tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[utils] get_string_script returns a struct rather than modifying a pointer for the length

Browse Source
This commit is contained in:
Al
2015-06-25 10:06:38 -04:00
parent bcee9832b3
commit c376bcef3d
2 changed files with 13 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
src/unicode_scripts.c
View File

@@ -11,14 +11,14 @@ inline script_languages_t get_script_languages(script_t script) {
return script_languages[script]; return script_languages[script];
} }
script_t string_script(char *str, size_t len, size_t *script_len) { string_script_t get_string_script(char *str, size_t len) {
int32_t ch; int32_t ch;
script_t last_script = SCRIPT_UNKNOWN; script_t last_script = SCRIPT_UNKNOWN;
script_t script = SCRIPT_UNKNOWN; script_t script = SCRIPT_UNKNOWN;
uint8_t *ptr = (uint8_t *)str; uint8_t *ptr = (uint8_t *)str;
*script_len = 0; size_t script_len = 0;
size_t idx = 0; size_t idx = 0;
while (idx < len) { while (idx < len) {
@@ -28,23 +28,23 @@ script_t string_script(char *str, size_t len, size_t *script_len) {
script = get_char_script((uint32_t)ch); script = get_char_script((uint32_t)ch);
if (script == SCRIPT_COMMON) { if (script == SCRIPT_COMMON && last_script != SCRIPT_UNKNOWN) {
script = last_script; script = last_script;
} }
if (last_script != script && last_script != SCRIPT_UNKNOWN) { if (last_script != script && last_script != SCRIPT_UNKNOWN && last_script != SCRIPT_COMMON) {
break; break;
} }
ptr += char_len; ptr += char_len;
idx += char_len; idx += char_len;
*script_len += char_len; script_len += char_len;
if (script != SCRIPT_UNKNOWN || script != SCRIPT_COMMON) { if (script != SCRIPT_UNKNOWN) {
last_script = script; last_script = script;
} }
} }
return last_script; return (string_script_t) {last_script, script_len};
} }

7
src/unicode_scripts.h
View File

@@ -15,9 +15,14 @@ typedef struct script_languages {
char *languages[MAX_LANGS]; char *languages[MAX_LANGS];
} script_languages_t; } script_languages_t;
typedef struct string_script {
script_t script;
size_t len;
} string_script_t;
script_t get_char_script(uint32_t ch); script_t get_char_script(uint32_t ch);
script_languages_t get_script_languages(script_t script); script_languages_t get_script_languages(script_t script);
script_t string_script(char *str, size_t len, size_t *script_len); string_script_t get_string_script(char *str, size_t len);
#endif #endif