/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2012 Index Data
+ * Copyright (C) Index Data
* See the file LICENSE for details.
*/
#include <yaz/log.h>
#include <yaz/nmem.h>
#include <yaz/nmem_xml.h>
+#include <yaz/xml_get.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
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) */
+ YAZ_chain_step_type_stemming, /* apply utf16 stemming (YAZ) */
+ ICU_chain_step_type_join
};
struct icu_chain_step
struct icu_transform *transform;
struct icu_tokenizer *tokenizer;
yaz_stemmer_p stemmer;
+ struct icu_buf_utf16 *join;
} u;
struct icu_chain_step *previous;
};
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)
+ const char *rule, UErrorCode *status)
{
struct icu_chain_step *step = 0;
- if (!chain || !type || !rule)
- return 0;
+ assert(chain);
+ assert(type);
step = (struct icu_chain_step *) xmalloc(sizeof(*step));
-
step->type = type;
- /* 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(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:
- step->u.stemmer = yaz_stemmer_create(chain->locale,
- (const char *) rule, status);
+ 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;
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;
}
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;
if (xml_node && xml_node->type == XML_ELEMENT_NODE)
{
- xmlChar *xml_locale = xmlGetProp((xmlNode *) xml_node,
- (xmlChar *) "locale");
+ const char *xml_locale = yaz_xml_get_prop((xmlNode *) xml_node,
+ "locale");
if (xml_locale)
- {
chain = icu_chain_create((const char *) xml_locale, sort, status);
- xmlFree(xml_locale);
- }
}
if (!chain)
{
char *rule = 0;
struct icu_chain_step *step = 0;
- struct _xmlAttr *attr;
+ const char *attr_str;
nmem_reset(nmem);
if (node->type != XML_ELEMENT_NODE)
continue;
-
- for (attr = node->properties; attr; attr = attr->next)
+ attr_str = yaz_xml_get_prop(node, "rule%s", &rule);
+ if (attr_str)
{
- 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++;
- continue;
- }
+ yaz_log(YLOG_WARN, "Unsupported attribute '%s' for "
+ "element '%s'", attr_str, 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 *) 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 *) 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 *) 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 *) rule, status);
+ 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);
+ rule, 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);
+ 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 *) rule, status);
+ rule, status);
}
else if (!strcmp((const char *) node->name, "index")
|| !strcmp((const char *) node->name, "sortkey"))
no_errors++;
continue;
}
+ if (!step)
+ {
+ yaz_log(YLOG_WARN, "Step not created for %s", node->name);
+ no_errors++;
+ }
if (step && U_FAILURE(*status))
{
+ yaz_log(YLOG_WARN, "ICU Error %d %s for element %s, rule %s",
+ *status, u_errorName(*status), node->name, rule ?
+ rule : "");
no_errors++;
break;
}
struct icu_iter {
struct icu_chain *chain;
struct icu_buf_utf16 *last;
+ struct icu_buf_utf16 *org;
+ struct icu_buf_utf8 *org8;
UErrorCode status;
struct icu_buf_utf8 *display;
struct icu_buf_utf8 *sort8;
struct icu_buf_utf8 *result;
- struct icu_buf_utf16 *input;
int token_count;
+ size_t org_start;
+ size_t org_len;
+ size_t utf8_base;
+ size_t utf16_base;
struct icu_chain_step *steps;
};
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");
struct icu_buf_utf16 *src = dst;
icu_tokenizer_attach(step->u.tokenizer, src, &iter->status);
+ if (step->previous)
+ { /* no need to copy if it's already the same */
+ iter->utf8_base = iter->utf16_base = 0;
+ icu_buf_utf16_copy(iter->org, src);
+ }
icu_buf_utf16_destroy(src);
}
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;
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);
}
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->org8 = 0;
iter->last = 0; /* no last returned string (yet) */
iter->steps = icu_chain_step_clone(chain->csteps);
- iter->input = 0;
+ iter->token_count = 0;
return iter;
}
void icu_iter_first(yaz_icu_iter_t iter, const char *src8cstr)
{
- if (iter->input)
- icu_buf_utf16_destroy(iter->input);
- iter->input = icu_buf_utf16_create(0);
+ 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;
- /* fill and assign input string.. It will be 0 after
- first iteration */
- icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
+ iter->org_start = 0;
+ iter->utf8_base = iter->utf16_base = 0;
+ iter->org_len = src->utf16_len;
+ iter->last = icu_iter_invoke(iter, iter->steps, src);
}
void icu_iter_destroy(yaz_icu_iter_t iter)
icu_buf_utf8_destroy(iter->display);
icu_buf_utf8_destroy(iter->sort8);
icu_buf_utf8_destroy(iter->result);
- if (iter->input)
- icu_buf_utf16_destroy(iter->input);
+ icu_buf_utf16_destroy(iter->org);
+ icu_buf_utf8_destroy(iter->org8);
icu_chain_step_destroy(iter->steps);
xfree(iter);
}
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->steps ?
- iter->steps : iter->chain->csteps,
- iter->input);
- iter->input = 0;
-
- if (!iter->last)
- return 0;
-
iter->token_count++;
-
if (iter->chain->sort)
{
icu_sortkey8_from_utf16(iter->chain->coll,
return iter->token_count;
}
+
+void icu_iter_get_org_info2(yaz_icu_iter_t iter, size_t *start, size_t *len,
+ const char **cstr)
+{
+ int32_t len1 = 0, len2 = 0;
+ UErrorCode status = U_ZERO_ERROR;
+
+ if (iter->org_start < iter->utf16_base)
+ {
+ iter->utf8_base = 0;
+ iter->utf16_base = 0;
+ }
+ u_strToUTF8(0, 0, &len1,
+ iter->org->utf16 + iter->utf16_base,
+ iter->org_start - iter->utf16_base,
+ &status);
+
+ status = U_ZERO_ERROR;
+
+ *start = len1 + iter->utf8_base;
+
+ u_strToUTF8(0, 0, &len2,
+ iter->org->utf16 + iter->utf16_base,
+ iter->org_start - iter->utf16_base + iter->org_len,
+ &status);
+
+ *len = len2 - len1;
+
+ if (cstr)
+ {
+ if (!iter->org8)
+ iter->org8 = icu_buf_utf8_create(0);
+ status = U_ZERO_ERROR;
+ icu_utf16_to_utf8(iter->org8, iter->org, &status);
+ *cstr = icu_buf_utf8_to_cstr(iter->org8);
+ }
+ iter->utf8_base = *start;
+ iter->utf16_base = iter->org_start;
+}
+
+void icu_iter_get_org_info(yaz_icu_iter_t iter, size_t *start, size_t *len)
+{
+ icu_iter_get_org_info2(iter, start, len, 0);
+}
+
int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr,
UErrorCode *status)
{
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);
+}
+
+void icu_chain_get_org_info2(struct icu_chain *chain, size_t *start,
+ size_t *len, const char **cstr)
+{
+ if (chain->iter)
+ icu_iter_get_org_info2(chain->iter, start, len, cstr);
+}
+
+
#endif /* YAZ_HAVE_ICU */
/*