From: Adam Dickmeiss <adam@indexdata.dk>
Date: Tue, 4 Jun 1996 10:20:02 +0000 (+0000)
Subject: Added support for character mapping.
X-Git-Tag: ZEBRA.1.0~432
X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=commitdiff_plain;h=7ed12b6cc4a492d6062dd56bb1ebc49c9bc75b4a

Added support for character mapping.
---

diff --git a/dfa/dfa.c b/dfa/dfa.c
index 4c7c864..ffe11f3 100644
--- a/dfa/dfa.c
+++ b/dfa/dfa.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1994, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dfa.c,v $
- * Revision 1.11  1996-01-08 19:15:24  adam
+ * Revision 1.12  1996-06-04 10:20:02  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.11  1996/01/08  19:15:24  adam
  * Allow single $ in expressions.
  *
  * Revision 1.10  1996/01/08  09:09:17  adam
@@ -428,6 +431,30 @@ static int read_charset (void)
     return L_CHARS;
 }
 
+static int map_l_char (void)
+{
+    char **mapto;
+    const char *cp0 = expr_ptr-1;
+    int i = 0, len = strlen(cp0);
+
+    if (cp0[0] == 1 && cp0[1])
+    {
+        expr_ptr++;
+        look_ch = cp0[1];
+        return L_CHAR;
+    }
+    if (!parse_info->cmap)
+        return L_CHAR;
+
+    mapto = (*parse_info->cmap) (&cp0, len);
+    assert (mapto);
+    
+    expr_ptr = cp0;
+    look_ch = mapto[i][0];
+    logf (LOG_DEBUG, "map from %c to %d", expr_ptr[-1], look_ch);
+    return L_CHAR;
+}
+
 static int lex_sub(void)
 {
     int esc;
@@ -435,11 +462,11 @@ static int lex_sub(void)
         if (look_ch == '\"')
         {
             if (esc)
-                return L_CHAR;
+                return map_l_char ();
             inside_string = !inside_string;
         }
         else if (esc || inside_string)
-            return L_CHAR;
+            return map_l_char ();
         else if (look_ch == '[')
             return read_charset();
         else 
@@ -452,7 +479,7 @@ static int lex_sub(void)
                         --expr_ptr;
                     return cc[1];
                 }
-            return L_CHAR;            
+            return map_l_char ();     
         }
     return 0;
 }
@@ -978,6 +1005,7 @@ static struct DFA_parse *dfa_parse_init (void)
     parse_info->use_Tnode = parse_info->max_Tnode = 0;
     parse_info->charMap = NULL;
     parse_info->charMapSize = 0;
+    parse_info->cmap = NULL;
     return parse_info;
 }
 
@@ -1032,6 +1060,11 @@ struct DFA *dfa_init (void)
     return dfa;
 }
 
+void dfa_set_cmap (struct DFA *dfa, char **(*cmap)(const char **from, int len))
+{
+    dfa->parse_info->cmap = cmap;
+}
+
 int dfa_parse (struct DFA *dfa, const char **pattern)
 {
     struct Tnode *top;
diff --git a/dfa/dfap.h b/dfa/dfap.h
index 2b7dbda..6265962 100644
--- a/dfa/dfap.h
+++ b/dfa/dfap.h
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dfap.h,v $
- * Revision 1.3  1996-01-08 09:09:19  adam
+ * Revision 1.4  1996-06-04 10:20:03  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.3  1996/01/08  09:09:19  adam
  * Function dfa_parse got 'const' string argument.
  * New functions to define char mappings made public.
  *
@@ -35,6 +38,7 @@ struct DFA_parse {
     struct Tblock *end;       /* end block of Tnodes */
     int *charMap;
     int charMapSize;
+    char **(*cmap)(const char **from, int len);
 };
 
 typedef struct DFA_stateb_ {
diff --git a/dict/lookgrep.c b/dict/lookgrep.c
index 5d8fe87..733be36 100644
--- a/dict/lookgrep.c
+++ b/dict/lookgrep.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1994, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: lookgrep.c,v $
- * Revision 1.16  1996-05-24 14:46:04  adam
+ * Revision 1.17  1996-06-04 10:20:06  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.16  1996/05/24  14:46:04  adam
  * Added dict_grep_cmap function to define user-mapping in grep lookups.
  *
  * Revision 1.15  1996/03/20  09:35:18  adam
@@ -404,6 +407,9 @@ int dict_lookup_grep (Dict dict, const char *pattern, int range, void *client,
     int i, d;
 
     logf (LOG_DEBUG, "dict_lookup_grep '%s' range=%d", pattern, range);
+   
+    dfa_set_cmap (dfa, dict->grep_cmap);
+
     i = dfa_parse (dfa, &this_pattern);
     if (i || *this_pattern)
     {
@@ -445,7 +451,7 @@ int dict_lookup_grep (Dict dict, const char *pattern, int range, void *client,
     return i;
 }
 
-void dict_grep_cmap (Dict dict, char **(*cmap)(const char **from))
+void dict_grep_cmap (Dict dict, char **(*cmap)(const char **from, int len))
 {
     dict->grep_cmap = cmap;
 }
diff --git a/include/dfa.h b/include/dfa.h
index b609433..d4a4a44 100644
--- a/include/dfa.h
+++ b/include/dfa.h
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1994, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dfa.h,v $
- * Revision 1.5  1996-01-08 09:09:48  adam
+ * Revision 1.6  1996-06-04 10:20:10  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.5  1996/01/08  09:09:48  adam
  * Function dfa_parse got 'const' string argument.
  *
  * Revision 1.4  1995/01/25  11:31:04  adam
@@ -59,6 +62,8 @@ struct DFA {
 };
 
 struct DFA *dfa_init (void);
+void dfa_set_cmap (struct DFA *dfa,
+                   char **(*cmap)(const char **from, int len));
 int dfa_parse (struct DFA *, const char **);
 void dfa_mkstate (struct DFA *);
 void dfa_delete (struct DFA **);
diff --git a/include/dict.h b/include/dict.h
index 84b6e48..4cd048c 100644
--- a/include/dict.h
+++ b/include/dict.h
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1994, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dict.h,v $
- * Revision 1.21  1996-05-24 14:46:07  adam
+ * Revision 1.22  1996-06-04 10:20:10  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.21  1996/05/24  14:46:07  adam
  * Added dict_grep_cmap function to define user-mapping in grep lookups.
  *
  * Revision 1.20  1996/03/20  09:35:23  adam
@@ -121,7 +124,7 @@ typedef struct Dict_file_struct
 typedef struct Dict_struct {
     int rw;
     Dict_BFile dbf;
-    char **(*grep_cmap)(const char **from);
+    char **(*grep_cmap)(const char **from, int len);
     struct Dict_head head;
 } *Dict;
 
@@ -154,7 +157,8 @@ int	   dict_scan (Dict dict, char *str,
 		      int (*f)(char *name, const char *info, int pos,
                                void *client));
 
-void       dict_grep_cmap (Dict dict, char **(*cmap)(const char **from));
+void       dict_grep_cmap (Dict dict,
+                           char **(*cmap)(const char **from, int len));
 
 
 #define DICT_EOS        0