/*
- * 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 1.10 1996-01-08 09:09:17 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
* Function dfa_parse got 'const' string argument.
* New functions to define char mappings made public.
*
#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
return t1;
}
-static void do_parse (struct DFA_parse *dfap, const char **s, struct Tnode **tnp)
+static void do_parse (struct DFA_parse *dfap, const char **s,
+ struct Tnode **tnp)
{
- int anchor_flag = 0;
int start_anchor_flag = 0;
struct Tnode *t1, *t2, *tn;
parse_info = dfap;
err_code = 0;
- expr_ptr = (unsigned char *) *s;
+ expr_ptr = (const unsigned char *) *s;
inside_string = 0;
lex ();
start_anchor_flag = 1;
lex ();
}
- t1 = expr_1 ();
- if (anchor_flag)
+ if (lookahead == L_END)
{
- tn = mk_Tnode ();
- tn->pos = CAT;
- tn->u.p[0] = t2;
- tn->u.p[1] = t1;
- t1 = tn;
+ t1 = mk_Tnode ();
+ t1->pos = ++parse_info->position;
+ t1->u.ch[1] = t1->u.ch[0] = '\n';
+ lex ();
}
- if (lookahead == L_END && t1)
+ else
{
- t2 = mk_Tnode ();
- t2->pos = ++parse_info->position;
- t2->u.ch[1] = t2->u.ch[0] = '\n';
-
- tn = mk_Tnode ();
- tn->pos = CAT;
- tn->u.p[0] = t1;
- tn->u.p[1] = t2;
- t1 = tn;
-
- anchor_flag |= 2;
- lex ();
+ t1 = expr_1 ();
+ if (t1 && lookahead == L_END)
+ {
+ t2 = mk_Tnode ();
+ t2->pos = ++parse_info->position;
+ t2->u.ch[1] = t2->u.ch[0] = '\n';
+
+ tn = mk_Tnode ();
+ tn->pos = CAT;
+ tn->u.p[0] = t1;
+ tn->u.p[1] = t2;
+ t1 = tn;
+
+ lex ();
+ }
}
- if (lookahead == 0 && t1)
+ if (t1 && lookahead == 0)
{
t2 = mk_Tnode();
t2->pos = ++parse_info->position;
err_code = DFA_ERR_SYNTAX;
}
}
- *s = (char *) expr_ptr;
+ *s = (const char *) expr_ptr;
}
static int nextchar (int *esc)
{
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
{
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;
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
{
const int *cc;
- if (look_ch == '/')
- logf (LOG_DEBUG, "xxxx / xxx");
for (cc = parse_info->charMap; *cc; cc += 2)
if (*cc == look_ch)
{
--expr_ptr;
return cc[1];
}
- return L_CHAR;
+ return map_l_char ();
}
return 0;
}
parse_info->use_Tnode = parse_info->max_Tnode = 0;
parse_info->charMap = NULL;
parse_info->charMapSize = 0;
+ parse_info->cmap = NULL;
return parse_info;
}
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;
projects / idzebra-moved-to-github.git / blobdiff