192986Sobrien#include <sys/cdefs.h>
292986Sobrien__FBSDID("$FreeBSD$");
392986Sobrien
462856Sdcs#include <stdio.h>
562856Sdcs#include <string.h>
662856Sdcs#include <sys/types.h>
762856Sdcs#include <regex.h>
862856Sdcs#include <assert.h>
962856Sdcs
1062856Sdcs#include "main.ih"
1162856Sdcs
1262856Sdcschar *progname;
1362856Sdcsint debug = 0;
1462856Sdcsint line = 0;
1562856Sdcsint status = 0;
1662856Sdcs
1762856Sdcsint copts = REG_EXTENDED;
1862856Sdcsint eopts = 0;
1962856Sdcsregoff_t startoff = 0;
2062856Sdcsregoff_t endoff = 0;
2162856Sdcs
2262856Sdcs
2362856Sdcsextern int split();
2462856Sdcsextern void regprint();
2562856Sdcs
2662856Sdcs/*
2762856Sdcs - main - do the simple case, hand off to regress() for regression
2862856Sdcs */
2962856Sdcsmain(argc, argv)
3062856Sdcsint argc;
3162856Sdcschar *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));
8362856Sdcs		fprintf(stderr, "error %s, %d/%d `%s'\n",
8462856Sdcs			eprint(err), len, sizeof(erbuf), erbuf);
8562856Sdcs		exit(status);
8662856Sdcs	}
8762856Sdcs	regprint(&re, stdout);
8862856Sdcs
8962856Sdcs	if (optind >= argc) {
9062856Sdcs		regfree(&re);
9162856Sdcs		exit(status);
9262856Sdcs	}
9362856Sdcs
9462856Sdcs	if (eopts&REG_STARTEND) {
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));
10162856Sdcs		fprintf(stderr, "error %s, %d/%d `%s'\n",
10262856Sdcs			eprint(err), len, sizeof(erbuf), erbuf);
10362856Sdcs		exit(status);
10462856Sdcs	}
10562856Sdcs	if (!(copts&REG_NOSUB)) {
10662856Sdcs		len = (int)(subs[0].rm_eo - subs[0].rm_so);
10762856Sdcs		if (subs[0].rm_so != -1) {
10862856Sdcs			if (len != 0)
10962856Sdcs				printf("match `%.*s'\n", len,
11062856Sdcs					argv[optind] + subs[0].rm_so);
11162856Sdcs			else
11262856Sdcs				printf("match `'@%.1s\n",
11362856Sdcs					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,
11862856Sdcs					(int)(subs[i].rm_eo - subs[i].rm_so),
11962856Sdcs					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
12962856Sdcsregress(in)
13062856SdcsFILE *in;
13162856Sdcs{
13262856Sdcs	char inbuf[1000];
13362856Sdcs#	define	MAXF	10
13462856Sdcs	char *f[MAXF];
13562856Sdcs	int nf;
13662856Sdcs	int i;
13762856Sdcs	char erbuf[100];
13862856Sdcs	size_t ne;
13962856Sdcs	char *badpat = "invalid regular expression";
14062856Sdcs#	define	SHORT	10
14162856Sdcs	char *bpname = "REG_BADPAT";
14262856Sdcs	regex_t re;
14362856Sdcs
14462856Sdcs	while (fgets(inbuf, sizeof(inbuf), in) != NULL) {
14562856Sdcs		line++;
14662856Sdcs		if (inbuf[0] == '#' || inbuf[0] == '\n')
14762856Sdcs			continue;			/* NOTE CONTINUE */
14862856Sdcs		inbuf[strlen(inbuf)-1] = '\0';	/* get rid of stupid \n */
14962856Sdcs		if (debug)
15062856Sdcs			fprintf(stdout, "%d:\n", line);
15162856Sdcs		nf = split(inbuf, f, MAXF, "\t\t");
15262856Sdcs		if (nf < 3) {
15362856Sdcs			fprintf(stderr, "bad input, line %d\n", line);
15462856Sdcs			exit(1);
15562856Sdcs		}
15662856Sdcs		for (i = 0; i < nf; i++)
15762856Sdcs			if (strcmp(f[i], "\"\"") == 0)
15862856Sdcs				f[i] = "";
15962856Sdcs		if (nf <= 3)
16062856Sdcs			f[3] = NULL;
16162856Sdcs		if (nf <= 4)
16262856Sdcs			f[4] = NULL;
16362856Sdcs		try(f[0], f[1], f[2], f[3], f[4], options('c', f[1]));
16462856Sdcs		if (opt('&', f[1]))	/* try with either type of RE */
16562856Sdcs			try(f[0], f[1], f[2], f[3], f[4],
16662856Sdcs					options('c', f[1]) &~ REG_EXTENDED);
16762856Sdcs	}
16862856Sdcs
16962856Sdcs	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
17062856Sdcs	if (strcmp(erbuf, badpat) != 0 || ne != strlen(badpat)+1) {
17162856Sdcs		fprintf(stderr, "end: regerror() test gave `%s' not `%s'\n",
17262856Sdcs							erbuf, badpat);
17362856Sdcs		status = 1;
17462856Sdcs	}
17562856Sdcs	ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, (size_t)SHORT);
17662856Sdcs	if (strncmp(erbuf, badpat, SHORT-1) != 0 || erbuf[SHORT-1] != '\0' ||
17762856Sdcs						ne != strlen(badpat)+1) {
17862856Sdcs		fprintf(stderr, "end: regerror() short test gave `%s' not `%.*s'\n",
17962856Sdcs						erbuf, SHORT-1, badpat);
18062856Sdcs		status = 1;
18162856Sdcs	}
18262856Sdcs	ne = regerror(REG_ITOA|REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf));
18362856Sdcs	if (strcmp(erbuf, bpname) != 0 || ne != strlen(bpname)+1) {
18462856Sdcs		fprintf(stderr, "end: regerror() ITOA test gave `%s' not `%s'\n",
18562856Sdcs						erbuf, bpname);
18662856Sdcs		status = 1;
18762856Sdcs	}
18862856Sdcs	re.re_endp = bpname;
18962856Sdcs	ne = regerror(REG_ATOI, &re, erbuf, sizeof(erbuf));
19062856Sdcs	if (atoi(erbuf) != (int)REG_BADPAT) {
19162856Sdcs		fprintf(stderr, "end: regerror() ATOI test gave `%s' not `%ld'\n",
19262856Sdcs						erbuf, (long)REG_BADPAT);
19362856Sdcs		status = 1;
19462856Sdcs	} else if (ne != strlen(erbuf)+1) {
19562856Sdcs		fprintf(stderr, "end: regerror() ATOI test len(`%s') = %ld\n",
19662856Sdcs						erbuf, (long)REG_BADPAT);
19762856Sdcs		status = 1;
19862856Sdcs	}
19962856Sdcs}
20062856Sdcs
20162856Sdcs/*
20262856Sdcs - try - try it, and report on problems
20362856Sdcs == void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts);
20462856Sdcs */
20562856Sdcsvoid
20662856Sdcstry(f0, f1, f2, f3, f4, opts)
20762856Sdcschar *f0;
20862856Sdcschar *f1;
20962856Sdcschar *f2;
21062856Sdcschar *f3;
21162856Sdcschar *f4;
21262856Sdcsint opts;			/* may not match f1 */
21362856Sdcs{
21462856Sdcs	regex_t re;
21562856Sdcs#	define	NSUBS	10
21662856Sdcs	regmatch_t subs[NSUBS];
21762856Sdcs#	define	NSHOULD	15
21862856Sdcs	char *should[NSHOULD];
21962856Sdcs	int nshould;
22062856Sdcs	char erbuf[100];
22162856Sdcs	int err;
22262856Sdcs	int len;
22362856Sdcs	char *type = (opts & REG_EXTENDED) ? "ERE" : "BRE";
22492889Sobrien	int i;
22562856Sdcs	char *grump;
22662856Sdcs	char f0copy[1000];
22762856Sdcs	char f2copy[1000];
22862856Sdcs
22962856Sdcs	strcpy(f0copy, f0);
23062856Sdcs	re.re_endp = (opts&REG_PEND) ? f0copy + strlen(f0copy) : NULL;
23162856Sdcs	fixstr(f0copy);
23262856Sdcs	err = regcomp(&re, f0copy, opts);
23362856Sdcs	if (err != 0 && (!opt('C', f1) || err != efind(f2))) {
23462856Sdcs		/* unexpected error or wrong error */
23562856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
23662856Sdcs		fprintf(stderr, "%d: %s error %s, %d/%d `%s'\n",
23762856Sdcs					line, type, eprint(err), len,
23862856Sdcs					sizeof(erbuf), erbuf);
23962856Sdcs		status = 1;
24062856Sdcs	} else if (err == 0 && opt('C', f1)) {
24162856Sdcs		/* unexpected success */
24262856Sdcs		fprintf(stderr, "%d: %s should have given REG_%s\n",
24362856Sdcs						line, type, f2);
24462856Sdcs		status = 1;
24562856Sdcs		err = 1;	/* so we won't try regexec */
24662856Sdcs	}
24762856Sdcs
24862856Sdcs	if (err != 0) {
24962856Sdcs		regfree(&re);
25062856Sdcs		return;
25162856Sdcs	}
25262856Sdcs
25362856Sdcs	strcpy(f2copy, f2);
25462856Sdcs	fixstr(f2copy);
25562856Sdcs
25662856Sdcs	if (options('e', f1)&REG_STARTEND) {
25762856Sdcs		if (strchr(f2, '(') == NULL || strchr(f2, ')') == NULL)
25862856Sdcs			fprintf(stderr, "%d: bad STARTEND syntax\n", line);
25962856Sdcs		subs[0].rm_so = strchr(f2, '(') - f2 + 1;
26062856Sdcs		subs[0].rm_eo = strchr(f2, ')') - f2;
26162856Sdcs	}
26262856Sdcs	err = regexec(&re, f2copy, NSUBS, subs, options('e', f1));
26362856Sdcs
26462856Sdcs	if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) {
26562856Sdcs		/* unexpected error or wrong error */
26662856Sdcs		len = regerror(err, &re, erbuf, sizeof(erbuf));
26762856Sdcs		fprintf(stderr, "%d: %s exec error %s, %d/%d `%s'\n",
26862856Sdcs					line, type, eprint(err), len,
26962856Sdcs					sizeof(erbuf), erbuf);
27062856Sdcs		status = 1;
27162856Sdcs	} else if (err != 0) {
27262856Sdcs		/* nothing more to check */
27362856Sdcs	} else if (f3 == NULL) {
27462856Sdcs		/* unexpected success */
27562856Sdcs		fprintf(stderr, "%d: %s exec should have failed\n",
27662856Sdcs						line, type);
27762856Sdcs		status = 1;
27862856Sdcs		err = 1;		/* just on principle */
27962856Sdcs	} else if (opts&REG_NOSUB) {
28062856Sdcs		/* nothing more to check */
28162856Sdcs	} else if ((grump = check(f2, subs[0], f3)) != NULL) {
28262856Sdcs		fprintf(stderr, "%d: %s %s\n", line, type, grump);
28362856Sdcs		status = 1;
28462856Sdcs		err = 1;
28562856Sdcs	}
28662856Sdcs
28762856Sdcs	if (err != 0 || f4 == NULL) {
28862856Sdcs		regfree(&re);
28962856Sdcs		return;
29062856Sdcs	}
29162856Sdcs
29262856Sdcs	for (i = 1; i < NSHOULD; i++)
29362856Sdcs		should[i] = NULL;
29462856Sdcs	nshould = split(f4, should+1, NSHOULD-1, ",");
29562856Sdcs	if (nshould == 0) {
29662856Sdcs		nshould = 1;
29762856Sdcs		should[1] = "";
29862856Sdcs	}
29962856Sdcs	for (i = 1; i < NSUBS; i++) {
30062856Sdcs		grump = check(f2, subs[i], should[i]);
30162856Sdcs		if (grump != NULL) {
30262856Sdcs			fprintf(stderr, "%d: %s $%d %s\n", line,
30362856Sdcs							type, i, grump);
30462856Sdcs			status = 1;
30562856Sdcs			err = 1;
30662856Sdcs		}
30762856Sdcs	}
30862856Sdcs
30962856Sdcs	regfree(&re);
31062856Sdcs}
31162856Sdcs
31262856Sdcs/*
31362856Sdcs - options - pick options out of a regression-test string
31462856Sdcs == int options(int type, char *s);
31562856Sdcs */
31662856Sdcsint
31762856Sdcsoptions(type, s)
31862856Sdcsint type;			/* 'c' compile, 'e' exec */
31962856Sdcschar *s;
32062856Sdcs{
32192889Sobrien	char *p;
32292889Sobrien	int o = (type == 'c') ? copts : eopts;
32392889Sobrien	char *legal = (type == 'c') ? "bisnmp" : "^$#tl";
32462856Sdcs
32562856Sdcs	for (p = s; *p != '\0'; p++)
32662856Sdcs		if (strchr(legal, *p) != NULL)
32762856Sdcs			switch (*p) {
32862856Sdcs			case 'b':
32962856Sdcs				o &= ~REG_EXTENDED;
33062856Sdcs				break;
33162856Sdcs			case 'i':
33262856Sdcs				o |= REG_ICASE;
33362856Sdcs				break;
33462856Sdcs			case 's':
33562856Sdcs				o |= REG_NOSUB;
33662856Sdcs				break;
33762856Sdcs			case 'n':
33862856Sdcs				o |= REG_NEWLINE;
33962856Sdcs				break;
34062856Sdcs			case 'm':
34162856Sdcs				o &= ~REG_EXTENDED;
34262856Sdcs				o |= REG_NOSPEC;
34362856Sdcs				break;
34462856Sdcs			case 'p':
34562856Sdcs				o |= REG_PEND;
34662856Sdcs				break;
34762856Sdcs			case '^':
34862856Sdcs				o |= REG_NOTBOL;
34962856Sdcs				break;
35062856Sdcs			case '$':
35162856Sdcs				o |= REG_NOTEOL;
35262856Sdcs				break;
35362856Sdcs			case '#':
35462856Sdcs				o |= REG_STARTEND;
35562856Sdcs				break;
35662856Sdcs			case 't':	/* trace */
35762856Sdcs				o |= REG_TRACE;
35862856Sdcs				break;
35962856Sdcs			case 'l':	/* force long representation */
36062856Sdcs				o |= REG_LARGE;
36162856Sdcs				break;
36262856Sdcs			case 'r':	/* force backref use */
36362856Sdcs				o |= REG_BACKR;
36462856Sdcs				break;
36562856Sdcs			}
36662856Sdcs	return(o);
36762856Sdcs}
36862856Sdcs
36962856Sdcs/*
37062856Sdcs - opt - is a particular option in a regression string?
37162856Sdcs == int opt(int c, char *s);
37262856Sdcs */
37362856Sdcsint				/* predicate */
37462856Sdcsopt(c, s)
37562856Sdcsint c;
37662856Sdcschar *s;
37762856Sdcs{
37862856Sdcs	return(strchr(s, c) != NULL);
37962856Sdcs}
38062856Sdcs
38162856Sdcs/*
38262856Sdcs - fixstr - transform magic characters in strings
38392889Sobrien == void fixstr(char *p);
38462856Sdcs */
38562856Sdcsvoid
38662856Sdcsfixstr(p)
38792889Sobrienchar *p;
38862856Sdcs{
38962856Sdcs	if (p == NULL)
39062856Sdcs		return;
39162856Sdcs
39262856Sdcs	for (; *p != '\0'; p++)
39362856Sdcs		if (*p == 'N')
39462856Sdcs			*p = '\n';
39562856Sdcs		else if (*p == 'T')
39662856Sdcs			*p = '\t';
39762856Sdcs		else if (*p == 'S')
39862856Sdcs			*p = ' ';
39962856Sdcs		else if (*p == 'Z')
40062856Sdcs			*p = '\0';
40162856Sdcs}
40262856Sdcs
40362856Sdcs/*
40462856Sdcs - check - check a substring match
40562856Sdcs == char *check(char *str, regmatch_t sub, char *should);
40662856Sdcs */
40762856Sdcschar *				/* NULL or complaint */
40862856Sdcscheck(str, sub, should)
40962856Sdcschar *str;
41062856Sdcsregmatch_t sub;
41162856Sdcschar *should;
41262856Sdcs{
41392889Sobrien	int len;
41492889Sobrien	int shlen;
41592889Sobrien	char *p;
41662856Sdcs	static char grump[500];
41792889Sobrien	char *at = NULL;
41862856Sdcs
41962856Sdcs	if (should != NULL && strcmp(should, "-") == 0)
42062856Sdcs		should = NULL;
42162856Sdcs	if (should != NULL && should[0] == '@') {
42262856Sdcs		at = should + 1;
42362856Sdcs		should = "";
42462856Sdcs	}
42562856Sdcs
42662856Sdcs	/* check rm_so and rm_eo for consistency */
42762856Sdcs	if (sub.rm_so > sub.rm_eo || (sub.rm_so == -1 && sub.rm_eo != -1) ||
42862856Sdcs				(sub.rm_so != -1 && sub.rm_eo == -1) ||
42962856Sdcs				(sub.rm_so != -1 && sub.rm_so < 0) ||
43062856Sdcs				(sub.rm_eo != -1 && sub.rm_eo < 0) ) {
43162856Sdcs		sprintf(grump, "start %ld end %ld", (long)sub.rm_so,
43262856Sdcs							(long)sub.rm_eo);
43362856Sdcs		return(grump);
43462856Sdcs	}
43562856Sdcs
43662856Sdcs	/* check for no match */
43762856Sdcs	if (sub.rm_so == -1 && should == NULL)
43862856Sdcs		return(NULL);
43962856Sdcs	if (sub.rm_so == -1)
44062856Sdcs		return("did not match");
44162856Sdcs
44262856Sdcs	/* check for in range */
44362856Sdcs	if (sub.rm_eo > strlen(str)) {
44462856Sdcs		sprintf(grump, "start %ld end %ld, past end of string",
44562856Sdcs					(long)sub.rm_so, (long)sub.rm_eo);
44662856Sdcs		return(grump);
44762856Sdcs	}
44862856Sdcs
44962856Sdcs	len = (int)(sub.rm_eo - sub.rm_so);
45062856Sdcs	shlen = (int)strlen(should);
45162856Sdcs	p = str + sub.rm_so;
45262856Sdcs
45362856Sdcs	/* check for not supposed to match */
45462856Sdcs	if (should == NULL) {
45562856Sdcs		sprintf(grump, "matched `%.*s'", len, p);
45662856Sdcs		return(grump);
45762856Sdcs	}
45862856Sdcs
45962856Sdcs	/* check for wrong match */
46062856Sdcs	if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) {
46162856Sdcs		sprintf(grump, "matched `%.*s' instead", len, p);
46262856Sdcs		return(grump);
46362856Sdcs	}
46462856Sdcs	if (shlen > 0)
46562856Sdcs		return(NULL);
46662856Sdcs
46762856Sdcs	/* check null match in right place */
46862856Sdcs	if (at == NULL)
46962856Sdcs		return(NULL);
47062856Sdcs	shlen = strlen(at);
47162856Sdcs	if (shlen == 0)
47262856Sdcs		shlen = 1;	/* force check for end-of-string */
47362856Sdcs	if (strncmp(p, at, shlen) != 0) {
47462856Sdcs		sprintf(grump, "matched null at `%.20s'", p);
47562856Sdcs		return(grump);
47662856Sdcs	}
47762856Sdcs	return(NULL);
47862856Sdcs}
47962856Sdcs
48062856Sdcs/*
48162856Sdcs - eprint - convert error number to name
48262856Sdcs == static char *eprint(int err);
48362856Sdcs */
48462856Sdcsstatic char *
48562856Sdcseprint(err)
48662856Sdcsint err;
48762856Sdcs{
48862856Sdcs	static char epbuf[100];
48962856Sdcs	size_t len;
49062856Sdcs
49162856Sdcs	len = regerror(REG_ITOA|err, (regex_t *)NULL, epbuf, sizeof(epbuf));
49262856Sdcs	assert(len <= sizeof(epbuf));
49362856Sdcs	return(epbuf);
49462856Sdcs}
49562856Sdcs
49662856Sdcs/*
49762856Sdcs - efind - convert error name to number
49862856Sdcs == static int efind(char *name);
49962856Sdcs */
50062856Sdcsstatic int
50162856Sdcsefind(name)
50262856Sdcschar *name;
50362856Sdcs{
50462856Sdcs	static char efbuf[100];
50562856Sdcs	size_t n;
50662856Sdcs	regex_t re;
50762856Sdcs
50862856Sdcs	sprintf(efbuf, "REG_%s", name);
50962856Sdcs	assert(strlen(efbuf) < sizeof(efbuf));
51062856Sdcs	re.re_endp = efbuf;
51162856Sdcs	(void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf));
51262856Sdcs	return(atoi(efbuf));
51362856Sdcs}
514