xref: /freebsd/usr.bin/grep/util.c (revision 069ac184)
1 /*	$NetBSD: util.c,v 1.9 2011/02/27 17:33:37 joerg Exp $	*/
2 /*	$OpenBSD: util.c,v 1.39 2010/07/02 22:18:03 tedu Exp $	*/
3 
4 /*-
5  * SPDX-License-Identifier: BSD-2-Clause
6  *
7  * Copyright (c) 1999 James Howard and Dag-Erling Smørgrav
8  * Copyright (C) 2008-2010 Gabor Kovesdan <gabor@FreeBSD.org>
9  * Copyright (C) 2017 Kyle Evans <kevans@FreeBSD.org>
10  * All rights reserved.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  */
33 
34 #include <sys/cdefs.h>
35 #include <sys/stat.h>
36 #include <sys/types.h>
37 
38 #include <ctype.h>
39 #include <err.h>
40 #include <errno.h>
41 #include <fnmatch.h>
42 #include <fts.h>
43 #include <libgen.h>
44 #include <stdbool.h>
45 #include <stdio.h>
46 #include <stdlib.h>
47 #include <string.h>
48 #include <unistd.h>
49 #include <wchar.h>
50 #include <wctype.h>
51 
52 #include "grep.h"
53 
54 static bool	 first_match = true;
55 
56 /*
57  * Match printing context
58  */
59 struct mprintc {
60 	long long	tail;		/* Number of trailing lines to record */
61 	int		last_outed;	/* Number of lines since last output */
62 	bool		doctx;		/* Printing context? */
63 	bool		printmatch;	/* Printing matches? */
64 	bool		same_file;	/* Same file as previously printed? */
65 };
66 
67 static void procmatch_match(struct mprintc *mc, struct parsec *pc);
68 static void procmatch_nomatch(struct mprintc *mc, struct parsec *pc);
69 static bool procmatches(struct mprintc *mc, struct parsec *pc, bool matched);
70 #ifdef WITH_INTERNAL_NOSPEC
71 static int litexec(const struct pat *pat, const char *string,
72     size_t nmatch, regmatch_t pmatch[]);
73 #endif
74 static bool procline(struct parsec *pc);
75 static void printline(struct parsec *pc, int sep);
76 static void printline_metadata(struct str *line, int sep);
77 
78 bool
79 file_matching(const char *fname)
80 {
81 	char *fname_base, *fname_buf;
82 	bool ret;
83 
84 	ret = finclude ? false : true;
85 	fname_buf = strdup(fname);
86 	if (fname_buf == NULL)
87 		err(2, "strdup");
88 	fname_base = basename(fname_buf);
89 
90 	for (unsigned int i = 0; i < fpatterns; ++i) {
91 		if (fnmatch(fpattern[i].pat, fname, 0) == 0 ||
92 		    fnmatch(fpattern[i].pat, fname_base, 0) == 0)
93 			/*
94 			 * The last pattern matched wins exclusion/inclusion
95 			 * rights, so we can't reasonably bail out early here.
96 			 */
97 			ret = (fpattern[i].mode != EXCL_PAT);
98 	}
99 	free(fname_buf);
100 	return (ret);
101 }
102 
103 static inline bool
104 dir_matching(const char *dname)
105 {
106 	bool ret;
107 
108 	ret = dinclude ? false : true;
109 
110 	for (unsigned int i = 0; i < dpatterns; ++i) {
111 		if (dname != NULL && fnmatch(dpattern[i].pat, dname, 0) == 0)
112 			/*
113 			 * The last pattern matched wins exclusion/inclusion
114 			 * rights, so we can't reasonably bail out early here.
115 			 */
116 			ret = (dpattern[i].mode != EXCL_PAT);
117 	}
118 	return (ret);
119 }
120 
121 /*
122  * Processes a directory when a recursive search is performed with
123  * the -R option.  Each appropriate file is passed to procfile().
124  */
125 bool
126 grep_tree(char **argv)
127 {
128 	FTS *fts;
129 	FTSENT *p;
130 	int fts_flags;
131 	bool matched, ok;
132 	const char *wd[] = { ".", NULL };
133 
134 	matched = false;
135 
136 	/* This switch effectively initializes 'fts_flags' */
137 	switch(linkbehave) {
138 	case LINK_EXPLICIT:
139 		fts_flags = FTS_COMFOLLOW;
140 		break;
141 	case LINK_SKIP:
142 		fts_flags = FTS_PHYSICAL;
143 		break;
144 	default:
145 		fts_flags = FTS_LOGICAL;
146 	}
147 
148 	fts_flags |= FTS_NOSTAT | FTS_NOCHDIR;
149 
150 	fts = fts_open((argv[0] == NULL) ?
151 	    __DECONST(char * const *, wd) : argv, fts_flags, NULL);
152 	if (fts == NULL)
153 		err(2, "fts_open");
154 	while (errno = 0, (p = fts_read(fts)) != NULL) {
155 		switch (p->fts_info) {
156 		case FTS_DNR:
157 			/* FALLTHROUGH */
158 		case FTS_ERR:
159 			file_err = true;
160 			if(!sflag)
161 				warnx("%s: %s", p->fts_path, strerror(p->fts_errno));
162 			break;
163 		case FTS_D:
164 			/* FALLTHROUGH */
165 		case FTS_DP:
166 			if (dexclude || dinclude)
167 				if (!dir_matching(p->fts_name) ||
168 				    !dir_matching(p->fts_path))
169 					fts_set(fts, p, FTS_SKIP);
170 			break;
171 		case FTS_DC:
172 			/* Print a warning for recursive directory loop */
173 			warnx("warning: %s: recursive directory loop",
174 			    p->fts_path);
175 			break;
176 		default:
177 			/* Check for file exclusion/inclusion */
178 			ok = true;
179 			if (fexclude || finclude)
180 				ok &= file_matching(p->fts_path);
181 
182 			if (ok && procfile(p->fts_path))
183 				matched = true;
184 			break;
185 		}
186 	}
187 	if (errno != 0)
188 		err(2, "fts_read");
189 
190 	fts_close(fts);
191 	return (matched);
192 }
193 
194 static void
195 procmatch_match(struct mprintc *mc, struct parsec *pc)
196 {
197 
198 	if (mc->doctx) {
199 		if (!first_match && (!mc->same_file || mc->last_outed > 0))
200 			printf("--\n");
201 		if (Bflag > 0)
202 			printqueue();
203 		mc->tail = Aflag;
204 	}
205 
206 	/* Print the matching line, but only if not quiet/binary */
207 	if (mc->printmatch) {
208 		printline(pc, ':');
209 		while (pc->matchidx >= MAX_MATCHES) {
210 			/* Reset matchidx and try again */
211 			pc->matchidx = 0;
212 			if (procline(pc) == !vflag)
213 				printline(pc, ':');
214 			else
215 				break;
216 		}
217 		first_match = false;
218 		mc->same_file = true;
219 		mc->last_outed = 0;
220 	}
221 }
222 
223 static void
224 procmatch_nomatch(struct mprintc *mc, struct parsec *pc)
225 {
226 
227 	/* Deal with any -A context as needed */
228 	if (mc->tail > 0) {
229 		grep_printline(&pc->ln, '-');
230 		mc->tail--;
231 		if (Bflag > 0)
232 			clearqueue();
233 	} else if (Bflag == 0 || (Bflag > 0 && enqueue(&pc->ln)))
234 		/*
235 		 * Enqueue non-matching lines for -B context. If we're not
236 		 * actually doing -B context or if the enqueue resulted in a
237 		 * line being rotated out, then go ahead and increment
238 		 * last_outed to signify a gap between context/match.
239 		 */
240 		++mc->last_outed;
241 }
242 
243 /*
244  * Process any matches in the current parsing context, return a boolean
245  * indicating whether we should halt any further processing or not. 'true' to
246  * continue processing, 'false' to halt.
247  */
248 static bool
249 procmatches(struct mprintc *mc, struct parsec *pc, bool matched)
250 {
251 
252 	if (mflag && mcount <= 0) {
253 		/*
254 		 * We already hit our match count, but we need to keep dumping
255 		 * lines until we've lost our tail.
256 		 */
257 		grep_printline(&pc->ln, '-');
258 		mc->tail--;
259 		return (mc->tail != 0);
260 	}
261 
262 	/*
263 	 * XXX TODO: This should loop over pc->matches and handle things on a
264 	 * line-by-line basis, setting up a `struct str` as needed.
265 	 */
266 	/* Deal with any -B context or context separators */
267 	if (matched) {
268 		procmatch_match(mc, pc);
269 
270 		/* Count the matches if we have a match limit */
271 		if (mflag) {
272 			/* XXX TODO: Decrement by number of matched lines */
273 			mcount -= 1;
274 			if (mcount <= 0)
275 				return (mc->tail != 0);
276 		}
277 	} else if (mc->doctx)
278 		procmatch_nomatch(mc, pc);
279 
280 	return (true);
281 }
282 
283 /*
284  * Opens a file and processes it.  Each file is processed line-by-line
285  * passing the lines to procline().
286  */
287 bool
288 procfile(const char *fn)
289 {
290 	struct parsec pc;
291 	struct mprintc mc;
292 	struct file *f;
293 	struct stat sb;
294 	mode_t s;
295 	int lines;
296 	bool line_matched;
297 
298 	if (strcmp(fn, "-") == 0) {
299 		fn = label != NULL ? label : errstr[1];
300 		f = grep_open(NULL);
301 	} else {
302 		if (stat(fn, &sb) == 0) {
303 			/* Check if we need to process the file */
304 			s = sb.st_mode & S_IFMT;
305 			if (dirbehave == DIR_SKIP && s == S_IFDIR)
306 				return (false);
307 			if (devbehave == DEV_SKIP && (s == S_IFIFO ||
308 			    s == S_IFCHR || s == S_IFBLK || s == S_IFSOCK))
309 				return (false);
310 		}
311 		f = grep_open(fn);
312 	}
313 	if (f == NULL) {
314 		file_err = true;
315 		if (!sflag)
316 			warn("%s", fn);
317 		return (false);
318 	}
319 
320 	pc.ln.file = grep_strdup(fn);
321 	pc.ln.line_no = 0;
322 	pc.ln.len = 0;
323 	pc.ln.boff = 0;
324 	pc.ln.off = -1;
325 	pc.binary = f->binary;
326 	pc.cntlines = false;
327 	memset(&mc, 0, sizeof(mc));
328 	mc.printmatch = true;
329 	if ((pc.binary && binbehave == BINFILE_BIN) || cflag || qflag ||
330 	    lflag || Lflag)
331 		mc.printmatch = false;
332 	if (mc.printmatch && (Aflag != 0 || Bflag != 0))
333 		mc.doctx = true;
334 	if (mc.printmatch && (Aflag != 0 || Bflag != 0 || mflag || nflag))
335 		pc.cntlines = true;
336 	mcount = mlimit;
337 
338 	for (lines = 0; lines == 0 || !(lflag || qflag); ) {
339 		/*
340 		 * XXX TODO: We need to revisit this in a chunking world. We're
341 		 * not going to be doing per-line statistics because of the
342 		 * overhead involved. procmatches can figure that stuff out as
343 		 * needed. */
344 		/* Reset per-line statistics */
345 		pc.printed = 0;
346 		pc.matchidx = 0;
347 		pc.lnstart = 0;
348 		pc.ln.boff = 0;
349 		pc.ln.off += pc.ln.len + 1;
350 		/* XXX TODO: Grab a chunk */
351 		if ((pc.ln.dat = grep_fgetln(f, &pc)) == NULL ||
352 		    pc.ln.len == 0)
353 			break;
354 
355 		if (pc.ln.len > 0 && pc.ln.dat[pc.ln.len - 1] == fileeol)
356 			--pc.ln.len;
357 		pc.ln.line_no++;
358 
359 		/* Return if we need to skip a binary file */
360 		if (pc.binary && binbehave == BINFILE_SKIP) {
361 			grep_close(f);
362 			free(pc.ln.file);
363 			free(f);
364 			return (0);
365 		}
366 
367 		if (mflag && mcount <= 0) {
368 			/*
369 			 * Short-circuit, already hit match count and now we're
370 			 * just picking up any remaining pieces.
371 			 */
372 			if (!procmatches(&mc, &pc, false))
373 				break;
374 			continue;
375 		}
376 		line_matched = procline(&pc) == !vflag;
377 		if (line_matched)
378 			++lines;
379 
380 		/* Halt processing if we hit our match limit */
381 		if (!procmatches(&mc, &pc, line_matched))
382 			break;
383 	}
384 	if (Bflag > 0)
385 		clearqueue();
386 	grep_close(f);
387 
388 	if (cflag && !qflag) {
389 		if (!hflag)
390 			printf("%s:", pc.ln.file);
391 		printf("%u\n", lines);
392 	}
393 	if (lflag && !qflag && lines != 0)
394 		printf("%s%c", fn, nullflag ? 0 : '\n');
395 	if (Lflag && !qflag && lines == 0)
396 		printf("%s%c", fn, nullflag ? 0 : '\n');
397 	if (lines != 0 && !cflag && !lflag && !Lflag &&
398 	    binbehave == BINFILE_BIN && f->binary && !qflag)
399 		printf(errstr[7], fn);
400 
401 	free(pc.ln.file);
402 	free(f);
403 	return (lines != 0);
404 }
405 
406 #ifdef WITH_INTERNAL_NOSPEC
407 /*
408  * Internal implementation of literal string search within a string, modeled
409  * after regexec(3), for use when the regex(3) implementation doesn't offer
410  * either REG_NOSPEC or REG_LITERAL. This does not apply in the default FreeBSD
411  * config, but in other scenarios such as building against libgnuregex or on
412  * some non-FreeBSD OSes.
413  */
414 static int
415 litexec(const struct pat *pat, const char *string, size_t nmatch,
416     regmatch_t pmatch[])
417 {
418 	char *(*strstr_fn)(const char *, const char *);
419 	char *sub, *subject;
420 	const char *search;
421 	size_t idx, n, ofs, stringlen;
422 
423 	if (cflags & REG_ICASE)
424 		strstr_fn = strcasestr;
425 	else
426 		strstr_fn = strstr;
427 	idx = 0;
428 	ofs = pmatch[0].rm_so;
429 	stringlen = pmatch[0].rm_eo;
430 	if (ofs >= stringlen)
431 		return (REG_NOMATCH);
432 	subject = strndup(string, stringlen);
433 	if (subject == NULL)
434 		return (REG_ESPACE);
435 	for (n = 0; ofs < stringlen;) {
436 		search = (subject + ofs);
437 		if ((unsigned long)pat->len > strlen(search))
438 			break;
439 		sub = strstr_fn(search, pat->pat);
440 		/*
441 		 * Ignoring the empty string possibility due to context: grep optimizes
442 		 * for empty patterns and will never reach this point.
443 		 */
444 		if (sub == NULL)
445 			break;
446 		++n;
447 		/* Fill in pmatch if necessary */
448 		if (nmatch > 0) {
449 			pmatch[idx].rm_so = ofs + (sub - search);
450 			pmatch[idx].rm_eo = pmatch[idx].rm_so + pat->len;
451 			if (++idx == nmatch)
452 				break;
453 			ofs = pmatch[idx].rm_so + 1;
454 		} else
455 			/* We only needed to know if we match or not */
456 			break;
457 	}
458 	free(subject);
459 	if (n > 0 && nmatch > 0)
460 		for (n = idx; n < nmatch; ++n)
461 			pmatch[n].rm_so = pmatch[n].rm_eo = -1;
462 
463 	return (n > 0 ? 0 : REG_NOMATCH);
464 }
465 #endif /* WITH_INTERNAL_NOSPEC */
466 
467 #define iswword(x)	(iswalnum((x)) || (x) == L'_')
468 
469 /*
470  * Processes a line comparing it with the specified patterns.  Each pattern
471  * is looped to be compared along with the full string, saving each and every
472  * match, which is necessary to colorize the output and to count the
473  * matches.  The matching lines are passed to printline() to display the
474  * appropriate output.
475  */
476 static bool
477 procline(struct parsec *pc)
478 {
479 	regmatch_t pmatch, lastmatch, chkmatch;
480 	wchar_t wbegin, wend;
481 	size_t st, nst;
482 	unsigned int i;
483 	int r = 0, leflags = eflags;
484 	size_t startm = 0, matchidx;
485 	unsigned int retry;
486 	bool lastmatched, matched;
487 
488 	matchidx = pc->matchidx;
489 
490 	/* Null pattern shortcuts. */
491 	if (matchall) {
492 		if (xflag && pc->ln.len == 0) {
493 			/* Matches empty lines (-x). */
494 			return (true);
495 		} else if (!wflag && !xflag) {
496 			/* Matches every line (no -w or -x). */
497 			return (true);
498 		}
499 
500 		/*
501 		 * If we only have the NULL pattern, whether we match or not
502 		 * depends on if we got here with -w or -x.  If either is set,
503 		 * the answer is no.  If we have other patterns, we'll defer
504 		 * to them.
505 		 */
506 		if (patterns == 0) {
507 			return (!(wflag || xflag));
508 		}
509 	} else if (patterns == 0) {
510 		/* Pattern file with no patterns. */
511 		return (false);
512 	}
513 
514 	matched = false;
515 	st = pc->lnstart;
516 	nst = 0;
517 	/* Initialize to avoid a false positive warning from GCC. */
518 	lastmatch.rm_so = lastmatch.rm_eo = 0;
519 
520 	/* Loop to process the whole line */
521 	while (st <= pc->ln.len) {
522 		lastmatched = false;
523 		startm = matchidx;
524 		retry = 0;
525 		if (st > 0 && pc->ln.dat[st - 1] != fileeol)
526 			leflags |= REG_NOTBOL;
527 		/* Loop to compare with all the patterns */
528 		for (i = 0; i < patterns; i++) {
529 			pmatch.rm_so = st;
530 			pmatch.rm_eo = pc->ln.len;
531 #ifdef WITH_INTERNAL_NOSPEC
532 			if (grepbehave == GREP_FIXED)
533 				r = litexec(&pattern[i], pc->ln.dat, 1, &pmatch);
534 			else
535 #endif
536 			r = regexec(&r_pattern[i], pc->ln.dat, 1, &pmatch,
537 			    leflags);
538 			if (r != 0)
539 				continue;
540 			/* Check for full match */
541 			if (xflag && (pmatch.rm_so != 0 ||
542 			    (size_t)pmatch.rm_eo != pc->ln.len))
543 				continue;
544 			/* Check for whole word match */
545 			if (wflag) {
546 				wbegin = wend = L' ';
547 				if (pmatch.rm_so != 0 &&
548 				    sscanf(&pc->ln.dat[pmatch.rm_so - 1],
549 				    "%lc", &wbegin) != 1)
550 					r = REG_NOMATCH;
551 				else if ((size_t)pmatch.rm_eo !=
552 				    pc->ln.len &&
553 				    sscanf(&pc->ln.dat[pmatch.rm_eo],
554 				    "%lc", &wend) != 1)
555 					r = REG_NOMATCH;
556 				else if (iswword(wbegin) ||
557 				    iswword(wend))
558 					r = REG_NOMATCH;
559 				/*
560 				 * If we're doing whole word matching and we
561 				 * matched once, then we should try the pattern
562 				 * again after advancing just past the start of
563 				 * the earliest match. This allows the pattern
564 				 * to  match later on in the line and possibly
565 				 * still match a whole word.
566 				 */
567 				if (r == REG_NOMATCH &&
568 				    (retry == pc->lnstart ||
569 				    (unsigned int)pmatch.rm_so + 1 < retry))
570 					retry = pmatch.rm_so + 1;
571 				if (r == REG_NOMATCH)
572 					continue;
573 			}
574 			lastmatched = true;
575 			lastmatch = pmatch;
576 
577 			if (matchidx == 0)
578 				matched = true;
579 
580 			/*
581 			 * Replace previous match if the new one is earlier
582 			 * and/or longer. This will lead to some amount of
583 			 * extra work if -o/--color are specified, but it's
584 			 * worth it from a correctness point of view.
585 			 */
586 			if (matchidx > startm) {
587 				chkmatch = pc->matches[matchidx - 1];
588 				if (pmatch.rm_so < chkmatch.rm_so ||
589 				    (pmatch.rm_so == chkmatch.rm_so &&
590 				    (pmatch.rm_eo - pmatch.rm_so) >
591 				    (chkmatch.rm_eo - chkmatch.rm_so))) {
592 					pc->matches[matchidx - 1] = pmatch;
593 					nst = pmatch.rm_eo;
594 				}
595 			} else {
596 				/* Advance as normal if not */
597 				pc->matches[matchidx++] = pmatch;
598 				nst = pmatch.rm_eo;
599 			}
600 			/* avoid excessive matching - skip further patterns */
601 			if ((color == NULL && !oflag) || qflag || lflag ||
602 			    matchidx >= MAX_MATCHES) {
603 				pc->lnstart = nst;
604 				lastmatched = false;
605 				break;
606 			}
607 		}
608 
609 		/*
610 		 * Advance to just past the start of the earliest match, try
611 		 * again just in case we still have a chance to match later in
612 		 * the string.
613 		 */
614 		if (!lastmatched && retry > pc->lnstart) {
615 			st = retry;
616 			continue;
617 		}
618 
619 		/* XXX TODO: We will need to keep going, since we're chunky */
620 		/* One pass if we are not recording matches */
621 		if (!wflag && ((color == NULL && !oflag) || qflag || lflag || Lflag))
622 			break;
623 
624 		/* If we didn't have any matches or REG_NOSUB set */
625 		if (!lastmatched || (cflags & REG_NOSUB))
626 			nst = pc->ln.len;
627 
628 		if (!lastmatched)
629 			/* No matches */
630 			break;
631 		else if (st == nst && lastmatch.rm_so == lastmatch.rm_eo)
632 			/* Zero-length match -- advance one more so we don't get stuck */
633 			nst++;
634 
635 		/* Advance st based on previous matches */
636 		st = nst;
637 		pc->lnstart = st;
638 	}
639 
640 	/* Reflect the new matchidx in the context */
641 	pc->matchidx = matchidx;
642 	return matched;
643 }
644 
645 /*
646  * Safe malloc() for internal use.
647  */
648 void *
649 grep_malloc(size_t size)
650 {
651 	void *ptr;
652 
653 	if (size == 0)
654 		return (NULL);
655 	if ((ptr = malloc(size)) == NULL)
656 		err(2, "malloc");
657 	return (ptr);
658 }
659 
660 /*
661  * Safe calloc() for internal use.
662  */
663 void *
664 grep_calloc(size_t nmemb, size_t size)
665 {
666 	void *ptr;
667 
668 	if (nmemb == 0 || size == 0)
669 		return (NULL);
670 	if ((ptr = calloc(nmemb, size)) == NULL)
671 		err(2, "calloc");
672 	return (ptr);
673 }
674 
675 /*
676  * Safe realloc() for internal use.
677  */
678 void *
679 grep_realloc(void *ptr, size_t size)
680 {
681 
682 	if ((ptr = realloc(ptr, size)) == NULL)
683 		err(2, "realloc");
684 	return (ptr);
685 }
686 
687 /*
688  * Safe strdup() for internal use.
689  */
690 char *
691 grep_strdup(const char *str)
692 {
693 	char *ret;
694 
695 	if ((ret = strdup(str)) == NULL)
696 		err(2, "strdup");
697 	return (ret);
698 }
699 
700 /*
701  * Print an entire line as-is, there are no inline matches to consider. This is
702  * used for printing context.
703  */
704 void grep_printline(struct str *line, int sep) {
705 	printline_metadata(line, sep);
706 	fwrite(line->dat, line->len, 1, stdout);
707 	putchar(fileeol);
708 }
709 
710 static void
711 printline_metadata(struct str *line, int sep)
712 {
713 	bool printsep;
714 
715 	printsep = false;
716 	if (!hflag) {
717 		if (!nullflag) {
718 			fputs(line->file, stdout);
719 			printsep = true;
720 		} else {
721 			printf("%s", line->file);
722 			putchar(0);
723 		}
724 	}
725 	if (nflag) {
726 		if (printsep)
727 			putchar(sep);
728 		printf("%d", line->line_no);
729 		printsep = true;
730 	}
731 	if (bflag) {
732 		if (printsep)
733 			putchar(sep);
734 		printf("%lld", (long long)(line->off + line->boff));
735 		printsep = true;
736 	}
737 	if (printsep)
738 		putchar(sep);
739 }
740 
741 /*
742  * Prints a matching line according to the command line options.
743  */
744 static void
745 printline(struct parsec *pc, int sep)
746 {
747 	size_t a = 0;
748 	size_t i, matchidx;
749 	regmatch_t match;
750 
751 	/* If matchall, everything matches but don't actually print for -o */
752 	if (oflag && matchall)
753 		return;
754 
755 	matchidx = pc->matchidx;
756 
757 	/* --color and -o */
758 	if ((oflag || color) && matchidx > 0) {
759 		/* Only print metadata once per line if --color */
760 		if (!oflag && pc->printed == 0)
761 			printline_metadata(&pc->ln, sep);
762 		for (i = 0; i < matchidx; i++) {
763 			match = pc->matches[i];
764 			/* Don't output zero length matches */
765 			if (match.rm_so == match.rm_eo)
766 				continue;
767 			/*
768 			 * Metadata is printed on a per-line basis, so every
769 			 * match gets file metadata with the -o flag.
770 			 */
771 			if (oflag) {
772 				pc->ln.boff = match.rm_so;
773 				printline_metadata(&pc->ln, sep);
774 			} else
775 				fwrite(pc->ln.dat + a, match.rm_so - a, 1,
776 				    stdout);
777 			if (color)
778 				fprintf(stdout, "\33[%sm\33[K", color);
779 			fwrite(pc->ln.dat + match.rm_so,
780 			    match.rm_eo - match.rm_so, 1, stdout);
781 			if (color)
782 				fprintf(stdout, "\33[m\33[K");
783 			a = match.rm_eo;
784 			if (oflag)
785 				putchar('\n');
786 		}
787 		if (!oflag) {
788 			if (pc->ln.len - a > 0)
789 				fwrite(pc->ln.dat + a, pc->ln.len - a, 1,
790 				    stdout);
791 			putchar('\n');
792 		}
793 	} else
794 		grep_printline(&pc->ln, sep);
795 	pc->printed++;
796 }
797