1/*	$OpenBSD: main.c,v 1.12 2022/12/04 23:50:46 cheloha Exp $	*/
2/*	$NetBSD: main.c,v 1.2 1995/04/20 22:39:51 cgd Exp $	*/
3
4#include <stdio.h>
5#include <stdlib.h>
6#include <string.h>
7#include <sys/types.h>
8#include <regex.h>
9#include <assert.h>
10#include <unistd.h>
11
12#include "main.ih"
13
14char *progname;
15int debug = 0;
16int line = 0;
17int status = 0;
18
19int copts = REG_EXTENDED;
20int eopts = 0;
21regoff_t startoff = 0;
22regoff_t endoff = 0;
23
24
25extern int split(char *, char *[], int, char *);
26extern void regprint(regex_t *, FILE *);
27
28/*
29 - main - do the simple case, hand off to regress() for regression
30 */
31int
32main(int argc, char *argv[])
33
34{
35	regex_t re;
36#	define	NS	10
37	regmatch_t subs[NS];
38	char erbuf[100];
39	int err;
40	size_t len;
41	int c;
42	int errflg = 0;
43	register int i;
44
45	progname = argv[0];
46
47	while ((c = getopt(argc, argv, "c:E:e:S:x")) != -1)
48		switch (c) {
49		case 'c':	/* compile options */
50			copts = options('c', optarg);
51			break;
52		case 'E':	/* end offset */
53			endoff = (regoff_t)atoi(optarg);
54			break;
55		case 'e':	/* execute options */
56			eopts = options('e', optarg);
57			break;
58		case 'S':	/* start offset */
59			startoff = (regoff_t)atoi(optarg);
60			break;
61		case 'x':	/* Debugging. */
62			debug++;
63			break;
64		default:
65			errflg++;
66			break;
67		}
68	if (errflg) {
69		fprintf(stderr, "usage: %s ", progname);
70		fprintf(stderr, "[-x] [-c copt] [-E endoff] [-e eopt] [-S startoff] [re]\n");
71		exit(2);
72	}
73
74	if (optind >= argc) {
75		regress(stdin);
76		exit(status);
77	}
78
79	err = regcomp(&re, argv[optind++], copts);
80	if (err) {
81		len = regerror(err, &re, erbuf, sizeof(erbuf));
82		fprintf(stderr, "error %s, %zu/%zu `%s'\n",
83			eprint(err), len, sizeof(erbuf), erbuf);
84		exit(status);
85	}
86	regprint(&re, stdout);
87
88	if (optind >= argc) {
89		regfree(&re);
90		exit(status);
91	}
92
93	if (eopts&REG_STARTEND) {
94		subs[0].rm_so = startoff;
95		subs[0].rm_eo = strlen(argv[optind]) - endoff;
96	}
97	err = regexec(&re, argv[optind], (size_t)NS, subs, eopts);
98	if (err) {
99		len = regerror(err, &re, erbuf, sizeof(erbuf));
100		fprintf(stderr, "error %s, %zu/%zu `%s'\n",
101			eprint(err), len, sizeof(erbuf), erbuf);
102		exit(status);
103	}
104	if (!(copts&REG_NOSUB)) {
105		len = (size_t)(subs[0].rm_eo - subs[0].rm_so);
106		if (subs[0].rm_so != -1) {
107			if (len != 0)
108				printf("match `%.*s'\n", (int)len,
109					argv[optind] + subs[0].rm_so);
110			else
111				printf("match `'@%.1s\n",
112					argv[optind] + subs[0].rm_so);
113		}
114		for (i = 1; i < NS; i++)
115			if (subs[i].rm_so != -1)
116				printf("(%d) `%.*s'\n", i,
117					(int)(subs[i].rm_eo - subs[i].rm_so),
118					argv[optind] + subs[i].rm_so);
119	}
120	exit(status);
121}
122
123/*
124 - regress - main loop of regression test
125 == void regress(FILE *in);
126 */
127void
128regress(in)
129FILE *in;
130{
131	char inbuf[1000];
132#	define	MAXF	10
133	char *f[MAXF];
134	int nf;
135	int i;
136	char erbuf[100];
137	size_t ne;
138	char *badpat = "invalid regular expression";
139#	define	SHORT	10
140	char *bpname = "REG_BADPAT";
141	regex_t re;
142
143	while (fgets(inbuf, sizeof(inbuf), in) != NULL) {
144		line++;
145		if (inbuf[0] == '#' || inbuf[0] == '\n')
146			continue;			/* NOTE CONTINUE */
147		inbuf[strcspn(inbuf, "\n")] = '\0';	/* get rid of stupid \n */
148		if (debug)
149			fprintf(stdout, "%d:\n", line);
150		nf = split(inbuf, f, MAXF, "\t\t");
151		if (nf < 3) {
152			fprintf(stderr, "bad input, line %d\n", line);
153			exit(1);
154		}
155		for (i = 0; i < nf; i++)
156			if (strcmp(f[i], "\"\"") == 0)
157				f[i] = "";
158		if (nf <= 3)
159			f[3] = NULL;
160		if (nf <= 4)
161			f[4] = NULL;
162		try(f[0], f[1], f[2], f[3], f[4], options('c', f[1]));
163		if (opt('&', f[1]))	/* try with either type of RE */
164			try(f[0], f[1], f[2], f[3], f[4],
165					options('c', f[1]) &~ REG_EXTENDED);
166	}
167
168	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
169	if (strcmp(erbuf, badpat) != 0 || ne != strlen(badpat)+1) {
170		fprintf(stderr, "end: regerror() test gave `%s' not `%s'\n",
171							erbuf, badpat);
172		status = 1;
173	}
174	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, (size_t)SHORT);
175	if (strncmp(erbuf, badpat, SHORT-1) != 0 || erbuf[SHORT-1] != '\0' ||
176						ne != strlen(badpat)+1) {
177		fprintf(stderr, "end: regerror() short test gave `%s' not `%.*s'\n",
178						erbuf, SHORT-1, badpat);
179		status = 1;
180	}
181	ne = regerror(REG_ITOA|REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
182	if (strcmp(erbuf, bpname) != 0 || ne != strlen(bpname)+1) {
183		fprintf(stderr, "end: regerror() ITOA test gave `%s' not `%s'\n",
184						erbuf, bpname);
185		status = 1;
186	}
187	re.re_endp = bpname;
188	ne = regerror(REG_ATOI, &re, erbuf, sizeof(erbuf));
189	if (atoi(erbuf) != (int)REG_BADPAT) {
190		fprintf(stderr, "end: regerror() ATOI test gave `%s' not `%ld'\n",
191						erbuf, (long)REG_BADPAT);
192		status = 1;
193	} else if (ne != strlen(erbuf)+1) {
194		fprintf(stderr, "end: regerror() ATOI test len(`%s') = %ld\n",
195						erbuf, (long)REG_BADPAT);
196		status = 1;
197	}
198}
199
200/*
201 - try - try it, and report on problems
202 == void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts);
203 */
204void
205try(f0, f1, f2, f3, f4, opts)
206char *f0;
207char *f1;
208char *f2;
209char *f3;
210char *f4;
211int opts;			/* may not match f1 */
212{
213	regex_t re;
214#	define	NSUBS	10
215	regmatch_t subs[NSUBS];
216#	define	NSHOULD	15
217	char *should[NSHOULD];
218	int nshould;
219	char erbuf[100];
220	int err;
221	int len;
222	char *type = (opts & REG_EXTENDED) ? "ERE" : "BRE";
223	register int i;
224	char *grump;
225	char f0copy[1000];
226	char f2copy[1000];
227
228	strlcpy(f0copy, f0, sizeof f0copy);
229	re.re_endp = (opts&REG_PEND) ? f0copy + strlen(f0copy) : NULL;
230	fixstr(f0copy);
231	err = regcomp(&re, f0copy, opts);
232	if (err != 0 && (!opt('C', f1) || err != efind(f2))) {
233		/* unexpected error or wrong error */
234		len = regerror(err, &re, erbuf, sizeof(erbuf));
235		fprintf(stderr, "%d: %s error %s, %d/%zu `%s'\n",
236					line, type, eprint(err), len,
237					sizeof(erbuf), erbuf);
238		status = 1;
239	} else if (err == 0 && opt('C', f1)) {
240		/* unexpected success */
241		fprintf(stderr, "%d: %s should have given REG_%s\n",
242						line, type, f2);
243		status = 1;
244		err = 1;	/* so we won't try regexec */
245	}
246
247	if (err != 0) {
248		regfree(&re);
249		return;
250	}
251
252	strlcpy(f2copy, f2, sizeof f2copy);
253	fixstr(f2copy);
254
255	if (options('e', f1)&REG_STARTEND) {
256		if (strchr(f2, '(') == NULL || strchr(f2, ')') == NULL)
257			fprintf(stderr, "%d: bad STARTEND syntax\n", line);
258		subs[0].rm_so = strchr(f2, '(') - f2 + 1;
259		subs[0].rm_eo = strchr(f2, ')') - f2;
260		/* the preceding character is relevant with REG_NOTBOL */
261		f2copy[subs[0].rm_so - 1] = subs[0].rm_so > 1 ?
262		    f2copy[subs[0].rm_so - 2] : 'X';
263	}
264	err = regexec(&re, f2copy, NSUBS, subs, options('e', f1));
265
266	if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) {
267		/* unexpected error or wrong error */
268		len = regerror(err, &re, erbuf, sizeof(erbuf));
269		fprintf(stderr, "%d: %s exec error %s, %d/%zu `%s'\n",
270					line, type, eprint(err), len,
271					sizeof(erbuf), erbuf);
272		status = 1;
273	} else if (err != 0) {
274		/* nothing more to check */
275	} else if (f3 == NULL) {
276		/* unexpected success */
277		fprintf(stderr, "%d: %s exec should have failed\n",
278						line, type);
279		status = 1;
280		err = 1;		/* just on principle */
281	} else if (opts&REG_NOSUB) {
282		/* nothing more to check */
283	} else if ((grump = check(f2, subs[0], f3)) != NULL) {
284		fprintf(stderr, "%d: %s %s\n", line, type, grump);
285		status = 1;
286		err = 1;
287	}
288
289	if (err != 0 || f4 == NULL) {
290		regfree(&re);
291		return;
292	}
293
294	for (i = 1; i < NSHOULD; i++)
295		should[i] = NULL;
296	nshould = split(f4, should+1, NSHOULD-1, ",");
297	if (nshould == 0) {
298		nshould = 1;
299		should[1] = "";
300	}
301	for (i = 1; i < NSUBS; i++) {
302		grump = check(f2, subs[i], should[i]);
303		if (grump != NULL) {
304			fprintf(stderr, "%d: %s $%d %s\n", line,
305							type, i, grump);
306			status = 1;
307			err = 1;
308		}
309	}
310
311	regfree(&re);
312}
313
314/*
315 - options - pick options out of a regression-test string
316 == int options(int type, char *s);
317 */
318int
319options(type, s)
320int type;			/* 'c' compile, 'e' exec */
321char *s;
322{
323	register char *p;
324	register int o = (type == 'c') ? copts : eopts;
325	register char *legal = (type == 'c') ? "bisnmp" : "^$#tl";
326
327	for (p = s; *p != '\0'; p++)
328		if (strchr(legal, *p) != NULL)
329			switch (*p) {
330			case 'b':
331				o &= ~REG_EXTENDED;
332				break;
333			case 'i':
334				o |= REG_ICASE;
335				break;
336			case 's':
337				o |= REG_NOSUB;
338				break;
339			case 'n':
340				o |= REG_NEWLINE;
341				break;
342			case 'm':
343				o &= ~REG_EXTENDED;
344				o |= REG_NOSPEC;
345				break;
346			case 'p':
347				o |= REG_PEND;
348				break;
349			case '^':
350				o |= REG_NOTBOL;
351				break;
352			case '$':
353				o |= REG_NOTEOL;
354				break;
355			case '#':
356				o |= REG_STARTEND;
357				break;
358			case 't':	/* trace */
359				o |= REG_TRACE;
360				break;
361			case 'l':	/* force long representation */
362				o |= REG_LARGE;
363				break;
364			case 'r':	/* force backref use */
365				o |= REG_BACKR;
366				break;
367			}
368	return(o);
369}
370
371/*
372 - opt - is a particular option in a regression string?
373 == int opt(int c, char *s);
374 */
375int				/* predicate */
376opt(c, s)
377int c;
378char *s;
379{
380	return(strchr(s, c) != NULL);
381}
382
383/*
384 - fixstr - transform magic characters in strings
385 == void fixstr(register char *p);
386 */
387void
388fixstr(p)
389register char *p;
390{
391	if (p == NULL)
392		return;
393
394	for (; *p != '\0'; p++)
395		if (*p == 'N')
396			*p = '\n';
397		else if (*p == 'T')
398			*p = '\t';
399		else if (*p == 'S')
400			*p = ' ';
401		else if (*p == 'Z')
402			*p = '\0';
403}
404
405/*
406 - check - check a substring match
407 == char *check(char *str, regmatch_t sub, char *should);
408 */
409char *				/* NULL or complaint */
410check(str, sub, should)
411char *str;
412regmatch_t sub;
413char *should;
414{
415	register int len;
416	register int shlen;
417	register char *p;
418	static char grump[500];
419	register char *at = NULL;
420
421	if (should != NULL && strcmp(should, "-") == 0)
422		should = NULL;
423	if (should != NULL && should[0] == '@') {
424		at = should + 1;
425		should = "";
426	}
427
428	/* check rm_so and rm_eo for consistency */
429	if (sub.rm_so > sub.rm_eo || (sub.rm_so == -1 && sub.rm_eo != -1) ||
430				(sub.rm_so != -1 && sub.rm_eo == -1) ||
431				(sub.rm_so != -1 && sub.rm_so < 0) ||
432				(sub.rm_eo != -1 && sub.rm_eo < 0) ) {
433		snprintf(grump, sizeof grump,
434		    "start %ld end %ld", (long)sub.rm_so,
435		    (long)sub.rm_eo);
436		return(grump);
437	}
438
439	/* check for no match */
440	if (sub.rm_so == -1 && should == NULL)
441		return(NULL);
442	if (sub.rm_so == -1)
443		return("did not match");
444
445	/* check for in range */
446	if (sub.rm_eo > strlen(str)) {
447		snprintf(grump, sizeof grump,
448			"start %ld end %ld, past end of string",
449			(long)sub.rm_so, (long)sub.rm_eo);
450		return(grump);
451	}
452
453	len = (int)(sub.rm_eo - sub.rm_so);
454	p = str + sub.rm_so;
455
456	/* check for not supposed to match */
457	if (should == NULL) {
458		snprintf(grump, sizeof grump, "matched `%.*s'", len, p);
459		return(grump);
460	}
461
462	/* check for wrong match */
463	shlen = (int)strlen(should);
464	if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) {
465		snprintf(grump, sizeof grump, "matched `%.*s' instead", len, p);
466		return(grump);
467	}
468	if (shlen > 0)
469		return(NULL);
470
471	/* check null match in right place */
472	if (at == NULL)
473		return(NULL);
474	shlen = strlen(at);
475	if (shlen == 0)
476		shlen = 1;	/* force check for end-of-string */
477	if (strncmp(p, at, shlen) != 0) {
478		snprintf(grump, sizeof grump, "matched null at `%.20s'", p);
479		return(grump);
480	}
481	return(NULL);
482}
483
484/*
485 - eprint - convert error number to name
486 == static char *eprint(int err);
487 */
488static char *
489eprint(err)
490int err;
491{
492	static char epbuf[100];
493	size_t len;
494
495	len = regerror(REG_ITOA|err, (regex_t *)NULL, epbuf, sizeof(epbuf));
496	assert(len <= sizeof(epbuf));
497	return(epbuf);
498}
499
500/*
501 - efind - convert error name to number
502 == static int efind(char *name);
503 */
504static int
505efind(name)
506char *name;
507{
508	static char efbuf[100];
509	regex_t re;
510
511	snprintf(efbuf, sizeof efbuf, "REG_%s", name);
512	assert(strlen(efbuf) < sizeof(efbuf));
513	re.re_endp = efbuf;
514	(void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf));
515	return(atoi(efbuf));
516}
517