added code for unac to perform pure case-folding

This commit is contained in:
Jean-Francois Dockes
2012-08-27 12:40:57 +02:00
parent 9327527101
commit 913dffc597
12 changed files with 2133 additions and 1972 deletions
+988 -946
View File
File diff suppressed because it is too large Load Diff
+24 -22
View File
@@ -35,7 +35,7 @@ extern "C" {
#define UNAC_BLOCK_SHIFT 4
#define UNAC_BLOCK_MASK ((1 << UNAC_BLOCK_SHIFT) - 1)
#define UNAC_BLOCK_SIZE (1 << UNAC_BLOCK_SHIFT)
#define UNAC_BLOCK_COUNT 418
#define UNAC_BLOCK_COUNT 422
#define UNAC_INDEXES_SIZE (0x10000 >> UNAC_BLOCK_SHIFT)
/* Generated by builder. Do not modify. End defines */
@@ -45,38 +45,31 @@ extern "C" {
* by <p> is returned in the <l> argument.
* The C++ prototype of this macro would be:
*
* void unac_char(const unsigned short c, unsigned short*& p, int& l)
* void unac_char(const unsigned short c, unsigned short*& p, int& l, int o)
*
* See unac(3) in IMPLEMENTATION NOTES for more information about the
* tables (unac_data_table, unac_positions) layout.
*
* Each transformed char has 3 possible outputs: unaccented, unaccented and
* folded, or just folded. These are kept at offset 0,1,2 in the position table
*/
#define unac_char_utf16(c,p,l) \
{ \
unsigned short index = unac_indexes[(c) >> UNAC_BLOCK_SHIFT]; \
unsigned char position = 2*((c) & UNAC_BLOCK_MASK); \
(p) = &(unac_data_table[index][unac_positions[index][position]]); \
(l) = unac_positions[index][position + 1] - unac_positions[index][position]; \
if((l) == 1 && *(p) == 0xFFFF) { \
(p) = 0; \
(l) = 0; \
} \
}
/*
* Save as unac_ but case-folded
*/
#define unacfold_char_utf16(c,p,l) \
#define unac_uf_char_utf16_(c,p,l,o) \
{ \
unsigned short index = unac_indexes[(c) >> UNAC_BLOCK_SHIFT]; \
unsigned char position = 2*((c) & UNAC_BLOCK_MASK)+1; \
unsigned short index = unac_indexes[(c) >> UNAC_BLOCK_SHIFT]; \
unsigned char position = 3*((c) & UNAC_BLOCK_MASK) + (o); \
(p) = &(unac_data_table[index][unac_positions[index][position]]); \
(l) = unac_positions[index][position + 1] - unac_positions[index][position]; \
(l) = unac_positions[index][position + 1] \
- unac_positions[index][position]; \
if((l) == 1 && *(p) == 0xFFFF) { \
(p) = 0; \
(l) = 0; \
} \
}
#define unac_char_utf16(c,p,l) unac_uf_char_utf16_((c),(p),(l),0)
#define unacfold_char_utf16(c,p,l) unac_uf_char_utf16_((c),(p),(l),1)
#define fold_char_utf16(c,p,l) unac_uf_char_utf16_((c),(p),(l),2)
/*
* Return the unaccented equivalent of the UTF-16 string <in> of
* length <in_length> in the pointer <out>. The length of the UTF-16
@@ -94,6 +87,8 @@ int unac_string_utf16(const char* in, size_t in_length,
char** out, size_t* out_length);
int unacfold_string_utf16(const char* in, size_t in_length,
char** out, size_t* out_length);
int fold_string_utf16(const char* in, size_t in_length,
char** out, size_t* out_length);
/*
* The semantic of this function is stricly equal to the function
@@ -112,6 +107,9 @@ int unac_string(const char* charset,
int unacfold_string(const char* charset,
const char* in, size_t in_length,
char** out, size_t* out_length);
int fold_string(const char* charset,
const char* in, size_t in_length,
char** out, size_t* out_length);
/* To be called before starting threads in mt programs */
void unac_init_mt();
@@ -180,7 +178,7 @@ void unac_debug_callback(int level, unac_debug_print_t function, void* data);
/* Generated by builder. Do not modify. Start declarations */
extern unsigned short unac_indexes[UNAC_INDEXES_SIZE];
extern unsigned char unac_positions[UNAC_BLOCK_COUNT][2*UNAC_BLOCK_SIZE + 1];
extern unsigned char unac_positions[UNAC_BLOCK_COUNT][3*UNAC_BLOCK_SIZE + 1];
extern unsigned short* unac_data_table[UNAC_BLOCK_COUNT];
extern unsigned short unac_data0[];
extern unsigned short unac_data1[];
@@ -600,6 +598,10 @@ extern unsigned short unac_data414[];
extern unsigned short unac_data415[];
extern unsigned short unac_data416[];
extern unsigned short unac_data417[];
extern unsigned short unac_data418[];
extern unsigned short unac_data419[];
extern unsigned short unac_data420[];
extern unsigned short unac_data421[];
/* Generated by builder. Do not modify. End declarations */
#ifdef __cplusplus
+1 -1
View File
@@ -1 +1 @@
#define UNAC_VERSION "1.0.7"
#define UNAC_VERSION "1.7.0"