diff src/lib/unichar.h @ 7185:6f014a866f38 HEAD

Replace invalid UTF8 input with a replacement character.
author Timo Sirainen <tss@iki.fi>
date Tue, 22 Jan 2008 09:31:59 +0200
parents dcbf6afdf931
children 81806d402514
line wrap: on
line diff
--- a/src/lib/unichar.h	Tue Jan 22 08:49:24 2008 +0200
+++ b/src/lib/unichar.h	Tue Jan 22 09:31:59 2008 +0200
@@ -1,6 +1,9 @@
 #ifndef UNICHAR_H
 #define UNICHAR_H
 
+/* Character used to replace invalid input. */
+#define UNICODE_REPLACEMENT_CHAR 0xfffd
+
 typedef uint32_t unichar_t;
 ARRAY_DEFINE_TYPE(unichars, unichar_t);
 
@@ -37,13 +40,14 @@
 
 /* Convert UTF-8 input to titlecase and decompose the titlecase characters to
    output buffer. Returns 0 if ok, -1 if input was invalid. This generates
-   output that's compatible with i;unicode-casemap comparator. */
+   output that's compatible with i;unicode-casemap comparator. Invalid input
+   is replaced with unicode replacement character (0xfffd). */
 int uni_utf8_to_decomposed_titlecase(const void *input, size_t max_len,
 				     buffer_t *output);
 
-/* If input contains only valid UTF-8 characters, return TRUE. If input
-   contains invalid UTF-8 characters, write only the valid ones to buf and
-   return FALSE. */
+/* If input contains only valid UTF-8 characters, return TRUE without updating
+   buf. If input contains invalid UTF-8 characters, replace them with unicode
+   replacement character (0xfffd), write the output to buf and return FALSE. */
 bool uni_utf8_get_valid_data(const unsigned char *input, size_t size,
 			     buffer_t *buf);