X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ficu_chain.c;h=d72e9cdbe587d92005cc3c1b100aeb7eecd01664;hb=419e0de8e395a9b43263d669eb83dc5bef0734a0;hp=73a7674c76e6d0632d381d1f99842aaa9c19b4fb;hpb=54bd147d353f7f340a48a9da5c6a568446223371;p=yaz-moved-to-github.git

diff --git a/src/icu_chain.c b/src/icu_chain.c
index 73a7674..d72e9cd 100644
--- a/src/icu_chain.c
+++ b/src/icu_chain.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.
  */
 
@@ -22,6 +22,7 @@
 #include <string.h>
 #include <stdlib.h>
 #include <stdio.h>
+#include <assert.h>
 
 #include <unicode/ustring.h>  /* some more string fcns*/
 #include <unicode/uchar.h>    /* char names           */
@@ -44,32 +45,19 @@ struct icu_chain_step
 	struct icu_transform * transform;
 	struct icu_tokenizer * tokenizer;  
     } u;
-    /* temprary post-action utf16 buffer */
-    struct icu_buf_utf16 * buf16;  
     struct icu_chain_step * previous;
-    int more_tokens;
-    int need_new_token;
 };
 
 struct icu_chain
 {
+    struct icu_iter *iter;
     char *locale;
     int sort;
 
-    const char * src8cstr;
-
     UCollator * coll;
     
-    /* number of tokens returned so far */
-    int32_t token_count;
-    
     /* utf8 output buffers */
-    struct icu_buf_utf8 * display8;
     struct icu_buf_utf8 * norm8;
-    struct icu_buf_utf8 * sort8;
-    
-    /* utf16 source buffer */
-    struct icu_buf_utf16 * src16;
     
     /* linked list of chain steps */
     struct icu_chain_step * steps;
@@ -87,7 +75,7 @@ int icu_check_status(UErrorCode status)
 
 static struct icu_chain_step *icu_chain_step_create(
     struct icu_chain * chain,  enum icu_chain_step_type type,
-    const uint8_t * rule, struct icu_buf_utf16 * buf16,
+    const uint8_t * rule, 
     UErrorCode *status)
 {
     struct icu_chain_step * step = 0;
@@ -98,9 +86,6 @@ static struct icu_chain_step *icu_chain_step_create(
     step = (struct icu_chain_step *) xmalloc(sizeof(struct icu_chain_step));
 
     step->type = type;
-
-    step->buf16 = buf16;
-
     /* create auxilary objects */
     switch (step->type)
     {
@@ -143,16 +128,13 @@ static void icu_chain_step_destroy(struct icu_chain_step * step)
         break;
     case ICU_chain_step_type_casemap:
         icu_casemap_destroy(step->u.casemap);
-        icu_buf_utf16_destroy(step->buf16);
         break;
     case ICU_chain_step_type_transform:
     case ICU_chain_step_type_transliterate:
         icu_transform_destroy(step->u.transform);
-        icu_buf_utf16_destroy(step->buf16);
         break;
     case ICU_chain_step_type_tokenize:
         icu_tokenizer_destroy(step->u.tokenizer);
-        icu_buf_utf16_destroy(step->buf16);
         break;
     default:
         break;
@@ -168,6 +150,7 @@ struct icu_chain *icu_chain_create(const char *locale, int sort,
 
     *status = U_ZERO_ERROR;
 
+    chain->iter = 0;
     chain->locale = xstrdup(locale);
 
     chain->sort = sort;
@@ -177,16 +160,7 @@ struct icu_chain *icu_chain_create(const char *locale, int sort,
     if (U_FAILURE(*status))
         return 0;
 
-    chain->token_count = 0;
-
-    chain->src8cstr = 0;
-
-    chain->display8 = icu_buf_utf8_create(0);
     chain->norm8 = icu_buf_utf8_create(0);
-    chain->sort8 = icu_buf_utf8_create(0);
-
-    chain->src16 = icu_buf_utf16_create(0);
-
     chain->steps = 0;
 
     return chain;
@@ -199,12 +173,9 @@ void icu_chain_destroy(struct icu_chain * chain)
         if (chain->coll)
             ucol_close(chain->coll);
 
-        icu_buf_utf8_destroy(chain->display8);
         icu_buf_utf8_destroy(chain->norm8);
-        icu_buf_utf8_destroy(chain->sort8);
-        
-        icu_buf_utf16_destroy(chain->src16);
-    
+        if (chain->iter)
+            icu_iter_destroy(chain->iter);
         icu_chain_step_destroy(chain->steps);
         xfree(chain->locale);
         xfree(chain);
@@ -300,42 +271,12 @@ static struct icu_chain_step *icu_chain_insert_step(
     const uint8_t * rule, UErrorCode *status)
 {    
     struct icu_chain_step * step = 0;
-    struct icu_buf_utf16 * src16 = 0;
-    struct icu_buf_utf16 * buf16 = 0;
-
     if (!chain || !type || !rule)
         return 0;
 
-    /* assign utf16 src buffers as needed */
-    if (chain->steps && chain->steps->buf16)
-        src16 = chain->steps->buf16;
-    else if (chain->src16)
-        src16 = chain->src16;
-    else
-        return 0;
-
-    /* create utf16 destination buffers as needed, or */
-    switch (type)
-    {
-    case ICU_chain_step_type_display:
-        buf16 = src16;
-        break;
-    case ICU_chain_step_type_casemap:
-        buf16 = icu_buf_utf16_create(0);
-        break;
-    case ICU_chain_step_type_transform:
-    case ICU_chain_step_type_transliterate:
-        buf16 = icu_buf_utf16_create(0);
-        break;
-    case ICU_chain_step_type_tokenize:
-        buf16 = icu_buf_utf16_create(0);
-        break;
-        break;
-    default:
-        break;
-    }
     /* create actual chain step with this buffer */
-    step = icu_chain_step_create(chain, type, rule, buf16, status);
+    step = icu_chain_step_create(chain, type, rule,
+                                 status);
 
     step->previous = chain->steps;
     chain->steps = step;
@@ -343,216 +284,210 @@ static struct icu_chain_step *icu_chain_insert_step(
     return step;
 }
 
-static int icu_chain_step_next_token(struct icu_chain * chain,
-                                     struct icu_chain_step * step,
-                                     UErrorCode *status)
-{
-    struct icu_buf_utf16 * src16 = 0;
-    int got_new_token = 0;
-
-    if (!chain || !chain->src16 || !step || !step->more_tokens)
-        return 0;
+struct icu_iter {
+    struct icu_chain *chain;
+    struct icu_buf_utf16 *last;
+    UErrorCode status;
+    struct icu_buf_utf8 *display;
+    struct icu_buf_utf8 *sort8;
+    struct icu_buf_utf16 *input;
+    int token_count;
+};
 
-    /* assign utf16 src buffers as needed, advance in previous steps
-       tokens until non-zero token met, and setting stop condition */
+void icu_utf16_print(struct icu_buf_utf16 *src16)
+{
+    UErrorCode status = U_ZERO_ERROR;
+    const char *p;
+    struct icu_buf_utf8 *dst8 = icu_buf_utf8_create(0);
+    icu_utf16_to_utf8(dst8, src16, &status);
 
-    if (step->previous)
+    assert(status != 1234);
+    if (U_FAILURE(status))
     {
-        src16 = step->previous->buf16;
-        /* tokens might be killed in previous steps, therefore looping */
-
-        while (step->need_new_token 
-               && step->previous->more_tokens
-               && !got_new_token)
-            got_new_token
-                = icu_chain_step_next_token(chain, step->previous, status);
+        printf("failure");
     }
-    else 
-    { /* first step can only work once on chain->src16 input buffer */
-        src16 = chain->src16;
-        step->more_tokens = 0;
-        got_new_token = 1;
-    }
-
-    if (!src16)
-        return 0;
-
-    /* stop if nothing to process */
-    if (step->need_new_token && !got_new_token)
+    else
     {
-        step->more_tokens = 0;
-        return 0;
+        p = icu_buf_utf8_to_cstr(dst8);
+        printf("%s", p);
     }
+    icu_buf_utf8_destroy(dst8);
+}
 
-    /* either an old token not finished yet, or a new token, thus
-       perform the work, eventually put this steps output in 
-       step->buf16 or the chains UTF8 output buffers  */
-
-    switch (step->type)
+struct icu_buf_utf16 *icu_iter_invoke(struct icu_iter *iter,
+                                      struct icu_chain_step *step,
+                                      struct icu_buf_utf16 *src)
+{
+    if (!step)
+        return src;
+    else
     {
-    case ICU_chain_step_type_display:
-        icu_utf16_to_utf8(chain->display8, src16, status);
-        break;
-    case ICU_chain_step_type_casemap:
-        icu_casemap_casemap(step->u.casemap,
-                            step->buf16, src16, status,
-                            chain->locale);
-        break;
-    case ICU_chain_step_type_transform:
-    case ICU_chain_step_type_transliterate:
-        icu_transform_trans(step->u.transform,
-                            step->buf16, src16, status);
-        break;
-    case ICU_chain_step_type_tokenize:
-        /* attach to new src16 token only first time during splitting */
-        if (step->need_new_token)
-        {
-            icu_tokenizer_attach(step->u.tokenizer, src16, status);
-            step->need_new_token = 0;
-        }
-
-        /* splitting one src16 token into multiple buf16 tokens */
-        step->more_tokens
-            = icu_tokenizer_next_token(step->u.tokenizer,
-                                       step->buf16, status);
-
-        /* make sure to get new previous token if this one had been used up
-           by recursive call to _same_ step */
-
-        if (!step->more_tokens)
+        struct icu_buf_utf16 *dst = icu_iter_invoke(iter, step->previous, src);
+        
+        switch (step->type)
         {
-            step->more_tokens = icu_chain_step_next_token(chain, step, status);
-            return step->more_tokens;  /* avoid one token count too much! */
+        case ICU_chain_step_type_casemap:
+            if (dst)
+            {
+                struct icu_buf_utf16 *src = dst;
+
+                dst = icu_buf_utf16_create(0);
+                icu_casemap_casemap(step->u.casemap, dst, src, &iter->status,
+                                    iter->chain->locale);
+                icu_buf_utf16_destroy(src);
+            }
+            break;
+        case ICU_chain_step_type_tokenize:
+            if (dst)
+            {
+                struct icu_buf_utf16 *src = dst;
+
+                icu_tokenizer_attach(step->u.tokenizer, src, &iter->status);
+                icu_buf_utf16_destroy(src);
+            }
+            dst = icu_buf_utf16_create(0);
+            iter->status = U_ZERO_ERROR;
+            if (!icu_tokenizer_next_token(step->u.tokenizer, dst, &iter->status))
+            {
+                icu_buf_utf16_destroy(dst);
+                dst = 0;
+            }
+            break;
+        case ICU_chain_step_type_transform:
+        case ICU_chain_step_type_transliterate:
+            if (dst)
+            {
+                struct icu_buf_utf16 *src = dst;
+                dst = icu_buf_utf16_create(0);
+                icu_transform_trans(step->u.transform, dst, src, &iter->status);
+                icu_buf_utf16_destroy(src);
+            }
+            break;
+        case ICU_chain_step_type_display:
+            if (dst)
+                icu_utf16_to_utf8(iter->display, dst, &iter->status);
+            break;
+        default:
+            assert(0);
         }
-        break;
-    default:
-        return 0;
-        break;
+        return dst;
     }
+}
 
-    if (U_FAILURE(*status))
+struct icu_iter *icu_iter_create(struct icu_chain *chain,
+                                 const char *src8cstr)
+{
+    if (!src8cstr)
         return 0;
+    else
+    {
+        struct icu_iter *iter = xmalloc(sizeof(*iter));
+        iter->chain = chain;
+        iter->status = U_ZERO_ERROR;
+        iter->display = icu_buf_utf8_create(0);
+        iter->sort8 = icu_buf_utf8_create(0);
+        iter->token_count = 0;
+        iter->last = 0; /* no last returned string (yet) */
+
+        /* fill and assign input string.. It will be 0 after
+           first iteration */
+        iter->input =  icu_buf_utf16_create(0);
+        icu_utf16_from_utf8_cstr(iter->input, src8cstr, &iter->status);
+        return iter;
 
-    /* if token disappered into thin air, tell caller */
-    /* if (!step->buf16->utf16_len && !step->more_tokens) */ 
-    /*    return 0; */ 
-
-    return 1;
+    }
 }
 
-int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr, 
-                          UErrorCode *status)
+void icu_iter_destroy(struct icu_iter *iter)
 {
-    struct icu_chain_step * stp = 0; 
+    if (iter)
+    {
+        icu_buf_utf8_destroy(iter->display);
+        icu_buf_utf8_destroy(iter->sort8);
+        if (iter->input)
+            icu_buf_utf16_destroy(iter->input);
+        xfree(iter);
+    }
+}
 
-    if (!chain || !src8cstr)
+int icu_iter_next(struct icu_iter *iter, struct icu_buf_utf8 *result)
+{
+    if (!iter->input && iter->last == 0)
         return 0;
+    else
+    {
+        /* on first call, iter->input is the input string. Thereafter: 0. */
+        iter->last = icu_iter_invoke(iter, iter->chain->steps, iter->input);
+        iter->input = 0;
+        
+        if (!iter->last)
+            return 0;
 
-    chain->src8cstr = src8cstr;
+        iter->token_count++;
 
-    stp = chain->steps;
-    
-    /* clear token count */
-    chain->token_count = 0;
+        if (iter->chain->sort)
+        {        
+            icu_sortkey8_from_utf16(iter->chain->coll,
+                                    iter->sort8, iter->last,
+                                    &iter->status);
+        }
+        icu_utf16_to_utf8(result, iter->last, &iter->status);
+        icu_buf_utf16_destroy(iter->last);
 
-    /* clear all steps stop states */
-    while (stp)
-    {
-        stp->more_tokens = 1;
-        stp->need_new_token = 1;
-        stp = stp->previous;
+        return 1;
     }
-    
-    /* finally convert UTF8 to UTF16 string if needed */
-    if (chain->steps || chain->sort)
-        icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status);
-            
-    if (U_FAILURE(*status))
-        return 0;
+}
+
+const char *icu_iter_get_sortkey(struct icu_iter *iter)
+{
+    return icu_buf_utf8_to_cstr(iter->sort8);
+}
 
+const char *icu_iter_get_display(struct icu_iter *iter)
+{ 
+    return icu_buf_utf8_to_cstr(iter->display);   
+}
+
+int icu_chain_assign_cstr(struct icu_chain * chain, const char * src8cstr, 
+                          UErrorCode *status)
+{
+    if (chain->iter)
+        icu_iter_destroy(chain->iter);
+    chain->iter = icu_iter_create(chain, src8cstr);
     return 1;
 }
 
 int icu_chain_next_token(struct icu_chain * chain, UErrorCode *status)
 {
-    int got_token = 0;
-    
     *status = U_ZERO_ERROR;
-
-    if (!chain)
-        return 0;
-
-    /* special case with no steps - same as index type binary */
-    if (!chain->steps)
-    {
-        if (chain->token_count)
-            return 0;
-        else
-        {
-            chain->token_count++;
-            
-            if (chain->sort)
-                icu_sortkey8_from_utf16(chain->coll,
-                                        chain->sort8, chain->steps->buf16,
-                                        status);
-            return chain->token_count;
-        }
-    }
-    /* usual case, one or more icu chain steps existing */
-    else 
-    {
-        while (!got_token && chain->steps && chain->steps->more_tokens)
-            got_token = icu_chain_step_next_token(chain, chain->steps, status);
-
-        if (got_token)
-        {
-            chain->token_count++;
-
-            icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status);
-            
-            if (chain->sort)
-                icu_sortkey8_from_utf16(chain->coll,
-                                        chain->sort8, chain->steps->buf16,
-                                        status);
-            return chain->token_count;
-        }
-    }
-        
-    return 0;
+    return icu_iter_next(chain->iter, chain->norm8);
 }
 
 int icu_chain_token_number(struct icu_chain * chain)
 {
-    if (!chain)
-        return 0;
-    
-    return chain->token_count;
+    if (chain && chain->iter)
+        return chain->iter->token_count;
+    return 0;
 }
 
 const char * icu_chain_token_display(struct icu_chain * chain)
 {
-    if (chain->display8)
-        return icu_buf_utf8_to_cstr(chain->display8);
-    
+    if (chain->iter)
+        return icu_iter_get_display(chain->iter);
     return 0;
 }
 
 const char * icu_chain_token_norm(struct icu_chain * chain)
 {
-    if (!chain->steps)
-        return chain->src8cstr;
-
     if (chain->norm8)
         return icu_buf_utf8_to_cstr(chain->norm8);
-    
     return 0;
 }
 
 const char * icu_chain_token_sortkey(struct icu_chain * chain)
 {
-    if (chain->sort8)
-        return icu_buf_utf8_to_cstr(chain->sort8);
-    
+    if (chain->iter)
+        return icu_iter_get_sortkey(chain->iter);
     return 0;
 }