X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Ficu_chain.c;h=bcb27b435fa61f1c0200950350e85b4f6fe406e1;hp=df313eb70c7a3a5e746e713736248fbf8cadb468;hb=be73a431cd4e151c77d3024b7ee09197e4a4910a;hpb=e662834ae7643141ab51c887afdbc40038494b1b diff --git a/src/icu_chain.c b/src/icu_chain.c index df313eb..bcb27b4 100644 --- a/src/icu_chain.c +++ b/src/icu_chain.c @@ -1,5 +1,5 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2009 Index Data + * Copyright (C) 1995-2013 Index Data * See the file LICENSE for details. */ @@ -17,8 +17,11 @@ #include -#include +#include +#include +#include +#include #include #include #include @@ -29,67 +32,39 @@ enum icu_chain_step_type { ICU_chain_step_type_none, - ICU_chain_step_type_display, /* convert to utf8 display format */ - ICU_chain_step_type_casemap, /* apply utf16 charmap */ - ICU_chain_step_type_transform, /* apply utf16 transform */ - ICU_chain_step_type_tokenize, /* apply utf16 tokenization */ - ICU_chain_step_type_transliterate /* apply utf16 tokenization */ + ICU_chain_step_type_display, /* convert to utf8 display format */ + ICU_chain_step_type_casemap, /* apply utf16 charmap */ + ICU_chain_step_type_transform, /* apply utf16 transform */ + ICU_chain_step_type_tokenize, /* apply utf16 tokenization */ + ICU_chain_step_type_transliterate, /* apply utf16 tokenization */ + YAZ_chain_step_type_stemming, /* apply utf16 stemming (YAZ) */ + ICU_chain_step_type_join }; -#define USE_ITER 1 - struct icu_chain_step { /* type and action object */ enum icu_chain_step_type type; union { - struct icu_casemap * casemap; - struct icu_transform * transform; - struct icu_tokenizer * tokenizer; + struct icu_casemap *casemap; + struct icu_transform *transform; + struct icu_tokenizer *tokenizer; + yaz_stemmer_p stemmer; + struct icu_buf_utf16 *join; } u; - struct icu_chain_step * previous; -#if USE_ITER -#else - /* temprary post-action utf16 buffer */ - struct icu_buf_utf16 * buf16; - int more_tokens; - int need_new_token; -#endif + struct icu_chain_step *previous; }; - struct icu_chain { -#if USE_ITER - struct icu_iter *iter; -#endif - + yaz_icu_iter_t iter; char *locale; int sort; - UCollator * coll; - -#if USE_ITER -#else - const char * src8cstr; + UCollator *coll; - /* number of tokens returned so far */ - int32_t token_count; -#endif - - /* utf8 output buffers */ - struct icu_buf_utf8 * norm8; -#if USE_ITER -#else - struct icu_buf_utf8 * display8; - struct icu_buf_utf8 * sort8; - - /* utf16 source buffer */ - struct icu_buf_utf16 * src16; -#endif - /* linked list of chain steps */ - struct icu_chain_step * steps; + struct icu_chain_step *csteps; }; int icu_check_status(UErrorCode status) @@ -97,62 +72,65 @@ int icu_check_status(UErrorCode status) if (U_FAILURE(status)) { yaz_log(YLOG_WARN, "ICU: %d %s\n", status, u_errorName(status)); - return 0; + return 0; } return 1; } -static struct icu_chain_step *icu_chain_step_create( - struct icu_chain * chain, enum icu_chain_step_type type, - const uint8_t * rule, -#if USE_ITER -#else - struct icu_buf_utf16 * buf16, -#endif - UErrorCode *status) +static struct icu_chain_step *icu_chain_insert_step( + struct icu_chain *chain, enum icu_chain_step_type type, + const char *rule, UErrorCode *status) { - struct icu_chain_step * step = 0; - - if(!chain || !type || !rule) - return 0; + struct icu_chain_step *step = 0; - step = (struct icu_chain_step *) xmalloc(sizeof(struct icu_chain_step)); + assert(chain); + assert(type); + step = (struct icu_chain_step *) xmalloc(sizeof(*step)); step->type = type; -#if USE_ITER -#else - step->buf16 = buf16; -#endif - /* create auxilary objects */ + switch (step->type) { case ICU_chain_step_type_display: break; case ICU_chain_step_type_casemap: + assert(rule); step->u.casemap = icu_casemap_create(rule[0], status); break; case ICU_chain_step_type_transform: + assert(rule); /* rule omitted. Only ID used */ - step->u.transform = icu_transform_create((const char *) rule, 'f', - 0, status); + step->u.transform = icu_transform_create(rule, 'f', 0, status); break; case ICU_chain_step_type_tokenize: - step->u.tokenizer = icu_tokenizer_create((char *) chain->locale, - (char) rule[0], status); + assert(rule); + step->u.tokenizer = icu_tokenizer_create(chain->locale, rule[0], status); break; case ICU_chain_step_type_transliterate: + assert(rule); /* we pass a dummy ID to utrans_openU.. */ - step->u.transform = icu_transform_create("custom", 'f', - (const char *) rule, status); + step->u.transform = icu_transform_create("custom", 'f', rule, status); + break; + case YAZ_chain_step_type_stemming: + assert(rule); + step->u.stemmer = yaz_stemmer_create(chain->locale, rule, status); + break; + case ICU_chain_step_type_join: + assert(rule); + step->u.join = icu_buf_utf16_create(0); + icu_utf16_from_utf8_cstr(step->u.join, rule, status); break; default: break; } + step->previous = chain->csteps; + chain->csteps = step; + return step; } -static void icu_chain_step_destroy(struct icu_chain_step * step) +static void icu_chain_step_destroy(struct icu_chain_step *step) { if (!step) return; @@ -165,25 +143,19 @@ static void icu_chain_step_destroy(struct icu_chain_step * step) break; case ICU_chain_step_type_casemap: icu_casemap_destroy(step->u.casemap); -#if USE_ITER -#else - icu_buf_utf16_destroy(step->buf16); -#endif break; case ICU_chain_step_type_transform: case ICU_chain_step_type_transliterate: icu_transform_destroy(step->u.transform); -#if USE_ITER -#else - icu_buf_utf16_destroy(step->buf16); -#endif break; case ICU_chain_step_type_tokenize: icu_tokenizer_destroy(step->u.tokenizer); -#if USE_ITER -#else - icu_buf_utf16_destroy(step->buf16); -#endif + break; + case YAZ_chain_step_type_stemming: + yaz_stemmer_destroy(step->u.stemmer); + break; + case ICU_chain_step_type_join: + icu_buf_utf16_destroy(step->u.join); break; default: break; @@ -191,128 +163,169 @@ static void icu_chain_step_destroy(struct icu_chain_step * step) xfree(step); } -struct icu_chain *icu_chain_create(const char *locale, int sort, - UErrorCode * status) +struct icu_chain_step *icu_chain_step_clone(struct icu_chain_step *old) { - struct icu_chain * chain - = (struct icu_chain *) xmalloc(sizeof(struct icu_chain)); - - *status = U_ZERO_ERROR; - -#if USE_ITER - chain->iter = 0; -#endif - chain->locale = xstrdup(locale); + struct icu_chain_step *step = 0; + struct icu_chain_step **sp = &step; + while (old) + { + *sp = (struct icu_chain_step *) xmalloc(sizeof(**sp)); + (*sp)->type = old->type; - chain->sort = sort; + switch ((*sp)->type) + { + case ICU_chain_step_type_display: + break; + case ICU_chain_step_type_casemap: + (*sp)->u.casemap = icu_casemap_clone(old->u.casemap); + break; + case ICU_chain_step_type_transform: + case ICU_chain_step_type_transliterate: + (*sp)->u.transform = icu_transform_clone(old->u.transform); + break; + case ICU_chain_step_type_tokenize: + (*sp)->u.tokenizer = icu_tokenizer_clone(old->u.tokenizer); + break; + case YAZ_chain_step_type_stemming: + (*sp)->u.stemmer = yaz_stemmer_clone(old->u.stemmer); + break; + case ICU_chain_step_type_none: + break; + case ICU_chain_step_type_join: + (*sp)->u.join = icu_buf_utf16_create(0); + (*sp)->u.join = icu_buf_utf16_copy((*sp)->u.join, old->u.join); + break; + } + old = old->previous; + sp = &(*sp)->previous; + } + *sp = 0; + return step; +} - chain->coll = ucol_open((const char *) chain->locale, status); +struct icu_chain *icu_chain_create(const char *locale, int sort, + UErrorCode *status) +{ + struct icu_chain *chain; + UCollator *coll = ucol_open(locale, status); if (U_FAILURE(*status)) return 0; -#if USE_ITER -#else - chain->token_count = 0; - chain->src8cstr = 0; -#endif - chain->norm8 = icu_buf_utf8_create(0); -#if USE_ITER -#else - chain->display8 = icu_buf_utf8_create(0); - chain->sort8 = icu_buf_utf8_create(0); - chain->src16 = icu_buf_utf16_create(0); -#endif - - chain->steps = 0; + chain = (struct icu_chain *) xmalloc(sizeof(*chain)); + chain->iter = 0; + chain->locale = xstrdup(locale); + chain->sort = sort; + chain->coll = coll; + chain->csteps = 0; return chain; } -void icu_chain_destroy(struct icu_chain * chain) +void icu_chain_destroy(struct icu_chain *chain) { if (chain) { if (chain->coll) ucol_close(chain->coll); - icu_buf_utf8_destroy(chain->norm8); -#if USE_ITER if (chain->iter) icu_iter_destroy(chain->iter); -#else - icu_buf_utf8_destroy(chain->display8); - icu_buf_utf8_destroy(chain->sort8); - icu_buf_utf16_destroy(chain->src16); -#endif - - icu_chain_step_destroy(chain->steps); + icu_chain_step_destroy(chain->csteps); xfree(chain->locale); xfree(chain); } } -static struct icu_chain_step *icu_chain_insert_step( - struct icu_chain * chain, enum icu_chain_step_type type, - const uint8_t * rule, UErrorCode *status); - -struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node, - int sort, - UErrorCode * status) +struct icu_chain *icu_chain_xml_config(const xmlNode *xml_node, + int sort, + UErrorCode *status) { xmlNode *node = 0; - struct icu_chain * chain = 0; - + int no_errors = 0; + struct icu_chain *chain = 0; + NMEM nmem = 0; + *status = U_ZERO_ERROR; - if (!xml_node ||xml_node->type != XML_ELEMENT_NODE) - return 0; - + if (xml_node && xml_node->type == XML_ELEMENT_NODE) { - xmlChar * xml_locale = xmlGetProp((xmlNode *) xml_node, - (xmlChar *) "locale"); - + xmlChar *xml_locale = xmlGetProp((xmlNode *) xml_node, + (xmlChar *) "locale"); if (xml_locale) { chain = icu_chain_create((const char *) xml_locale, sort, status); xmlFree(xml_locale); } - } + if (!chain) return 0; + nmem = nmem_create(); for (node = xml_node->children; node; node = node->next) { - xmlChar *xml_rule; - struct icu_chain_step * step = 0; + char *rule = 0; + struct icu_chain_step *step = 0; + struct _xmlAttr *attr; + nmem_reset(nmem); if (node->type != XML_ELEMENT_NODE) continue; - xml_rule = xmlGetProp(node, (xmlChar *) "rule"); + for (attr = node->properties; attr; attr = attr->next) + { + if (!strcmp((const char *) attr->name, "rule")) + { + rule = nmem_text_node_cdata(attr->children, nmem); + } + else + { + yaz_log(YLOG_WARN, "Unsupported attribute '%s' for " + "element '%s'", attr->name, node->name); + no_errors++; + } + } + if (!rule && node->children) + rule = nmem_text_node_cdata(node->children, nmem); + if (!rule && strcmp((const char *) node->name, "display")) + { + yaz_log(YLOG_WARN, "Missing attribute rule for element %s", + (const char *) node->name); + no_errors++; + continue; + } if (!strcmp((const char *) node->name, "casemap")) - step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap, - (const uint8_t *) xml_rule, status); + step = icu_chain_insert_step(chain, + ICU_chain_step_type_casemap, + rule, status); else if (!strcmp((const char *) node->name, "transform")) - step = icu_chain_insert_step(chain, ICU_chain_step_type_transform, - (const uint8_t *) xml_rule, status); + step = icu_chain_insert_step(chain, + ICU_chain_step_type_transform, + rule, status); else if (!strcmp((const char *) node->name, "transliterate")) - step = icu_chain_insert_step(chain, ICU_chain_step_type_transliterate, - (const uint8_t *) xml_rule, status); + step = icu_chain_insert_step(chain, + ICU_chain_step_type_transliterate, + rule, status); else if (!strcmp((const char *) node->name, "tokenize")) - step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize, - (const uint8_t *) xml_rule, status); + step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize, + rule, status); else if (!strcmp((const char *) node->name, "display")) - step = icu_chain_insert_step(chain, ICU_chain_step_type_display, - (const uint8_t *) "", status); + step = icu_chain_insert_step(chain, ICU_chain_step_type_display, + rule, status); + else if (!strcmp((const char *) node->name, "stemming")) + step = icu_chain_insert_step(chain, YAZ_chain_step_type_stemming, + rule, status); + else if (!strcmp((const char *) node->name, "join")) + step = icu_chain_insert_step(chain, ICU_chain_step_type_join, + rule, status); else if (!strcmp((const char *) node->name, "normalize")) { yaz_log(YLOG_WARN, "Element %s is deprecated. " "Use transform instead", node->name); - step = icu_chain_insert_step(chain, ICU_chain_step_type_transform, - (const uint8_t *) xml_rule, status); + step = icu_chain_insert_step(chain, ICU_chain_step_type_transform, + rule, status); } else if (!strcmp((const char *) node->name, "index") || !strcmp((const char *) node->name, "sortkey")) @@ -323,185 +336,41 @@ struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node, else { yaz_log(YLOG_WARN, "Unknown element %s", node->name); - icu_chain_destroy(chain); - return 0; + no_errors++; + continue; + } + if (!step) + { + yaz_log(YLOG_WARN, "Step not created for %s", node->name); + no_errors++; } - xmlFree(xml_rule); if (step && U_FAILURE(*status)) { - icu_chain_destroy(chain); - return 0; + no_errors++; + break; } } - return chain; -} - -static struct icu_chain_step *icu_chain_insert_step( - struct icu_chain * chain, enum icu_chain_step_type type, - const uint8_t * rule, UErrorCode *status) -{ - struct icu_chain_step * step = 0; -#if USE_ITER -#else - struct icu_buf_utf16 * src16 = 0; - struct icu_buf_utf16 * buf16 = 0; -#endif - if (!chain || !type || !rule) - return 0; - -#if USE_ITER -#else - /* assign utf16 src buffers as needed */ - if (chain->steps && chain->steps->buf16) - src16 = chain->steps->buf16; - else if (chain->src16) - src16 = chain->src16; - else - return 0; - - /* create utf16 destination buffers as needed, or */ - switch (type) - { - case ICU_chain_step_type_display: - buf16 = src16; - break; - case ICU_chain_step_type_casemap: - buf16 = icu_buf_utf16_create(0); - break; - case ICU_chain_step_type_transform: - case ICU_chain_step_type_transliterate: - buf16 = icu_buf_utf16_create(0); - break; - case ICU_chain_step_type_tokenize: - buf16 = icu_buf_utf16_create(0); - break; - break; - default: - break; - } -#endif - /* create actual chain step with this buffer */ - step = icu_chain_step_create(chain, type, rule, -#if USE_ITER -#else - buf16, -#endif - status); - - step->previous = chain->steps; - chain->steps = step; - - return step; -} - -#if USE_ITER -#else -static int icu_chain_step_next_token(struct icu_chain * chain, - struct icu_chain_step * step, - UErrorCode *status) -{ - struct icu_buf_utf16 * src16 = 0; - int got_new_token = 0; - - if (!chain || !chain->src16 || !step || !step->more_tokens) - return 0; - - /* assign utf16 src buffers as needed, advance in previous steps - tokens until non-zero token met, and setting stop condition */ - - if (step->previous) - { - src16 = step->previous->buf16; - /* tokens might be killed in previous steps, therefore looping */ - - while (step->need_new_token - && step->previous->more_tokens - && !got_new_token) - got_new_token - = icu_chain_step_next_token(chain, step->previous, status); - } - else - { /* first step can only work once on chain->src16 input buffer */ - src16 = chain->src16; - step->more_tokens = 0; - got_new_token = 1; - } - - if (!src16) - return 0; - - /* stop if nothing to process */ - if (step->need_new_token && !got_new_token) - { - step->more_tokens = 0; - return 0; - } - - /* either an old token not finished yet, or a new token, thus - perform the work, eventually put this steps output in - step->buf16 or the chains UTF8 output buffers */ - - switch (step->type) + nmem_destroy(nmem); + if (no_errors) { - case ICU_chain_step_type_display: - icu_utf16_to_utf8(chain->display8, src16, status); - break; - case ICU_chain_step_type_casemap: - icu_casemap_casemap(step->u.casemap, - step->buf16, src16, status, - chain->locale); - break; - case ICU_chain_step_type_transform: - case ICU_chain_step_type_transliterate: - icu_transform_trans(step->u.transform, - step->buf16, src16, status); - break; - case ICU_chain_step_type_tokenize: - /* attach to new src16 token only first time during splitting */ - if (step->need_new_token) - { - icu_tokenizer_attach(step->u.tokenizer, src16, status); - step->need_new_token = 0; - } - - /* splitting one src16 token into multiple buf16 tokens */ - step->more_tokens - = icu_tokenizer_next_token(step->u.tokenizer, - step->buf16, status); - - /* make sure to get new previous token if this one had been used up - by recursive call to _same_ step */ - - if (!step->more_tokens) - { - step->more_tokens = icu_chain_step_next_token(chain, step, status); - return step->more_tokens; /* avoid one token count too much! */ - } - break; - default: + icu_chain_destroy(chain); return 0; - break; } - - if (U_FAILURE(*status)) - return 0; - - /* if token disappered into thin air, tell caller */ - /* if (!step->buf16->utf16_len && !step->more_tokens) */ - /* return 0; */ - - return 1; + return chain; } -#endif struct icu_iter { struct icu_chain *chain; struct icu_buf_utf16 *last; + struct icu_buf_utf16 *org; UErrorCode status; struct icu_buf_utf8 *display; struct icu_buf_utf8 *sort8; - struct icu_buf_utf16 *input; + struct icu_buf_utf8 *result; int token_count; + size_t org_start; + size_t org_len; + struct icu_chain_step *steps; }; void icu_utf16_print(struct icu_buf_utf16 *src16) @@ -511,7 +380,6 @@ void icu_utf16_print(struct icu_buf_utf16 *src16) struct icu_buf_utf8 *dst8 = icu_buf_utf8_create(0); icu_utf16_to_utf8(dst8, src16, &status); - assert(status != 1234); if (U_FAILURE(status)) { printf("failure"); @@ -524,7 +392,7 @@ 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) { @@ -533,7 +401,7 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, else { struct icu_buf_utf16 *dst = icu_iter_invoke(iter, step->previous, src); - + switch (step->type) { case ICU_chain_step_type_casemap: @@ -557,7 +425,8 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, } dst = icu_buf_utf16_create(0); iter->status = U_ZERO_ERROR; - if (!icu_tokenizer_next_token(step->u.tokenizer, dst, &iter->status)) + if (!icu_tokenizer_next_token(step->u.tokenizer, dst, &iter->status, + &iter->org_start, &iter->org_len)) { icu_buf_utf16_destroy(dst); dst = 0; @@ -577,6 +446,31 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, if (dst) icu_utf16_to_utf8(iter->display, dst, &iter->status); break; + case YAZ_chain_step_type_stemming: + if (dst) + { + struct icu_buf_utf16 *src = dst; + dst = icu_buf_utf16_create(0); + yaz_stemmer_stem(step->u.stemmer, dst, src, &iter->status); + icu_buf_utf16_destroy(src); + } + break; + case ICU_chain_step_type_join: + if (dst) + { + while (1) + { + struct icu_buf_utf16 *dst1 = + icu_iter_invoke(iter, step->previous, 0); + + if (!dst1) + break; + dst = icu_buf_utf16_append(dst, step->u.join); + dst = icu_buf_utf16_append(dst, dst1); + icu_buf_utf16_destroy(dst1); + } + } + break; default: assert(0); } @@ -584,226 +478,163 @@ struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter, } } -struct icu_iter *icu_iter_create(struct icu_chain *chain, - const char *src8cstr) +yaz_icu_iter_t icu_iter_create(struct icu_chain *chain) { - if (!src8cstr) - return 0; - else - { - struct icu_iter *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->token_count = 0; - iter->last = 0; /* no last returned string (yet) */ - - /* fill and assign input string.. It will be 0 after - first iteration */ - iter->input = icu_buf_utf16_create(0); - icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status); - return 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->org = icu_buf_utf16_create(0); + iter->last = 0; /* no last returned string (yet) */ + iter->steps = icu_chain_step_clone(chain->csteps); + iter->token_count = 0; + + return iter; +} - } +void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr) +{ + struct icu_buf_utf16 *src = icu_buf_utf16_create(0); + icu_utf16_from_utf8_cstr(src, src8cstr, &iter->status); + icu_buf_utf16_copy(iter->org, src); + iter->token_count = 0; + iter->org_start = 0; + iter->org_len = src->utf16_len; + iter->last = icu_iter_invoke(iter, iter->steps, src); } -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); - if (iter->input) - icu_buf_utf16_destroy(iter->input); + icu_buf_utf8_destroy(iter->result); + icu_buf_utf16_destroy(iter->org); + icu_chain_step_destroy(iter->steps); xfree(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) + if (iter->token_count && iter->last) + iter->last = icu_iter_invoke(iter, iter->steps, 0); + if (!iter->last) return 0; else { - /* on first call, iter->input is the input string. Thereafter: 0. */ - iter->last = icu_iter_invoke(iter, iter->chain->steps, iter->input); - iter->input = 0; - - if (!iter->last) - return 0; - iter->token_count++; - if (iter->chain->sort) - { + { icu_sortkey8_from_utf16(iter->chain->coll, 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) -{ - return icu_buf_utf8_to_cstr(iter->display); +const char *icu_iter_get_display(yaz_icu_iter_t iter) +{ + return icu_buf_utf8_to_cstr(iter->display); } -int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr, - UErrorCode *status) +int icu_iter_get_token_number(yaz_icu_iter_t iter) { -#if USE_ITER - if (chain->iter) - icu_iter_destroy(chain->iter); - chain->iter = icu_iter_create(chain, src8cstr); - return 1; -#else - struct icu_chain_step * stp = 0; + return iter->token_count; +} - if (!chain || !src8cstr) - return 0; - chain->src8cstr = src8cstr; +void icu_iter_get_org_info(yaz_icu_iter_t iter, size_t *start, size_t *len) +{ + /* save full length of org since we're gonna cut it */ + int32_t save_len = iter->org->utf16_len; - stp = chain->steps; - - /* clear token count */ - chain->token_count = 0; + struct icu_buf_utf8 *tmp = icu_buf_utf8_create(0); + UErrorCode status = U_ZERO_ERROR; - /* clear all steps stop states */ - while (stp) - { - stp->more_tokens = 1; - stp->need_new_token = 1; - stp = stp->previous; - } - - /* finally convert UTF8 to UTF16 string if needed */ - if (chain->steps || chain->sort) - icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status); - - if (U_FAILURE(*status)) - return 0; + iter->org->utf16_len = iter->org_start; + icu_utf16_to_utf8(tmp, iter->org, &status); + if (U_SUCCESS(status)) + *start = tmp->utf8_len; + else + *start = 0; + iter->org->utf16_len = iter->org_start + iter->org_len; + icu_utf16_to_utf8(tmp, iter->org, &status); + if (U_SUCCESS(status)) + *len = tmp->utf8_len - *start; + else + *len = 0; + iter->org->utf16_len = save_len; +} +int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr, + UErrorCode *status) +{ + if (chain->iter) + icu_iter_destroy(chain->iter); + chain->iter = icu_iter_create(chain); + icu_iter_first(chain->iter, src8cstr); return 1; -#endif } -int icu_chain_next_token(struct icu_chain * chain, UErrorCode *status) +int icu_chain_next_token(struct icu_chain *chain, UErrorCode *status) { -#if USE_ITER - *status = U_ZERO_ERROR; - return icu_iter_next(chain->iter, chain->norm8); -#else - int got_token = 0; - *status = U_ZERO_ERROR; - - if (!chain) - return 0; - - /* special case with no steps - same as index type binary */ - if (!chain->steps) - { - if (chain->token_count) - return 0; - else - { - chain->token_count++; - - if (chain->sort) - icu_sortkey8_from_utf16(chain->coll, - chain->sort8, chain->steps->buf16, - status); - return chain->token_count; - } - } - /* usual case, one or more icu chain steps existing */ - else - { - while (!got_token && chain->steps && chain->steps->more_tokens) - got_token = icu_chain_step_next_token(chain, chain->steps, status); - - if (got_token) - { - chain->token_count++; - - icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status); - - if (chain->sort) - icu_sortkey8_from_utf16(chain->coll, - chain->sort8, chain->steps->buf16, - status); - return chain->token_count; - } - } - - return 0; -#endif + return icu_iter_next(chain->iter); } -int icu_chain_token_number(struct icu_chain * chain) +int icu_chain_token_number(struct icu_chain *chain) { -#if USE_ITER if (chain && chain->iter) return chain->iter->token_count; return 0; -#else - if (!chain) - return 0; - - return chain->token_count; -#endif } -const char * icu_chain_token_display(struct icu_chain * chain) +const char *icu_chain_token_display(struct icu_chain *chain) { -#if USE_ITER if (chain->iter) return icu_iter_get_display(chain->iter); -#else - if (chain->display8) - return icu_buf_utf8_to_cstr(chain->display8); -#endif return 0; } -const char * icu_chain_token_norm(struct icu_chain * chain) +const char *icu_chain_token_norm(struct icu_chain *chain) { -#if USE_ITER - if (chain->norm8) - return icu_buf_utf8_to_cstr(chain->norm8); -#else - if (!chain->steps) - return chain->src8cstr; - - if (chain->norm8) - return icu_buf_utf8_to_cstr(chain->norm8); -#endif + if (chain->iter) + return icu_iter_get_norm(chain->iter); return 0; } -const char * icu_chain_token_sortkey(struct icu_chain * chain) +const char *icu_chain_token_sortkey(struct icu_chain *chain) { -#if USE_ITER if (chain->iter) return icu_iter_get_sortkey(chain->iter); -#else - if (chain->sort8) - return icu_buf_utf8_to_cstr(chain->sort8); -#endif return 0; } +void icu_chain_get_org_info(struct icu_chain *chain, size_t *start, size_t *len) +{ + if (chain->iter) + icu_iter_get_org_info(chain->iter, start, len); +} + + #endif /* YAZ_HAVE_ICU */ /*