New berbatim hook in regular expressions. "[]n ..".

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

diff --git a/dfa/dfa.c b/dfa/dfa.c
index 0104852..368d275 100644 (file)
--- a/dfa/dfa.c
+++ b/dfa/dfa.c
@@ -4,7 +4,22 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dfa.c,v $
- * Revision 1.2  1995-01-25 11:30:50  adam
+ * Revision 1.7  1995-11-27 09:23:02  adam
+ * New berbatim hook in regular expressions. "[]n ..".
+ *
+ * Revision 1.6  1995/10/16  09:31:25  adam
+ * Bug fix.
+ *
+ * Revision 1.5  1995/10/02  15:17:58  adam
+ * Bug fix in dfa_delete.
+ *
+ * Revision 1.4  1995/09/28  09:18:52  adam
+ * Removed various preprocessor defines.
+ *
+ * Revision 1.3  1995/09/04  12:33:26  adam
+ * Various cleanup. YAZ util used instead.
+ *
+ * Revision 1.2  1995/01/25  11:30:50  adam
  * Simple error reporting when parsing regular expressions.
  * Memory usage reduced.
  *
@@ -20,7 +35,7 @@
 #include <string.h>
 #include <ctype.h>
 
-#include <util.h>
+#include <alexutil.h>
 #include "dfap.h"
 #include "imalloc.h"
 
@@ -57,13 +72,13 @@ int debug_dfa_trav  = 0;
 int debug_dfa_tran  = 0;
 int debug_dfa_followpos = 0;
 int dfa_verbose = 0;
-int yydebug = 0;
 
 static struct DFA_parse *parse_info = NULL;
 
 static int err_code;
 static int inside_string;
 static const unsigned char *expr_ptr;
+static int expr_verbatim;
 static unsigned short *ctrl_chars;
 static struct Tnode **posar;
 
@@ -305,6 +320,7 @@ struct Tnode **tnp;
     parse_info = dfap;
     err_code = 0;
     expr_ptr = (unsigned char *) *s;
+    expr_verbatim = 0;
 
     inside_string = 0;
     lex ();
@@ -341,8 +357,30 @@ static int nextchar (int *esc)
     *esc = 0;
     if (*expr_ptr == '\0' || isspace(*expr_ptr))
         return 0;
-    else if (*expr_ptr != '\\')
+    else if (*expr_ptr != '\\' || expr_verbatim)
+    {
+        if (*expr_ptr == '[' && expr_ptr[1] == ']' && !expr_verbatim)
+        {
+            int i = 2;
+            int val = 0;
+            while (expr_ptr[i] >= '0' && expr_ptr[i] <= '9')
+                val = val*10 + expr_ptr[i++]-'0';
+            if (i > 2)
+            {
+                if (expr_ptr[i] == ' ')
+                    i++;
+                expr_verbatim = val;
+                expr_ptr += i;
+            }
+        }
+        if (expr_verbatim)
+        {
+            assert (expr_verbatim > 0);
+            *esc = 1;
+            --expr_verbatim;
+        }
         return *expr_ptr++;
+    }
     *esc = 1;
     switch (*++expr_ptr)
     {
@@ -459,7 +497,7 @@ static int lex_sub(void)
             return read_charset();
         else if (look_ch == ' ')
             return 0;
-        else
+        else 
         {
             for (cc = ctrl_chars; *cc; cc += 2)
                 if (*cc == look_ch)
@@ -755,27 +793,28 @@ static void mk_dfa_tran (struct DFA_states *dfas)
         {
             char_0 = max_char+1;
             for (pos_i = pos; (i = *pos_i) != -1; ++pos_i)
-                if (posar[i]->u.ch[1] >= char_1)
-                    if ((c=posar[i]->u.ch[0]) < char_0)
-                        if (c < char_1)
-                            char_0 = char_1;
-                        else
-                            char_0 = c;
+                if (posar[i]->u.ch[1] >= char_1 
+                    && (c=posar[i]->u.ch[0]) < char_0)
+                    if (c < char_1)
+                        char_0 = char_1;
+                    else
+                        char_0 = c;
 
-            char_1 = max_char;
             if (char_0 > max_char)
                 break;
+
+            char_1 = max_char;
+                
             tran_set = mk_Set (poset);
             for (pos_i = pos; (i = *pos_i) != -1; ++pos_i)
-                if ((c=posar[i]->u.ch[1]) >= char_0)
-                    if (posar[i]->u.ch[0] <= char_0)
-                    {
-                        if (c < char_1)
-                            char_1 = c;
-                        tran_set = union_Set (poset, tran_set, followpos[i]);
-                    }
-                    else if (c <= char_1)
-                        char_1 = c-1;
+            {
+                if ((c=posar[i]->u.ch[0]) > char_0 && c <= char_1)
+                    char_1 = c - 1;                /* forward chunk */
+                else if ((c=posar[i]->u.ch[1]) >= char_0 && c < char_1)
+                    char_1 = c;                    /* backward chunk */
+                if (posar[i]->u.ch[1] >= char_0 && posar[i]->u.ch[0] <= char_0)
+                    tran_set = union_Set (poset, tran_set, followpos[i]);
+            }
             if (tran_set)
             {
                 add_DFA_state (dfas, &tran_set, &dfa_to);
@@ -967,7 +1006,8 @@ void dfa_delete (struct DFA **dfap)
     assert (*dfap);
     if ((*dfap)->parse_info)
         rm_dfa_parse (&(*dfap)->parse_info);
-    rm_DFA_states (&(*dfap)->state_info);
+    if ((*dfap)->state_info)
+        rm_DFA_states (&(*dfap)->state_info);
     ifree (*dfap);
     *dfap = NULL;
 }