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;
yaz_log(YLOG_WARN, "Unsupported attribute '%s' for "
"element '%s'", attr->name, node->name);
no_errors++;
- continue;
}
}
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))
{
no_errors++;
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");
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);
}
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);
+ assert(iter->steps || !iter->chain->csteps);
+ iter->last = icu_iter_invoke(iter, iter->steps, iter->input);
iter->input = 0;
if (!iter->last)