xref: /openbsd/regress/lib/libc/regex/debug.c (revision 7b36286a)
1 /*	$OpenBSD: debug.c,v 1.4 2003/07/31 21:48:03 deraadt Exp $	*/
2 /*	$NetBSD: debug.c,v 1.2 1995/04/20 22:39:42 cgd Exp $	*/
3 
4 #include <stdio.h>
5 #include <string.h>
6 #include <ctype.h>
7 #include <limits.h>
8 #include <stdlib.h>
9 #include <sys/types.h>
10 #include <regex.h>
11 
12 #include "utils.h"
13 #include "regex2.h"
14 #include "debug.ih"
15 
16 /*
17  - regprint - print a regexp for debugging
18  == void regprint(regex_t *r, FILE *d);
19  */
20 void
21 regprint(r, d)
22 regex_t *r;
23 FILE *d;
24 {
25 	register struct re_guts *g = r->re_g;
26 	register int i;
27 	register int c;
28 	register int last;
29 	int nincat[NC];
30 
31 	fprintf(d, "%ld states, %d categories", (long)g->nstates,
32 							g->ncategories);
33 	fprintf(d, ", first %ld last %ld", (long)g->firststate,
34 						(long)g->laststate);
35 	if (g->iflags&USEBOL)
36 		fprintf(d, ", USEBOL");
37 	if (g->iflags&USEEOL)
38 		fprintf(d, ", USEEOL");
39 	if (g->iflags&BAD)
40 		fprintf(d, ", BAD");
41 	if (g->nsub > 0)
42 		fprintf(d, ", nsub=%ld", (long)g->nsub);
43 	if (g->must != NULL)
44 		fprintf(d, ", must(%ld) `%*s'", (long)g->mlen, (int)g->mlen,
45 								g->must);
46 	if (g->backrefs)
47 		fprintf(d, ", backrefs");
48 	if (g->nplus > 0)
49 		fprintf(d, ", nplus %ld", (long)g->nplus);
50 	fprintf(d, "\n");
51 	s_print(g, d);
52 	for (i = 0; i < g->ncategories; i++) {
53 		nincat[i] = 0;
54 		for (c = CHAR_MIN; c <= CHAR_MAX; c++)
55 			if (g->categories[c] == i)
56 				nincat[i]++;
57 	}
58 	fprintf(d, "cc0#%d", nincat[0]);
59 	for (i = 1; i < g->ncategories; i++)
60 		if (nincat[i] == 1) {
61 			for (c = CHAR_MIN; c <= CHAR_MAX; c++)
62 				if (g->categories[c] == i)
63 					break;
64 			fprintf(d, ", %d=%s", i, regchar(c));
65 		}
66 	fprintf(d, "\n");
67 	for (i = 1; i < g->ncategories; i++)
68 		if (nincat[i] != 1) {
69 			fprintf(d, "cc%d\t", i);
70 			last = -1;
71 			for (c = CHAR_MIN; c <= CHAR_MAX+1; c++)	/* +1 does flush */
72 				if (c <= CHAR_MAX && g->categories[c] == i) {
73 					if (last < 0) {
74 						fprintf(d, "%s", regchar(c));
75 						last = c;
76 					}
77 				} else {
78 					if (last >= 0) {
79 						if (last != c-1)
80 							fprintf(d, "-%s",
81 								regchar(c-1));
82 						last = -1;
83 					}
84 				}
85 			fprintf(d, "\n");
86 		}
87 }
88 
89 /*
90  - s_print - print the strip for debugging
91  == static void s_print(register struct re_guts *g, FILE *d);
92  */
93 static void
94 s_print(g, d)
95 register struct re_guts *g;
96 FILE *d;
97 {
98 	register sop *s;
99 	register cset *cs;
100 	register int i;
101 	register int done = 0;
102 	register sop opnd;
103 	register int col = 0;
104 	register int last;
105 	register sopno offset = 2;
106 #	define	GAP()	{	if (offset % 5 == 0) { \
107 					if (col > 40) { \
108 						fprintf(d, "\n\t"); \
109 						col = 0; \
110 					} else { \
111 						fprintf(d, " "); \
112 						col++; \
113 					} \
114 				} else \
115 					col++; \
116 				offset++; \
117 			}
118 
119 	if (OP(g->strip[0]) != OEND)
120 		fprintf(d, "missing initial OEND!\n");
121 	for (s = &g->strip[1]; !done; s++) {
122 		opnd = OPND(*s);
123 		switch (OP(*s)) {
124 		case OEND:
125 			fprintf(d, "\n");
126 			done = 1;
127 			break;
128 		case OCHAR:
129 			if (strchr("\\|()^$.[+*?{}!<> ", (char)opnd) != NULL)
130 				fprintf(d, "\\%c", (char)opnd);
131 			else
132 				fprintf(d, "%s", regchar((char)opnd));
133 			break;
134 		case OBOL:
135 			fprintf(d, "^");
136 			break;
137 		case OEOL:
138 			fprintf(d, "$");
139 			break;
140 		case OBOW:
141 			fprintf(d, "\\{");
142 			break;
143 		case OEOW:
144 			fprintf(d, "\\}");
145 			break;
146 		case OANY:
147 			fprintf(d, ".");
148 			break;
149 		case OANYOF:
150 			fprintf(d, "[(%ld)", (long)opnd);
151 			cs = &g->sets[opnd];
152 			last = -1;
153 			for (i = 0; i < g->csetsize+1; i++)	/* +1 flushes */
154 				if (CHIN(cs, i) && i < g->csetsize) {
155 					if (last < 0) {
156 						fprintf(d, "%s", regchar(i));
157 						last = i;
158 					}
159 				} else {
160 					if (last >= 0) {
161 						if (last != i-1)
162 							fprintf(d, "-%s",
163 								regchar(i-1));
164 						last = -1;
165 					}
166 				}
167 			fprintf(d, "]");
168 			break;
169 		case OBACK_:
170 			fprintf(d, "(\\<%ld>", (long)opnd);
171 			break;
172 		case O_BACK:
173 			fprintf(d, "<%ld>\\)", (long)opnd);
174 			break;
175 		case OPLUS_:
176 			fprintf(d, "(+");
177 			if (OP(*(s+opnd)) != O_PLUS)
178 				fprintf(d, "<%ld>", (long)opnd);
179 			break;
180 		case O_PLUS:
181 			if (OP(*(s-opnd)) != OPLUS_)
182 				fprintf(d, "<%ld>", (long)opnd);
183 			fprintf(d, "+)");
184 			break;
185 		case OQUEST_:
186 			fprintf(d, "(?");
187 			if (OP(*(s+opnd)) != O_QUEST)
188 				fprintf(d, "<%ld>", (long)opnd);
189 			break;
190 		case O_QUEST:
191 			if (OP(*(s-opnd)) != OQUEST_)
192 				fprintf(d, "<%ld>", (long)opnd);
193 			fprintf(d, "?)");
194 			break;
195 		case OLPAREN:
196 			fprintf(d, "((<%ld>", (long)opnd);
197 			break;
198 		case ORPAREN:
199 			fprintf(d, "<%ld>))", (long)opnd);
200 			break;
201 		case OCH_:
202 			fprintf(d, "<");
203 			if (OP(*(s+opnd)) != OOR2)
204 				fprintf(d, "<%ld>", (long)opnd);
205 			break;
206 		case OOR1:
207 			if (OP(*(s-opnd)) != OOR1 && OP(*(s-opnd)) != OCH_)
208 				fprintf(d, "<%ld>", (long)opnd);
209 			fprintf(d, "|");
210 			break;
211 		case OOR2:
212 			fprintf(d, "|");
213 			if (OP(*(s+opnd)) != OOR2 && OP(*(s+opnd)) != O_CH)
214 				fprintf(d, "<%ld>", (long)opnd);
215 			break;
216 		case O_CH:
217 			if (OP(*(s-opnd)) != OOR1)
218 				fprintf(d, "<%ld>", (long)opnd);
219 			fprintf(d, ">");
220 			break;
221 		default:
222 			fprintf(d, "!%ld(%ld)!", (long)OP(*s), (long)opnd);
223 			break;
224 		}
225 		if (!done)
226 			GAP();
227 	}
228 }
229 
230 /*
231  - regchar - make a character printable
232  == static char *regchar(int ch);
233  */
234 static char *			/* -> representation */
235 regchar(ch)
236 int ch;
237 {
238 	static char buf[10];
239 
240 	if (isprint(ch) || ch == ' ')
241 		snprintf(buf, sizeof buf, "%c", ch);
242 	else
243 		snprintf(buf, sizeof buf, "\\%o", ch);
244 	return(buf);
245 }
246