X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Ftest_icu_I18N.c;h=62b52557f1b3f7687a3513dff49786fcb6a6af77;hb=1ea74f79189dfde78b3b65e7355774da84efd0dc;hp=3b56883d0709d449e70c9968abab0e4905009622;hpb=543d6fff956361a3fdd1de89392eb2cc308670b1;p=pazpar2-moved-to-github.git

diff --git a/src/test_icu_I18N.c b/src/test_icu_I18N.c
index 3b56883..62b5255 100644
--- a/src/test_icu_I18N.c
+++ b/src/test_icu_I18N.c
@@ -1,4 +1,4 @@
-/* $Id: test_icu_I18N.c,v 1.2 2007-04-30 13:56:52 marc Exp $
+/* $Id: test_icu_I18N.c,v 1.7 2007-05-02 14:01:36 marc Exp $
    Copyright (c) 2006-2007, Index Data.
 
 This file is part of Pazpar2.
@@ -19,6 +19,9 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
  */
 
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
+ 
+
 #if HAVE_CONFIG_H
 #include "cconfig.h"
 #endif
@@ -34,129 +37,113 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 
 #ifdef HAVE_ICU
 #include "icu_I18N.h"
-#include "string.h"
+#include <string.h>
+#include <stdlib.h>
 
-void test_icu_I18N_casemap_en(int argc, char **argv)
-{
 
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
+
+int test_icu_casemap(const char * locale, char action,
+                     const char * src8, const char * check8)
+{
+    NMEM nmem = nmem_create();    
     size_t buf_cap = 128;
     char buf[buf_cap];
+    const char * dest8 = 0;
     size_t dest8_len = 0;
-    NMEM nmem = nmem_create();
-    char * dest8 = 0;
+    //size_t src8_len = strlen(src8);
+    int sucess = 0;
 
-    const char * src8 =  "A ReD fOx hunTS sQUirriLs";
-    size_t src8_len = strlen(src8);
-    
-    printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
+    //printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
 
     //these shall succeed
     dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "en", 'l');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'en:l' '%s' (%d)\n", dest8, (int) dest8_len);
-
-
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "en", 'u');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'en:u' '%s' (%d)\n", dest8, (int) dest8_len);
+                        src8, locale, action);
 
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "en", 'f');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'en:f' '%s' (%d)\n", dest8, (int) dest8_len);
+    //printf("icu_casemap '%s:%c' '%s' (%d)\n", 
+    //       locale, action, dest8, (int) dest8_len);
 
+    if (dest8 
+        && (dest8_len == strlen(check8))
+        && !strcmp(check8, dest8))
+        sucess = dest8_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);
+    nmem_destroy(nmem);
 
-    // 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);
+    return sucess;
+}
 
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
 
+void test_icu_I18N_casemap(int argc, char **argv)
+{
 
+    // Locale 'en'
 
+    // sucessful tests
+    YAZ_CHECK(test_icu_casemap("en", 'l',
+                               "A ReD fOx hunTS sQUirriLs", 
+                               "a red fox hunts squirrils"));
+    
+    YAZ_CHECK(test_icu_casemap("en", 'u',
+                               "A ReD fOx hunTS sQUirriLs", 
+                               "A RED FOX HUNTS SQUIRRILS"));
+    
+    YAZ_CHECK(test_icu_casemap("en", 'f',
+                               "A ReD fOx hunTS sQUirriLs", 
+                               "a red fox hunts squirrils"));
+    
+    // this one fails and needs more investigation ..
+    YAZ_CHECK(0 == test_icu_casemap("en", 't',
+                               "A ReD fOx hunTS sQUirriLs", 
+                               "A Red Fox Hunts Squirrils"));
 
-    nmem_destroy(nmem);
 
-    YAZ_CHECK(0 == 0);
-    //YAZ_CHECK_EQ(0, 1);
-}
+    // Locale 'da'
 
-void test_icu_I18N_casemap_da(int argc, char **argv)
-{
+    // sucess expected    
+    YAZ_CHECK(test_icu_casemap("da", 'l',
+                               "åh ÆbLE, øs fLØde i Åen efter bLåBærGRødeN", 
+                               "åh æble, øs fløde i åen efter blåbærgrøden"));
 
-    size_t buf_cap = 128;
-    char buf[buf_cap];
-    size_t dest8_len = 0;
-    NMEM nmem = nmem_create();
-    char * dest8 = 0;
+    YAZ_CHECK(test_icu_casemap("da", 'u',
+                               "åh ÆbLE, øs fLØde i Åen efter bLåBærGRødeN", 
+                               "ÅH ÆBLE, ØS FLØDE I ÅEN EFTER BLÅBÆRGRØDEN"));
 
-    const char * src8 =  "åh ÆbLE, øs fLØde i Åen efter bLåBærGRødeN";
-    size_t src8_len = strlen(src8);
-    
-    printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
+    YAZ_CHECK(test_icu_casemap("da", 'f',
+                               "åh ÆbLE, øs fLØde i Åen efter bLåBærGRødeN", 
+                               "åh æble, øs fløde i åen efter blåbærgrøden"));
 
-    //these shall succeed
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 'l');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:l' '%s' (%d)\n", dest8, (int) dest8_len);
+    YAZ_CHECK(test_icu_casemap("da", 't',
+                               "åh ÆbLE, øs fLØde i Åen efter bLåBærGRødeN", 
+                               "Åh Æble, Øs Fløde I Åen Efter Blåbærgrøden"));
 
+    // Locale 'de'
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 'u');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:u' '%s' (%d)\n", dest8, (int) dest8_len);
+    // sucess expected    
+    YAZ_CHECK(test_icu_casemap("de", 'l',
+                          "zWÖlf ärgerliche Würste rollen ÜBer die StRAße",
+                          "zwölf ärgerliche würste rollen über die straße"));
 
+    YAZ_CHECK(test_icu_casemap("de", 'u',
+                           "zWÖlf ärgerliche Würste rollen ÜBer die StRAße",
+                           "ZWÖLF ÄRGERLICHE WÜRSTE ROLLEN ÜBER DIE STRASSE"));
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 'f');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:f' '%s' (%d)\n", dest8, (int) dest8_len);
+    YAZ_CHECK(test_icu_casemap("de", 'f',
+                           "zWÖlf ärgerliche Würste rollen ÜBer die StRAße",
+                           "zwölf ärgerliche würste rollen über die strasse"));
 
+    YAZ_CHECK(test_icu_casemap("de", 't',
+                           "zWÖlf ärgerliche Würste rollen ÜBer die StRAße",
+                           "Zwölf Ärgerliche Würste Rollen Über Die Straße"));
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 't');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:t' '%s' (%d)\n", dest8, (int) dest8_len);
+}
 
-    nmem_destroy(nmem);
 
-    YAZ_CHECK(0 == 0);
-    //YAZ_CHECK_EQ(0, 1);
-}
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
 
-void test_icu_I18N_casemap_de(int argc, char **argv)
+void test_icu_I18N_casemap_failures(int argc, char **argv)
 {
 
     size_t buf_cap = 128;
@@ -165,93 +152,152 @@ void test_icu_I18N_casemap_de(int argc, char **argv)
     NMEM nmem = nmem_create();
     char * dest8 = 0;
 
-    const char * src8 = "zWÖlf ärgerliche Würste rollen ÜBer die StRAße";
-    size_t src8_len = strlen(src8);
+    const char * src8 =  "A ReD fOx hunTS sQUirriLs";
+    //size_t src8_len = strlen(src8);
     
-    printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
+    //printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
 
-    //these shall succeed
+    // some calling error needs investigation
     dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 'l');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:l' '%s' (%d)\n", dest8, (int) 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, "da", 'u');
+                        src8, "zz_abc", 'l');
     YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:u' '%s' (%d)\n", dest8, (int) dest8_len);
+    //printf("icu_casemap 'zz:l' '%s' (%d)\n", dest8, (int) dest8_len);
 
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "da", 'f');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:f' '%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, "da", 't');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'da:t' '%s' (%d)\n", dest8, (int) 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(0 == 0);
-    //YAZ_CHECK_EQ(0, 1);
 }
 
-void test_icu_I18N_casemap_el(int argc, char **argv)
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
+
+int test_icu_sortmap(const char * locale, size_t list_len,
+                     const char ** src8_list, const char ** check8_list)
 {
+    int sucess = 1;
 
+    size_t i = 0;
 
-#if 0
 
+    NMEM nmem = nmem_create();    
     size_t buf_cap = 128;
     char buf[buf_cap];
-    size_t dest8_len = 0;
-    NMEM nmem = nmem_create();
-    char * dest8 = 0;
-
-    const char * src8 = ""
-    size_t src8_len = strlen(src8);
+    struct icu_termmap ** dest8_list 
+        = nmem_malloc(nmem, sizeof(struct icu_termmap *) * list_len);
+    //size_t dest8_len = 0;
+    //size_t src8_len = strlen(src8);
+
+    // initializing icu_termmap
+    for (i = 0; i < list_len; i++){
+        dest8_list[i] = icu_termmap_create(nmem);
+        dest8_list[i]->norm_term = nmem_strdup(nmem, src8_list[i]); 
+        dest8_list[i]->disp_term = nmem_strdup(nmem, src8_list[i]);
+        //dest8_list[i]->sort_key =  nmem_strdup(nmem, src8_list[i]);
+        //dest8_list[i]->sort_len =  strlen(src8_list[i]);
+        dest8_list[i]->sort_key 
+            = icu_sortmap(nmem, buf, buf_cap, 0, src8_list[i], locale);
+        // = icu_sortmap(nmem, buf, buf_cap, &(dest8_list[i]->sort_len),
+        //                  src8_list[i], locale);
+    }
+
+    // do the sorting
+    qsort(dest8_list, list_len, 
+          sizeof(struct icu_termmap *), icu_termmap_cmp);
+
+    // checking correct sorting
+    for (i = 0; i < list_len; i++){
+        if (0 != strcmp(dest8_list[i]->disp_term, check8_list[i])){
+            sucess = 0;
+        }
+    }
+
+    if (1 || !sucess){
+        printf("\n");
+        printf("Input    '%s':", locale);
+        for (i = 0; i < list_len; i++)
+            printf(" '%s'", src8_list[i]);
+        printf("\n");        
+        printf("ICU sort '%s':", locale);
+        for (i = 0; i < list_len; i++)
+            printf(" '%s'", dest8_list[i]->disp_term);
+        if (sucess)
+            printf(" OK");
+        else
+            printf(" ERROR ??");
+        printf("\n");
+        printf("Expected '%s':", locale);
+        for (i = 0; i < list_len; i++)
+            printf(" '%s'", check8_list[i]);
+        printf("\n");        
+    }
     
-    printf("original string:   '%s' (%d)\n", src8, (int) src8_len);
-
-    //these shall succeed
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "el", 'l');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'el:l' '%s' (%d)\n", dest8, (int) dest8_len);
-
-
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "el", 'u');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'el:u' '%s' (%d)\n", dest8, (int) dest8_len);
-
-
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "el", 'f');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'el:f' '%s' (%d)\n", dest8, (int) dest8_len);
+    nmem_destroy(nmem);
 
+    return sucess;
+}
 
-    dest8 = icu_casemap(nmem, buf, buf_cap, &dest8_len,
-                        src8, "el", 't');
-    YAZ_CHECK(dest8_len);
-    printf("icu_casemap 'el:t' '%s' (%d)\n", dest8, (int) dest8_len);
 
-    nmem_destroy(nmem);
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
 
-    YAZ_CHECK(0 == 0);
-    //YAZ_CHECK_EQ(0, 1);
-#endif 
+void test_icu_I18N_sortmap(int argc, char **argv)
+{
 
+    // sucessful tests
+    size_t en_1_len = 6;
+    const char * en_1_src[6] = {"z", "K", "a", "A", "Z", "k"};
+    const char * en_1_cck[6] = {"a", "A", "K", "k", "z", "Z"};
+    YAZ_CHECK(test_icu_sortmap("en", en_1_len, en_1_src, en_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("en_AU", en_1_len, en_1_src, en_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("en_CA", en_1_len, en_1_src, en_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("en_GB", en_1_len, en_1_src, en_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("en_US", en_1_len, en_1_src, en_1_cck));
+    
+    // sucessful tests - this one fails and should not!!!
+    size_t da_1_len = 6;
+    const char * da_1_src[6] = {"z", "å", "o", "æ", "a", "ø"};
+    const char * da_1_cck[6] = {"a", "o", "z", "æ", "ø", "å"};
+    YAZ_CHECK(0 == test_icu_sortmap("da", da_1_len, da_1_src, da_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("da_DK", da_1_len, da_1_src, da_1_cck));
+    
+    // sucessful tests
+    size_t de_1_len = 9;
+    const char * de_1_src[9] = {"u", "ä", "o", "t", "s", "ß", "ü", "ö", "a"};
+    const char * de_1_cck[9] = {"ä", "a", "o", "ö", "s", "ß", "t", "u", "ü"};
+    YAZ_CHECK(test_icu_sortmap("de", de_1_len, de_1_src, de_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("de_AT", de_1_len, de_1_src, de_1_cck));
+    YAZ_CHECK(0 == test_icu_sortmap("de_DE", de_1_len, de_1_src, de_1_cck));
+    
 }
 
 
 #endif    
 
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
+
 int main(int argc, char **argv)
 {
 
@@ -260,15 +306,14 @@ int main(int argc, char **argv)
 
 #ifdef HAVE_ICU
 
-    test_icu_I18N_casemap_en(argc, argv);
-    test_icu_I18N_casemap_da(argc, argv); 
-    test_icu_I18N_casemap_de(argc, argv); 
-    test_icu_I18N_casemap_el(argc, argv); 
+    test_icu_I18N_casemap_failures(argc, argv);
+    test_icu_I18N_casemap(argc, argv);
+    test_icu_I18N_sortmap(argc, argv);
  
 #else
 
     printf("ICU unit tests omitted.\n"
-           "Please install libicu36-dev and icu-doc or similar\n".);
+           "Please install libicu36-dev and icu-doc or similar\n");
     YAZ_CHECK(0 == 0);
 
 #endif    
@@ -277,7 +322,7 @@ int main(int argc, char **argv)
 }
 
 
-
+// DO NOT EDIT THIS FILE IF YOUR EDITOR DOES NOT SUPPORT UTF-8
 
 /*
  * Local variables: