192986Sobrien#include <sys/cdefs.h>
292986Sobrien__FBSDID("$FreeBSD$");
392986Sobrien
4292144Sngie#include <sys/types.h>
5292144Sngie#include <assert.h>
6292144Sngie#include <regex.h>
762856Sdcs#include <stdio.h>
8292144Sngie#include <stdlib.h>
962856Sdcs#include <string.h>
10292144Sngie#include <unistd.h>
1162856Sdcs
12292170Sngie#include "debug.ih"
1362856Sdcs#include "main.ih"
14292170Sngie#include "split.ih"
1562856Sdcs
1662856Sdcschar *progname;
1762856Sdcsint debug = 0;
1862856Sdcsint line = 0;
1962856Sdcsint status = 0;
2062856Sdcs
2162856Sdcsint copts = REG_EXTENDED;
2262856Sdcsint eopts = 0;
2362856Sdcsregoff_t startoff = 0;
2462856Sdcsregoff_t endoff = 0;
2562856Sdcs
2662856Sdcs
2762856Sdcs/*
2862856Sdcs - main - do the simple case, hand off to regress() for regression
2962856Sdcs */
30292144Sngieint
31292144Sngiemain(int argc, char **argv)
3262856Sdcs{
3362856Sdcs	regex_t re;
3462856Sdcs#	define	NS	10
3562856Sdcs	regmatch_t subs[NS];
3662856Sdcs	char erbuf[100];
3762856Sdcs	int err;
3862856Sdcs	size_t len;
3962856Sdcs	int c;
4062856Sdcs	int errflg = 0;
4192889Sobrien	int i;
4262856Sdcs	extern int optind;
4362856Sdcs	extern char *optarg;
4462856Sdcs
4562856Sdcs	progname = argv[0];
4662856Sdcs
47176380Skevlo	while ((c = getopt(argc, argv, "c:e:S:E:x")) != -1)
4862856Sdcs		switch (c) {
4962856Sdcs		case 'c':	/* compile options */
5062856Sdcs			copts = options('c', optarg);
5162856Sdcs			break;
5262856Sdcs		case 'e':	/* execute options */
5362856Sdcs			eopts = options('e', optarg);
5462856Sdcs			break;
5562856Sdcs		case 'S':	/* start offset */
5662856Sdcs			startoff = (regoff_t)atoi(optarg);
5762856Sdcs			break;
5862856Sdcs		case 'E':	/* end offset */
5962856Sdcs			endoff = (regoff_t)atoi(optarg);
6062856Sdcs			break;
6162856Sdcs		case 'x':	/* Debugging. */
6262856Sdcs			debug++;
6362856Sdcs			break;
6462856Sdcs		case '?':
6562856Sdcs		default:
6662856Sdcs			errflg++;
6762856Sdcs			break;
6862856Sdcs		}
6962856Sdcs	if (errflg) {
7062856Sdcs		fprintf(stderr, "usage: %s ", progname);
7162856Sdcs		fprintf(stderr, "[-c copt][-C][-d] [re]\n");
7262856Sdcs		exit(2);
7362856Sdcs	}
7462856Sdcs
7562856Sdcs	if (optind >= argc) {
7662856Sdcs		regress(stdin);
7762856Sdcs		exit(status);
7862856Sdcs	}
7962856Sdcs
8062856Sdcs	err = regcomp(&re, argv[optind++], copts);
8162856Sdcs	if (err) {
8262856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
83292144Sngie		fprintf(stderr, "error %s, %zu/%zu `%s'\n",
84292144Sngie		    eprint(err), len, sizeof(erbuf), erbuf);
8562856Sdcs		exit(status);
8662856Sdcs	}
87292144Sngie	regprint(&re, stdout);
8862856Sdcs
8962856Sdcs	if (optind >= argc) {
9062856Sdcs		regfree(&re);
9162856Sdcs		exit(status);
9262856Sdcs	}
9362856Sdcs
94292144Sngie	if ((eopts & REG_STARTEND) != 0) {
9562856Sdcs		subs[0].rm_so = startoff;
9662856Sdcs		subs[0].rm_eo = strlen(argv[optind]) - endoff;
9762856Sdcs	}
9862856Sdcs	err = regexec(&re, argv[optind], (size_t)NS, subs, eopts);
9962856Sdcs	if (err) {
10062856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
101292144Sngie		fprintf(stderr, "error %s, %zu/%zu `%s'\n",
102292144Sngie		    eprint(err), len, sizeof(erbuf), erbuf);
10362856Sdcs		exit(status);
10462856Sdcs	}
105292144Sngie	if ((copts & REG_NOSUB) == 0) {
10662856Sdcs		len = (int)(subs[0].rm_eo - subs[0].rm_so);
10762856Sdcs		if (subs[0].rm_so != -1) {
10862856Sdcs			if (len != 0)
109292144Sngie				printf("match `%.*s'\n", (int)len,
110292144Sngie				    argv[optind] + subs[0].rm_so);
11162856Sdcs			else
11262856Sdcs				printf("match `'@%.1s\n",
113292144Sngie				    argv[optind] + subs[0].rm_so);
11462856Sdcs		}
11562856Sdcs		for (i = 1; i < NS; i++)
11662856Sdcs			if (subs[i].rm_so != -1)
11762856Sdcs				printf("(%d) `%.*s'\n", i,
118292144Sngie				    (int)(subs[i].rm_eo - subs[i].rm_so),
119292144Sngie				    argv[optind] + subs[i].rm_so);
12062856Sdcs	}
12162856Sdcs	exit(status);
12262856Sdcs}
12362856Sdcs
12462856Sdcs/*
12562856Sdcs - regress - main loop of regression test
12662856Sdcs == void regress(FILE *in);
12762856Sdcs */
12862856Sdcsvoid
129292144Sngieregress(FILE *in)
13062856Sdcs{
13162856Sdcs	char inbuf[1000];
13262856Sdcs#	define	MAXF	10
13362856Sdcs	char *f[MAXF];
13462856Sdcs	int nf;
13562856Sdcs	int i;
13662856Sdcs	char erbuf[100];
13762856Sdcs	size_t ne;
13862856Sdcs	char *badpat = "invalid regular expression";
13962856Sdcs#	define	SHORT	10
14062856Sdcs	char *bpname = "REG_BADPAT";
14162856Sdcs	regex_t re;
14262856Sdcs
14362856Sdcs	while (fgets(inbuf, sizeof(inbuf), in) != NULL) {
14462856Sdcs		line++;
14562856Sdcs		if (inbuf[0] == '#' || inbuf[0] == '\n')
14662856Sdcs			continue;			/* NOTE CONTINUE */
14762856Sdcs		inbuf[strlen(inbuf)-1] = '\0';	/* get rid of stupid \n */
14862856Sdcs		if (debug)
14962856Sdcs			fprintf(stdout, "%d:\n", line);
15062856Sdcs		nf = split(inbuf, f, MAXF, "\t\t");
15162856Sdcs		if (nf < 3) {
15262856Sdcs			fprintf(stderr, "bad input, line %d\n", line);
15362856Sdcs			exit(1);
15462856Sdcs		}
15562856Sdcs		for (i = 0; i < nf; i++)
15662856Sdcs			if (strcmp(f[i], "\"\"") == 0)
15762856Sdcs				f[i] = "";
15862856Sdcs		if (nf <= 3)
15962856Sdcs			f[3] = NULL;
16062856Sdcs		if (nf <= 4)
16162856Sdcs			f[4] = NULL;
16262856Sdcs		try(f[0], f[1], f[2], f[3], f[4], options('c', f[1]));
16362856Sdcs		if (opt('&', f[1]))	/* try with either type of RE */
16462856Sdcs			try(f[0], f[1], f[2], f[3], f[4],
16562856Sdcs					options('c', f[1]) &~ REG_EXTENDED);
16662856Sdcs	}
16762856Sdcs
16862856Sdcs	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
16962856Sdcs	if (strcmp(erbuf, badpat) != 0 || ne != strlen(badpat)+1) {
17062856Sdcs		fprintf(stderr, "end: regerror() test gave `%s' not `%s'\n",
17162856Sdcs							erbuf, badpat);
17262856Sdcs		status = 1;
17362856Sdcs	}
17462856Sdcs	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, (size_t)SHORT);
17562856Sdcs	if (strncmp(erbuf, badpat, SHORT-1) != 0 || erbuf[SHORT-1] != '\0' ||
176292144Sngie	    ne != strlen(badpat)+1) {
17762856Sdcs		fprintf(stderr, "end: regerror() short test gave `%s' not `%.*s'\n",
17862856Sdcs						erbuf, SHORT-1, badpat);
17962856Sdcs		status = 1;
18062856Sdcs	}
18162856Sdcs	ne = regerror(REG_ITOA|REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
182292144Sngie	if (strcmp(erbuf, bpname) != 0 || ne != strlen(bpname) + 1) {
18362856Sdcs		fprintf(stderr, "end: regerror() ITOA test gave `%s' not `%s'\n",
18462856Sdcs						erbuf, bpname);
18562856Sdcs		status = 1;
18662856Sdcs	}
18762856Sdcs	re.re_endp = bpname;
18862856Sdcs	ne = regerror(REG_ATOI, &re, erbuf, sizeof(erbuf));
18962856Sdcs	if (atoi(erbuf) != (int)REG_BADPAT) {
19062856Sdcs		fprintf(stderr, "end: regerror() ATOI test gave `%s' not `%ld'\n",
19162856Sdcs						erbuf, (long)REG_BADPAT);
19262856Sdcs		status = 1;
193292144Sngie	} else if (ne != strlen(erbuf) + 1) {
19462856Sdcs		fprintf(stderr, "end: regerror() ATOI test len(`%s') = %ld\n",
19562856Sdcs						erbuf, (long)REG_BADPAT);
19662856Sdcs		status = 1;
19762856Sdcs	}
19862856Sdcs}
19962856Sdcs
20062856Sdcs/*
20162856Sdcs - try - try it, and report on problems
20262856Sdcs == void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts);
203292144Sngie - opts: may not match f1
20462856Sdcs */
20562856Sdcsvoid
206292144Sngietry(char *f0, char *f1, char *f2, char *f3, char *f4, int opts)
20762856Sdcs{
20862856Sdcs	regex_t re;
20962856Sdcs#	define	NSUBS	10
21062856Sdcs	regmatch_t subs[NSUBS];
21162856Sdcs#	define	NSHOULD	15
21262856Sdcs	char *should[NSHOULD];
21362856Sdcs	char erbuf[100];
214292144Sngie	size_t len;
215292144Sngie	int err, i, nshould;
216292144Sngie	char *grump;
21762856Sdcs	char *type = (opts & REG_EXTENDED) ? "ERE" : "BRE";
21862856Sdcs	char f0copy[1000];
21962856Sdcs	char f2copy[1000];
22062856Sdcs
22162856Sdcs	strcpy(f0copy, f0);
22262856Sdcs	re.re_endp = (opts&REG_PEND) ? f0copy + strlen(f0copy) : NULL;
22362856Sdcs	fixstr(f0copy);
22462856Sdcs	err = regcomp(&re, f0copy, opts);
22562856Sdcs	if (err != 0 && (!opt('C', f1) || err != efind(f2))) {
22662856Sdcs		/* unexpected error or wrong error */
22762856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
228292144Sngie		fprintf(stderr, "%d: %s error %s, %zu/%zu `%s'\n",
229292144Sngie		    line, type, eprint(err), len, sizeof(erbuf), erbuf);
23062856Sdcs		status = 1;
23162856Sdcs	} else if (err == 0 && opt('C', f1)) {
23262856Sdcs		/* unexpected success */
23362856Sdcs		fprintf(stderr, "%d: %s should have given REG_%s\n",
23462856Sdcs						line, type, f2);
23562856Sdcs		status = 1;
23662856Sdcs		err = 1;	/* so we won't try regexec */
23762856Sdcs	}
23862856Sdcs
23962856Sdcs	if (err != 0) {
24062856Sdcs		regfree(&re);
24162856Sdcs		return;
24262856Sdcs	}
24362856Sdcs
24462856Sdcs	strcpy(f2copy, f2);
24562856Sdcs	fixstr(f2copy);
24662856Sdcs
24762856Sdcs	if (options('e', f1)&REG_STARTEND) {
24862856Sdcs		if (strchr(f2, '(') == NULL || strchr(f2, ')') == NULL)
24962856Sdcs			fprintf(stderr, "%d: bad STARTEND syntax\n", line);
25062856Sdcs		subs[0].rm_so = strchr(f2, '(') - f2 + 1;
25162856Sdcs		subs[0].rm_eo = strchr(f2, ')') - f2;
25262856Sdcs	}
25362856Sdcs	err = regexec(&re, f2copy, NSUBS, subs, options('e', f1));
25462856Sdcs
25562856Sdcs	if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) {
25662856Sdcs		/* unexpected error or wrong error */
25762856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
258292144Sngie		fprintf(stderr, "%d: %s exec error %s, %zu/%zu `%s'\n",
259292144Sngie		    line, type, eprint(err), len, sizeof(erbuf), erbuf);
26062856Sdcs		status = 1;
26162856Sdcs	} else if (err != 0) {
26262856Sdcs		/* nothing more to check */
26362856Sdcs	} else if (f3 == NULL) {
26462856Sdcs		/* unexpected success */
26562856Sdcs		fprintf(stderr, "%d: %s exec should have failed\n",
266292144Sngie		    line, type);
26762856Sdcs		status = 1;
26862856Sdcs		err = 1;		/* just on principle */
26962856Sdcs	} else if (opts&REG_NOSUB) {
27062856Sdcs		/* nothing more to check */
27162856Sdcs	} else if ((grump = check(f2, subs[0], f3)) != NULL) {
27262856Sdcs		fprintf(stderr, "%d: %s %s\n", line, type, grump);
27362856Sdcs		status = 1;
27462856Sdcs		err = 1;
27562856Sdcs	}
27662856Sdcs
27762856Sdcs	if (err != 0 || f4 == NULL) {
27862856Sdcs		regfree(&re);
27962856Sdcs		return;
28062856Sdcs	}
28162856Sdcs
28262856Sdcs	for (i = 1; i < NSHOULD; i++)
28362856Sdcs		should[i] = NULL;
28462856Sdcs	nshould = split(f4, should+1, NSHOULD-1, ",");
28562856Sdcs	if (nshould == 0) {
28662856Sdcs		nshould = 1;
28762856Sdcs		should[1] = "";
28862856Sdcs	}
28962856Sdcs	for (i = 1; i < NSUBS; i++) {
29062856Sdcs		grump = check(f2, subs[i], should[i]);
29162856Sdcs		if (grump != NULL) {
29262856Sdcs			fprintf(stderr, "%d: %s $%d %s\n", line,
293292144Sngie			    type, i, grump);
29462856Sdcs			status = 1;
29562856Sdcs			err = 1;
29662856Sdcs		}
29762856Sdcs	}
29862856Sdcs
29962856Sdcs	regfree(&re);
30062856Sdcs}
30162856Sdcs
30262856Sdcs/*
30362856Sdcs - options - pick options out of a regression-test string
304292144Sngie - type: 'c' - compile, 'e' - exec
30562856Sdcs == int options(int type, char *s);
30662856Sdcs */
30762856Sdcsint
308292144Sngieoptions(int type, char *s)
30962856Sdcs{
31092889Sobrien	char *p;
31192889Sobrien	int o = (type == 'c') ? copts : eopts;
31292889Sobrien	char *legal = (type == 'c') ? "bisnmp" : "^$#tl";
31362856Sdcs
31462856Sdcs	for (p = s; *p != '\0'; p++)
31562856Sdcs		if (strchr(legal, *p) != NULL)
31662856Sdcs			switch (*p) {
31762856Sdcs			case 'b':
31862856Sdcs				o &= ~REG_EXTENDED;
31962856Sdcs				break;
32062856Sdcs			case 'i':
32162856Sdcs				o |= REG_ICASE;
32262856Sdcs				break;
32362856Sdcs			case 's':
32462856Sdcs				o |= REG_NOSUB;
32562856Sdcs				break;
32662856Sdcs			case 'n':
32762856Sdcs				o |= REG_NEWLINE;
32862856Sdcs				break;
32962856Sdcs			case 'm':
33062856Sdcs				o &= ~REG_EXTENDED;
33162856Sdcs				o |= REG_NOSPEC;
33262856Sdcs				break;
33362856Sdcs			case 'p':
33462856Sdcs				o |= REG_PEND;
33562856Sdcs				break;
33662856Sdcs			case '^':
33762856Sdcs				o |= REG_NOTBOL;
33862856Sdcs				break;
33962856Sdcs			case '$':
34062856Sdcs				o |= REG_NOTEOL;
34162856Sdcs				break;
34262856Sdcs			case '#':
34362856Sdcs				o |= REG_STARTEND;
34462856Sdcs				break;
34562856Sdcs			case 't':	/* trace */
34662856Sdcs				o |= REG_TRACE;
34762856Sdcs				break;
34862856Sdcs			case 'l':	/* force long representation */
34962856Sdcs				o |= REG_LARGE;
35062856Sdcs				break;
35162856Sdcs			case 'r':	/* force backref use */
35262856Sdcs				o |= REG_BACKR;
35362856Sdcs				break;
35462856Sdcs			}
35562856Sdcs	return(o);
35662856Sdcs}
35762856Sdcs
35862856Sdcs/*
35962856Sdcs - opt - is a particular option in a regression string?
36062856Sdcs == int opt(int c, char *s);
36162856Sdcs */
36262856Sdcsint				/* predicate */
363292144Sngieopt(int c, char *s)
36462856Sdcs{
36562856Sdcs	return(strchr(s, c) != NULL);
36662856Sdcs}
36762856Sdcs
36862856Sdcs/*
36962856Sdcs - fixstr - transform magic characters in strings
37092889Sobrien == void fixstr(char *p);
37162856Sdcs */
37262856Sdcsvoid
373292144Sngiefixstr(char *p)
37462856Sdcs{
37562856Sdcs	if (p == NULL)
37662856Sdcs		return;
37762856Sdcs
37862856Sdcs	for (; *p != '\0'; p++)
37962856Sdcs		if (*p == 'N')
38062856Sdcs			*p = '\n';
38162856Sdcs		else if (*p == 'T')
38262856Sdcs			*p = '\t';
38362856Sdcs		else if (*p == 'S')
38462856Sdcs			*p = ' ';
38562856Sdcs		else if (*p == 'Z')
38662856Sdcs			*p = '\0';
38762856Sdcs}
38862856Sdcs
38962856Sdcs/*
39062856Sdcs - check - check a substring match
39162856Sdcs == char *check(char *str, regmatch_t sub, char *should);
39262856Sdcs */
39362856Sdcschar *				/* NULL or complaint */
394292144Sngiecheck(char *str, regmatch_t sub, char *should)
39562856Sdcs{
39692889Sobrien	int len;
39792889Sobrien	int shlen;
39892889Sobrien	char *p;
39962856Sdcs	static char grump[500];
40092889Sobrien	char *at = NULL;
40162856Sdcs
40262856Sdcs	if (should != NULL && strcmp(should, "-") == 0)
40362856Sdcs		should = NULL;
40462856Sdcs	if (should != NULL && should[0] == '@') {
40562856Sdcs		at = should + 1;
40662856Sdcs		should = "";
40762856Sdcs	}
40862856Sdcs
40962856Sdcs	/* check rm_so and rm_eo for consistency */
41062856Sdcs	if (sub.rm_so > sub.rm_eo || (sub.rm_so == -1 && sub.rm_eo != -1) ||
41162856Sdcs				(sub.rm_so != -1 && sub.rm_eo == -1) ||
41262856Sdcs				(sub.rm_so != -1 && sub.rm_so < 0) ||
41362856Sdcs				(sub.rm_eo != -1 && sub.rm_eo < 0) ) {
41462856Sdcs		sprintf(grump, "start %ld end %ld", (long)sub.rm_so,
41562856Sdcs							(long)sub.rm_eo);
41662856Sdcs		return(grump);
41762856Sdcs	}
41862856Sdcs
41962856Sdcs	/* check for no match */
42062856Sdcs	if (sub.rm_so == -1 && should == NULL)
42162856Sdcs		return(NULL);
42262856Sdcs	if (sub.rm_so == -1)
42362856Sdcs		return("did not match");
42462856Sdcs
42562856Sdcs	/* check for in range */
42662856Sdcs	if (sub.rm_eo > strlen(str)) {
42762856Sdcs		sprintf(grump, "start %ld end %ld, past end of string",
428292144Sngie		    (long)sub.rm_so, (long)sub.rm_eo);
42962856Sdcs		return(grump);
43062856Sdcs	}
43162856Sdcs
43262856Sdcs	len = (int)(sub.rm_eo - sub.rm_so);
43362856Sdcs	shlen = (int)strlen(should);
43462856Sdcs	p = str + sub.rm_so;
43562856Sdcs
43662856Sdcs	/* check for not supposed to match */
43762856Sdcs	if (should == NULL) {
43862856Sdcs		sprintf(grump, "matched `%.*s'", len, p);
43962856Sdcs		return(grump);
44062856Sdcs	}
44162856Sdcs
44262856Sdcs	/* check for wrong match */
44362856Sdcs	if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) {
44462856Sdcs		sprintf(grump, "matched `%.*s' instead", len, p);
44562856Sdcs		return(grump);
44662856Sdcs	}
44762856Sdcs	if (shlen > 0)
44862856Sdcs		return(NULL);
44962856Sdcs
45062856Sdcs	/* check null match in right place */
45162856Sdcs	if (at == NULL)
45262856Sdcs		return(NULL);
45362856Sdcs	shlen = strlen(at);
45462856Sdcs	if (shlen == 0)
45562856Sdcs		shlen = 1;	/* force check for end-of-string */
45662856Sdcs	if (strncmp(p, at, shlen) != 0) {
45762856Sdcs		sprintf(grump, "matched null at `%.20s'", p);
45862856Sdcs		return(grump);
45962856Sdcs	}
46062856Sdcs	return(NULL);
46162856Sdcs}
46262856Sdcs
46362856Sdcs/*
46462856Sdcs - eprint - convert error number to name
46562856Sdcs == static char *eprint(int err);
46662856Sdcs */
46762856Sdcsstatic char *
468292144Sngieeprint(int err)
46962856Sdcs{
47062856Sdcs	static char epbuf[100];
47162856Sdcs	size_t len;
47262856Sdcs
47362856Sdcs	len = regerror(REG_ITOA|err, (regex_t *)NULL, epbuf, sizeof(epbuf));
47462856Sdcs	assert(len <= sizeof(epbuf));
47562856Sdcs	return(epbuf);
47662856Sdcs}
47762856Sdcs
47862856Sdcs/*
47962856Sdcs - efind - convert error name to number
48062856Sdcs == static int efind(char *name);
48162856Sdcs */
48262856Sdcsstatic int
483292144Sngieefind(char *name)
48462856Sdcs{
48562856Sdcs	static char efbuf[100];
48662856Sdcs	size_t n;
48762856Sdcs	regex_t re;
48862856Sdcs
48962856Sdcs	sprintf(efbuf, "REG_%s", name);
49062856Sdcs	assert(strlen(efbuf) < sizeof(efbuf));
49162856Sdcs	re.re_endp = efbuf;
49262856Sdcs	(void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf));
49362856Sdcs	return(atoi(efbuf));
49462856Sdcs}
495