1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2010 Index Data
3 * See the file LICENSE for details.
16 #include <yaz/xmalloc.h>
18 #include <yaz/icu_I18N.h>
22 #include <yaz/nmem_xml.h>
28 #include <unicode/ustring.h> /* some more string fcns*/
29 #include <unicode/uchar.h> /* char names */
31 enum icu_chain_step_type {
32 ICU_chain_step_type_none,
33 ICU_chain_step_type_display, /* convert to utf8 display format */
34 ICU_chain_step_type_casemap, /* apply utf16 charmap */
35 ICU_chain_step_type_transform, /* apply utf16 transform */
36 ICU_chain_step_type_tokenize, /* apply utf16 tokenization */
37 ICU_chain_step_type_transliterate /* apply utf16 tokenization */
42 /* type and action object */
43 enum icu_chain_step_type type;
45 struct icu_casemap *casemap;
46 struct icu_transform *transform;
47 struct icu_tokenizer *tokenizer;
49 struct icu_chain_step *previous;
60 /* linked list of chain steps */
61 struct icu_chain_step *csteps;
64 int icu_check_status(UErrorCode status)
66 if (U_FAILURE(status))
68 yaz_log(YLOG_WARN, "ICU: %d %s\n", status, u_errorName(status));
74 static struct icu_chain_step *icu_chain_step_create(
75 struct icu_chain *chain, enum icu_chain_step_type type,
76 const uint8_t *rule, UErrorCode *status)
78 struct icu_chain_step *step = 0;
80 if (!chain || !type || !rule)
83 step = (struct icu_chain_step *) xmalloc(sizeof(*step));
86 /* create auxilary objects */
89 case ICU_chain_step_type_display:
91 case ICU_chain_step_type_casemap:
92 step->u.casemap = icu_casemap_create(rule[0], status);
94 case ICU_chain_step_type_transform:
95 /* rule omitted. Only ID used */
96 step->u.transform = icu_transform_create((const char *) rule, 'f',
99 case ICU_chain_step_type_tokenize:
100 step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
101 (char) rule[0], status);
103 case ICU_chain_step_type_transliterate:
104 /* we pass a dummy ID to utrans_openU.. */
105 step->u.transform = icu_transform_create("custom", 'f',
106 (const char *) rule, status);
115 static void icu_chain_step_destroy(struct icu_chain_step *step)
120 icu_chain_step_destroy(step->previous);
124 case ICU_chain_step_type_display:
126 case ICU_chain_step_type_casemap:
127 icu_casemap_destroy(step->u.casemap);
129 case ICU_chain_step_type_transform:
130 case ICU_chain_step_type_transliterate:
131 icu_transform_destroy(step->u.transform);
133 case ICU_chain_step_type_tokenize:
134 icu_tokenizer_destroy(step->u.tokenizer);
142 struct icu_chain_step *icu_chain_step_clone(struct icu_chain_step *old)
144 struct icu_chain_step *step = 0;
145 struct icu_chain_step **sp = &step;
148 *sp = (struct icu_chain_step *) xmalloc(sizeof(**sp));
149 (*sp)->type = old->type;
153 case ICU_chain_step_type_display:
155 case ICU_chain_step_type_casemap:
156 (*sp)->u.casemap = icu_casemap_clone(old->u.casemap);
158 case ICU_chain_step_type_transform:
159 case ICU_chain_step_type_transliterate:
160 (*sp)->u.transform = icu_transform_clone(old->u.transform);
162 case ICU_chain_step_type_tokenize:
163 (*sp)->u.tokenizer = icu_tokenizer_clone(old->u.tokenizer);
165 case ICU_chain_step_type_none:
169 sp = &(*sp)->previous;
175 struct icu_chain *icu_chain_create(const char *locale, int sort,
178 struct icu_chain *chain
179 = (struct icu_chain *) xmalloc(sizeof(*chain));
181 *status = U_ZERO_ERROR;
184 chain->locale = xstrdup(locale);
188 chain->coll = ucol_open((const char *) chain->locale, status);
190 if (U_FAILURE(*status))
198 void icu_chain_destroy(struct icu_chain *chain)
203 ucol_close(chain->coll);
206 icu_iter_destroy(chain->iter);
207 icu_chain_step_destroy(chain->csteps);
208 xfree(chain->locale);
213 static struct icu_chain_step *icu_chain_insert_step(
214 struct icu_chain *chain, enum icu_chain_step_type type,
215 const uint8_t *rule, UErrorCode *status);
217 struct icu_chain *icu_chain_xml_config(const xmlNode *xml_node,
223 struct icu_chain *chain = 0;
226 *status = U_ZERO_ERROR;
228 if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)
232 xmlChar *xml_locale = xmlGetProp((xmlNode *) xml_node,
233 (xmlChar *) "locale");
237 chain = icu_chain_create((const char *) xml_locale, sort, status);
245 nmem = nmem_create();
246 for (node = xml_node->children; node; node = node->next)
249 struct icu_chain_step *step = 0;
250 struct _xmlAttr *attr;
253 if (node->type != XML_ELEMENT_NODE)
256 for (attr = node->properties; attr; attr = attr->next)
258 if (!strcmp((const char *) attr->name, "rule"))
260 rule = nmem_text_node_cdata(attr->children, nmem);
264 yaz_log(YLOG_WARN, "Unsupported attribute '%s' for "
265 "element '%s'", attr->name, node->name);
270 if (!rule && node->children)
271 rule = nmem_text_node_cdata(node->children, nmem);
273 if (!strcmp((const char *) node->name, "casemap"))
274 step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
275 (const uint8_t *) rule, status);
276 else if (!strcmp((const char *) node->name, "transform"))
277 step = icu_chain_insert_step(chain, ICU_chain_step_type_transform,
278 (const uint8_t *) rule, status);
279 else if (!strcmp((const char *) node->name, "transliterate"))
280 step = icu_chain_insert_step(chain, ICU_chain_step_type_transliterate,
281 (const uint8_t *) rule, status);
282 else if (!strcmp((const char *) node->name, "tokenize"))
283 step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
284 (const uint8_t *) rule, status);
285 else if (!strcmp((const char *) node->name, "display"))
286 step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
287 (const uint8_t *) "", status);
288 else if (!strcmp((const char *) node->name, "normalize"))
290 yaz_log(YLOG_WARN, "Element %s is deprecated. "
291 "Use transform instead", node->name);
292 step = icu_chain_insert_step(chain, ICU_chain_step_type_transform,
293 (const uint8_t *) rule, status);
295 else if (!strcmp((const char *) node->name, "index")
296 || !strcmp((const char *) node->name, "sortkey"))
298 yaz_log(YLOG_WARN, "Element %s is no longer needed. "
299 "Remove it from the configuration", node->name);
303 yaz_log(YLOG_WARN, "Unknown element %s", node->name);
307 if (step && U_FAILURE(*status))
316 icu_chain_destroy(chain);
323 static struct icu_chain_step *icu_chain_insert_step(
324 struct icu_chain *chain, enum icu_chain_step_type type,
325 const uint8_t *rule, UErrorCode *status)
327 struct icu_chain_step *step = 0;
328 if (!chain || !type || !rule)
331 /* create actual chain step with this buffer */
332 step = icu_chain_step_create(chain, type, rule,
335 step->previous = chain->csteps;
336 chain->csteps = step;
342 struct icu_chain *chain;
343 struct icu_buf_utf16 *last;
345 struct icu_buf_utf8 *display;
346 struct icu_buf_utf8 *sort8;
347 struct icu_buf_utf8 *result;
348 struct icu_buf_utf16 *input;
350 struct icu_chain_step *steps;
353 void icu_utf16_print(struct icu_buf_utf16 *src16)
355 UErrorCode status = U_ZERO_ERROR;
357 struct icu_buf_utf8 *dst8 = icu_buf_utf8_create(0);
358 icu_utf16_to_utf8(dst8, src16, &status);
360 assert(status != 1234);
361 if (U_FAILURE(status))
367 p = icu_buf_utf8_to_cstr(dst8);
370 icu_buf_utf8_destroy(dst8);
373 struct icu_buf_utf16 *icu_iter_invoke(yaz_icu_iter_t iter,
374 struct icu_chain_step *step,
375 struct icu_buf_utf16 *src)
381 struct icu_buf_utf16 *dst = icu_iter_invoke(iter, step->previous, src);
385 case ICU_chain_step_type_casemap:
388 struct icu_buf_utf16 *src = dst;
390 dst = icu_buf_utf16_create(0);
391 icu_casemap_casemap(step->u.casemap, dst, src, &iter->status,
392 iter->chain->locale);
393 icu_buf_utf16_destroy(src);
396 case ICU_chain_step_type_tokenize:
399 struct icu_buf_utf16 *src = dst;
401 icu_tokenizer_attach(step->u.tokenizer, src, &iter->status);
402 icu_buf_utf16_destroy(src);
404 dst = icu_buf_utf16_create(0);
405 iter->status = U_ZERO_ERROR;
406 if (!icu_tokenizer_next_token(step->u.tokenizer, dst, &iter->status))
408 icu_buf_utf16_destroy(dst);
412 case ICU_chain_step_type_transform:
413 case ICU_chain_step_type_transliterate:
416 struct icu_buf_utf16 *src = dst;
417 dst = icu_buf_utf16_create(0);
418 icu_transform_trans(step->u.transform, dst, src, &iter->status);
419 icu_buf_utf16_destroy(src);
422 case ICU_chain_step_type_display:
424 icu_utf16_to_utf8(iter->display, dst, &iter->status);
433 yaz_icu_iter_t icu_iter_create(struct icu_chain *chain)
435 yaz_icu_iter_t iter = xmalloc(sizeof(*iter));
437 iter->status = U_ZERO_ERROR;
438 iter->display = icu_buf_utf8_create(0);
439 iter->sort8 = icu_buf_utf8_create(0);
440 iter->result = icu_buf_utf8_create(0);
441 iter->last = 0; /* no last returned string (yet) */
442 iter->steps = icu_chain_step_clone(chain->csteps);
448 void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr)
451 icu_buf_utf16_destroy(iter->input);
452 iter->input = icu_buf_utf16_create(0);
453 iter->token_count = 0;
454 /* fill and assign input string.. It will be 0 after
456 icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
459 void icu_iter_destroy(yaz_icu_iter_t iter)
463 icu_buf_utf8_destroy(iter->display);
464 icu_buf_utf8_destroy(iter->sort8);
465 icu_buf_utf8_destroy(iter->result);
467 icu_buf_utf16_destroy(iter->input);
468 icu_chain_step_destroy(iter->steps);
473 int icu_iter_next(yaz_icu_iter_t iter)
475 if (!iter->input && iter->last == 0)
479 /* on first call, iter->input is the input string. Thereafter: 0. */
480 iter->last = icu_iter_invoke(iter, iter->steps ?
481 iter->steps : iter->chain->csteps,
490 if (iter->chain->sort)
492 icu_sortkey8_from_utf16(iter->chain->coll,
493 iter->sort8, iter->last,
496 icu_utf16_to_utf8(iter->result, iter->last, &iter->status);
497 icu_buf_utf16_destroy(iter->last);
503 const char *icu_iter_get_norm(yaz_icu_iter_t iter)
505 return icu_buf_utf8_to_cstr(iter->result);
508 const char *icu_iter_get_sortkey(yaz_icu_iter_t iter)
510 return icu_buf_utf8_to_cstr(iter->sort8);
513 const char *icu_iter_get_display(yaz_icu_iter_t iter)
515 return icu_buf_utf8_to_cstr(iter->display);
518 int icu_iter_get_token_number(yaz_icu_iter_t iter)
520 return iter->token_count;
523 int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr,
527 icu_iter_destroy(chain->iter);
528 chain->iter = icu_iter_create(chain);
529 icu_iter_first(chain->iter, src8cstr);
533 int icu_chain_next_token(struct icu_chain *chain, UErrorCode *status)
535 *status = U_ZERO_ERROR;
536 return icu_iter_next(chain->iter);
539 int icu_chain_token_number(struct icu_chain *chain)
541 if (chain && chain->iter)
542 return chain->iter->token_count;
546 const char *icu_chain_token_display(struct icu_chain *chain)
549 return icu_iter_get_display(chain->iter);
553 const char *icu_chain_token_norm(struct icu_chain *chain)
556 return icu_iter_get_norm(chain->iter);
560 const char *icu_chain_token_sortkey(struct icu_chain *chain)
563 return icu_iter_get_sortkey(chain->iter);
567 #endif /* YAZ_HAVE_ICU */
572 * c-file-style: "Stroustrup"
573 * indent-tabs-mode: nil
575 * vim: shiftwidth=4 tabstop=8 expandtab