X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Ficu_tokenizer.c;h=d6ec7fee7f6aa710a4cb6b7fa3131a7dd10c7264;hp=a79fd6266f6d11dd26359aebc6714ca8bcebc5cf;hb=df84e4897abc4565fe11adead30b99d9634b41b1;hpb=3a98cd4b508df847d4aa2b7ed7c1a7213766bad3 diff --git a/src/icu_tokenizer.c b/src/icu_tokenizer.c index a79fd62..d6ec7fe 100644 --- a/src/icu_tokenizer.c +++ b/src/icu_tokenizer.c @@ -1,5 +1,5 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2009 Index Data + * Copyright (C) 1995-2010 Index Data * See the file LICENSE for details. */ @@ -19,6 +19,7 @@ #include +#include #include #include #include @@ -26,12 +27,28 @@ #include /* some more string fcns*/ #include /* char names */ -struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action, - UErrorCode *status) +struct icu_tokenizer +{ + char action; + UBreakIterator* bi; + struct icu_buf_utf16 * buf16; + int32_t token_count; + int32_t token_id; + int32_t token_start; + int32_t token_end; +/* + keep always invariant + 0 <= token_start + <= token_end + <= buf16->utf16_len + and invariant + 0 <= token_id <= token_count +*/ +}; + +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); @@ -39,7 +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) +{ + int32_t bufferSize = U_BRK_SAFECLONE_BUFFERSIZE; + UErrorCode status = U_ZERO_ERROR; + 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':