Context Navigation

← Previous Revision
Latest Revision
Next Revision →
Normal
Revision Log

grep.c@ 3621

Visit:

Last change on this file since 3621 was 3621, checked in by Bruno Cornec, 7 years ago
New 3?3 banch for incorporation of latest busybox 1.25. Changing minor version to handle potential incompatibilities.
File size: 25.2 KB

Rev	Line
[821]	1	/* vi: set sw=4 ts=4: */
	2	/*
	3	* Mini grep implementation for busybox using libc regex.
	4	*
	5	* Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
	6	* Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
	7	*
[2725]	8	* Licensed under GPLv2 or later, see file LICENSE in this source tree.
[821]	9	*/
[2725]	10	/* BB_AUDIT SUSv3 defects - unsupported option -x "match whole line only". */
[821]	11	/* BB_AUDIT GNU defects - always acts as -a. */
	12	/* http://www.opengroup.org/onlinepubs/007904975/utilities/grep.html */
	13	/*
	14	* 2004,2006 (C) Vladimir Oleynik <dzo@simtreas.ru> -
	15	* correction "-e pattern1 -e pattern2" logic and more optimizations.
	16	* precompiled regex
[2725]	17	*
[1765]	18	* (C) 2006 Jac Goudsmit added -o option
	19	*/
[821]	20
[3232]	21	//applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP))
	22	//applet:IF_FEATURE_GREP_EGREP_ALIAS(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep))
	23	//applet:IF_FEATURE_GREP_FGREP_ALIAS(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep))
[2725]	24
	25	//kbuild:lib-$(CONFIG_GREP) += grep.o
	26
	27	//config:config GREP
	28	//config: bool "grep"
	29	//config: default y
	30	//config: help
	31	//config: grep is used to search files for a specified pattern.
	32	//config:
	33	//config:config FEATURE_GREP_EGREP_ALIAS
	34	//config: bool "Enable extended regular expressions (egrep & grep -E)"
	35	//config: default y
	36	//config: depends on GREP
	37	//config: help
	38	//config: Enabled support for extended regular expressions. Extended
	39	//config: regular expressions allow for alternation (foo\|bar), grouping,
	40	//config: and various repetition operators.
	41	//config:
	42	//config:config FEATURE_GREP_FGREP_ALIAS
	43	//config: bool "Alias fgrep to grep -F"
	44	//config: default y
	45	//config: depends on GREP
	46	//config: help
	47	//config: fgrep sees the search pattern as a normal string rather than
	48	//config: regular expressions.
	49	//config: grep -F always works, this just creates the fgrep alias.
	50	//config:
	51	//config:config FEATURE_GREP_CONTEXT
	52	//config: bool "Enable before and after context flags (-A, -B and -C)"
	53	//config: default y
	54	//config: depends on GREP
	55	//config: help
	56	//config: Print the specified number of leading (-B) and/or trailing (-A)
	57	//config: context surrounding our matching lines.
	58	//config: Print the specified number of context lines (-C).
	59
[1765]	60	#include "libbb.h"
[3621]	61	#include "common_bufsiz.h"
[821]	62	#include "xregex.h"
	63
[2725]	64
[821]	65	/* options */
[2725]	66	//usage:#define grep_trivial_usage
	67	//usage: "[-HhnlLoqvsriw"
	68	//usage: "F"
	69	//usage: IF_FEATURE_GREP_EGREP_ALIAS("E")
	70	//usage: IF_EXTRA_COMPAT("z")
	71	//usage: "] [-m N] "
	72	//usage: IF_FEATURE_GREP_CONTEXT("[-A/B/C N] ")
	73	//usage: "PATTERN/-e PATTERN.../-f FILE [FILE]..."
	74	//usage:#define grep_full_usage "\n\n"
	75	//usage: "Search for PATTERN in FILEs (or stdin)\n"
	76	//usage: "\n -H Add 'filename:' prefix"
	77	//usage: "\n -h Do not add 'filename:' prefix"
	78	//usage: "\n -n Add 'line_no:' prefix"
	79	//usage: "\n -l Show only names of files that match"
	80	//usage: "\n -L Show only names of files that don't match"
	81	//usage: "\n -c Show only count of matching lines"
	82	//usage: "\n -o Show only the matching part of line"
	83	//usage: "\n -q Quiet. Return 0 if PATTERN is found, 1 otherwise"
	84	//usage: "\n -v Select non-matching lines"
	85	//usage: "\n -s Suppress open and read errors"
	86	//usage: "\n -r Recurse"
	87	//usage: "\n -i Ignore case"
	88	//usage: "\n -w Match whole words only"
[3232]	89	//usage: "\n -x Match whole lines only"
[2725]	90	//usage: "\n -F PATTERN is a literal (not regexp)"
	91	//usage: IF_FEATURE_GREP_EGREP_ALIAS(
	92	//usage: "\n -E PATTERN is an extended regexp"
	93	//usage: )
	94	//usage: IF_EXTRA_COMPAT(
	95	//usage: "\n -z Input is NUL terminated"
	96	//usage: )
	97	//usage: "\n -m N Match up to N times per file"
	98	//usage: IF_FEATURE_GREP_CONTEXT(
	99	//usage: "\n -A N Print N lines of trailing context"
	100	//usage: "\n -B N Print N lines of leading context"
	101	//usage: "\n -C N Same as '-A N -B N'"
	102	//usage: )
	103	//usage: "\n -e PTRN Pattern to match"
	104	//usage: "\n -f FILE Read pattern from file"
	105	//usage:
	106	//usage:#define grep_example_usage
	107	//usage: "$ grep root /etc/passwd\n"
	108	//usage: "root:x:0:0:root:/root:/bin/bash\n"
	109	//usage: "$ grep ^[rR]oo. /etc/passwd\n"
	110	//usage: "root:x:0:0:root:/root:/bin/bash\n"
	111	//usage:
	112	//usage:#define egrep_trivial_usage NOUSAGE_STR
	113	//usage:#define egrep_full_usage ""
	114	//usage:#define fgrep_trivial_usage NOUSAGE_STR
	115	//usage:#define fgrep_full_usage ""
	116
[1765]	117	#define OPTSTR_GREP \
[3232]	118	"lnqvscFiHhe:f:Lorm:wx" \
[2725]	119	IF_FEATURE_GREP_CONTEXT("A:B:C:") \
	120	IF_FEATURE_GREP_EGREP_ALIAS("E") \
	121	IF_EXTRA_COMPAT("z") \
[1765]	122	"aI"
	123	/* ignored: -a "assume all files to be text" */
	124	/* ignored: -I "assume binary files have no matches" */
	125	enum {
	126	OPTBIT_l, /* list matched file names only */
	127	OPTBIT_n, /* print line# */
[2725]	128	OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */
[1765]	129	OPTBIT_v, /* invert the match, to select non-matching lines */
	130	OPTBIT_s, /* suppress errors about file open errors */
	131	OPTBIT_c, /* count matches per file (suppresses normal output) */
	132	OPTBIT_F, /* literal match */
	133	OPTBIT_i, /* case-insensitive */
	134	OPTBIT_H, /* force filename display */
	135	OPTBIT_h, /* inhibit filename display */
	136	OPTBIT_e, /* -e PATTERN */
	137	OPTBIT_f, /* -f FILE_WITH_PATTERNS */
	138	OPTBIT_L, /* list unmatched file names only */
	139	OPTBIT_o, /* show only matching parts of lines */
	140	OPTBIT_r, /* recurse dirs */
	141	OPTBIT_m, /* -m MAX_MATCHES */
[2725]	142	OPTBIT_w, /* -w whole word match */
[3232]	143	OPTBIT_x, /* -x whole line match */
[2725]	144	IF_FEATURE_GREP_CONTEXT( OPTBIT_A ,) /* -A NUM: after-match context */
	145	IF_FEATURE_GREP_CONTEXT( OPTBIT_B ,) /* -B NUM: before-match context */
	146	IF_FEATURE_GREP_CONTEXT( OPTBIT_C ,) /* -C NUM: -A and -B combined */
	147	IF_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
	148	IF_EXTRA_COMPAT( OPTBIT_z ,) /* input is NUL terminated */
[1765]	149	OPT_l = 1 << OPTBIT_l,
	150	OPT_n = 1 << OPTBIT_n,
	151	OPT_q = 1 << OPTBIT_q,
	152	OPT_v = 1 << OPTBIT_v,
	153	OPT_s = 1 << OPTBIT_s,
	154	OPT_c = 1 << OPTBIT_c,
	155	OPT_F = 1 << OPTBIT_F,
	156	OPT_i = 1 << OPTBIT_i,
	157	OPT_H = 1 << OPTBIT_H,
	158	OPT_h = 1 << OPTBIT_h,
	159	OPT_e = 1 << OPTBIT_e,
	160	OPT_f = 1 << OPTBIT_f,
	161	OPT_L = 1 << OPTBIT_L,
	162	OPT_o = 1 << OPTBIT_o,
	163	OPT_r = 1 << OPTBIT_r,
	164	OPT_m = 1 << OPTBIT_m,
[2725]	165	OPT_w = 1 << OPTBIT_w,
[3232]	166	OPT_x = 1 << OPTBIT_x,
[2725]	167	OPT_A = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_A)) + 0,
	168	OPT_B = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_B)) + 0,
	169	OPT_C = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_C)) + 0,
	170	OPT_E = IF_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
	171	OPT_z = IF_EXTRA_COMPAT( (1 << OPTBIT_z)) + 0,
[1765]	172	};
	173
	174	#define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
	175	#define PRINT_LINE_NUM (option_mask32 & OPT_n)
	176	#define BE_QUIET (option_mask32 & OPT_q)
	177	#define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
	178	#define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
	179	#define FGREP_FLAG (option_mask32 & OPT_F)
	180	#define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
[2725]	181	#define NUL_DELIMITED (option_mask32 & OPT_z)
[1765]	182
[2725]	183	struct globals {
	184	int max_matches;
	185	#if !ENABLE_EXTRA_COMPAT
	186	int reflags;
	187	#else
	188	RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
	189	#endif
	190	smalluint invert_search;
	191	smalluint print_filename;
	192	smalluint open_errors;
	193	#if ENABLE_FEATURE_GREP_CONTEXT
	194	smalluint did_print_line;
	195	int lines_before;
	196	int lines_after;
	197	char **before_buf;
	198	IF_EXTRA_COMPAT(size_t *before_buf_size;)
	199	int last_line_printed;
	200	#endif
	201	/* globals used internally */
	202	llist_t pattern_head; / growable list of patterns to match */
	203	const char cur_file; / the current file we are reading */
	204	} FIX_ALIASING;
[3621]	205	#define G ((struct globals)bb_common_bufsiz1)
[2725]	206	#define INIT_G() do { \
[3621]	207	setup_common_bufsiz(); \
	208	BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
[2725]	209	} while (0)
	210	#define max_matches (G.max_matches )
	211	#if !ENABLE_EXTRA_COMPAT
	212	# define reflags (G.reflags )
	213	#else
	214	# define case_fold (G.case_fold )
	215	/* http://www.delorie.com/gnu/docs/regex/regex_46.html */
	216	# define reflags re_syntax_options
	217	# undef REG_NOSUB
	218	# undef REG_EXTENDED
	219	# undef REG_ICASE
	220	# define REG_NOSUB bug:is:here /* should not be used */
	221	/* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */
	222	# define REG_EXTENDED (RE_SYNTAX_EGREP \| RE_INTERVALS \| RE_NO_BK_BRACES)
	223	# define REG_ICASE bug:is:here /* should not be used */
	224	#endif
	225	#define invert_search (G.invert_search )
	226	#define print_filename (G.print_filename )
	227	#define open_errors (G.open_errors )
	228	#define did_print_line (G.did_print_line )
	229	#define lines_before (G.lines_before )
	230	#define lines_after (G.lines_after )
	231	#define before_buf (G.before_buf )
	232	#define before_buf_size (G.before_buf_size )
	233	#define last_line_printed (G.last_line_printed )
	234	#define pattern_head (G.pattern_head )
	235	#define cur_file (G.cur_file )
[1765]	236
[821]	237
[1765]	238	typedef struct grep_list_data_t {
[821]	239	char *pattern;
[2725]	240	/* for GNU regex, matched_range must be persistent across grep_file() calls */
	241	#if !ENABLE_EXTRA_COMPAT
	242	regex_t compiled_regex;
	243	regmatch_t matched_range;
	244	#else
	245	struct re_pattern_buffer compiled_regex;
	246	struct re_registers matched_range;
	247	#endif
	248	#define ALLOCATED 1
[821]	249	#define COMPILED 2
	250	int flg_mem_alocated_compiled;
	251	} grep_list_data_t;
	252
[2725]	253	#if !ENABLE_EXTRA_COMPAT
	254	#define print_line(line, line_len, linenum, decoration) \
	255	print_line(line, linenum, decoration)
	256	#endif
	257	static void print_line(const char *line, size_t line_len, int linenum, char decoration)
[821]	258	{
	259	#if ENABLE_FEATURE_GREP_CONTEXT
[1765]	260	/* Happens when we go to next file, immediately hit match
	261	* and try to print prev context... from prev file! Don't do it */
	262	if (linenum < 1)
	263	return;
[821]	264	/* possibly print the little '--' separator */
[2725]	265	if ((lines_before \|\| lines_after) && did_print_line
	266	&& last_line_printed != linenum - 1
	267	) {
[821]	268	puts("--");
	269	}
[1765]	270	/* guard against printing "--" before first line of first file */
	271	did_print_line = 1;
[821]	272	last_line_printed = linenum;
	273	#endif
[1765]	274	if (print_filename)
[821]	275	printf("%s%c", cur_file, decoration);
	276	if (PRINT_LINE_NUM)
	277	printf("%i%c", linenum, decoration);
[1765]	278	/* Emulate weird GNU grep behavior with -ov */
[2725]	279	if ((option_mask32 & (OPT_v\|OPT_o)) != (OPT_v\|OPT_o)) {
	280	#if !ENABLE_EXTRA_COMPAT
[1765]	281	puts(line);
[2725]	282	#else
	283	fwrite(line, 1, line_len, stdout);
	284	putchar(NUL_DELIMITED ? '\0' : '\n');
	285	#endif
	286	}
[821]	287	}
	288
[2725]	289	#if ENABLE_EXTRA_COMPAT
	290	/* Unlike getline, this one removes trailing '\n' */
	291	static ssize_t FAST_FUNC bb_getline(char *line_ptr, size_t line_alloc_len, FILE *file)
	292	{
	293	ssize_t res_sz;
	294	char *line;
	295	int delim = (NUL_DELIMITED ? '\0' : '\n');
	296
	297	res_sz = getdelim(line_ptr, line_alloc_len, delim, file);
	298	line = *line_ptr;
	299
	300	if (res_sz > 0) {
	301	if (line[res_sz - 1] == delim)
	302	line[--res_sz] = '\0';
	303	} else {
	304	free(line); /* uclibc allocates a buffer even on EOF. WTF? */
	305	}
	306	return res_sz;
	307	}
	308	#endif
	309
[821]	310	static int grep_file(FILE *file)
	311	{
[2725]	312	smalluint found;
[821]	313	int linenum = 0;
	314	int nmatches = 0;
[2725]	315	#if !ENABLE_EXTRA_COMPAT
	316	char *line;
	317	#else
	318	char *line = NULL;
	319	ssize_t line_len;
	320	size_t line_alloc_len;
	321	# define rm_so start[0]
	322	# define rm_eo end[0]
	323	#endif
[821]	324	#if ENABLE_FEATURE_GREP_CONTEXT
	325	int print_n_lines_after = 0;
	326	int curpos = 0; /* track where we are in the circular 'before' buffer */
	327	int idx = 0; /* used for iteration through the circular buffer */
[1765]	328	#else
	329	enum { print_n_lines_after = 0 };
[2725]	330	#endif
[821]	331
[2725]	332	while (
	333	#if !ENABLE_EXTRA_COMPAT
	334	(line = xmalloc_fgetline(file)) != NULL
	335	#else
	336	(line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
	337	#endif
	338	) {
[821]	339	llist_t *pattern_ptr = pattern_head;
[2725]	340	grep_list_data_t gl = gl; / for gcc */
[821]	341
	342	linenum++;
[2725]	343	found = 0;
[821]	344	while (pattern_ptr) {
	345	gl = (grep_list_data_t *)pattern_ptr->data;
	346	if (FGREP_FLAG) {
[3232]	347	char *match;
	348	char *str = line;
	349	opt_f_again:
	350	match = ((option_mask32 & OPT_i)
	351	? strcasestr(str, gl->pattern)
	352	: strstr(str, gl->pattern)
	353	);
	354	if (match) {
	355	if (option_mask32 & OPT_x) {
	356	if (match != str)
	357	goto opt_f_not_found;
	358	if (str[strlen(gl->pattern)] != '\0')
	359	goto opt_f_not_found;
	360	} else
	361	if (option_mask32 & OPT_w) {
	362	char c = (match != str) ? match[-1] : ' ';
	363	if (!isalnum(c) && c != '_') {
	364	c = match[strlen(gl->pattern)];
	365	if (!c \|\| (!isalnum(c) && c != '_'))
	366	goto opt_f_found;
	367	}
	368	str = match + 1;
	369	goto opt_f_again;
	370	}
	371	opt_f_found:
	372	found = 1;
	373	opt_f_not_found: ;
	374	}
[821]	375	} else {
[3621]	376	#if ENABLE_EXTRA_COMPAT
	377	unsigned start_pos;
	378	#else
	379	int match_flg;
	380	#endif
	381	char *match_at;
	382
[1765]	383	if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
[821]	384	gl->flg_mem_alocated_compiled \|= COMPILED;
[2725]	385	#if !ENABLE_EXTRA_COMPAT
	386	xregcomp(&gl->compiled_regex, gl->pattern, reflags);
	387	#else
	388	memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
	389	gl->compiled_regex.translate = case_fold; /* for -i */
	390	if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
	391	bb_error_msg_and_die("bad regex '%s'", gl->pattern);
	392	#endif
[821]	393	}
[2725]	394	#if !ENABLE_EXTRA_COMPAT
	395	gl->matched_range.rm_so = 0;
	396	gl->matched_range.rm_eo = 0;
[3621]	397	match_flg = 0;
	398	#else
	399	start_pos = 0;
[2725]	400	#endif
[3621]	401	match_at = line;
	402	opt_w_again:
	403	//bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
[2725]	404	if (
	405	#if !ENABLE_EXTRA_COMPAT
[3621]	406	regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
[2725]	407	#else
[3621]	408	re_search(&gl->compiled_regex, match_at, line_len,
	409	start_pos, /range:/ line_len,
[2725]	410	&gl->matched_range) >= 0
	411	#endif
	412	) {
[3232]	413	if (option_mask32 & OPT_x) {
	414	found = (gl->matched_range.rm_so == 0
[3621]	415	&& match_at[gl->matched_range.rm_eo] == '\0');
[3232]	416	} else
	417	if (!(option_mask32 & OPT_w)) {
[2725]	418	found = 1;
[3232]	419	} else {
[1765]	420	char c = ' ';
[3621]	421	if (match_at > line \|\| gl->matched_range.rm_so != 0) {
	422	c = match_at[gl->matched_range.rm_so - 1];
	423	}
[1765]	424	if (!isalnum(c) && c != '_') {
[3621]	425	c = match_at[gl->matched_range.rm_eo];
[1765]	426	}
[3621]	427	if (!isalnum(c) && c != '_') {
	428	found = 1;
	429	} else {
	430	/*
	431	* Why check gl->matched_range.rm_eo?
	432	* Zero-length match makes -w skip the line:
	433	* "echo foo \| grep ^" prints "foo",
	434	* "echo foo \| grep -w ^" prints nothing.
	435	* Without such check, we can loop forever.
	436	*/
	437	#if !ENABLE_EXTRA_COMPAT
	438	if (gl->matched_range.rm_eo != 0) {
	439	match_at += gl->matched_range.rm_eo;
	440	match_flg \|= REG_NOTBOL;
	441	goto opt_w_again;
	442	}
	443	#else
	444	if (gl->matched_range.rm_eo > start_pos) {
	445	start_pos = gl->matched_range.rm_eo;
	446	goto opt_w_again;
	447	}
	448	#endif
	449	}
[1765]	450	}
	451	}
[821]	452	}
[2725]	453	/* If it's non-inverted search, we can stop
	454	* at first match */
	455	if (found && !invert_search)
	456	goto do_found;
[821]	457	pattern_ptr = pattern_ptr->link;
	458	} /* while (pattern_ptr) */
	459
[2725]	460	if (found ^ invert_search) {
	461	do_found:
[1765]	462	/* keep track of matches */
	463	nmatches++;
[821]	464
[1765]	465	/* quiet/print (non)matching file names only? */
	466	if (option_mask32 & (OPT_q\|OPT_l\|OPT_L)) {
	467	free(line); /* we don't need line anymore */
	468	if (BE_QUIET) {
	469	/* manpage says about -q:
	470	* "exit immediately with zero status
	471	* if any match is found,
	472	* even if errors were detected" */
[2725]	473	exit(EXIT_SUCCESS);
[1765]	474	}
[821]	475	/* if we're just printing filenames, we stop after the first match */
[1765]	476	if (PRINT_FILES_WITH_MATCHES) {
	477	puts(cur_file);
[2725]	478	/* fall through to "return 1" */
[1765]	479	}
	480	/* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
	481	return 1; /* one match */
	482	}
[821]	483
	484	#if ENABLE_FEATURE_GREP_CONTEXT
[1765]	485	/* Were we printing context and saw next (unwanted) match? */
	486	if ((option_mask32 & OPT_m) && nmatches > max_matches)
	487	break;
	488	#endif
[821]	489
[1765]	490	/* print the matched line */
	491	if (PRINT_MATCH_COUNTS == 0) {
	492	#if ENABLE_FEATURE_GREP_CONTEXT
	493	int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
[821]	494
[1765]	495	/* if we were told to print 'before' lines and there is at least
	496	* one line in the circular buffer, print them */
	497	if (lines_before && before_buf[prevpos] != NULL) {
	498	int first_buf_entry_line_num = linenum - lines_before;
[821]	499
[1765]	500	/* advance to the first entry in the circular buffer, and
	501	* figure out the line number is of the first line in the
	502	* buffer */
	503	idx = curpos;
	504	while (before_buf[idx] == NULL) {
	505	idx = (idx + 1) % lines_before;
	506	first_buf_entry_line_num++;
[821]	507	}
	508
[1765]	509	/* now print each line in the buffer, clearing them as we go */
	510	while (before_buf[idx] != NULL) {
[2725]	511	print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
[1765]	512	free(before_buf[idx]);
	513	before_buf[idx] = NULL;
	514	idx = (idx + 1) % lines_before;
	515	first_buf_entry_line_num++;
	516	}
	517	}
	518
	519	/* make a note that we need to print 'after' lines */
	520	print_n_lines_after = lines_after;
[821]	521	#endif
[1765]	522	if (option_mask32 & OPT_o) {
[2725]	523	if (FGREP_FLAG) {
	524	/* -Fo just prints the pattern
	525	* (unless -v: -Fov doesnt print anything at all) */
	526	if (found)
	527	print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
	528	} else while (1) {
	529	unsigned start = gl->matched_range.rm_so;
	530	unsigned end = gl->matched_range.rm_eo;
	531	unsigned len = end - start;
	532	char old = line[end];
	533	line[end] = '\0';
	534	/* Empty match is not printed: try "echo test \| grep -o ''" */
	535	if (len != 0)
	536	print_line(line + start, len, linenum, ':');
	537	if (old == '\0')
	538	break;
	539	line[end] = old;
	540	if (len == 0)
	541	end++;
	542	#if !ENABLE_EXTRA_COMPAT
	543	if (regexec(&gl->compiled_regex, line + end,
	544	1, &gl->matched_range, REG_NOTBOL) != 0)
	545	break;
	546	gl->matched_range.rm_so += end;
	547	gl->matched_range.rm_eo += end;
	548	#else
	549	if (re_search(&gl->compiled_regex, line, line_len,
	550	end, line_len - end,
	551	&gl->matched_range) < 0)
	552	break;
	553	#endif
	554	}
[1765]	555	} else {
[2725]	556	print_line(line, line_len, linenum, ':');
[821]	557	}
	558	}
[1765]	559	}
[821]	560	#if ENABLE_FEATURE_GREP_CONTEXT
[1765]	561	else { /* no match */
[821]	562	/* if we need to print some context lines after the last match, do so */
[1765]	563	if (print_n_lines_after) {
[2725]	564	print_line(line, strlen(line), linenum, '-');
[821]	565	print_n_lines_after--;
[1765]	566	} else if (lines_before) {
	567	/* Add the line to the circular 'before' buffer */
	568	free(before_buf[curpos]);
	569	before_buf[curpos] = line;
[2725]	570	IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
[1765]	571	curpos = (curpos + 1) % lines_before;
[2725]	572	/* avoid free(line) - we took the line */
[1765]	573	line = NULL;
[821]	574	}
[1765]	575	}
	576
[821]	577	#endif /* ENABLE_FEATURE_GREP_CONTEXT */
[2725]	578	#if !ENABLE_EXTRA_COMPAT
[821]	579	free(line);
[2725]	580	#endif
[1765]	581	/* Did we print all context after last requested match? */
	582	if ((option_mask32 & OPT_m)
[2725]	583	&& !print_n_lines_after
	584	&& nmatches == max_matches
	585	) {
[1765]	586	break;
[2725]	587	}
	588	} /* while (read line) */
[821]	589
	590	/* special-case file post-processing for options where we don't print line
	591	* matches, just filenames and possibly match counts */
	592
	593	/* grep -c: print [filename:]count, even if count is zero */
	594	if (PRINT_MATCH_COUNTS) {
[1765]	595	if (print_filename)
[821]	596	printf("%s:", cur_file);
[1765]	597	printf("%d\n", nmatches);
[821]	598	}
	599
[1765]	600	/* grep -L: print just the filename */
	601	if (PRINT_FILES_WITHOUT_MATCHES) {
	602	/* nmatches is zero, no need to check it:
	603	* we return 1 early if we detected a match
	604	* and PRINT_FILES_WITHOUT_MATCHES is set */
[821]	605	puts(cur_file);
	606	}
	607
	608	return nmatches;
	609	}
	610
	611	#if ENABLE_FEATURE_CLEAN_UP
	612	#define new_grep_list_data(p, m) add_grep_list_data(p, m)
[2725]	613	static char add_grep_list_data(char pattern, int flg_used_mem)
[821]	614	#else
	615	#define new_grep_list_data(p, m) add_grep_list_data(p)
[2725]	616	static char add_grep_list_data(char pattern)
[821]	617	#endif
	618	{
[2725]	619	grep_list_data_t gl = xzalloc(sizeof(gl));
[821]	620	gl->pattern = pattern;
	621	#if ENABLE_FEATURE_CLEAN_UP
	622	gl->flg_mem_alocated_compiled = flg_used_mem;
	623	#else
[2725]	624	/gl->flg_mem_alocated_compiled = 0;/
[821]	625	#endif
	626	return (char *)gl;
	627	}
	628
	629	static void load_regexes_from_file(llist_t *fopt)
	630	{
[1765]	631	while (fopt) {
[3232]	632	char *line;
	633	FILE *fp;
[821]	634	llist_t *cur = fopt;
	635	char *ffile = cur->data;
	636
	637	fopt = cur->link;
	638	free(cur);
[3232]	639	fp = xfopen_stdin(ffile);
	640	while ((line = xmalloc_fgetline(fp)) != NULL) {
[821]	641	llist_add_to(&pattern_head,
[2725]	642	new_grep_list_data(line, ALLOCATED));
[821]	643	}
[3232]	644	fclose_if_not_stdin(fp);
[821]	645	}
	646	}
	647
[2725]	648	static int FAST_FUNC file_action_grep(const char *filename,
	649	struct stat *statbuf UNUSED_PARAM,
	650	void* matched,
	651	int depth UNUSED_PARAM)
[1765]	652	{
[2725]	653	FILE *file = fopen_for_read(filename);
[1765]	654	if (file == NULL) {
	655	if (!SUPPRESS_ERR_MSGS)
[2725]	656	bb_simple_perror_msg(filename);
[1765]	657	open_errors = 1;
	658	return 0;
	659	}
	660	cur_file = filename;
	661	(int)matched += grep_file(file);
	662	fclose(file);
	663	return 1;
	664	}
[821]	665
[1765]	666	static int grep_dir(const char *dir)
	667	{
	668	int matched = 0;
	669	recursive_action(dir,
	670	/* recurse=yes */ ACTION_RECURSE \|
[3621]	671	/* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 \|
[1765]	672	/* depthFirst=yes */ ACTION_DEPTHFIRST,
	673	/* fileAction= */ file_action_grep,
	674	/* dirAction= */ NULL,
	675	/* userData= */ &matched,
	676	/* depth= */ 0);
	677	return matched;
	678	}
	679
[2725]	680	int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
	681	int grep_main(int argc UNUSED_PARAM, char **argv)
[821]	682	{
	683	FILE *file;
	684	int matched;
	685	llist_t *fopt = NULL;
[3621]	686	#if ENABLE_FEATURE_GREP_CONTEXT
	687	int Copt, opts;
	688	#endif
	689	INIT_G();
[821]	690
[3621]	691	/* For grep, exitcode of 1 is "not found". Other errors are 2: */
	692	xfunc_error_retval = 2;
	693
[821]	694	/* do normal option parsing */
	695	#if ENABLE_FEATURE_GREP_CONTEXT
[2725]	696	/* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
	697	* -m,-A,-B,-C have numeric param */
	698	opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
	699	opts = getopt32(argv,
[1765]	700	OPTSTR_GREP,
[2725]	701	&pattern_head, &fopt, &max_matches,
	702	&lines_after, &lines_before, &Copt);
[821]	703
[2725]	704	if (opts & OPT_C) {
[1765]	705	/* -C unsets prev -A and -B, but following -A or -B
[3232]	706	* may override it */
[2725]	707	if (!(opts & OPT_A)) /* not overridden */
	708	lines_after = Copt;
	709	if (!(opts & OPT_B)) /* not overridden */
	710	lines_before = Copt;
[821]	711	}
[1765]	712	/* sanity checks */
[2725]	713	if (opts & (OPT_c\|OPT_q\|OPT_l\|OPT_L)) {
[1765]	714	option_mask32 &= ~OPT_n;
[821]	715	lines_before = 0;
	716	lines_after = 0;
[2725]	717	} else if (lines_before > 0) {
	718	if (lines_before > INT_MAX / sizeof(long long))
	719	lines_before = INT_MAX / sizeof(long long);
	720	/* overflow in (lines_before * sizeof(x)) is prevented (above) */
	721	before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
	722	IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
	723	}
[821]	724	#else
	725	/* with auto sanity checks */
[2725]	726	/* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
	727	opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
[1765]	728	getopt32(argv, OPTSTR_GREP,
[2725]	729	&pattern_head, &fopt, &max_matches);
[821]	730	#endif
[1765]	731	invert_search = ((option_mask32 & OPT_v) != 0); /* 0 \| 1 */
[821]	732
[3232]	733	{ /* convert char *argv to grep_list_data_t /
[821]	734	llist_t *cur;
[1765]	735	for (cur = pattern_head; cur; cur = cur->link)
[821]	736	cur->data = new_grep_list_data(cur->data, 0);
	737	}
[3232]	738	if (option_mask32 & OPT_f) {
[821]	739	load_regexes_from_file(fopt);
[3232]	740	if (!pattern_head) { /* -f EMPTY_FILE? */
	741	/* GNU grep treats it as "nothing matches" */
	742	llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0));
	743	invert_search ^= 1;
	744	}
	745	}
[821]	746
[1765]	747	if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
	748	option_mask32 \|= OPT_F;
[821]	749
[2725]	750	#if !ENABLE_EXTRA_COMPAT
[3621]	751	if (!(option_mask32 & (OPT_o \| OPT_w \| OPT_x)))
[821]	752	reflags = REG_NOSUB;
[2725]	753	#endif
[821]	754
[2725]	755	if (ENABLE_FEATURE_GREP_EGREP_ALIAS
	756	&& (applet_name[0] == 'e' \|\| (option_mask32 & OPT_E))
	757	) {
[1765]	758	reflags \|= REG_EXTENDED;
[2725]	759	}
	760	#if ENABLE_EXTRA_COMPAT
	761	else {
	762	reflags = RE_SYNTAX_GREP;
	763	}
	764	#endif
[1765]	765
[2725]	766	if (option_mask32 & OPT_i) {
	767	#if !ENABLE_EXTRA_COMPAT
[821]	768	reflags \|= REG_ICASE;
[2725]	769	#else
	770	int i;
	771	case_fold = xmalloc(256);
	772	for (i = 0; i < 256; i++)
	773	case_fold[i] = (unsigned char)i;
	774	for (i = 'a'; i <= 'z'; i++)
	775	case_fold[i] = (unsigned char)(i - ('a' - 'A'));
	776	#endif
	777	}
[821]	778
	779	argv += optind;
	780
[2725]	781	/* if we didn't get a pattern from -e and no command file was specified,
	782	* first parameter should be the pattern. no pattern, no worky */
[821]	783	if (pattern_head == NULL) {
[1765]	784	char *pattern;
[821]	785	if (*argv == NULL)
	786	bb_show_usage();
[1765]	787	pattern = new_grep_list_data(*argv++, 0);
	788	llist_add_to(&pattern_head, pattern);
[821]	789	}
	790
[2725]	791	/* argv[0..(argc-1)] should be names of file to grep through. If
[1765]	792	* there is more than one file to grep, we will print the filenames. */
[2725]	793	if (argv[0] && argv[1])
[1765]	794	print_filename = 1;
	795	/* -H / -h of course override */
	796	if (option_mask32 & OPT_H)
	797	print_filename = 1;
	798	if (option_mask32 & OPT_h)
	799	print_filename = 0;
[821]	800
	801	/* If no files were specified, or '-' was specified, take input from
	802	* stdin. Otherwise, we grep through all the files specified. */
	803	matched = 0;
[2725]	804	do {
	805	cur_file = *argv;
[1765]	806	file = stdin;
[2725]	807	if (!cur_file \|\| LONE_DASH(cur_file)) {
[821]	808	cur_file = "(standard input)";
	809	} else {
[1765]	810	if (option_mask32 & OPT_r) {
	811	struct stat st;
	812	if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
	813	if (!(option_mask32 & OPT_h))
	814	print_filename = 1;
	815	matched += grep_dir(cur_file);
	816	goto grep_done;
	817	}
	818	}
	819	/* else: fopen(dir) will succeed, but reading won't */
[2725]	820	file = fopen_for_read(cur_file);
[1765]	821	if (file == NULL) {
	822	if (!SUPPRESS_ERR_MSGS)
[2725]	823	bb_simple_perror_msg(cur_file);
[1765]	824	open_errors = 1;
	825	continue;
[821]	826	}
	827	}
[1765]	828	matched += grep_file(file);
	829	fclose_if_not_stdin(file);
	830	grep_done: ;
[2725]	831	} while (argv && ++argv);
[821]	832
	833	/* destroy all the elments in the pattern list */
	834	if (ENABLE_FEATURE_CLEAN_UP) {
	835	while (pattern_head) {
	836	llist_t *pattern_head_ptr = pattern_head;
[2725]	837	grep_list_data_t gl = (grep_list_data_t )pattern_head_ptr->data;
[821]	838
	839	pattern_head = pattern_head->link;
[2725]	840	if (gl->flg_mem_alocated_compiled & ALLOCATED)
[821]	841	free(gl->pattern);
[2725]	842	if (gl->flg_mem_alocated_compiled & COMPILED)
	843	regfree(&gl->compiled_regex);
[1765]	844	free(gl);
[821]	845	free(pattern_head_ptr);
	846	}
	847	}
	848	/* 0 = success, 1 = failed, 2 = error */
[1765]	849	if (open_errors)
[821]	850	return 2;
[2725]	851	return !matched; /* invert return value: 0 = success, 1 = failed */
[821]	852	}

Note: See TracBrowser for help on using the repository browser.

Context Navigation

source: MondoRescue/branches/3.3/mindi-busybox/findutils/grep.c@ 3621

Download in other formats: