X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcharsets.c;h=b36d7e2573214462393857098517b29f284bc531;hb=c1de9c944c1383e173558b15bc693bd0eae2fba5;hp=39ede26ab299d228b2df65e2f90ae242d8f4c9bf;hpb=9267d02c3601122f74bdf3a9d21f30080a4a1a9a;p=pazpar2-moved-to-github.git diff --git a/src/charsets.c b/src/charsets.c index 39ede26..b36d7e2 100644 --- a/src/charsets.c +++ b/src/charsets.c @@ -1,5 +1,5 @@ /* This file is part of Pazpar2. - Copyright (C) 2006-2009 Index Data + Copyright (C) Index Data Pazpar2 is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free @@ -28,159 +28,367 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include +#include +#include #include #include +#include #include "charsets.h" #include "normalize7bit.h" +typedef struct pp2_charset_s *pp2_charset_t; +static pp2_charset_t pp2_charset_create_xml(xmlNode *xml_node); +static pp2_charset_t pp2_charset_create(void); +static pp2_charset_t pp2_charset_create_a_to_z(void); +static void pp2_charset_destroy(pp2_charset_t pct); +static pp2_charset_token_t pp2_charset_tokenize(pp2_charset_t pct); + #if YAZ_HAVE_ICU #include +static pp2_charset_t pp2_charset_create_icu(struct icu_chain *icu_chn); #endif - /* charset handle */ struct pp2_charset_s { - const char *(*token_next_handler)(pp2_relevance_token_t prt); - const char *(*get_sort_handler)(pp2_relevance_token_t prt, int skip); + const char *(*token_next_handler)(pp2_charset_token_t prt); + const char *(*get_sort_handler)(pp2_charset_token_t prt); + const char *(*get_display_handler)(pp2_charset_token_t prt); + void (*get_org_handler)(pp2_charset_token_t ptr, + size_t *start, size_t *len); #if YAZ_HAVE_ICU struct icu_chain * icu_chn; UErrorCode icu_sts; #endif }; -static const char *pp2_relevance_token_a_to_z(pp2_relevance_token_t prt); -static const char *pp2_get_sort_ascii(pp2_relevance_token_t prt, int skip_article); +static const char *pp2_charset_token_null(pp2_charset_token_t prt); +static const char *pp2_charset_token_a_to_z(pp2_charset_token_t prt); +static const char *pp2_get_sort_ascii(pp2_charset_token_t prt); +static const char *pp2_get_display_ascii(pp2_charset_token_t prt); +static void pp2_get_org_ascii(pp2_charset_token_t prt, + size_t *start, size_t *len); #if YAZ_HAVE_ICU -static const char *pp2_relevance_token_icu(pp2_relevance_token_t prt); -static const char *pp2_get_sort_icu(pp2_relevance_token_t prt, int skip_article); +static const char *pp2_charset_token_icu(pp2_charset_token_t prt); +static const char *pp2_get_sort_icu(pp2_charset_token_t prt); +static const char *pp2_get_display_icu(pp2_charset_token_t prt); +static void pp2_get_org_icu(pp2_charset_token_t prt, + size_t *start, size_t *len); #endif /* tokenzier handle */ -struct pp2_relevance_token_s { +struct pp2_charset_token_s { const char *cp; /* unnormalized buffer we're tokenizing */ const char *last_cp; /* pointer to last token we're dealing with */ pp2_charset_t pct; /* our main charset handle (type+config) */ WRBUF norm_str; /* normized string we return (temporarily) */ WRBUF sort_str; /* sort string we return (temporarily) */ +#if YAZ_HAVE_ICU + yaz_icu_iter_t iter; +#endif + const char *cp0; + size_t start; + size_t len; +}; + +struct pp2_charset_fact_s { + struct pp2_charset_entry *list; + int ref_count; +}; + +struct pp2_charset_entry { + struct pp2_charset_entry *next; + pp2_charset_t pct; + char *name; }; +static int pp2_charset_fact_add(pp2_charset_fact_t pft, + pp2_charset_t pct, const char *default_id); + +pp2_charset_fact_t pp2_charset_fact_create(void) +{ + pp2_charset_fact_t pft = xmalloc(sizeof(*pft)); + pft->list = 0; + pft->ref_count = 1; + + pp2_charset_fact_add(pft, pp2_charset_create_a_to_z(), "relevance"); + pp2_charset_fact_add(pft, pp2_charset_create_a_to_z(), "sort"); + pp2_charset_fact_add(pft, pp2_charset_create_a_to_z(), "mergekey"); + pp2_charset_fact_add(pft, pp2_charset_create(), "facet"); + return pft; +} + +void pp2_charset_fact_destroy(pp2_charset_fact_t pft) +{ + if (pft) + { + assert(pft->ref_count >= 1); + --(pft->ref_count); + if (pft->ref_count == 0) + { + struct pp2_charset_entry *pce = pft->list; + while (pce) + { + struct pp2_charset_entry *next = pce->next; + pp2_charset_destroy(pce->pct); + xfree(pce->name); + xfree(pce); + pce = next; + } + xfree(pft); + } + } +} + +int pp2_charset_fact_add(pp2_charset_fact_t pft, + pp2_charset_t pct, const char *default_id) +{ + struct pp2_charset_entry *pce; + + for (pce = pft->list; pce; pce = pce->next) + if (!strcmp(default_id, pce->name)) + break; + + if (!pce) + { + pce = xmalloc(sizeof(*pce)); + pce->name = xstrdup(default_id); + pce->next = pft->list; + pft->list = pce; + } + else + { + pp2_charset_destroy(pce->pct); + } + pce->pct = pct; + return 0; +} + +int pp2_charset_fact_define(pp2_charset_fact_t pft, + xmlNode *xml_node, const char *id) +{ + int r; + pp2_charset_t pct; + + assert(xml_node); + + if (strcmp((const char *) xml_node->name, "icu_chain")) + { + yaz_log(YLOG_WARN, "Wrapper element <%s> deprecated", xml_node->name); + yaz_log(YLOG_LOG, "Use .. only", xml_node->name); + xml_node = xml_node->children; + while (xml_node && xml_node->type != XML_ELEMENT_NODE) + xml_node = xml_node->next; + } + if (!xml_node) + { + yaz_log(YLOG_FATAL, "Missing icu_chain element"); + return -1; + } + pct = pp2_charset_create_xml(xml_node); + if (!pct) + return -1; + if (!id) + { + id = yaz_xml_get_prop(xml_node, "id"); + if (!id) + { + yaz_log(YLOG_WARN, "Missing id for icu_chain"); + pp2_charset_destroy(pct); + return -1; + } + } + r = pp2_charset_fact_add(pft, pct, id); + return r; +} + +void pp2_charset_fact_incref(pp2_charset_fact_t pft) +{ + (pft->ref_count)++; +} + pp2_charset_t pp2_charset_create_xml(xmlNode *xml_node) { #if YAZ_HAVE_ICU UErrorCode status = U_ZERO_ERROR; struct icu_chain *chain = 0; - if (xml_node) - xml_node = xml_node->children; - while (xml_node && xml_node->type != XML_ELEMENT_NODE) - xml_node = xml_node->next; chain = icu_chain_xml_config(xml_node, 1, &status); - if (!chain || U_FAILURE(status)){ - //xmlDocPtr icu_doc = 0; - //xmlChar *xmlstr = 0; - //int size = 0; - //xmlDocDumpMemory(icu_doc, size); - + if (!chain || U_FAILURE(status)) + { yaz_log(YLOG_FATAL, "Could not parse ICU chain config:\n" "<%s>\n ... \n", xml_node->name, xml_node->name); return 0; } - return pp2_charset_create(chain); + return pp2_charset_create_icu(chain); #else // YAZ_HAVE_ICU yaz_log(YLOG_FATAL, "Error: ICU support requested with element:\n" "<%s>\n ... \n", xml_node->name, xml_node->name); - yaz_log(YLOG_FATAL, + yaz_log(YLOG_FATAL, "But no ICU support is compiled into the YAZ library."); return 0; #endif // YAZ_HAVE_ICU } - -pp2_charset_t pp2_charset_create(struct icu_chain * icu_chn) +pp2_charset_t pp2_charset_create(void) { pp2_charset_t pct = xmalloc(sizeof(*pct)); - pct->token_next_handler = pp2_relevance_token_a_to_z; + pct->token_next_handler = pp2_charset_token_null; pct->get_sort_handler = pp2_get_sort_ascii; + pct->get_display_handler = pp2_get_display_ascii; + pct->get_org_handler = pp2_get_org_ascii; #if YAZ_HAVE_ICU pct->icu_chn = 0; +#endif // YAZ_HAVE_ICU + return pct; +} + +pp2_charset_t pp2_charset_create_a_to_z(void) +{ + pp2_charset_t pct = pp2_charset_create(); + pct->token_next_handler = pp2_charset_token_a_to_z; + return pct; +} + +#if YAZ_HAVE_ICU +pp2_charset_t pp2_charset_create_icu(struct icu_chain *icu_chn) +{ + pp2_charset_t pct = pp2_charset_create(); if (icu_chn) { pct->icu_chn = icu_chn; pct->icu_sts = U_ZERO_ERROR; - pct->token_next_handler = pp2_relevance_token_icu; + pct->token_next_handler = pp2_charset_token_icu; pct->get_sort_handler = pp2_get_sort_icu; + pct->get_display_handler = pp2_get_display_icu; + pct->get_org_handler = pp2_get_org_icu; } -#endif // YAZ_HAVE_ICU return pct; } +#endif // YAZ_HAVE_ICU void pp2_charset_destroy(pp2_charset_t pct) { +#if YAZ_HAVE_ICU + icu_chain_destroy(pct->icu_chn); +#endif xfree(pct); } -pp2_relevance_token_t pp2_relevance_tokenize(pp2_charset_t pct, - const char *buf) +pp2_charset_token_t pp2_charset_token_create(pp2_charset_fact_t pft, + const char *id) +{ + struct pp2_charset_entry *pce; + for (pce = pft->list; pce; pce = pce->next) + if (!strcmp(id, pce->name)) + return pp2_charset_tokenize(pce->pct); + return 0; +} + +pp2_charset_token_t pp2_charset_tokenize(pp2_charset_t pct) { - pp2_relevance_token_t prt = xmalloc(sizeof(*prt)); + pp2_charset_token_t prt = xmalloc(sizeof(*prt)); assert(pct); prt->norm_str = wrbuf_alloc(); prt->sort_str = wrbuf_alloc(); - prt->cp = buf; + prt->cp = 0; prt->last_cp = 0; prt->pct = pct; #if YAZ_HAVE_ICU + prt->iter = 0; if (pct->icu_chn) + prt->iter = icu_iter_create(pct->icu_chn); +#endif + prt->start = 0; + prt->len = 0; + return prt; +} + +void pp2_charset_token_first(pp2_charset_token_t prt, + const char *buf, int skip_article) +{ + if (skip_article) { - int ok = 0; - pct->icu_sts = U_ZERO_ERROR; - ok = icu_chain_assign_cstr(pct->icu_chn, buf, &pct->icu_sts); - //printf("\nfield ok: %d '%s'\n", ok, buf); - prt->pct = pct; + const char *p = buf; + char firstword[64]; + char *pout = firstword; + char articles[] = "the den der die des an a "; // must end in space + + for (; *p && *p != ' ' && pout - firstword < (sizeof(firstword)-2); p++) + *pout++ = tolower(*(unsigned char *)p); + *pout++ = ' '; + *pout++ = '\0'; + if (strstr(articles, firstword)) + buf = p; + } + + wrbuf_rewind(prt->norm_str); + wrbuf_rewind(prt->sort_str); + prt->cp0 = buf; + prt->cp = buf; + prt->last_cp = 0; + +#if YAZ_HAVE_ICU + if (prt->iter) + { + icu_iter_first(prt->iter, buf); } #endif // YAZ_HAVE_ICU - return prt; } - -void pp2_relevance_token_destroy(pp2_relevance_token_t prt) +void pp2_charset_token_destroy(pp2_charset_token_t prt) { assert(prt); - if(prt->norm_str) +#if YAZ_HAVE_ICU + if (prt->iter) + icu_iter_destroy(prt->iter); +#endif + if(prt->norm_str) wrbuf_destroy(prt->norm_str); - if(prt->sort_str) + if(prt->sort_str) wrbuf_destroy(prt->sort_str); xfree(prt); } -const char *pp2_relevance_token_next(pp2_relevance_token_t prt) +const char *pp2_charset_token_next(pp2_charset_token_t prt) { assert(prt); return (prt->pct->token_next_handler)(prt); } -const char *pp2_get_sort(pp2_relevance_token_t prt, int skip) +const char *pp2_get_sort(pp2_charset_token_t prt) +{ + return prt->pct->get_sort_handler(prt); +} + +const char *pp2_get_display(pp2_charset_token_t prt) { - return prt->pct->get_sort_handler(prt, skip); + return prt->pct->get_display_handler(prt); } +void pp2_get_org(pp2_charset_token_t prt, size_t *start, size_t *len) +{ + prt->pct->get_org_handler(prt, start, len); +} + + #define raw_char(c) (((c) >= 'a' && (c) <= 'z') ? (c) : -1) /* original tokenizer with our tokenize interface, but we add +1 to ensure no '\0' are in our string (except for EOF) */ -static const char *pp2_relevance_token_a_to_z(pp2_relevance_token_t prt) +static const char *pp2_charset_token_a_to_z(pp2_charset_token_t prt) { const char *cp = prt->cp; int c; + prt->start = cp - prt->cp0; /* skip white space */ while (*cp && (c = raw_char(tolower(*(const unsigned char *)cp))) < 0) cp++; @@ -199,12 +407,12 @@ static const char *pp2_relevance_token_a_to_z(pp2_relevance_token_t prt) wrbuf_putc(prt->norm_str, c); cp++; } + prt->len = (cp - prt->cp0) - prt->start; prt->cp = cp; return wrbuf_cstr(prt->norm_str); } -static const char *pp2_get_sort_ascii(pp2_relevance_token_t prt, - int skip_article) +static const char *pp2_get_sort_ascii(pp2_charset_token_t prt) { if (prt->last_cp == 0) return 0; @@ -212,8 +420,8 @@ static const char *pp2_get_sort_ascii(pp2_relevance_token_t prt, { char *tmp = xstrdup(prt->last_cp); char *result = 0; - result = normalize7bit_mergekey(tmp, skip_article); - + result = normalize7bit_mergekey(tmp); + wrbuf_rewind(prt->sort_str); wrbuf_puts(prt->sort_str, result); xfree(tmp); @@ -221,25 +429,58 @@ static const char *pp2_get_sort_ascii(pp2_relevance_token_t prt, } } +static const char *pp2_get_display_ascii(pp2_charset_token_t prt) +{ + if (prt->last_cp == 0) + return 0; + else + { + return wrbuf_cstr(prt->norm_str); + } +} + +static void pp2_get_org_ascii(pp2_charset_token_t prt, + size_t *start, size_t *len) +{ + *start = prt->start; + *len = prt->len; +} + +static const char *pp2_charset_token_null(pp2_charset_token_t prt) +{ + const char *cp = prt->cp; + + prt->last_cp = *cp ? cp : 0; + while (*cp) + cp++; + prt->cp = cp; + prt->len = cp - prt->cp0; + return prt->last_cp; +} #if YAZ_HAVE_ICU -static const char *pp2_relevance_token_icu(pp2_relevance_token_t prt) +static const char *pp2_charset_token_icu(pp2_charset_token_t prt) { - if (icu_chain_next_token(prt->pct->icu_chn, &prt->pct->icu_sts)) + if (icu_iter_next(prt->iter)) { - if (U_FAILURE(prt->pct->icu_sts)) - { - return 0; - } - return icu_chain_token_norm(prt->pct->icu_chn); + return icu_iter_get_norm(prt->iter); } return 0; } -static const char *pp2_get_sort_icu(pp2_relevance_token_t prt, - int skip_article) +static const char *pp2_get_sort_icu(pp2_charset_token_t prt) +{ + return icu_iter_get_sortkey(prt->iter); +} + +static const char *pp2_get_display_icu(pp2_charset_token_t prt) +{ + return icu_iter_get_display(prt->iter); +} + +static void pp2_get_org_icu(pp2_charset_token_t prt, size_t *start, size_t *len) { - return icu_chain_token_sortkey(prt->pct->icu_chn); + icu_iter_get_org_info(prt->iter, start, len); } #endif // YAZ_HAVE_ICU