First go at returning start+offset

[yaz-moved-to-github.git] / src / icu_tokenizer.c

diff --git a/src/icu_tokenizer.c b/src/icu_tokenizer.c
index 0697029..7e2fc3f 100644 (file)
--- a/src/icu_tokenizer.c
+++ b/src/icu_tokenizer.c
@@ -1,5 +1,5 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2012 Index Data
+ * Copyright (C) 1995-2013 Index Data
  * See the file LICENSE for details.
  */
 
@@ -78,7 +78,7 @@ struct icu_tokenizer *icu_tokenizer_clone(struct icu_tokenizer *old)
 struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action,
                                            UErrorCode *status)
 {
-    struct icu_tokenizer * tokenizer
+    struct icu_tokenizer *tokenizer
         = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer));
 
     icu_tokenizer_reset(tokenizer, action);
@@ -119,7 +119,7 @@ struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action,
     return 0;
 }
 
-void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
+void icu_tokenizer_destroy(struct icu_tokenizer *tokenizer)
 {
     if (tokenizer)
     {
@@ -130,8 +130,8 @@ void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
     }
 }
 
-int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
-                         struct icu_buf_utf16 * src16,
+int icu_tokenizer_attach(struct icu_tokenizer *tokenizer,
+                         struct icu_buf_utf16 *src16,
                          UErrorCode *status)
 {
     if (!tokenizer || !tokenizer->bi || !src16)
@@ -153,9 +153,10 @@ int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
     return 1;
 }
 
-int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
-                                 struct icu_buf_utf16 * tkn16,
-                                 UErrorCode *status)
+int32_t icu_tokenizer_next_token(struct icu_tokenizer *tokenizer,
+                                 struct icu_buf_utf16 *tkn16,
+                                 UErrorCode *status,
+                                 size_t *start, size_t *len)
 {
     int32_t tkn_start = 0;
     int32_t tkn_end = 0;
@@ -202,6 +203,9 @@ int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
     tokenizer->token_start = tkn_start;
     tokenizer->token_end = tkn_end;
 
+    *start = tkn_start;
+    *len = tkn_end - tkn_start;
+
     /* copying into token buffer if it exists */
     if (tkn16)
     {
@@ -217,7 +221,7 @@ int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
     return tkn_len;
 }
 
-int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
+int32_t icu_tokenizer_token_count(struct icu_tokenizer *tokenizer)
 {
     return tokenizer->token_count;
 }