Added facility for open character sets, eg [a-].

[idzebra-moved-to-github.git] / dfa / dfa.c

diff --git a/dfa/dfa.c b/dfa/dfa.c
index 4c7c864..199ebea 100644 (file)
--- a/dfa/dfa.c
+++ b/dfa/dfa.c
@@ -1,10 +1,22 @@
 /*
- * 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 1.15  1997-02-10 10:19:20  adam
+ * Added facility for open character sets, eg [a-].
+ *
+ * Revision 1.14  1996/10/29 13:57:22  adam
+ * Include of zebrautl.h instead of alexutil.h.
+ *
+ * Revision 1.13  1996/06/17 14:24:08  adam
+ * Bug fix: read_charset didn't handle character mapping.
+ *
+ * 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
@@ -48,10 +60,12 @@
 #include <string.h>
 #include <ctype.h>
 
-#include <alexutil.h>
+#include <zebrautl.h>
 #include "dfap.h"
 #include "imalloc.h"
 
+#define DFA_OPEN_RANGE 1
+
 #define CAT     16000
 #define OR      16001
 #define STAR    16002
@@ -402,20 +416,50 @@ static int read_charset (void)
     {
         if (!esc0 && ch0 == ']')
             break;
+        if (parse_info->cmap)
+        {
+            char **mapto, mapfrom[2];
+            const char *mcp = mapfrom;
+            mapfrom[0] = ch0;
+            mapto = (*parse_info->cmap)(&mcp, 1);
+            assert (mapto);
+            ch0 = mapto[0][0];
+        }
         add_BSet (parse_info->charset, look_chars, ch0);
         ch1 = nextchar_set (&esc1);
         if (!esc1 && ch1 == '-')
         {
+            int open_range = 0;
             if ((ch1 = nextchar_set (&esc1)) == 0)
                 break;
+#if DFA_OPEN_RANGE
+            if (!esc1 && ch1 == ']')
+            {
+                ch1 = 255;
+                open_range = 1;
+            }
+#else
             if (!esc1 && ch1 == ']')
             {
                 add_BSet (parse_info->charset, look_chars, '-');
                 break;
             }
+#endif
+            if (!open_range && parse_info->cmap)
+            {
+                char **mapto, mapfrom[2];
+                const char *mcp = mapfrom;
+                mapfrom[0] = ch1;
+                mapto = (*parse_info->cmap) (&mcp, 1);
+                assert (mapto);
+                ch1 = mapto[0][0];
+            }
             for (i=ch0; ++i<=ch1;)
                 add_BSet (parse_info->charset, look_chars, i);
-            ch0 = nextchar_set (&esc0);
+            if (!open_range)
+                ch0 = nextchar_set (&esc0);
+            else
+                break;
         }
         else
         {
@@ -428,6 +472,30 @@ static int read_charset (void)
     return L_CHARS;
 }
 
+static int map_l_char (void)
+{
+    char **mapto;
+    const char *cp0 = (const char *) (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 = (const unsigned char *) 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 +503,11 @@ 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 +520,7 @@ static int lex_sub(void)
                         --expr_ptr;
                     return cc[1];
                 }
-            return L_CHAR;            
+            return map_l_char ();     
         }
     return 0;
 }
@@ -978,6 +1046,7 @@ 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 +1101,11 @@ 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;