#include <yaz/icu_I18N.h>
+#include <yaz/stemmer.h>
+
#include <yaz/log.h>
#include <yaz/nmem.h>
#include <yaz/nmem_xml.h>
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) */
};
struct icu_chain_step
/* type and action object */
enum icu_chain_step_type type;
union {
- struct icu_casemap *casemap;
+ struct icu_casemap *casemap;
struct icu_transform *transform;
struct icu_tokenizer *tokenizer;
+ yaz_stemmer_p stemmer;
} u;
struct icu_chain_step *previous;
};
return 1;
}
-static struct icu_chain_step *icu_chain_step_create(
- struct icu_chain *chain, enum icu_chain_step_type type,
+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;
step->u.transform = icu_transform_create("custom", 'f',
(const char *) rule, status);
break;
+ case YAZ_chain_step_type_stemming:
+ step->u.stemmer = yaz_stemmer_create((char *) chain->locale, (const char *) rule, status);
+ break;
default:
break;
}
+ step->previous = chain->csteps;
+ chain->csteps = step;
+
return step;
}
case ICU_chain_step_type_tokenize:
icu_tokenizer_destroy(step->u.tokenizer);
break;
+ case YAZ_chain_step_type_stemming:
+ yaz_stemmer_destroy(step->u.stemmer);
+ break;
default:
break;
}
case ICU_chain_step_type_tokenize:
(*sp)->u.tokenizer = icu_tokenizer_clone(old->u.tokenizer);
break;
+ case YAZ_chain_step_type_stemming:
+ yaz_stemmer_clone(step->u.stemmer);
+ break;
case ICU_chain_step_type_none:
break;
}
if (!rule && node->children)
rule = nmem_text_node_cdata(node->children, nmem);
- if (!rule || !*rule)
- {
- yaz_log(YLOG_WARN, "Missing rule for element '%s'", 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 *) 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);
+ else if (!strcmp((const char *) node->name, "stemming"))
+ step = icu_chain_insert_step(chain, YAZ_chain_step_type_stemming,
+ (const uint8_t *) rule, status);
else if (!strcmp((const char *) node->name, "normalize"))
{
yaz_log(YLOG_WARN, "Element %s is deprecated. "
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 (!chain || !type || !rule)
- return 0;
-
- /* create actual chain step with this buffer */
- step = icu_chain_step_create(chain, type, rule,
- status);
-
- step->previous = chain->csteps;
- chain->csteps = step;
-
- return step;
-}
-
struct icu_iter {
struct icu_chain *chain;
struct icu_buf_utf16 *last;
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;
default:
assert(0);
}