X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=test%2Ftst_icu_I18N.c;h=848da691b785f84d16eba00cc548bd4860ba27a2;hp=23fb1d5a2a61e0e6212dce3e941a74d567b49a29;hb=88d3bedf772316f87e1996f655ccf8d1e2589755;hpb=362a3c56893f05cb1417cc23e24fc165303486dc diff --git a/test/tst_icu_I18N.c b/test/tst_icu_I18N.c index 23fb1d5..848da69 100644 --- a/test/tst_icu_I18N.c +++ b/test/tst_icu_I18N.c @@ -1,5 +1,5 @@ /* This file is part of the YAZ toolkit. - * Copyright (C) 1995-2009 Index Data + * Copyright (C) 1995-2010 Index Data * See the file LICENSE for details. */ @@ -23,9 +23,6 @@ #include #include -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - - #define MAX_KEY_SIZE 256 struct icu_termmap { @@ -34,7 +31,6 @@ struct icu_termmap }; - int icu_termmap_cmp(const void *vp1, const void *vp2) { struct icu_termmap *itmp1 = *(struct icu_termmap **) vp1; @@ -45,9 +41,7 @@ int icu_termmap_cmp(const void *vp1, const void *vp2) cmp = strcmp((const char *)itmp1->sort_key, (const char *)itmp2->sort_key); return cmp; -}; - - +} int test_icu_casemap(const char * locale, char action, @@ -73,9 +67,7 @@ int test_icu_casemap(const char * locale, char action, /* converting to UTF8 */ icu_utf16_to_utf8(dest8, dest16, &status); - - /* determine success */ if (dest8->utf8 && (dest8->utf8_len == strlen(chk8cstr)) @@ -98,18 +90,12 @@ int test_icu_casemap(const char * locale, char action, icu_buf_utf8_destroy(dest8); icu_buf_utf16_destroy(src16); icu_buf_utf16_destroy(dest16); - - + return success; } - - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - void test_icu_I18N_casemap(int argc, char **argv) { - /* Locale 'en' */ /* successful tests */ @@ -129,7 +115,6 @@ void test_icu_I18N_casemap(int argc, char **argv) "A ReD fOx hunTS sQUirriLs", "A Red Fox Hunts Squirrils")); - /* Locale 'da' */ /* success expected */ @@ -170,9 +155,6 @@ void test_icu_I18N_casemap(int argc, char **argv) } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - int test_icu_sortmap(const char * locale, int src_list_len, const char ** src_list, const char ** chk_list) { @@ -194,39 +176,39 @@ int test_icu_sortmap(const char * locale, int src_list_len, return 0; /* assigning display terms and sort keys using buf 8 and buf16 */ - for( i = 0; i < src_list_len; i++) - { + for (i = 0; i < src_list_len; i++) + { - list[i] = (struct icu_termmap *) malloc(sizeof(struct icu_termmap)); + list[i] = (struct icu_termmap *) malloc(sizeof(struct icu_termmap)); - /* copy display term */ - strcpy(list[i]->disp_term, src_list[i]); + /* copy display term */ + strcpy(list[i]->disp_term, src_list[i]); - /* transforming to UTF16 */ - icu_utf16_from_utf8_cstr(buf16, list[i]->disp_term, &status); - icu_check_status(status); + /* transforming to UTF16 */ + icu_utf16_from_utf8_cstr(buf16, list[i]->disp_term, &status); + icu_check_status(status); - /* computing sortkeys */ - icu_sortkey8_from_utf16(coll, buf8, buf16, &status); - icu_check_status(status); + /* computing sortkeys */ + icu_sortkey8_from_utf16(coll, buf8, buf16, &status); + icu_check_status(status); - /* assigning sortkeys */ - memcpy(list[i]->sort_key, buf8->utf8, buf8->utf8_len); - } - + /* assigning sortkeys */ + memcpy(list[i]->sort_key, buf8->utf8, buf8->utf8_len); + } /* do the sorting */ - qsort(list, src_list_len, - sizeof(struct icu_termmap *), icu_termmap_cmp); + qsort(list, src_list_len, sizeof(struct icu_termmap *), icu_termmap_cmp); /* checking correct sorting */ - for (i = 0; i < src_list_len; i++){ + for (i = 0; i < src_list_len; i++) + { if (0 != strcmp(list[i]->disp_term, chk_list[i])){ success = 0; } } - if(!success){ + if (!success) + { printf("\nERROR\n"); printf("Input str: '%s' : ", locale); for (i = 0; i < src_list_len; i++) { @@ -245,12 +227,9 @@ int test_icu_sortmap(const char * locale, int src_list_len, printf("\n"); } - - - for( i = 0; i < src_list_len; i++) + for (i = 0; i < src_list_len; i++) free(list[i]); - - + ucol_close(coll); icu_buf_utf8_destroy(buf8); @@ -259,12 +238,8 @@ int test_icu_sortmap(const char * locale, int src_list_len, return success; } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - void test_icu_I18N_sortmap(int argc, char **argv) { - /* successful tests */ size_t en_1_len = 6; const char * en_1_src[6] = {"z", "K", "a", "A", "Z", "k"}; @@ -295,15 +270,9 @@ void test_icu_I18N_sortmap(int argc, char **argv) } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - - - - int test_icu_normalizer(const char * rules8cstr, - const char * src8cstr, - const char * chk8cstr) + const char * src8cstr, + const char * chk8cstr) { int success = 0; @@ -313,7 +282,7 @@ int test_icu_normalizer(const char * rules8cstr, struct icu_buf_utf16 * dest16 = icu_buf_utf16_create(0); struct icu_buf_utf8 * dest8 = icu_buf_utf8_create(0); struct icu_transform * transform - = icu_transform_create(rules8cstr, 'f', &status); + = icu_transform_create(rules8cstr, 'f', 0, &status); icu_check_status(status); icu_utf16_from_utf8_cstr(src16, src8cstr, &status); @@ -329,7 +298,8 @@ int test_icu_normalizer(const char * rules8cstr, if(!strcmp((const char *) dest8->utf8, (const char *) chk8cstr)) success = 1; - else { + else + { success = 0; printf("Normalization\n"); printf("Rules: '%s'\n", rules8cstr); @@ -337,7 +307,6 @@ int test_icu_normalizer(const char * rules8cstr, printf("Normalized: '%s'\n", dest8->utf8); printf("Expected: '%s'\n", chk8cstr); } - icu_transform_destroy(transform); icu_buf_utf16_destroy(src16); @@ -345,14 +314,10 @@ int test_icu_normalizer(const char * rules8cstr, icu_buf_utf8_destroy(dest8); return success; -}; - - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ +} void test_icu_I18N_normalizer(int argc, char **argv) { - YAZ_CHECK(test_icu_normalizer("[:Punctuation:] Any-Remove", "Don't shoot!", "Dont shoot")); @@ -378,18 +343,13 @@ void test_icu_I18N_normalizer(int argc, char **argv) " word4you? ", "word4you")); - YAZ_CHECK(test_icu_normalizer("NFD; [:Nonspacing Mark:] Remove; NFC", "à côté de l'alcôve ovoïde", "a cote de l'alcove ovoide")); - } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - int test_icu_tokenizer(const char * locale, char action, - const char * src8cstr, int count) + const char * src8cstr, int count) { int success = 1; @@ -437,13 +397,8 @@ int test_icu_tokenizer(const char * locale, char action, return success; } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - void test_icu_I18N_tokenizer(int argc, char **argv) { - - const char * en_str = "O Romeo, Romeo! wherefore art thou Romeo?"; @@ -452,8 +407,6 @@ void test_icu_I18N_tokenizer(int argc, char **argv) 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. " @@ -464,10 +417,8 @@ void test_icu_I18N_tokenizer(int argc, char **argv) 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 @@ -476,7 +427,6 @@ void test_icu_I18N_chain(int argc, char **argv) UErrorCode status = U_ZERO_ERROR; struct icu_chain * chain = 0; - const char * xml_str = "" "" "" @@ -497,12 +447,13 @@ void test_icu_I18N_chain(int argc, char **argv) YAZ_CHECK(icu_chain_assign_cstr(chain, en_str, &status)); - while (icu_chain_next_token(chain, &status)){ + while (icu_chain_next_token(chain, &status)) + { ; /* printf("%d '%s' '%s'\n", - icu_chain_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } YAZ_CHECK_EQ(icu_chain_token_number(chain), 7); @@ -510,12 +461,13 @@ void test_icu_I18N_chain(int argc, char **argv) YAZ_CHECK(icu_chain_assign_cstr(chain, "what is this?", &status)); - while (icu_chain_next_token(chain, &status)){ + while (icu_chain_next_token(chain, &status)) + { ; /* printf("%d '%s' '%s'\n", - icu_chain_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } @@ -556,13 +508,13 @@ void test_bug_1140(void) chain, "O Romeo, Romeo! wherefore art thou\t Romeo?", &status)); - while (icu_chain_next_token(chain, &status)){ + while (icu_chain_next_token(chain, &status)) + { ; /* printf("%d '%s' '%s'\n", - icu_chain_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ - + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } @@ -570,12 +522,13 @@ void test_bug_1140(void) 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_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ + while (icu_chain_next_token(chain, &status)) + { + ; + /* printf("%d '%s' '%s'\n", + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } /* we expect 'what' 'is' 'this', i.e. 3 tokens */ @@ -585,7 +538,6 @@ void test_bug_1140(void) } - void test_chain_empty_token(void) { UErrorCode status = U_ZERO_ERROR; @@ -609,12 +561,13 @@ void test_chain_empty_token(void) chain, "a string with 15 tokenss and 8 displays", &status)); - while (icu_chain_next_token(chain, &status)){ + while (icu_chain_next_token(chain, &status)) + { ; /* printf("%d '%s' '%s'\n", - icu_chain_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } YAZ_CHECK_EQ(icu_chain_token_number(chain), 15); @@ -646,12 +599,13 @@ void test_chain_empty_chain(void) chain, src8, &status)); - while (icu_chain_next_token(chain, &status)){ + while (icu_chain_next_token(chain, &status)) + { ; /* printf("%d '%s' '%s'\n", - icu_chain_token_number(chain), - icu_chain_token_norm(chain), - icu_chain_token_display(chain)); */ + icu_chain_token_number(chain), + icu_chain_token_norm(chain), + icu_chain_token_display(chain)); */ } YAZ_CHECK_EQ(icu_chain_token_number(chain), 1); @@ -659,17 +613,104 @@ void test_chain_empty_chain(void) dest8 = (char *) icu_chain_token_norm(chain); YAZ_CHECK_EQ(strcmp(src8, dest8), 0); + icu_chain_destroy(chain); +} + +void test_icu_iter1(void) +{ + UErrorCode status = U_ZERO_ERROR; + struct icu_chain * chain = 0; + xmlNode *xml_node; + struct icu_iter *iter; + struct icu_buf_utf8 *token; + + const char * xml_str = "" + "" + "" + ""; + + xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str)); + YAZ_CHECK(doc); + if (!doc) + return; + xml_node = xmlDocGetRootElement(doc); + YAZ_CHECK(xml_node); + if (!xml_node) + return ; + chain = icu_chain_xml_config(xml_node, 0, &status); + + xmlFreeDoc(doc); + YAZ_CHECK(chain); + + iter = icu_iter_create(chain, "a string with 15 tokens and 8 displays"); + YAZ_CHECK(iter); + if (!iter) + return; +#if 1 + token = icu_buf_utf8_create(0); + while (icu_iter_next(iter, token)) + { + printf("[%.*s]", (int) token->utf8_len, token->utf8); + } + icu_buf_utf8_destroy(token); +#endif + + icu_iter_destroy(iter); icu_chain_destroy(chain); } -#endif /* YAZ_HAVE_ICU */ -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ +void test_icu_iter2(void) +{ + UErrorCode status = U_ZERO_ERROR; + struct icu_chain * chain = 0; + xmlNode *xml_node; + struct icu_iter *iter; + struct icu_buf_utf8 *token; + + const char * xml_str = "" + "" + "" + "" + "" + "" + "" + ""; + + xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str)); + YAZ_CHECK(doc); + if (!doc) + return; + xml_node = xmlDocGetRootElement(doc); + YAZ_CHECK(xml_node); + if (!xml_node) + return ; + + chain = icu_chain_xml_config(xml_node, 0, &status); + + xmlFreeDoc(doc); + YAZ_CHECK(chain); + + iter = icu_iter_create(chain, "Adobe Acrobat Reader, 1991-1999."); + YAZ_CHECK(iter); + if (!iter) + return; + token = icu_buf_utf8_create(0); + while (icu_iter_next(iter, token)) + { + printf("[%.*s]", (int) token->utf8_len, token->utf8); + } + icu_buf_utf8_destroy(token); + + icu_iter_destroy(iter); + icu_chain_destroy(chain); +} + +#endif /* YAZ_HAVE_ICU */ int main(int argc, char **argv) { - YAZ_CHECK_INIT(argc, argv); YAZ_CHECK_LOG(); @@ -682,6 +723,9 @@ int main(int argc, char **argv) test_icu_I18N_chain(argc, argv); test_chain_empty_token(); test_chain_empty_chain(); + test_icu_iter1(); + test_icu_iter2(); + test_bug_1140(); #else /* YAZ_HAVE_ICU */ @@ -695,11 +739,6 @@ int main(int argc, char **argv) YAZ_CHECK_TERM; } - -/* DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8 */ - - - /* * Local variables: * c-basic-offset: 4