From b5b083d6c36e61c77b8b67980c155dde0a539ca5 Mon Sep 17 00:00:00 2001
From: Dennis Schafroth <dennis@indexdata.com>
Date: Wed, 10 Nov 2010 14:41:16 +0100
Subject: [PATCH] First implementation of yaz_stemmer API. Can compile.

---
 src/stemmer.c |   77 +++++++++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 70 insertions(+), 7 deletions(-)

diff --git a/src/stemmer.c b/src/stemmer.c
index b544ebc..dd596f7 100644
--- a/src/stemmer.c
+++ b/src/stemmer.c
@@ -10,32 +10,95 @@
 
 #include <yaz/stemmer.h>
 
+#include <yaz/xmalloc.h>
+
+#include <libstemmer.h>
+
 #include <unicode/ustring.h>  /* some more string fcns*/
 #include <unicode/uchar.h>    /* char names           */
 
+enum stemmer_implementation {
+    yaz_snowball
+};
 struct yaz_stemmer_t
 {
     int implementation;
+    // Required for cloning.
+    const char *locale;
+    const char *rule;
     union {
-        struct sb_stemmer *snowballer;
+        struct sb_stemmer *sb_stemmer;
     };
 };
 
-yaz_stemmer_p yaz_stemmer_create(const char *locale, const char *rule, UErrorCode *status) {
-    return 0;
+const char* yaz_stemmer_lookup_charenc(const char *charenc) {
+    return charenc;
 }
 
-yaz_stemmer_p yaz_stemmer_clone(yaz_stemmer_p stemmer) {
-    return 0;
+const char* yaz_stemmer_lookup_algorithm(const char *locale, const char *rule) {
+    return rule;
 }
 
-void yaz_stemmer_stem(yaz_stemmer_p stemmer, struct icu_buf_utf16 *dst, struct icu_buf_utf16* src, UErrorCode *status) {
+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 *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;
+    }
+    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;
 
+    return yaz_stemmer;
 }
 
-void yaz_stemmer_destroy(yaz_stemmer_p 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.
+    return yaz_stemmer_snowball_create(locale, rule, status);
+}
 
+yaz_stemmer_p yaz_stemmer_clone(yaz_stemmer_p stemmer) {
+    UErrorCode error = U_ZERO_ERROR;
+    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);
+            }
+        }
+        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);
 }
 
 #endif /* YAZ_HAVE_ICU */
-- 
1.7.10.4