- //printf("original string: '%s' (%d)\n", src8, (int) src8_len);
-
- // some calling error needs investigation
- dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
- src8, "en", 't');
- YAZ_CHECK(0 == dest8_len);
- //printf("icu_casemap 'en:t' '%s' (%d)\n", dest8, (int) dest8_len);
-
-
- // attention: does not fail even if no locale 'xy_zz' defined
- // it seems to default to english locale
- dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
- src8, "zz_abc", 'l');
- YAZ_CHECK(dest8_len);
- //printf("icu_casemap 'zz:l' '%s' (%d)\n", dest8, (int) dest8_len);
-
-
- // shall fail - no buf buffer defined
- dest8 = icu_casemap(nmem, 0, buf_cap, &dest8_len,
- src8, "en", 'l');
- YAZ_CHECK(0 == dest8_len);
- //printf("icu_casemap 'en:l' '%s' (%d)\n", dest8, (int) dest8_len);
-
- // shall fail - no buf_cap defined
- dest8 = icu_casemap(nmem, buf, 0, &dest8_len,
- src8, "en", 'l');
- YAZ_CHECK(0 == dest8_len);
- //printf("icu_casemap 'en:l' '%s' (%d)\n", dest8, (int) dest8_len);
-
- // shall fail - no action 'x' defined
- dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
- src8, "en", 'x');
- YAZ_CHECK(0 == dest8_len);
- //printf("icu_casemap 'en:x' '%s' (%d)\n", dest8, (int) dest8_len);
-
- nmem_destroy(nmem);
+ YAZ_CHECK(test_icu_tokenizer("en", 's', en_str, 2));
+ YAZ_CHECK(test_icu_tokenizer("en", 'l', en_str, 7));
+ YAZ_CHECK(test_icu_tokenizer("en", 'w', en_str, 16));
+ YAZ_CHECK(test_icu_tokenizer("en", 'c', en_str, 41));
+
+
+
+ const char * da_str
+ = "Blåbærtærte. Denne kage stammer fra Finland. "
+ "Den er med blåbær, men alle sommerens forskellige bær kan bruges.";
+
+ YAZ_CHECK(test_icu_tokenizer("da", 's', da_str, 3));
+ YAZ_CHECK(test_icu_tokenizer("dar", 'l', da_str, 17));
+ YAZ_CHECK(test_icu_tokenizer("da", 'w', da_str, 37));
+ YAZ_CHECK(test_icu_tokenizer("da", 'c', da_str, 110));
+
+}
+
+
+void test_icu_I18N_chain(int argc, char **argv)
+{
+ const char * en_str
+ = "O Romeo, Romeo! wherefore art thou\t Romeo?";
+
+ printf("ICU chain:\ninput: '%s'\n", en_str);
+
+ UErrorCode status = U_ZERO_ERROR;
+ //struct icu_chain_step * step = 0;
+ struct icu_chain * chain = 0;
+
+
+ const char * xml_str = "<icu_chain id=\"en:word\" locale=\"en\">"
+ "<normalize rule=\"[:Control:] Any-Remove\"/>"
+ "<tokenize rule=\"l\"/>"
+ "<normalize rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
+ "<display/>"
+ "<casemap rule=\"l\"/>"
+ "<index/>"
+ "<sortkey/>"
+ "</icu_chain>";
+
+
+ xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str));
+ xmlNode *xml_node = xmlDocGetRootElement(doc);
+ YAZ_CHECK(xml_node);
+
+
+ chain = icu_chain_xml_config(xml_node, &status);
+
+#if 0
+ chain = icu_chain_create((uint8_t *) "en:word", (uint8_t *) "en");
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
+ (const uint8_t *) "[:Control:] Any-Remove",
+ &status);
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
+ (const uint8_t *) "s",
+ &status);
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
+ (const uint8_t *) "l",
+ &status);
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
+ (const uint8_t *)
+ "[[:WhiteSpace:][:Punctuation:]] Any-Remove",
+ &status);
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
+ (const uint8_t *)"",
+ &status);
+/* step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize, */
+/* (const uint8_t *) "Lower", */
+/* &status); */
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
+ (const uint8_t *) "l",
+ &status);
+ step = icu_chain_insert_step(chain, ICU_chain_step_type_index,
+ (const uint8_t *)"",
+ &status);
+/* step = icu_chain_insert_step(chain, ICU_chain_step_type_sortkey, */
+/* (const uint8_t *)"", */
+/* &status); */
+
+#endif
+
+ xmlFreeDoc(doc);
+ YAZ_CHECK(chain);
+
+ YAZ_CHECK(icu_chain_assign_cstr(chain, en_str, &status));
+
+ while (icu_chain_next_token(chain, &status)){
+ printf("%d '%s' '%s'\n",
+ icu_chain_get_token_count(chain),
+ icu_chain_get_norm(chain),
+ icu_chain_get_display(chain));
+ }
+
+ YAZ_CHECK_EQ(icu_chain_get_token_count(chain), 7);
+
+
+ YAZ_CHECK(icu_chain_assign_cstr(chain, "what is this?", &status));
+
+ while (icu_chain_next_token(chain, &status)){
+ printf("%d '%s' '%s'\n",
+ icu_chain_get_token_count(chain),
+ icu_chain_get_norm(chain),
+ icu_chain_get_display(chain));
+ }
+
+
+ YAZ_CHECK_EQ(icu_chain_get_token_count(chain), 3);
+
+ icu_chain_destroy(chain);