(struct squat_trie, 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
663 trie->fd = -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
664 trie->filepath = i_strdup(path);
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
665 trie->uidvalidity = uidvalidity;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
666 trie->lock_method = lock_method;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
667 trie->mmap_disable = mmap_disable;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
668 trie->buf = buffer_create_dynamic(default_pool, 1024);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
669
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
670 trie->uidlist_filepath = i_strconcat(path, ".uids", NULL);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
671 trie->uidlist =
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
672 squat_uidlist_init(trie, trie->uidlist_filepath,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
673 uidvalidity, mmap_disable);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
674 return trie;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
675 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
676
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
677 void squat_trie_close(struct squat_trie *trie)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
678 {
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
679 squat_trie_unmap(trie);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
680 buffer_free(trie->buf);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
681 squat_uidlist_deinit(trie->uidlist);
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
682 i_free(trie->uidlist_filepath);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
683 i_free(trie->filepath);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
684 i_free(trie);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
685 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
686
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
687 int squat_trie_get_last_uid(struct squat_trie *trie, uint32_t *uid_r)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
688 {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
689 int ret;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
690
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
691 if (trie->fd == -1) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
692 if ((ret = trie_file_open(trie, FALSE)) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
693 return ret;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
694 if (ret == 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
695 *uid_r = 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
696 return 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
697 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
698 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
699
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
700 if (squat_trie_lock(trie, F_RDLCK) <= 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
701 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
702
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
703 ret = squat_uidlist_get_last_uid(trie->uidlist, uid_r);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
704 squat_trie_unlock(trie);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
705 return ret;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
706 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
707
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
708 static int squat_trie_is_file_stale(struct squat_trie *trie)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
709 {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
710 struct stat st;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
711
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
712 if (stat(trie->filepath, &st) < 0) {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
713 if (errno == ENOENT)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
714 return 1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
715
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
716 squat_trie_set_syscall_error(trie, "stat()");
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
717 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
718 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
719
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
720 return st.st_ino == trie->ino &&
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
721 CMP_DEV_T(st.st_dev, trie->dev) ? 0 : 1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
722 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
723
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
724 static int
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
725 squat_trie_file_lock(struct squat_trie *trie, int fd, const char *path,
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
726 int lock_type, struct file_lock **lock_r)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
727 {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
728 int ret;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
729
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
730 ret = file_wait_lock(fd, path, lock_type, trie->lock_method,
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
731 SQUAT_TRIE_LOCK_TIMEOUT, lock_r);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
732 if (ret == 0)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
733 squat_trie_set_syscall_error(trie, "file_wait_lock()");
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
734 return ret;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
735 }
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
736
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
737 int squat_trie_lock(struct squat_trie *trie, int lock_type)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
738 {
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
739 bool created = FALSE;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
740 int ret;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
741
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
742 i_assert(lock_type == F_RDLCK || lock_type == F_WRLCK);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
743
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
744 if (trie->lock_count > 0) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
745 /* read lock -> write lock would deadlock */
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
746 i_assert(trie->lock_type == lock_type || lock_type == F_RDLCK);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
747
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
748 trie->lock_count++;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
749 return 1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
750 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
751
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
752 if (trie->fd == -1 || trie->corrupted) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
753 trie_file_close(trie);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
754 if (lock_type == F_WRLCK) {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
755 if ((ret = trie_file_open(trie, FALSE)) < 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
756 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
757 if (ret == 0) {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
758 if (trie_file_open(trie, TRUE) < 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
759 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
760 created = TRUE;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
761 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
762 } else {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
763 if (trie_file_open(trie, FALSE) <= 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
764 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
765 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
766 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
767
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
768 for (;;) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
769 i_assert(trie->file_lock == NULL);
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
770 ret = squat_trie_file_lock(trie, trie->fd, trie->filepath,
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
771 lock_type, &trie->file_lock);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
772 if (ret <= 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
773 return ret;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
774
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
775 /* if the trie has been compressed, we need to reopen the
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
776 file and try to lock again */
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
777 ret = squat_trie_is_file_stale(trie);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
778 if (ret == 0)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
779 break;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
780
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
781 file_unlock(&trie->file_lock);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
782 if (ret < 0)
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
783 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
784
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
785 trie_file_close(trie);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
786 if (trie_file_open(trie, FALSE) <= 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
787 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
788 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
789
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
790 if (created) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
791 /* we possibly created this file. now that we've locked the
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
792 file, we can safely check if someone else already wrote the
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
793 header or if we should do it now */
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
794 if (trie_file_create_finish(trie) < 0) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
795 file_unlock(&trie->file_lock);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
796 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
797 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
798 }
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
799
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
800 if (squat_trie_map(trie) <= 0) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
801 file_unlock(&trie->file_lock);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
802 return -1;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
803 }
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
804 if (squat_uidlist_refresh(trie->uidlist) < 0) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
805 file_unlock(&trie->file_lock);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
806 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
807 }
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
808
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
809 trie->lock_count++;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
810 trie->lock_type = lock_type;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
811 return 1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
812 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
813
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
814 void squat_trie_unlock(struct squat_trie *trie)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
815 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
816 i_assert(trie->lock_count > 0);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
817
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
818 if (--trie->lock_count > 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
819 return;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
820
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
821 file_unlock(&trie->file_lock);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
822 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
823
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
824 static struct trie_node *
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
825 node_alloc(uint16_t chr, unsigned int level)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
826 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
827 struct trie_node *node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
828 unsigned int idx_size, idx_offset = sizeof(*node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
829
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
830 idx_size = level < BLOCK_SIZE ?
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
831 sizeof(struct trie_node *) : sizeof(uint32_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
832
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
833 if (chr < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
834 uint8_t *chrp;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
835
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
836 idx_offset += ALIGN(sizeof(*chrp));
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
837 node = i_malloc(idx_offset + idx_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
838 node->chars_8bit_count = 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
839
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
840 chrp = PTR_OFFSET(node, sizeof(*node));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
841 *chrp = chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
842 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
843 uint16_t *chrp;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
844
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
845 idx_offset += ALIGN(sizeof(*chrp));
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
846 node = i_malloc(idx_offset + idx_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
847 node->chars_16bit_count = 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
848
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
849 chrp = PTR_OFFSET(node, sizeof(*node));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
850 *chrp = chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
851 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
852
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
853 node->modified = TRUE;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
854 node->resized = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
855 return node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
856 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
857
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
858 static struct trie_node *
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
859 node_realloc(struct trie_node *node, uint32_t char_idx, uint16_t chr,
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
860 unsigned int level)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
861 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
862 struct trie_node *new_node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
863 unsigned int old_size_8bit, old_size_16bit, old_idx_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
864 unsigned int idx_size, old_size, new_size, new_idx_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
865 unsigned int hole1_pos, hole2_pos, skip;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
866
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
867 idx_size = level < BLOCK_SIZE ?
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
868 sizeof(struct trie_node *) : sizeof(uint32_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
869
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
870 old_size_8bit = ALIGN(node->chars_8bit_count) +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
871 node->chars_8bit_count * idx_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
872 old_size_16bit = ALIGN(sizeof(uint16_t) * node->chars_16bit_count) +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
873 node->chars_16bit_count * idx_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
874 old_size = sizeof(*node) + old_size_8bit + old_size_16bit;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
875
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
876 if (chr < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
877 new_idx_offset = sizeof(*node) +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
878 ALIGN(node->chars_8bit_count + sizeof(uint8_t));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
879 new_size = new_idx_offset + old_size_16bit +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
880 (node->chars_8bit_count + 1) * idx_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
881 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
882 new_idx_offset = sizeof(*node) + old_size_8bit +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
883 ALIGN((node->chars_16bit_count + 1) * sizeof(uint16_t));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
884 new_size = new_idx_offset +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
885 (node->chars_16bit_count + 1) * idx_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
886 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
887
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
888 new_node = i_malloc(new_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
889 if (chr < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
890 hole1_pos = sizeof(*node) + char_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
891 old_idx_offset = sizeof(*node) + ALIGN(node->chars_8bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
892 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
893 hole1_pos = sizeof(*node) + old_size_8bit +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
894 char_idx * sizeof(uint16_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
895 old_idx_offset = sizeof(*node) + old_size_8bit +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
896 ALIGN(node->chars_16bit_count * sizeof(uint16_t));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
897 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
898 hole2_pos = old_idx_offset + idx_size * char_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
899
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
900 memcpy(new_node, node, hole1_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
901 if (chr < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
902 uint8_t *chrp = PTR_OFFSET(new_node, hole1_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
903 *chrp = chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
904 new_node->chars_8bit_count++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
905
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
906 memcpy(PTR_OFFSET(new_node, hole1_pos + sizeof(uint8_t)),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
907 PTR_OFFSET(node, hole1_pos), old_idx_offset - hole1_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
908 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
909 uint16_t *chrp = PTR_OFFSET(new_node, hole1_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
910 *chrp = chr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
911 new_node->chars_16bit_count++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
912
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
913 memcpy(PTR_OFFSET(new_node, hole1_pos + sizeof(uint16_t)),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
914 PTR_OFFSET(node, hole1_pos), old_idx_offset - hole1_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
915 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
916
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
917 memcpy(PTR_OFFSET(new_node, new_idx_offset),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
918 PTR_OFFSET(node, old_idx_offset),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
919 hole2_pos - old_idx_offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
920
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
921 skip = new_idx_offset - old_idx_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
922 memset(PTR_OFFSET(new_node, hole2_pos + skip), 0, idx_size);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
923 skip += sizeof(uint32_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
924 memcpy(PTR_OFFSET(new_node, hole2_pos + skip),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
925 PTR_OFFSET(node, hole2_pos),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
926 old_size - hole2_pos);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
927
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
928 new_node->resized = TRUE;
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
929 i_free(node);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
930 return new_node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
931 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
932
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
933 static int
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
934 trie_insert_node(struct squat_trie_build_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
935 struct trie_node **parent,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
936 const uint16_t *data, uint32_t uid, unsigned int level)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
937 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
938 struct squat_trie *trie = ctx->trie;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
939 struct trie_node *node = *parent;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
940 uint32_t char_idx, idx_base_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
941 bool modified = FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
942 int ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
943
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
944 if (*data < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
945 unsigned int count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
946
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
947 if (node == NULL) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
948 ctx->node_count++;
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
949 node = *parent = node_alloc(*data, level);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
950 char_idx = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
951 count = 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
952 modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
953 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
954 uint8_t *chars = PTR_OFFSET(node, sizeof(*node));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
955 uint8_t *pos;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
956
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
957 count = node->chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
958 pos = bsearch_insert_pos(data, chars, count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
959 sizeof(chars[0]),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
960 chr_8bit_cmp);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
961 char_idx = pos - chars;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
962 if (char_idx == count || *pos != *data) {
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
963 node = node_realloc(node, char_idx,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
964 *data, level);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
965 *parent = node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
966 modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
967 count++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
968 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
969 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
970 idx_base_offset = sizeof(*node) + ALIGN(count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
971 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
972 unsigned int offset = sizeof(*node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
973 unsigned int count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
974
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
975 if (node == NULL) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
976 ctx->node_count++;
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
977 node = *parent = node_alloc(*data, level);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
978 char_idx = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
979 count = 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
980 modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
981 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
982 unsigned int idx_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
983 uint16_t *chars, *pos;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
984
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
985 idx_size = level < BLOCK_SIZE ?
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
986 sizeof(struct trie_node *) : sizeof(uint32_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
987 offset += ALIGN(node->chars_8bit_count) +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
988 idx_size * node->chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
989 chars = PTR_OFFSET(node, offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
990
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
991 count = node->chars_16bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
992 pos = bsearch_insert_pos(data, chars, count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
993 sizeof(chars[0]),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
994 chr_16bit_cmp);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
995 char_idx = pos - chars;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
996 if (char_idx == count || *pos != *data) {
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
997 node = node_realloc(node, char_idx,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
998 *data, level);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
999 *parent = node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1000 modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1001 count++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1002 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1003 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1004
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1005 idx_base_offset = offset + ALIGN(sizeof(uint16_t) * count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1006 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1007
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1008 if (level < BLOCK_SIZE) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1009 struct trie_node **children = PTR_OFFSET(node, idx_base_offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1010 size_t child_idx = POINTER_CAST_TO(children[char_idx], size_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1011
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1012 if ((child_idx & 1) != 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1013 if (trie_map_node(trie, child_idx & ~1, level + 1,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1014 &children[char_idx]) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1015 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1016 }
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1017 ret = trie_insert_node(ctx, &children[char_idx],
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1018 data + 1, uid, level + 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1019 if (ret < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1020 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1021 if (ret > 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1022 node->modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1023 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1024 uint32_t *uid_lists = PTR_OFFSET(node, idx_base_offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1025 if (squat_uidlist_add(trie->uidlist, &uid_lists[char_idx],
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1026 uid) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1027 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1028
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1029 node->modified = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1030 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1031 return modified ? 1 : 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1032 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1033
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1034 static uint32_t
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1035 trie_lookup_node(struct squat_trie *trie, struct trie_node *node,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1036 const uint16_t *data, unsigned int level)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1037 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1038 uint32_t char_idx, idx_base_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1039
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1040 if (*data < 256) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1041 const uint8_t *chars, *pos;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1042 unsigned int count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1043
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1044 if (node == NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1045 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1046
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1047 chars = CONST_PTR_OFFSET(node, sizeof(*node));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1048 count = node->chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1049 pos = bsearch(data, chars, count, sizeof(chars[0]),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1050 chr_8bit_cmp);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1051 if (pos == NULL || *pos != *data)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1052 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1053
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1054 char_idx = pos - chars;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1055 idx_base_offset = sizeof(*node) + ALIGN(count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1056 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1057 const uint16_t *chars, *pos;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1058 unsigned int count, idx_size, offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1059
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1060 if (node == NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1061 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1062
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1063 idx_size = level < BLOCK_SIZE ?
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1064 sizeof(struct trie_node *) : sizeof(uint32_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1065 offset = sizeof(*node) + ALIGN(node->chars_8bit_count) +
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1066 idx_size * node->chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1067 chars = PTR_OFFSET(node, offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1068
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1069 count = node->chars_16bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1070 pos = bsearch(data, chars, count, sizeof(chars[0]),
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1071 chr_16bit_cmp);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1072 if (pos == NULL || *pos != *data)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1073 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1074
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1075 char_idx = pos - chars;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1076 idx_base_offset = offset + ALIGN(sizeof(uint16_t) * count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1077 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1078
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1079 if (level < BLOCK_SIZE) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1080 struct trie_node **children = PTR_OFFSET(node, idx_base_offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1081 size_t child_idx = POINTER_CAST_TO(children[char_idx], size_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1082
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1083 if ((child_idx & 1) != 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1084 /* not mapped to memory yet. do it. */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1085 if (trie_map_node(trie, child_idx & ~1, level + 1,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1086 &children[char_idx]) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1087 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1088 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1089
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1090 return trie_lookup_node(trie, children[char_idx],
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1091 data + 1, level + 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1092 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1093 const uint32_t *uid_lists =
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1094 CONST_PTR_OFFSET(node, idx_base_offset);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1095
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1096 return uid_lists[char_idx];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1097 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1098 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1099
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1100 static bool block_want_add(const uint16_t *data)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1101 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1102 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1103
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1104 /* skip all blocks that contain spaces or control characters.
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1105 no-one searches them anyway */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1106 for (i = 0; i < BLOCK_SIZE; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1107 if (data[i] == 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1108 return FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1109 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1110 return TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1111 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1112
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1113 struct squat_trie_build_context *
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1114 squat_trie_build_init(struct squat_trie *trie, uint32_t *last_uid_r)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1115 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1116 struct squat_trie_build_context *ctx;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1117
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1118 ctx = i_new(struct squat_trie_build_context, 1);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1119 ctx->trie = trie;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1120
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1121 if (squat_trie_lock(trie, F_WRLCK) <= 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1122 ctx->failed = TRUE;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1123 else {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1124 ctx->locked = TRUE;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1125 ctx->node_count = trie->hdr->node_count;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1126
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1127 if (squat_uidlist_get_last_uid(trie->uidlist, last_uid_r) < 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1128 ctx->failed = TRUE;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1129 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1130
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1131 if (ctx->failed)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1132 *last_uid_r = 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1133 return ctx;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1134 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1135
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1136 int squat_trie_build_deinit(struct squat_trie_build_context *ctx)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1137 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1138 int ret = ctx->failed ? -1 : 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1139
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1140 if (ret == 0)
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1141 ret = squat_trie_build_flush(ctx, TRUE);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1142
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1143 if (ctx->locked)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1144 squat_trie_unlock(ctx->trie);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1145
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1146 i_free(ctx);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1147 return ret;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1148 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1149
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1150 int squat_trie_build_more(struct squat_trie_build_context *ctx, uint32_t uid,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1151 const void *data, size_t size)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1152 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1153 const uint16_t *str;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1154 uint16_t buf[(BLOCK_SIZE-1)*2];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1155 unsigned int i, tmp_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1156
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1157 if (ctx->failed)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1158 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1159
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1160 t_push();
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1161 str = data_normalize(data, size, ctx->trie->buf);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1162
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1163 if (uid == ctx->prev_uid) {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1164 /* @UNSAFE: continue from last block */
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1165 memcpy(buf, ctx->prev_added,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1166 sizeof(buf[0]) * ctx->prev_added_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1167 tmp_size = I_MIN(size, BLOCK_SIZE-1);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1168 memcpy(buf + ctx->prev_added_size, str,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1169 sizeof(buf[0]) * tmp_size);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1170
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1171 tmp_size += ctx->prev_added_size;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1172 for (i = 0; i + BLOCK_SIZE <= tmp_size; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1173 if (block_want_add(buf+i)) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1174 if (trie_insert_node(ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1175 &ctx->trie->root,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1176 buf + i, uid, 1) < 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1177 t_pop();
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1178 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1179 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1180 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1181 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1182
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1183 if (size < BLOCK_SIZE) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1184 ctx->prev_added_size = I_MIN(tmp_size, BLOCK_SIZE-1);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1185 memcpy(ctx->prev_added, buf + i,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1186 sizeof(buf[0]) * ctx->prev_added_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1187 t_pop();
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1188 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1189 }
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1190 } else if (squat_uidlist_want_flush(ctx->trie->uidlist)) {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1191 if (squat_trie_build_flush(ctx, FALSE) < 0) {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1192 ctx->failed = TRUE;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1193 t_pop();
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1194 return -1;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1195 }
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1196 str = data_normalize(data, size, ctx->trie->buf);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1197 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1198
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1199 ctx->prev_uid = uid;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1200 for (i = 0; i + BLOCK_SIZE <= size; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1201 if (block_want_add(str+i)) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1202 if (trie_insert_node(ctx, &ctx->trie->root,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1203 str + i, uid, 1) < 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1204 t_pop();
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1205 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1206 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1207 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1208 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1209
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1210 ctx->prev_added_size = I_MIN(size, BLOCK_SIZE-1);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1211 memcpy(ctx->prev_added, str + i,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1212 sizeof(ctx->prev_added[0]) * ctx->prev_added_size);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1213 t_pop();
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1214 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1215 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1216
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1217 static void node_pack_children(buffer_t *buf, struct trie_node **children,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1218 unsigned int count)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1219 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1220 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1221 size_t child_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1222 uint32_t idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1223
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1224 for (i = 0; i < count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1225 child_idx = POINTER_CAST_TO(children[i], size_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1226 if ((child_idx & 1) != 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1227 idx = child_idx & ~1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1228 else
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1229 idx = children[i]->file_offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1230 buffer_append(buf, &idx, sizeof(idx));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1231 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1232 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1233
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1234 static void node_pack(buffer_t *buf, struct trie_node *node)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1235 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1236 uint8_t *chars8 = NODE_CHARS8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1237 uint16_t *chars16 = NODE_CHARS16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1238 struct trie_node **children8 = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1239 struct trie_node **children16 = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1240
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1241 buffer_set_used_size(buf, 0);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1242 _squat_trie_pack_num(buf, (node->chars_8bit_count << 1) |
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1243 (node->chars_16bit_count > 0 ? 1 : 0));
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1244 buffer_append(buf, chars8, node->chars_8bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1245 node_pack_children(buf, children8, node->chars_8bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1246
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1247 if (node->chars_16bit_count > 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1248 _squat_trie_pack_num(buf, node->chars_16bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1249 buffer_append(buf, chars16,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1250 sizeof(*chars16) * node->chars_16bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1251 node_pack_children(buf, children16, node->chars_16bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1252 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1253 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1254
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1255 static int node_leaf_finish(struct squat_trie *trie, struct trie_node *node)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1256 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1257 uint32_t *idx8 = (uint32_t *)NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1258 uint32_t *idx16 = (uint32_t *)NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1259 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1260
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1261 for (i = 0; i < node->chars_8bit_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1262 if (squat_uidlist_finish_list(trie->uidlist, &idx8[i]) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1263 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1264 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1265 for (i = 0; i < node->chars_16bit_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1266 if (squat_uidlist_finish_list(trie->uidlist, &idx16[i]) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1267 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1268 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1269 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1270 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1271
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1272 static void node_pack_leaf(buffer_t *buf, struct trie_node *node)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1273 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1274 uint8_t *chars8 = NODE_CHARS8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1275 uint16_t *chars16 = NODE_CHARS16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1276 uint32_t *idx8 = (uint32_t *)NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1277 uint32_t *idx16 = (uint32_t *)NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1278
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1279 buffer_set_used_size(buf, 0);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1280 _squat_trie_pack_num(buf, (node->chars_8bit_count << 1) |
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1281 (node->chars_16bit_count > 0 ? 1 : 0));
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1282 buffer_append(buf, chars8, node->chars_8bit_count);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1283 buffer_append(buf, idx8, sizeof(*idx8) * node->chars_8bit_count);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1284
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1285 if (node->chars_16bit_count > 0) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1286 _squat_trie_pack_num(buf, node->chars_16bit_count);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1287 buffer_append(buf, chars16,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1288 sizeof(*chars16) * node->chars_16bit_count);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1289 buffer_append(buf, idx16,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1290 sizeof(*idx16) * node->chars_16bit_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1291 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1292 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1293
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1294 static int
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1295 trie_write_node_children(struct squat_trie_build_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1296 unsigned int level, struct trie_node **children,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1297 unsigned int count)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1298 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1299 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1300 size_t child_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1301
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1302 for (i = 0; i < count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1303 child_idx = POINTER_CAST_TO(children[i], size_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1304 if ((child_idx & 1) == 0) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1305 if (trie_write_node(ctx, level, children[i]) < 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1306 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1307 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1308 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1309 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1310 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1311
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1312 static int trie_write_node(struct squat_trie_build_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1313 unsigned int level, struct trie_node *node)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1314 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1315 struct squat_trie *trie = ctx->trie;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1316 uoff_t offset;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1317
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1318 if (level < BLOCK_SIZE) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1319 struct trie_node **children8 = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1320 struct trie_node **children16 = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1321
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1322 if (trie_write_node_children(ctx, level + 1,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1323 children8,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1324 node->chars_8bit_count) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1325 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1326 if (trie_write_node_children(ctx, level + 1,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1327 children16,
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1328 node->chars_16bit_count) < 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1329 return -1;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1330 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1331
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1332 if (!node->modified)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1333 return 0;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1334
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1335 if (level < BLOCK_SIZE)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1336 node_pack(trie->buf, node);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1337 else {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1338 if (node_leaf_finish(trie, node) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1339 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1340
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1341 node_pack_leaf(trie->buf, node);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1342 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1343
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1344 offset = ctx->output->offset;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1345 if ((offset & 1) != 0) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1346 o_stream_send(ctx->output, "", 1);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1347 offset++;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1348 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1349
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1350 if (node->resized && node->orig_size != trie->buf->used) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1351 /* append to end of file. the parent node is written later. */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1352 node->file_offset = offset;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1353 o_stream_send(ctx->output, trie->buf->data, trie->buf->used);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1354
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1355 ctx->deleted_space += node->orig_size;
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1356 } else {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1357 /* overwrite node's contents */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1358 i_assert(node->file_offset != 0);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1359 i_assert(trie->buf->used <= node->orig_size);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1360
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1361 /* FIXME: write only the indexes if !node->resized */
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1362 o_stream_seek(ctx->output, node->file_offset);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1363 o_stream_send(ctx->output, trie->buf->data, trie->buf->used);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1364 o_stream_seek(ctx->output, offset);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1365
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1366 ctx->deleted_space += trie->buf->used - node->orig_size;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1367 }
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1368
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1369 ctx->modified = TRUE;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1370 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1371 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1372
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1373 static int
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1374 trie_nodes_write(struct squat_trie_build_context *ctx, uint32_t *uidvalidity_r)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1375 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1376 struct squat_trie *trie = ctx->trie;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1377 struct squat_trie_header hdr;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1378
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1379 hdr = *trie->hdr;
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1380 if (lseek(trie->fd, hdr.used_file_size, SEEK_SET) < 0) {
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1381 squat_trie_set_syscall_error(trie, "lseek()");
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1382 return -1;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1383 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1384
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1385 ctx->output = o_stream_create_file(trie->fd, default_pool, 0, FALSE);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1386 if (hdr.used_file_size == 0) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1387 o_stream_send(ctx->output, &hdr, sizeof(hdr));
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1388 ctx->modified = TRUE;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1389 }
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1390
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1391 ctx->deleted_space = 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1392 if (trie_write_node(ctx, 1, trie->root) < 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1393 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1394
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1395 if (ctx->modified) {
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1396 /* update the header */
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1397 hdr.root_offset = trie->root->file_offset;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1398 hdr.used_file_size = ctx->output->offset;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1399 hdr.deleted_space += ctx->deleted_space;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1400 hdr.node_count = ctx->node_count;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1401 hdr.modify_counter++;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1402 o_stream_seek(ctx->output, 0);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1403 o_stream_send(ctx->output, &hdr, sizeof(hdr));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1404 }
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1405
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1406 o_stream_destroy(&ctx->output);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1407 *uidvalidity_r = hdr.uidvalidity;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1408 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1409 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1410
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1411 static bool squat_trie_need_compress(struct squat_trie *trie,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1412 unsigned int current_message_count)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1413 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1414 uint32_t max_del_space;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1415
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1416 if (trie->hdr->used_file_size >= TRIE_COMPRESS_MIN_SIZE) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1417 /* see if we've reached the max. deleted space in file */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1418 max_del_space = trie->hdr->used_file_size / 100 *
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1419 TRIE_COMPRESS_PERCENTAGE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1420 if (trie->hdr->deleted_space > max_del_space)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1421 return TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1422 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1423
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1424 return squat_uidlist_need_compress(trie->uidlist,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1425 current_message_count);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1426 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1427
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1428 static int
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1429 squat_trie_build_flush(struct squat_trie_build_context *ctx, bool finish)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1430 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1431 struct squat_trie *trie = ctx->trie;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1432 uint32_t uidvalidity;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1433
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1434 if (trie->root == NULL) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1435 /* nothing changed */
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1436 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1437 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1438
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1439 if (trie->corrupted)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1440 return -1;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1441
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1442 if (trie_nodes_write(ctx, &uidvalidity) < 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1443 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1444 if (squat_uidlist_flush(trie->uidlist, uidvalidity) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1445 return -1;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1446
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1447 squat_trie_unmap(trie);
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1448 if (squat_trie_map(trie) <= 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1449 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1450
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
1451 if (squat_trie_need_compress(trie, (unsigned int)-1)) {
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1452 if (ctx->locked && finish) {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1453 squat_trie_unlock(ctx->trie);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1454 ctx->locked = FALSE;
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1455 }
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1456
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1457 if (squat_trie_compress(trie, NULL) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1458 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1459 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1460 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1461 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1462
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1463 static void squat_trie_compress_chars8(struct trie_node *node)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1464 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1465 uint8_t *chars = NODE_CHARS8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1466 struct trie_node **child_src = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1467 struct trie_node **child_dest;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1468 unsigned int i, j, old_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1469
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1470 old_count = node->chars_8bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1471 for (i = j = 0; i < old_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1472 if (child_src[i] != NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1473 chars[j++] = chars[i];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1474 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1475
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1476 node->chars_8bit_count = j;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1477 child_dest = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1478
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1479 for (i = j = 0; i < old_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1480 if (child_src[i] != NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1481 child_dest[j++] = child_src[i];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1482 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1483 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1484
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1485 static void squat_trie_compress_chars16(struct trie_node *node)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1486 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1487 uint16_t *chars = NODE_CHARS16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1488 struct trie_node **child_src = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1489 struct trie_node **child_dest;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1490 unsigned int i, j, old_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1491
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1492 old_count = node->chars_16bit_count;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1493 for (i = j = 0; i < old_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1494 if (child_src[i] != NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1495 chars[j++] = chars[i];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1496 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1497
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1498 node->chars_16bit_count = j;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1499 child_dest = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1500
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1501 for (i = j = 0; i < old_count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1502 if (child_src[i] != NULL)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1503 child_dest[j++] = child_src[i];
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1504 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1505 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1506
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1507 static int
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1508 squat_trie_compress_children(struct squat_trie_compress_context *ctx,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1509 struct trie_node **children, unsigned int count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1510 unsigned int level)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1511 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1512 struct trie_node *child_node;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1513 size_t child_idx;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1514 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1515 int ret = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1516 bool need_char_compress = FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1517
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1518 for (i = 0; i < count; i++) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1519 child_idx = POINTER_CAST_TO(children[i], size_t);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1520 i_assert((child_idx & 1) != 0);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1521 child_idx &= ~1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1522
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1523 if (trie_map_node(ctx->trie, child_idx, level, &child_node) < 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1524 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1525
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1526 ret = squat_trie_compress_node(ctx, child_node, level);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1527 if (child_node->file_offset != 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1528 children[i] = POINTER_CAST(child_node->file_offset | 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1529 else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1530 children[i] = NULL;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1531 need_char_compress = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1532 }
4887
463a4ebba685 Another try at memory reduction / leak fixes
Timo Sirainen <tss@iki.fi>
parents: 4886
diff changeset
1533 i_free(child_node);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1534
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1535 if (ret < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1536 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1537 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1538 return need_char_compress ? 0 : 1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1539 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1540
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1541 static int
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1542 squat_trie_compress_leaf_uidlist(struct squat_trie_compress_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1543 struct trie_node *node)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1544 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1545 uint32_t *idx8 = (uint32_t *)NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1546 uint32_t *idx16 = (uint32_t *)NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1547 unsigned int i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1548 int ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1549 bool compress_chars = FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1550
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1551 for (i = 0; i < node->chars_8bit_count; i++) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1552 ret = squat_uidlist_compress_next(ctx->uidlist_ctx, &idx8[i]);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1553 if (ret < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1554 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1555 if (ret == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1556 idx8[i] = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1557 compress_chars = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1558 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1559 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1560 if (compress_chars) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1561 squat_trie_compress_chars8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1562 compress_chars = FALSE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1563 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1564 for (i = 0; i < node->chars_16bit_count; i++) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1565 ret = squat_uidlist_compress_next(ctx->uidlist_ctx, &idx16[i]);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1566 if (ret < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1567 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1568 if (ret == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1569 idx16[i] = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1570 compress_chars = TRUE;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1571 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1572 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1573 if (compress_chars) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1574 squat_trie_compress_chars16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1575 node->chars_16bit_count = i;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1576 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1577 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1578 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1579
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1580 static int
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1581 squat_trie_compress_node(struct squat_trie_compress_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1582 struct trie_node *node, unsigned int level)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1583 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1584 struct squat_trie *trie = ctx->trie;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1585 int ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1586
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1587 if (level == BLOCK_SIZE) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1588 if (squat_trie_compress_leaf_uidlist(ctx, node))
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1589 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1590
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1591 if (node->chars_8bit_count == 0 &&
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1592 node->chars_16bit_count == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1593 /* everything expunged */
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1594 ctx->node_count--;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1595 node->file_offset = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1596 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1597 }
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1598 node_pack_leaf(trie->buf, node);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1599 } else {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1600 struct trie_node **children8 = NODE_CHILDREN8(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1601 struct trie_node **children16 = NODE_CHILDREN16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1602
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1603 if ((ret = squat_trie_compress_children(ctx, children8,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1604 node->chars_8bit_count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1605 level + 1)) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1606 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1607 if (ret == 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1608 squat_trie_compress_chars8(node);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1609 if ((ret = squat_trie_compress_children(ctx, children16,
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1610 node->chars_16bit_count,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1611 level + 1)) < 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1612 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1613 if (ret == 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1614 squat_trie_compress_chars16(node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1615
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1616 if (node->chars_8bit_count == 0 &&
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1617 node->chars_16bit_count == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1618 /* everything expunged */
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1619 ctx->node_count--;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1620 node->file_offset = 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1621 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1622 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1623 node_pack(trie->buf, node);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1624 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1625
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1626 if ((ctx->output->offset & 1) != 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1627 o_stream_send(ctx->output, "", 1);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1628 node->file_offset = ctx->output->offset;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1629
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1630 o_stream_send(ctx->output, trie->buf->data, trie->buf->used);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1631 return 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1632 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1633
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1634 static int squat_trie_compress_init(struct squat_trie_compress_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1635 struct squat_trie *trie)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1636 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1637 struct squat_trie_header hdr;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1638
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1639 memset(ctx, 0, sizeof(*ctx));
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1640
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1641 ctx->tmp_path = t_strconcat(trie->filepath, ".tmp", NULL);
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1642 ctx->fd = open(ctx->tmp_path, O_RDWR | O_CREAT | O_TRUNC, 0600);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1643 if (ctx->fd == -1) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1644 i_error("open(%s, O_CREAT) failed: %m", ctx->tmp_path);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1645 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1646 }
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1647
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1648 ctx->trie = trie;
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1649 ctx->output = o_stream_create_file(ctx->fd, default_pool, 0, FALSE);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1650 ctx->node_count = trie->hdr->node_count;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1651
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1652 /* write a dummy header first */
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1653 memset(&hdr, 0, sizeof(hdr));
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1654 o_stream_send(ctx->output, &hdr, sizeof(hdr));
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1655 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1656 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1657
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1658 static void
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1659 squat_trie_compress_write_header(struct squat_trie_compress_context *ctx,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1660 struct trie_node *root_node)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1661 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1662 struct squat_trie_header hdr;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1663
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1664 memset(&hdr, 0, sizeof(hdr));
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1665 hdr.version = SQUAT_TRIE_VERSION;
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1666 hdr.uidvalidity = ctx->trie->uidvalidity;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1667 hdr.root_offset = root_node->file_offset;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1668 hdr.used_file_size = ctx->output->offset;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1669 hdr.node_count = ctx->node_count;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1670
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1671 o_stream_seek(ctx->output, 0);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1672 o_stream_send(ctx->output, &hdr, sizeof(hdr));
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1673 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1674
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1675 int squat_trie_compress(struct squat_trie *trie,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1676 const ARRAY_TYPE(seq_range) *existing_uids)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1677 {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1678 struct squat_trie_compress_context ctx;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1679 struct trie_node *node;
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1680 struct file_lock *file_lock = NULL;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1681 unsigned int orig_lock_count;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1682 int ret;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1683
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1684 orig_lock_count = trie->lock_count;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1685 if (squat_trie_lock(trie, F_WRLCK) <= 0)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1686 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1687
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1688 if (squat_trie_compress_init(&ctx, trie) < 0) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1689 squat_trie_unlock(trie);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1690 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1691 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1692
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1693 ret = trie_map_node(trie, trie->hdr->root_offset, 1, &node);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1694 if (ret == 0) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1695 /* do the compression */
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1696 ctx.uidlist_ctx = squat_uidlist_compress_begin(trie->uidlist,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1697 existing_uids);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1698 if ((ret = squat_trie_compress_node(&ctx, node, 1)) < 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1699 squat_uidlist_compress_rollback(&ctx.uidlist_ctx);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1700 else {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1701 ret = squat_uidlist_compress_commit(&ctx.uidlist_ctx);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1702
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1703 squat_trie_compress_write_header(&ctx, node);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1704 }
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1705 }
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1706
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1707 if (ret == 0 && orig_lock_count > 0) {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1708 /* lock the file before renaming so we can keep it locked. */
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1709 if (squat_trie_file_lock(trie, ctx.fd, ctx.tmp_path, F_WRLCK,
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1710 &file_lock) <= 0)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1711 ret = -1;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1712 }
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1713
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1714 if (ret == 0) {
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1715 if (rename(ctx.tmp_path, trie->filepath) < 0) {
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1716 i_error("rename(%s, %s) failed: %m",
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1717 ctx.tmp_path, trie->filepath);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1718 ret = -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1719 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1720 }
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1721
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1722 o_stream_destroy(&ctx.output);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1723 squat_trie_unlock(trie);
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1724
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1725 if (ret < 0) {
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1726 if (file_lock != NULL)
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1727 file_lock_free(&file_lock);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1728 (void)close(ctx.fd);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1729 (void)unlink(ctx.tmp_path);
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1730 } else {
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1731 trie_file_close(trie);
4886
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1732 trie_file_open_fd(trie, ctx.fd);
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1733
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1734 trie->file_lock = file_lock;
ae2114f11a0e Memory leak fixes. Also when building a large mailbox flush once in a while
Timo Sirainen <tss@iki.fi>
parents: 4883
diff changeset
1735 if (squat_trie_map(trie) <= 0)
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1736 return -1;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1737 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1738 return ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1739 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1740
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1741 int squat_trie_mark_having_expunges(struct squat_trie *trie,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1742 const ARRAY_TYPE(seq_range) *existing_uids,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1743 unsigned int current_message_count)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1744 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1745 bool compress;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1746 int ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1747
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1748 if ((ret = squat_trie_lock(trie, F_RDLCK)) <= 0)
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1749 return ret;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1750 compress = squat_trie_need_compress(trie, current_message_count);
4883
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1751 squat_trie_unlock(trie);
d8adbe93c969 Added support for mmap_disable=yes and several other fixes.
Timo Sirainen <tss@iki.fi>
parents: 4879
diff changeset
1752
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1753 ret = squat_uidlist_mark_having_expunges(trie->uidlist, compress);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1754
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1755 if (compress)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1756 ret = squat_trie_compress(trie, existing_uids);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1757 return ret;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1758 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1759
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1760 size_t squat_trie_mem_used(struct squat_trie *trie, unsigned int *count_r)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1761 {
4879
Timo Sirainen <tss@iki.fi>
parents: 4878
diff changeset
1762 *count_r = trie->hdr == NULL ? 0 : trie->hdr->node_count;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1763
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1764 return trie->mmap_size;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1765 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1766
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1767 static int squat_trie_lookup_init(struct squat_trie *trie, const char *str,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1768 const uint16_t **data_r, unsigned int *len_r)
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1769 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1770 const uint16_t *data;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1771 unsigned int len = strlen(str);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1772
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1773 if (len < BLOCK_SIZE)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1774 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1775
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1776 data = data_normalize(str, len, trie->buf);
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1777
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1778 /* skip the blocks that can't exist */
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1779 while (!block_want_add(data + len - BLOCK_SIZE)) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1780 if (--len < BLOCK_SIZE)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1781 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1782 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1783
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1784 if (squat_trie_lock(trie, F_RDLCK) <= 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1785 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1786
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1787 *data_r = data;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1788 *len_r = len;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1789 return 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1790 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1791
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1792 static int
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1793 squat_trie_lookup_locked(struct squat_trie *trie, ARRAY_TYPE(seq_range) *result,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1794 const uint16_t *data, unsigned int len)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1795 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1796 uint32_t list;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1797
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1798 list = trie_lookup_node(trie, trie->root, data + len - BLOCK_SIZE, 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1799 if (list == 0)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1800 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1801
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1802 if (squat_uidlist_get(trie->uidlist, list, result) < 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1803 squat_trie_set_corrupted(trie, "uidlist offset broken");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1804 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1805 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1806 while (len > BLOCK_SIZE) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1807 len--;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1808
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1809 if (!block_want_add(data + len - BLOCK_SIZE))
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1810 continue;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1811
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1812 list = trie_lookup_node(trie, trie->root,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1813 data + len - BLOCK_SIZE, 1);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1814 if (list == 0) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1815 array_clear(result);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1816 return 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1817 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1818 if (squat_uidlist_filter(trie->uidlist, list, result) < 0) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1819 squat_trie_set_corrupted(trie, "uidlist offset broken");
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1820 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1821 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1822 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1823 return array_count(result) > 0 ? 1 : 0;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1824 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1825
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1826 int squat_trie_lookup(struct squat_trie *trie, ARRAY_TYPE(seq_range) *result,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1827 const char *str)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1828 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1829 const uint16_t *data;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1830 unsigned int len;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1831 int ret;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1832
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1833 if (squat_trie_lookup_init(trie, str, &data, &len) < 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1834 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1835
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1836 ret = squat_trie_lookup_locked(trie, result, data, len);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1837 squat_trie_unlock(trie);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1838 return ret;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1839 }
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1840
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1841 static int
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1842 squat_trie_filter_locked(struct squat_trie *trie, ARRAY_TYPE(seq_range) *result,
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1843 const uint16_t *data, unsigned int len)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1844 {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1845 uint32_t list;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1846
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1847 for (; len >= BLOCK_SIZE; len--) {
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1848 if (!block_want_add(data + len - BLOCK_SIZE))
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1849 continue;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1850
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1851 list = trie_lookup_node(trie, trie->root,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1852 data + len - BLOCK_SIZE, 1);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1853 if (list == 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1854 array_clear(result);
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1855 return 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1856 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1857 if (squat_uidlist_filter(trie->uidlist, list, result) < 0) {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1858 squat_trie_set_corrupted(trie, "uidlist offset broken");
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1859 return -1;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1860 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1861 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1862 return array_count(result) > 0 ? 1 : 0;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1863 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1864
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1865 int squat_trie_filter(struct squat_trie *trie, ARRAY_TYPE(seq_range) *result,
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1866 const char *str)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1867 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1868 const uint16_t *data;
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1869 unsigned int len;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1870 int ret;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1871
4878
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1872 if (squat_trie_lookup_init(trie, str, &data, &len) < 0)
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1873 return -1;
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1874 ret = squat_trie_filter_locked(trie, result, data, len);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1875 squat_trie_unlock(trie);
88a91d9a867b Fixes. Should be pretty much working now.
Timo Sirainen <tss@iki.fi>
parents: 4855
diff changeset
1876 return ret;
4855
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1877 }
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1878
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1879 struct squat_uidlist *_squat_trie_get_uidlist(struct squat_trie *trie)
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1880 {
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1881 return trie->uidlist;
5bc593f1a8f6 Added "squat" full text search indexer backend. Its name and basic ideas
Timo Sirainen <tss@iki.fi>
parents:
diff changeset
1882 }