From: Adam Dickmeiss Date: Tue, 2 Feb 2010 12:54:07 +0000 (+0100) Subject: New icu_iter-functions X-Git-Tag: v4.0.2~18 X-Git-Url: http://lists.indexdata.dk/?a=commitdiff_plain;h=fbb0389c468253ae678c1169fe0d7a8d8d54e8cb;p=yaz-moved-to-github.git New icu_iter-functions The new icu_iter-functions makes it possible to use ICU chains from many threads. --- diff --git a/include/yaz/icu.h b/include/yaz/icu.h index ddf9f63..4abba69 100644 --- a/include/yaz/icu.h +++ b/include/yaz/icu.h @@ -109,6 +109,67 @@ YAZ_EXPORT const char * icu_chain_token_norm(yaz_icu_chain_t chain); */ YAZ_EXPORT const char * icu_chain_token_sortkey(yaz_icu_chain_t chain); +/** \brief ICU tokenizer iterator type (opaque) */ +typedef struct icu_iter *yaz_icu_iter_t; + +/** \brief create ICU tokenizer iterator from chain + \param chain ICU chain + \returns ICU iterator +*/ +YAZ_EXPORT +yaz_icu_iter_t icu_iter_create(struct icu_chain *chain); + +/** \brief starts iteration over string + \param iter ICU tokenizer iterator + \param src8cstr input string (0-terminated) + + Call icu_iter_next to iterate over each token. +*/ +YAZ_EXPORT +void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr); + +/** \brief iterates over one token + \param iter ICU tokenizer iterator + \retval 0 no more tokens (EOF) + \retval 1 got one token (use icu_iter_get..-functions) +*/ +YAZ_EXPORT +int icu_iter_next(yaz_icu_iter_t iter); + +/** \brief destroy ICU tokenizer iterator + \param iter ICU tokenizer iterator +*/ +YAZ_EXPORT +void icu_iter_destroy(yaz_icu_iter_t iter); + +/** \brief returns ICU normalized token + \param iter ICU tokenizer iterator + \returns string (0-terminated) +*/ +YAZ_EXPORT +const char *icu_iter_get_norm(yaz_icu_iter_t iter); + +/** \brief returns ICU sortkey string + \param iter ICU tokenizer iterator + \returns string (0-terminated) +*/ +YAZ_EXPORT +const char *icu_iter_get_sortkey(yaz_icu_iter_t iter); + +/** \brief returns ICU display string + \param iter ICU tokenizer iterator + \returns string (0-terminated) +*/ +YAZ_EXPORT +const char *icu_iter_get_display(yaz_icu_iter_t iter); + +/** \brief returns ICU token count for iterator + \param iter ICU tokenizer iterator + \returns token count (1, 2, 3...) +*/ +YAZ_EXPORT +const char *icu_iter_get_token_number(yaz_icu_iter_t iter); + YAZ_END_CDECL #endif /* YAZ_ICU_H */ diff --git a/include/yaz/icu_I18N.h b/include/yaz/icu_I18N.h index 932dab6..eb0f22d 100644 --- a/include/yaz/icu_I18N.h +++ b/include/yaz/icu_I18N.h @@ -159,15 +159,6 @@ yaz_icu_chain_t icu_chain_create(const char * locale, int sort, UErrorCode * status); - -struct icu_iter; -struct icu_iter *icu_iter_create(struct icu_chain *chain); -void icu_iter_first(struct icu_iter *iter, const char *src8cstr); -void icu_iter_destroy(struct icu_iter *iter); -int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result); -const char *icu_iter_get_sortkey(struct icu_iter *iter); -const char *icu_iter_get_display(struct icu_iter *iter); - #endif /* ICU_I18NL_H */ /* diff --git a/src/icu_chain.c b/src/icu_chain.c index 4abbf68..e1addb7 100644 --- a/src/icu_chain.c +++ b/src/icu_chain.c @@ -50,15 +50,12 @@ struct icu_chain_step struct icu_chain { - struct icu_iter *iter; + yaz_icu_iter_t iter; char *locale; int sort; UCollator * coll; - /* utf8 output buffers */ - struct icu_buf_utf8 * norm8; - /* linked list of chain steps */ struct icu_chain_step * csteps; }; @@ -83,7 +80,7 @@ static struct icu_chain_step *icu_chain_step_create( if (!chain || !type || !rule) return 0; - step = (struct icu_chain_step *) xmalloc(sizeof(struct icu_chain_step)); + step = (struct icu_chain_step *) xmalloc(sizeof(*step)); step->type = type; /* create auxilary objects */ @@ -179,7 +176,7 @@ struct icu_chain *icu_chain_create(const char *locale, int sort, UErrorCode * status) { struct icu_chain * chain - = (struct icu_chain *) xmalloc(sizeof(struct icu_chain)); + = (struct icu_chain *) xmalloc(sizeof(*chain)); *status = U_ZERO_ERROR; @@ -193,7 +190,6 @@ struct icu_chain *icu_chain_create(const char *locale, int sort, if (U_FAILURE(*status)) return 0; - chain->norm8 = icu_buf_utf8_create(0); chain->csteps = 0; return chain; @@ -206,7 +202,6 @@ void icu_chain_destroy(struct icu_chain * chain) if (chain->coll) ucol_close(chain->coll); - icu_buf_utf8_destroy(chain->norm8); if (chain->iter) icu_iter_destroy(chain->iter); icu_chain_step_destroy(chain->csteps); @@ -324,6 +319,7 @@ struct icu_iter { UErrorCode status; struct icu_buf_utf8 *display; struct icu_buf_utf8 *sort8; + struct icu_buf_utf8 *result; struct icu_buf_utf16 *input; int token_count; struct icu_chain_step *steps; @@ -349,7 +345,7 @@ void icu_utf16_print(struct icu_buf_utf16 *src16) icu_buf_utf8_destroy(dst8); } -struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, +struct icu_buf_utf16 *icu_iter_invoke(yaz_icu_iter_t iter, struct icu_chain_step *step, struct icu_buf_utf16 *src) { @@ -409,13 +405,14 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, } } -struct icu_iter *icu_iter_create(struct icu_chain *chain) +yaz_icu_iter_t icu_iter_create(struct icu_chain *chain) { - struct icu_iter *iter = xmalloc(sizeof(*iter)); + yaz_icu_iter_t iter = xmalloc(sizeof(*iter)); iter->chain = chain; iter->status = U_ZERO_ERROR; iter->display = icu_buf_utf8_create(0); iter->sort8 = icu_buf_utf8_create(0); + iter->result = icu_buf_utf8_create(0); iter->last = 0; /* no last returned string (yet) */ iter->steps = icu_chain_step_clone(chain->csteps); iter->input = 0; @@ -423,7 +420,7 @@ struct icu_iter *icu_iter_create(struct icu_chain *chain) return iter; } -void icu_iter_first(struct icu_iter *iter, const char *src8cstr) +void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr) { if (iter->input) icu_buf_utf16_destroy(iter->input); @@ -434,12 +431,13 @@ void icu_iter_first(struct icu_iter *iter, const char *src8cstr) icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status); } -void icu_iter_destroy(struct icu_iter *iter) +void icu_iter_destroy(yaz_icu_iter_t iter) { if (iter) { icu_buf_utf8_destroy(iter->display); icu_buf_utf8_destroy(iter->sort8); + icu_buf_utf8_destroy(iter->result); if (iter->input) icu_buf_utf16_destroy(iter->input); icu_chain_step_destroy(iter->steps); @@ -447,7 +445,7 @@ void icu_iter_destroy(struct icu_iter *iter) } } -int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result) +int icu_iter_next(yaz_icu_iter_t iter) { if (!iter->input && iter->last == 0) return 0; @@ -470,23 +468,33 @@ int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result) iter->sort8, iter->last, &iter->status); } - icu_utf16_to_utf8(result, iter->last, &iter->status); + icu_utf16_to_utf8(iter->result, iter->last, &iter->status); icu_buf_utf16_destroy(iter->last); return 1; } } -const char *icu_iter_get_sortkey(struct icu_iter *iter) +const char *icu_iter_get_norm(yaz_icu_iter_t iter) +{ + return icu_buf_utf8_to_cstr(iter->result); +} + +const char *icu_iter_get_sortkey(yaz_icu_iter_t iter) { return icu_buf_utf8_to_cstr(iter->sort8); } -const char *icu_iter_get_display(struct icu_iter *iter) +const char *icu_iter_get_display(yaz_icu_iter_t iter) { return icu_buf_utf8_to_cstr(iter->display); } +const char *icu_iter_get_token_number(yaz_icu_iter_t iter) +{ + return iter->token_count; +} + int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr, UErrorCode *status) { @@ -500,7 +508,7 @@ int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr, int icu_chain_next_token(struct icu_chain * chain, UErrorCode *status) { *status = U_ZERO_ERROR; - return icu_iter_next(chain->iter, chain->norm8); + return icu_iter_next(chain->iter); } int icu_chain_token_number(struct icu_chain * chain) @@ -519,8 +527,8 @@ const char * icu_chain_token_display(struct icu_chain * chain) const char * icu_chain_token_norm(struct icu_chain * chain) { - if (chain->norm8) - return icu_buf_utf8_to_cstr(chain->norm8); + if (chain->iter) + return icu_iter_get_norm(chain->iter); return 0; } diff --git a/test/tst_icu_I18N.c b/test/tst_icu_I18N.c index 4ffc9de..2ad005c 100644 --- a/test/tst_icu_I18N.c +++ b/test/tst_icu_I18N.c @@ -625,8 +625,7 @@ static void check_icu_iter1(void) UErrorCode status = U_ZERO_ERROR; struct icu_chain * chain = 0; xmlNode *xml_node; - struct icu_iter *iter; - struct icu_buf_utf8 *token; + yaz_icu_iter_t iter; const char * xml_str = "" "" @@ -652,13 +651,10 @@ static void check_icu_iter1(void) YAZ_CHECK(iter); if (!iter) return; - token = icu_buf_utf8_create(0); - while (icu_iter_next(iter, token)) + while (icu_iter_next(iter)) { - yaz_log(YLOG_LOG, "[%.*s]", (int) token->utf8_len, token->utf8); + yaz_log(YLOG_LOG, "[%s]", icu_iter_get_norm(iter)); } - icu_buf_utf8_destroy(token); - icu_iter_destroy(iter); icu_chain_destroy(chain); } @@ -666,10 +662,9 @@ static void check_icu_iter1(void) static int test_iter(struct icu_chain *chain, const char *input, const char *expected) { - struct icu_iter *iter = icu_iter_create(chain); + yaz_icu_iter_t iter = icu_iter_create(chain); WRBUF result, second; int success = 1; - struct icu_buf_utf8 *token; if (!iter) { @@ -677,9 +672,7 @@ static int test_iter(struct icu_chain *chain, const char *input, return 0; } - token = icu_buf_utf8_create(0); - - if (icu_iter_next(iter, token)) + if (icu_iter_next(iter)) { yaz_log(YLOG_WARN, "test_iter: expecting 0 before icu_iter_first"); return 0; @@ -687,24 +680,22 @@ static int test_iter(struct icu_chain *chain, const char *input, result = wrbuf_alloc(); icu_iter_first(iter, input); - while (icu_iter_next(iter, token)) + while (icu_iter_next(iter)) { wrbuf_puts(result, "["); - wrbuf_write(result, (const char *) token->utf8, (int) token->utf8_len); + wrbuf_puts(result, icu_iter_get_norm(iter)); wrbuf_puts(result, "]"); } - second = wrbuf_alloc(); icu_iter_first(iter, input); - while (icu_iter_next(iter, token)) + while (icu_iter_next(iter)) { wrbuf_puts(second, "["); - wrbuf_write(second, (const char *) token->utf8, (int) token->utf8_len); + wrbuf_puts(second, icu_iter_get_norm(iter)); wrbuf_puts(second, "]"); } - icu_buf_utf8_destroy(token); icu_iter_destroy(iter); if (strcmp(expected, wrbuf_cstr(result)))