X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fstemmer.c;h=7ee26339444939e954fcb342c760166dbdf7abff;hp=dd596f7c70804c91ffad41dd3ae53eff8cb12a49;hb=554423d357ed2bcf4ece977f85de3bc1307d6536;hpb=b5b083d6c36e61c77b8b67980c155dde0a539ca5

diff --git a/src/stemmer.c b/src/stemmer.c
index dd596f7..7ee2633 100644
--- a/src/stemmer.c
+++ b/src/stemmer.c
@@ -18,87 +18,110 @@
 #include <unicode/uchar.h>    /* char names           */
 
 enum stemmer_implementation {
+    yaz_no_operation,
     yaz_snowball
 };
 struct yaz_stemmer_t
 {
     int implementation;
     // Required for cloning.
-    const char *locale;
-    const char *rule;
-    union {
-        struct sb_stemmer *sb_stemmer;
-    };
+    char *locale;
+    char *rule;
+    struct sb_stemmer *sb_stemmer;
 };
 
-const char* yaz_stemmer_lookup_charenc(const char *charenc) {
-    return charenc;
+const char* yaz_stemmer_lookup_charenc(const char *charenc, const char *rule) {
+    return "UTF_8";
 }
 
 const char* yaz_stemmer_lookup_algorithm(const char *locale, const char *rule) {
-    return rule;
+    return locale;
 }
 
 yaz_stemmer_p yaz_stemmer_snowball_create(const char *locale, const char *rule, UErrorCode *status) {
-    const char *charenc = yaz_stemmer_lookup_charenc(locale);
+    const char *charenc = yaz_stemmer_lookup_charenc(locale, rule);
     const char *algorithm = yaz_stemmer_lookup_algorithm(locale,rule);
     struct sb_stemmer *stemmer = sb_stemmer_new(algorithm, charenc);
     yaz_stemmer_p yaz_stemmer;
     if (stemmer == 0) {
-        *status = U_ARGUMENT_TYPE_MISMATCH;
-        return 0;
+        *status = U_ILLEGAL_ARGUMENT_ERROR;
+        yaz_log(YLOG_FATAL, "yaz_stemmer: Failed to create snowball stemmer from locale %srule %s. Showball: charenc %s algorithm %s ", 
+		locale, rule, charenc, algorithm);
+	return 0;
     }
+    yaz_log(YLOG_DEBUG, "created snowball stemmer: algoritm %s charenc %s ", algorithm, charenc);
     yaz_stemmer = xmalloc(sizeof(*yaz_stemmer));
     yaz_stemmer->implementation = yaz_snowball;
+      
     yaz_stemmer->locale = xstrdup(locale);
     yaz_stemmer->rule = xstrdup(rule);
     yaz_stemmer->sb_stemmer = stemmer;
-
+    yaz_log(YLOG_DEBUG, "created snowball stemmer: algoritm %s charenc %s ", algorithm, charenc);
     return yaz_stemmer;
 }
 
 yaz_stemmer_p yaz_stemmer_create(const char *locale, const char *rule, UErrorCode *status) {
     *status = U_ZERO_ERROR;
     // dispatch logic required if more algorithms is implemented.
+    yaz_log(YLOG_DEBUG, "create stemmer: locale %s rule %s ", locale, rule);
     return yaz_stemmer_snowball_create(locale, rule, status);
 }
 
 yaz_stemmer_p yaz_stemmer_clone(yaz_stemmer_p stemmer) {
     UErrorCode error = U_ZERO_ERROR;
+    if (stemmer == 0)
+      return 0;
     return yaz_stemmer_create(stemmer->locale, stemmer->rule, &error);
 }
 
 void yaz_stemmer_stem(yaz_stemmer_p stemmer, struct icu_buf_utf16 *dst, struct icu_buf_utf16* src, UErrorCode *status)
 {
     switch(stemmer->implementation) {
-    case yaz_snowball: {
-        int length;
-        struct icu_buf_utf8 *utf8_buf = icu_buf_utf8_create(0);
-        icu_utf16_to_utf8(utf8_buf, src, status);
-        if (*status == U_ZERO_ERROR) {
-            const char *sb_symbol = sb_stemmer_stem(stemmer->sb_stemmer, icu_buf_utf8_to_cstr(utf8_buf), length);
-            if (sb_symbol == 0) {
-                icu_buf_utf16_copy(dst, src);
-            }
-            else {
-                icu_utf16_from_utf8_cstr(dst, sb_symbol, status);
+        case yaz_snowball: {
+            struct icu_buf_utf8 *utf8_buf = icu_buf_utf8_create(0);
+            icu_utf16_to_utf8(utf8_buf, src, status);
+            if (*status == U_ZERO_ERROR) {
+                const sb_symbol *cstr = (const sb_symbol*) icu_buf_utf8_to_cstr(utf8_buf);
+                const sb_symbol *sb_symbol = sb_stemmer_stem(stemmer->sb_stemmer, cstr, utf8_buf->utf8_len);
+                if (sb_symbol == 0) {
+                    icu_buf_utf16_copy(dst, src);
+                }
+                else {
+
+                    const char *cstr2 = (const char *) sb_symbol;
+                    icu_utf16_from_utf8_cstr(dst, cstr2 , status);
+#if 0		  
+		    yaz_log(YLOG_DEBUG, "stemming %s to %s ", cstr, cstr2);
+#endif
+                }
             }
+            icu_buf_utf8_destroy(utf8_buf);
+            return ;
+            break;
+        }
+    case yaz_no_operation:
+      yaz_log(YLOG_DEBUG, "Stemmer (No operation) called");
+    default: {
+            // Default return the same as given.
+            icu_buf_utf16_copy(dst, src);
         }
-        return ;
-        break;
-    }
     }
 }
 
-void yaz_stemmer_destroy(yaz_stemmer_p stemmer) {
-    switch (stemmer->implementation) {
-    case yaz_snowball:
-        sb_stemmer_delete(stemmer->sb_stemmer);
-        break;
-    }
-    free(stemmer->locale);
-    free(stemmer->rule);
-    free(stemmer);
+void yaz_stemmer_destroy(yaz_stemmer_p stemmer) 
+{
+  /* Handle no stemmer correctly */
+  if (stemmer == 0)
+    return ;
+
+  switch (stemmer->implementation) {
+  case yaz_snowball:
+    sb_stemmer_delete(stemmer->sb_stemmer);
+    break;
+  }
+  xfree(stemmer->locale);
+  xfree(stemmer->rule);
+  xfree(stemmer);
 }
 
 #endif /* YAZ_HAVE_ICU */