X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Ficu_tokenizer.c;fp=src%2Ficu_tokenizer.c;h=e8d3123270f3ee2be2777f3066f5aa18f3d43001;hp=f9b4926c8ae410aacac32ab99fa519b3887bdd10;hb=527a008f2f3b1a19a968a7a1275132a04aa4d945;hpb=ded9dadedab82a379e0e52d78a61fe9632ac870c diff --git a/src/icu_tokenizer.c b/src/icu_tokenizer.c index f9b4926..e8d3123 100644 --- a/src/icu_tokenizer.c +++ b/src/icu_tokenizer.c @@ -19,6 +19,7 @@ #include +#include #include #include #include @@ -45,12 +46,9 @@ struct icu_tokenizer */ }; -struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action, - UErrorCode *status) +static void icu_tokenizer_reset(struct icu_tokenizer *tokenizer, + char action) { - struct icu_tokenizer * tokenizer - = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer)); - tokenizer->action = action; tokenizer->bi = 0; tokenizer->buf16 = icu_buf_utf16_create(0); @@ -58,7 +56,32 @@ struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action, tokenizer->token_id = 0; tokenizer->token_start = 0; tokenizer->token_end = 0; + tokenizer->bi = 0; +} + +struct icu_tokenizer *icu_tokenizer_clone(struct icu_tokenizer *old) +{ + uint32_t bufferSize = 10000; + UErrorCode status = 0; + struct icu_tokenizer * tokenizer + = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer)); + + assert(old); + icu_tokenizer_reset(tokenizer, old->action); + assert(old->bi); + tokenizer->bi = ubrk_safeClone(old->bi, NULL, &bufferSize, &status); + if (U_SUCCESS(status)) + return tokenizer; + return tokenizer; +} + +struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action, + UErrorCode *status) +{ + struct icu_tokenizer * tokenizer + = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer)); + icu_tokenizer_reset(tokenizer, action); switch (tokenizer->action) { case 'l':