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_insert_step(
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);
111 step->previous = chain->csteps;
112 chain->csteps = step;
118 static void icu_chain_step_destroy(struct icu_chain_step *step)
123 icu_chain_step_destroy(step->previous);
127 case ICU_chain_step_type_display:
129 case ICU_chain_step_type_casemap:
130 icu_casemap_destroy(step->u.casemap);
132 case ICU_chain_step_type_transform:
133 case ICU_chain_step_type_transliterate:
134 icu_transform_destroy(step->u.transform);
136 case ICU_chain_step_type_tokenize:
137 icu_tokenizer_destroy(step->u.tokenizer);
145 struct icu_chain_step *icu_chain_step_clone(struct icu_chain_step *old)
147 struct icu_chain_step *step = 0;
148 struct icu_chain_step **sp = &step;
151 *sp = (struct icu_chain_step *) xmalloc(sizeof(**sp));
152 (*sp)->type = old->type;
156 case ICU_chain_step_type_display:
158 case ICU_chain_step_type_casemap:
159 (*sp)->u.casemap = icu_casemap_clone(old->u.casemap);
161 case ICU_chain_step_type_transform:
162 case ICU_chain_step_type_transliterate:
163 (*sp)->u.transform = icu_transform_clone(old->u.transform);
165 case ICU_chain_step_type_tokenize:
166 (*sp)->u.tokenizer = icu_tokenizer_clone(old->u.tokenizer);
168 case ICU_chain_step_type_none:
172 sp = &(*sp)->previous;
178 struct icu_chain *icu_chain_create(const char *locale, int sort,
181 struct icu_chain *chain
182 = (struct icu_chain *) xmalloc(sizeof(*chain));
184 *status = U_ZERO_ERROR;
187 chain->locale = xstrdup(locale);
191 chain->coll = ucol_open((const char *) chain->locale, status);
193 if (U_FAILURE(*status))
201 void icu_chain_destroy(struct icu_chain *chain)
206 ucol_close(chain->coll);
209 icu_iter_destroy(chain->iter);
210 icu_chain_step_destroy(chain->csteps);
211 xfree(chain->locale);
216 static struct icu_chain_step *icu_chain_insert_step(
217 struct icu_chain *chain, enum icu_chain_step_type type,
218 const uint8_t *rule, UErrorCode *status);
220 struct icu_chain *icu_chain_xml_config(const xmlNode *xml_node,
226 struct icu_chain *chain = 0;
229 *status = U_ZERO_ERROR;
231 if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)
235 xmlChar *xml_locale = xmlGetProp((xmlNode *) xml_node,
236 (xmlChar *) "locale");
240 chain = icu_chain_create((const char *) xml_locale, sort, status);
248 nmem = nmem_create();
249 for (node = xml_node->children; node; node = node->next)
252 struct icu_chain_step *step = 0;
253 struct _xmlAttr *attr;
256 if (node->type != XML_ELEMENT_NODE)
259 for (attr = node->properties; attr; attr = attr->next)
261 if (!strcmp((const char *) attr->name, "rule"))
263 rule = nmem_text_node_cdata(attr->children, nmem);
267 yaz_log(YLOG_WARN, "Unsupported attribute '%s' for "
268 "element '%s'", attr->name, node->name);
273 if (!rule && node->children)
274 rule = nmem_text_node_cdata(node->children, nmem);
276 if (!strcmp((const char *) node->name, "casemap"))
277 step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
278 (const uint8_t *) rule, status);
279 else if (!strcmp((const char *) node->name, "transform"))
280 step = icu_chain_insert_step(chain, ICU_chain_step_type_transform,
281 (const uint8_t *) rule, status);
282 else if (!strcmp((const char *) node->name, "transliterate"))
283 step = icu_chain_insert_step(chain, ICU_chain_step_type_transliterate,
284 (const uint8_t *) rule, status);
285 else if (!strcmp((const char *) node->name, "tokenize"))
286 step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
287 (const uint8_t *) rule, status);
288 else if (!strcmp((const char *) node->name, "display"))
289 step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
290 (const uint8_t *) "", status);
291 else if (!strcmp((const char *) node->name, "normalize"))
293 yaz_log(YLOG_WARN, "Element %s is deprecated. "
294 "Use transform instead", node->name);
295 step = icu_chain_insert_step(chain, ICU_chain_step_type_transform,
296 (const uint8_t *) rule, status);
298 else if (!strcmp((const char *) node->name, "index")
299 || !strcmp((const char *) node->name, "sortkey"))
301 yaz_log(YLOG_WARN, "Element %s is no longer needed. "
302 "Remove it from the configuration", node->name);
306 yaz_log(YLOG_WARN, "Unknown element %s", node->name);
310 if (step && U_FAILURE(*status))
319 icu_chain_destroy(chain);
326 struct icu_chain *chain;
327 struct icu_buf_utf16 *last;
329 struct icu_buf_utf8 *display;
330 struct icu_buf_utf8 *sort8;
331 struct icu_buf_utf8 *result;
332 struct icu_buf_utf16 *input;
334 struct icu_chain_step *steps;
337 void icu_utf16_print(struct icu_buf_utf16 *src16)
339 UErrorCode status = U_ZERO_ERROR;
341 struct icu_buf_utf8 *dst8 = icu_buf_utf8_create(0);
342 icu_utf16_to_utf8(dst8, src16, &status);
344 assert(status != 1234);
345 if (U_FAILURE(status))
351 p = icu_buf_utf8_to_cstr(dst8);
354 icu_buf_utf8_destroy(dst8);
357 struct icu_buf_utf16 *icu_iter_invoke(yaz_icu_iter_t iter,
358 struct icu_chain_step *step,
359 struct icu_buf_utf16 *src)
365 struct icu_buf_utf16 *dst = icu_iter_invoke(iter, step->previous, src);
369 case ICU_chain_step_type_casemap:
372 struct icu_buf_utf16 *src = dst;
374 dst = icu_buf_utf16_create(0);
375 icu_casemap_casemap(step->u.casemap, dst, src, &iter->status,
376 iter->chain->locale);
377 icu_buf_utf16_destroy(src);
380 case ICU_chain_step_type_tokenize:
383 struct icu_buf_utf16 *src = dst;
385 icu_tokenizer_attach(step->u.tokenizer, src, &iter->status);
386 icu_buf_utf16_destroy(src);
388 dst = icu_buf_utf16_create(0);
389 iter->status = U_ZERO_ERROR;
390 if (!icu_tokenizer_next_token(step->u.tokenizer, dst, &iter->status))
392 icu_buf_utf16_destroy(dst);
396 case ICU_chain_step_type_transform:
397 case ICU_chain_step_type_transliterate:
400 struct icu_buf_utf16 *src = dst;
401 dst = icu_buf_utf16_create(0);
402 icu_transform_trans(step->u.transform, dst, src, &iter->status);
403 icu_buf_utf16_destroy(src);
406 case ICU_chain_step_type_display:
408 icu_utf16_to_utf8(iter->display, dst, &iter->status);
417 yaz_icu_iter_t icu_iter_create(struct icu_chain *chain)
419 yaz_icu_iter_t iter = xmalloc(sizeof(*iter));
421 iter->status = U_ZERO_ERROR;
422 iter->display = icu_buf_utf8_create(0);
423 iter->sort8 = icu_buf_utf8_create(0);
424 iter->result = icu_buf_utf8_create(0);
425 iter->last = 0; /* no last returned string (yet) */
426 iter->steps = icu_chain_step_clone(chain->csteps);
432 void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr)
435 icu_buf_utf16_destroy(iter->input);
436 iter->input = icu_buf_utf16_create(0);
437 iter->token_count = 0;
438 /* fill and assign input string.. It will be 0 after
440 icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
443 void icu_iter_destroy(yaz_icu_iter_t iter)
447 icu_buf_utf8_destroy(iter->display);
448 icu_buf_utf8_destroy(iter->sort8);
449 icu_buf_utf8_destroy(iter->result);
451 icu_buf_utf16_destroy(iter->input);
452 icu_chain_step_destroy(iter->steps);
457 int icu_iter_next(yaz_icu_iter_t iter)
459 if (!iter->input && iter->last == 0)
463 /* on first call, iter->input is the input string. Thereafter: 0. */
464 iter->last = icu_iter_invoke(iter, iter->steps ?
465 iter->steps : iter->chain->csteps,
474 if (iter->chain->sort)
476 icu_sortkey8_from_utf16(iter->chain->coll,
477 iter->sort8, iter->last,
480 icu_utf16_to_utf8(iter->result, iter->last, &iter->status);
481 icu_buf_utf16_destroy(iter->last);
487 const char *icu_iter_get_norm(yaz_icu_iter_t iter)
489 return icu_buf_utf8_to_cstr(iter->result);
492 const char *icu_iter_get_sortkey(yaz_icu_iter_t iter)
494 return icu_buf_utf8_to_cstr(iter->sort8);
497 const char *icu_iter_get_display(yaz_icu_iter_t iter)
499 return icu_buf_utf8_to_cstr(iter->display);
502 int icu_iter_get_token_number(yaz_icu_iter_t iter)
504 return iter->token_count;
507 int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr,
511 icu_iter_destroy(chain->iter);
512 chain->iter = icu_iter_create(chain);
513 icu_iter_first(chain->iter, src8cstr);
517 int icu_chain_next_token(struct icu_chain *chain, UErrorCode *status)
519 *status = U_ZERO_ERROR;
520 return icu_iter_next(chain->iter);
523 int icu_chain_token_number(struct icu_chain *chain)
525 if (chain && chain->iter)
526 return chain->iter->token_count;
530 const char *icu_chain_token_display(struct icu_chain *chain)
533 return icu_iter_get_display(chain->iter);
537 const char *icu_chain_token_norm(struct icu_chain *chain)
540 return icu_iter_get_norm(chain->iter);
544 const char *icu_chain_token_sortkey(struct icu_chain *chain)
547 return icu_iter_get_sortkey(chain->iter);
551 #endif /* YAZ_HAVE_ICU */
556 * c-file-style: "Stroustrup"
557 * indent-tabs-mode: nil
559 * vim: shiftwidth=4 tabstop=8 expandtab