Mercurial > dovecot > original-hg > dovecot-1.2
annotate src/plugins/fts-solr/fts-backend-solr.c @ 8493:149c6ccc3df4 HEAD
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
author | Timo Sirainen <tss@iki.fi> |
---|---|
date | Sat, 29 Nov 2008 19:24:59 +0200 |
parents | ee1ecdda8a6b |
children | ffb37c392166 |
rev | line source |
---|---|
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
1 /* Copyright (c) 2006-2008 Dovecot authors, see the included COPYING file */ |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
2 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
3 #include "lib.h" |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
4 #include "array.h" |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
5 #include "str.h" |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
6 #include "mail-storage-private.h" |
8109
e7929190cd32
fts-solr: Fixed compiling with recent struct mail_user changes.
Timo Sirainen <tss@iki.fi>
parents:
8006
diff
changeset
|
7 #include "mail-namespace.h" |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
8 #include "solr-connection.h" |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
9 #include "fts-solr-plugin.h" |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
10 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
11 #include <curl/curl.h> |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
12 |
8001
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
13 #define SOLR_CMDBUF_SIZE (1024*64) |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
14 #define SOLR_MAX_ROWS 100000 |
8001
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
15 |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
16 struct solr_fts_backend { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
17 struct fts_backend backend; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
18 char *id_username; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
19 }; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
20 |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
21 struct solr_fts_backend_build_context { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
22 struct fts_backend_build_context ctx; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
23 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
24 struct solr_connection_post *post; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
25 uint32_t prev_uid, uid_validity; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
26 string_t *cmd; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
27 bool headers; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
28 }; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
29 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
30 static struct solr_connection *solr_conn = NULL; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
31 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
32 static void solr_quote_str(string_t *dest, const char *str) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
33 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
34 solr_connection_quote_str(solr_conn, dest, str); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
35 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
36 |
8000
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
37 static void |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
38 xml_encode_data(string_t *dest, const unsigned char *data, unsigned int len) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
39 { |
8000
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
40 unsigned int i; |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
41 |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
42 for (i = 0; i < len; i++) { |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
43 switch (data[i]) { |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
44 case '&': |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
45 str_append(dest, "&"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
46 break; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
47 case '<': |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
48 str_append(dest, "<"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
49 break; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
50 case '>': |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
51 str_append(dest, ">"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
52 break; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
53 default: |
8000
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
54 str_append_c(dest, data[i]); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
55 break; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
56 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
57 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
58 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
59 |
8000
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
60 static void xml_encode(string_t *dest, const char *str) |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
61 { |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
62 xml_encode_data(dest, (const unsigned char *)str, strlen(str)); |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
63 } |
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
64 |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
65 static const char *solr_escape_id_str(const char *str) |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
66 { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
67 string_t *tmp; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
68 const char *p; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
69 |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
70 for (p = str; *p != '\0'; p++) { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
71 if (*p == '/' || *p == '!') |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
72 break; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
73 } |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
74 if (*p == '\0') |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
75 return str; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
76 |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
77 tmp = t_str_new(64); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
78 for (p = str; *p != '\0'; p++) { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
79 switch (*p) { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
80 case '/': |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
81 str_append(tmp, "!\\"); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
82 break; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
83 case '!': |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
84 str_append(tmp, "!!"); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
85 break; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
86 default: |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
87 str_append_c(tmp, *p); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
88 break; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
89 } |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
90 } |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
91 return str_c(tmp); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
92 } |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
93 |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
94 static struct fts_backend * |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
95 fts_backend_solr_init(struct mailbox *box) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
96 { |
8002
e814614ffb65
fts-solr: Added break-imap-search setting to use Solr with SEARCH TEXT/BODY
Timo Sirainen <tss@iki.fi>
parents:
8001
diff
changeset
|
97 const struct fts_solr_settings *set = &fts_solr_settings; |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
98 struct solr_fts_backend *backend; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
99 const char *username = box->storage->ns->user->username; |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
100 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
101 if (solr_conn == NULL) |
8002
e814614ffb65
fts-solr: Added break-imap-search setting to use Solr with SEARCH TEXT/BODY
Timo Sirainen <tss@iki.fi>
parents:
8001
diff
changeset
|
102 solr_conn = solr_connection_init(set->url, set->debug); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
103 |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
104 backend = i_new(struct solr_fts_backend, 1); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
105 backend->id_username = i_strdup(solr_escape_id_str(username)); |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
106 backend->backend = fts_backend_solr; |
8002
e814614ffb65
fts-solr: Added break-imap-search setting to use Solr with SEARCH TEXT/BODY
Timo Sirainen <tss@iki.fi>
parents:
8001
diff
changeset
|
107 |
e814614ffb65
fts-solr: Added break-imap-search setting to use Solr with SEARCH TEXT/BODY
Timo Sirainen <tss@iki.fi>
parents:
8001
diff
changeset
|
108 if (set->substring_search) |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
109 backend->backend.flags |= FTS_BACKEND_FLAG_SUBSTRING_LOOKUPS; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
110 return &backend->backend; |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
111 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
112 |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
113 static void fts_backend_solr_deinit(struct fts_backend *_backend) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
114 { |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
115 struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
116 |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
117 i_free(backend->id_username); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
118 i_free(backend); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
119 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
120 |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
121 static const char *fts_backend_solr_username(struct fts_backend *_backend) |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
122 { |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
123 struct solr_fts_backend *backend = (struct solr_fts_backend *)_backend; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
124 |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
125 return backend->id_username; |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
126 } |
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
127 |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
128 static int fts_backend_solr_get_last_uid_fallback(struct fts_backend *backend, |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
129 uint32_t *last_uid_r) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
130 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
131 struct mailbox_status status; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
132 ARRAY_TYPE(seq_range) uids; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
133 const struct seq_range *uidvals; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
134 unsigned int count; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
135 string_t *str; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
136 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
137 str = t_str_new(256); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
138 str_append(str, "fl=uid&rows=1&sort=uid%20desc&q="); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
139 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
140 mailbox_get_status(backend->box, STATUS_UIDVALIDITY, &status); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
141 str_printfa(str, "uidv:%u%%20box:", status.uidvalidity); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
142 solr_quote_str(str, backend->box->name); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
143 str_append(str, "%20user:"); |
8109
e7929190cd32
fts-solr: Fixed compiling with recent struct mail_user changes.
Timo Sirainen <tss@iki.fi>
parents:
8006
diff
changeset
|
144 solr_quote_str(str, backend->box->storage->ns->user->username); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
145 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
146 t_array_init(&uids, 1); |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
147 if (solr_connection_select(solr_conn, str_c(str), |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
148 NULL, NULL, &uids, NULL) < 0) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
149 return -1; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
150 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
151 uidvals = array_get(&uids, &count); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
152 if (count == 0) { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
153 /* nothing indexed yet for this mailbox */ |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
154 *last_uid_r = 0; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
155 } else if (count == 1 && uidvals[0].seq1 == uidvals[0].seq2) { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
156 *last_uid_r = uidvals[0].seq1; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
157 } else { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
158 i_error("fts_solr: Last UID lookup returned multiple rows"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
159 return -1; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
160 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
161 return 0; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
162 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
163 |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
164 static int fts_backend_solr_get_last_uid(struct fts_backend *backend, |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
165 uint32_t *last_uid_r) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
166 { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
167 struct mailbox_status status; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
168 ARRAY_TYPE(seq_range) uids; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
169 const struct seq_range *uidvals; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
170 unsigned int count; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
171 string_t *str; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
172 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
173 str = t_str_new(256); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
174 str_append(str, "fl=uid&rows=1&q=last_uid:TRUE%20"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
175 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
176 mailbox_get_status(backend->box, STATUS_UIDVALIDITY, &status); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
177 str_printfa(str, "uidv:%u%%20box:", status.uidvalidity); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
178 solr_quote_str(str, backend->box->name); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
179 str_append(str, "%20user:"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
180 solr_quote_str(str, backend->box->storage->ns->user->username); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
181 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
182 t_array_init(&uids, 1); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
183 if (solr_connection_select(solr_conn, str_c(str), |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
184 NULL, NULL, &uids, NULL) < 0) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
185 return -1; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
186 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
187 uidvals = array_get(&uids, &count); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
188 if (count == 0) { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
189 /* either nothing is indexed or we're converting from an |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
190 older database format without the last_uid fields */ |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
191 return fts_backend_solr_get_last_uid_fallback(backend, |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
192 last_uid_r); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
193 } else if (count == 1 && uidvals[0].seq1 == uidvals[0].seq2) { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
194 *last_uid_r = uidvals[0].seq1; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
195 } else { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
196 i_error("fts_solr: Last UID lookup returned multiple rows"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
197 return -1; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
198 } |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
199 return 0; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
200 } |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
201 |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
202 static int |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
203 fts_backend_solr_build_init(struct fts_backend *backend, uint32_t *last_uid_r, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
204 struct fts_backend_build_context **ctx_r) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
205 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
206 struct solr_fts_backend_build_context *ctx; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
207 struct mailbox_status status; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
208 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
209 *last_uid_r = (uint32_t)-1; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
210 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
211 ctx = i_new(struct solr_fts_backend_build_context, 1); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
212 ctx->ctx.backend = backend; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
213 ctx->post = solr_connection_post_begin(solr_conn); |
8001
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
214 ctx->cmd = str_new(default_pool, SOLR_CMDBUF_SIZE); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
215 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
216 mailbox_get_status(backend->box, STATUS_UIDVALIDITY, &status); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
217 ctx->uid_validity = status.uidvalidity; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
218 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
219 *ctx_r = &ctx->ctx; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
220 return 0; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
221 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
222 |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
223 static void |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
224 fts_backend_solr_add_doc_prefix(struct solr_fts_backend_build_context *ctx, |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
225 uint32_t uid) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
226 { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
227 struct mailbox *box = ctx->ctx.backend->box; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
228 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
229 str_printfa(ctx->cmd, "<doc>" |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
230 "<field name=\"uid\">%u</field>" |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
231 "<field name=\"uidv\">%u</field>", |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
232 uid, ctx->uid_validity); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
233 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
234 str_append(ctx->cmd, "<field name=\"box\">"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
235 xml_encode(ctx->cmd, box->name); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
236 str_append(ctx->cmd, "</field><field name=\"user\">"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
237 xml_encode(ctx->cmd, box->storage->ns->user->username); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
238 str_append(ctx->cmd, "</field>"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
239 } |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
240 |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
241 static int |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
242 fts_backend_solr_build_more(struct fts_backend_build_context *_ctx, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
243 uint32_t uid, const unsigned char *data, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
244 size_t size, bool headers) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
245 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
246 struct solr_fts_backend_build_context *ctx = |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
247 (struct solr_fts_backend_build_context *)_ctx; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
248 struct mailbox *box = _ctx->backend->box; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
249 string_t *cmd = ctx->cmd; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
250 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
251 /* body comes first, then headers */ |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
252 if (ctx->prev_uid != uid) { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
253 /* uid changed */ |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
254 if (ctx->prev_uid == 0) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
255 str_append(cmd, "<add>"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
256 else |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
257 str_append(cmd, "</field></doc>"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
258 ctx->prev_uid = uid; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
259 |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
260 fts_backend_solr_add_doc_prefix(ctx, uid); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
261 str_printfa(cmd, "<field name=\"id\">%u/%u/", |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
262 uid, ctx->uid_validity); |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
263 xml_encode(cmd, fts_backend_solr_username(ctx->ctx.backend)); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
264 str_append_c(cmd, '/'); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
265 xml_encode(cmd, box->name); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
266 str_append(cmd, "</field>"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
267 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
268 ctx->headers = headers; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
269 if (headers) { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
270 str_append(cmd, "<field name=\"hdr\">"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
271 } else { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
272 str_append(cmd, "<field name=\"body\">"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
273 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
274 } else if (headers && !ctx->headers) { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
275 str_append(cmd, "</field><field name=\"hdr\">"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
276 } else { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
277 i_assert(!(!headers && ctx->headers)); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
278 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
279 |
8000
61be6a4722c5
fts-solr: Encode entities when indexing data.
Timo Sirainen <tss@iki.fi>
parents:
7999
diff
changeset
|
280 xml_encode_data(cmd, data, size); |
8001
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
281 if (str_len(cmd) > SOLR_CMDBUF_SIZE-128) { |
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
282 solr_connection_post_more(ctx->post, str_data(cmd), |
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
283 str_len(cmd)); |
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
284 str_truncate(cmd, 0); |
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
285 } |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
286 return 0; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
287 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
288 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
289 static int |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
290 fts_backed_solr_build_commit(struct solr_fts_backend_build_context *ctx) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
291 { |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
292 struct mailbox *box = ctx->ctx.backend->box; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
293 int ret; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
294 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
295 if (ctx->prev_uid == 0) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
296 return 0; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
297 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
298 str_append(ctx->cmd, "</field></doc>"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
299 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
300 /* Update the mailbox's last_uid field, replacing the existing |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
301 document. Note that since there is no locking, it's possible that |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
302 if another session is indexing at the same time, the last_uid value |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
303 may shrink. This doesn't really matter, we'll simply do more work |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
304 in future by reindexing some messages. */ |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
305 fts_backend_solr_add_doc_prefix(ctx, ctx->prev_uid); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
306 str_printfa(ctx->cmd, "<field name=\"last_uid\">TRUE</field>" |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
307 "<field name=\"id\">L/%u/", ctx->uid_validity); |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
308 xml_encode(ctx->cmd, fts_backend_solr_username(ctx->ctx.backend)); |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
309 str_append_c(ctx->cmd, '/'); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
310 xml_encode(ctx->cmd, box->name); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
311 str_append(ctx->cmd, "</field></doc></add>"); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
312 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
313 solr_connection_post_more(ctx->post, str_data(ctx->cmd), |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
314 str_len(ctx->cmd)); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
315 ret = solr_connection_post_end(ctx->post); |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
316 /* commit and wait until the documents we just indexed are |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
317 visible to the following search */ |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
318 if (solr_connection_post(solr_conn, "<commit waitFlush=\"false\" " |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
319 "waitSearcher=\"true\"/>") < 0) |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
320 ret = -1; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
321 return ret; |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
322 } |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
323 |
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
324 static int |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
325 fts_backend_solr_build_deinit(struct fts_backend_build_context *_ctx) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
326 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
327 struct solr_fts_backend_build_context *ctx = |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
328 (struct solr_fts_backend_build_context *)_ctx; |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
329 int ret; |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
330 |
8492
ee1ecdda8a6b
fts-solr: Keep last_uid field updated containing the mailbox's last added UID.
Timo Sirainen <tss@iki.fi>
parents:
8484
diff
changeset
|
331 ret = fts_backed_solr_build_commit(ctx); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
332 str_free(&ctx->cmd); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
333 i_free(ctx); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
334 return ret; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
335 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
336 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
337 static void |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
338 fts_backend_solr_expunge(struct fts_backend *backend, struct mail *mail) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
339 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
340 struct mailbox_status status; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
341 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
342 mailbox_get_status(mail->box, STATUS_UIDVALIDITY, &status); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
343 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
344 T_BEGIN { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
345 string_t *cmd; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
346 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
347 cmd = t_str_new(256); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
348 str_printfa(cmd, "<delete><id>%u/%u/", |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
349 mail->uid, status.uidvalidity); |
8493
149c6ccc3df4
fts-solr: Use '!' as escape character in usernames. Escape '/' and '!'.
Timo Sirainen <tss@iki.fi>
parents:
8492
diff
changeset
|
350 xml_encode(cmd, fts_backend_solr_username(backend)); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
351 str_append_c(cmd, '/'); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
352 xml_encode(cmd, mail->box->name); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
353 str_append(cmd, "</id></delete>"); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
354 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
355 (void)solr_connection_post(solr_conn, str_c(cmd)); |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
356 } T_END; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
357 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
358 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
359 static void |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
360 fts_backend_solr_expunge_finish(struct fts_backend *backend ATTR_UNUSED, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
361 struct mailbox *box ATTR_UNUSED, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
362 bool committed ATTR_UNUSED) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
363 { |
8001
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
364 solr_connection_post(solr_conn, |
3fbfdda3e5d3
fts-solr: Send data to Solr in larger packets. Commit after expunges.
Timo Sirainen <tss@iki.fi>
parents:
8000
diff
changeset
|
365 "<commit waitFlush=\"false\" waitSearcher=\"false\"/>"); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
366 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
367 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
368 static int fts_backend_solr_lock(struct fts_backend *backend ATTR_UNUSED) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
369 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
370 return 1; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
371 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
372 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
373 static void fts_backend_solr_unlock(struct fts_backend *backend ATTR_UNUSED) |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
374 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
375 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
376 |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
377 static bool solr_virtual_uid_map(const char *mailbox, uint32_t uidvalidity, |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
378 uint32_t *uid, void *context) |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
379 { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
380 struct mailbox *box = context; |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
381 |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
382 return mailbox_get_virtual_uid(box, mailbox, uidvalidity, |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
383 *uid, uid); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
384 } |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
385 |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
386 static int fts_backend_solr_lookup(struct fts_backend_lookup_context *ctx, |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
387 ARRAY_TYPE(seq_range) *definite_uids, |
8006
0a3b5ddb0869
Added support for sorting by X-SCORE. Currently it's only set by fts-solr.
Timo Sirainen <tss@iki.fi>
parents:
8004
diff
changeset
|
388 ARRAY_TYPE(seq_range) *maybe_uids, |
0a3b5ddb0869
Added support for sorting by X-SCORE. Currently it's only set by fts-solr.
Timo Sirainen <tss@iki.fi>
parents:
8004
diff
changeset
|
389 ARRAY_TYPE(fts_score_map) *scores) |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
390 { |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
391 struct mailbox *box = ctx->backend->box; |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
392 const struct fts_backend_lookup_field *fields; |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
393 unsigned int i, count; |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
394 struct mailbox_status status; |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
395 string_t *str; |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
396 bool virtual; |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
397 |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
398 virtual = strcmp(box->storage->name, "virtual") == 0; |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
399 mailbox_get_status(box, STATUS_UIDVALIDITY, &status); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
400 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
401 str = t_str_new(256); |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
402 if (!virtual) { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
403 str_printfa(str, "fl=uid,score&rows=%u&sort=uid%%20asc&q=", |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
404 status.uidnext); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
405 } else { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
406 str_printfa(str, "fl=uid,score,box,uidv&rows=%u" |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
407 "&sort=box%%20asc,uid%%20asc&q=", |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
408 SOLR_MAX_ROWS); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
409 } |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
410 |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
411 /* build a lucene search query from the fields */ |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
412 fields = array_get(&ctx->fields, &count); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
413 for (i = 0; i < count; i++) { |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
414 if (i > 0) |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
415 str_append(str, "%20"); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
416 |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
417 if ((fields[i].flags & FTS_LOOKUP_FLAG_INVERT) != 0) |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
418 str_append_c(str, '-'); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
419 |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
420 if ((fields[i].flags & FTS_LOOKUP_FLAG_HEADER) == 0) { |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
421 /* body only */ |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
422 i_assert((fields[i].flags & FTS_LOOKUP_FLAG_BODY) != 0); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
423 str_append(str, "body:"); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
424 } else if ((fields[i].flags & FTS_LOOKUP_FLAG_BODY) == 0) { |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
425 /* header only */ |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
426 str_append(str, "hdr:"); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
427 } else { |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
428 /* both */ |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
429 str_append(str, "any:"); |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
430 } |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
431 solr_quote_str(str, fields[i].key); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
432 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
433 |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
434 /* use a separate filter query for selecting the mailbox. it shouldn't |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
435 affect the score and there could be some caching benefits too. */ |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
436 str_append(str, "&fq=user:"); |
8109
e7929190cd32
fts-solr: Fixed compiling with recent struct mail_user changes.
Timo Sirainen <tss@iki.fi>
parents:
8006
diff
changeset
|
437 solr_quote_str(str, box->storage->ns->user->username); |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
438 |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
439 /* FIXME: limit what mailboxes to search with virtual storage */ |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
440 if (!virtual) { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
441 str_printfa(str, "%%20uidv:%u%%20box:", status.uidvalidity); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
442 solr_quote_str(str, box->name); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
443 } |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
444 |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
445 array_clear(maybe_uids); |
8484
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
446 if (!virtual) { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
447 return solr_connection_select(solr_conn, str_c(str), NULL, NULL, |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
448 definite_uids, scores); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
449 } else { |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
450 return solr_connection_select(solr_conn, str_c(str), |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
451 solr_virtual_uid_map, box, |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
452 definite_uids, scores); |
db78eab2ba5d
Initial support for getting full text search working nicely with virtual mailboxes.
Timo Sirainen <tss@iki.fi>
parents:
8109
diff
changeset
|
453 } |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
454 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
455 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
456 struct fts_backend fts_backend_solr = { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
457 MEMBER(name) "solr", |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
458 MEMBER(flags) 0, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
459 |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
460 { |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
461 fts_backend_solr_init, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
462 fts_backend_solr_deinit, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
463 fts_backend_solr_get_last_uid, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
464 fts_backend_solr_build_init, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
465 fts_backend_solr_build_more, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
466 fts_backend_solr_build_deinit, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
467 fts_backend_solr_expunge, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
468 fts_backend_solr_expunge_finish, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
469 fts_backend_solr_lock, |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
470 fts_backend_solr_unlock, |
8004
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
471 NULL, |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
472 NULL, |
6418d4c0ff16
fts: Added a new lookup API where the backend can look up all the fields
Timo Sirainen <tss@iki.fi>
parents:
8002
diff
changeset
|
473 fts_backend_solr_lookup |
7990
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
474 } |
662172573fe1
Initial code to support Apache Solr (Lucene indexing server).
Timo Sirainen <tss@iki.fi>
parents:
diff
changeset
|
475 }; |