busybox/findutils/grep.c

509 lines
14 KiB
C
Raw Normal View History

/* vi: set sw=4 ts=4: */
1999-10-05 21:54:54 +05:30
/*
* Mini grep implementation for busybox using libc regex.
1999-10-21 03:38:37 +05:30
*
* Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
* Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
*
* Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
1999-10-05 21:54:54 +05:30
*/
/* BB_AUDIT SUSv3 defects - unsupported option -x. */
/* BB_AUDIT GNU defects - always acts as -a. */
/* http://www.opengroup.org/onlinepubs/007904975/utilities/grep.html */
/*
* 2004,2006 (C) Vladimir Oleynik <dzo@simtreas.ru> -
* correction "-e pattern1 -e pattern2" logic and more optimizations.
* precompiled regex
*/
/*
* (C) 2006 Jac Goudsmit added -o option
*/
1999-10-05 21:54:54 +05:30
#include "libbb.h"
#include "xregex.h"
/* options */
2007-02-25 08:08:20 +05:30
#define OPTSTR_GREP \
"lnqvscFiHhe:f:Lor" \
USE_FEATURE_GREP_CONTEXT("A:B:C:") \
USE_FEATURE_GREP_EGREP_ALIAS("E") \
USE_DESKTOP("w") \
"aI"
/* ignored: -a "assume all files to be text" */
/* ignored: -I "assume binary files have no matches" */
2007-02-25 08:08:20 +05:30
enum {
OPTBIT_l,
OPTBIT_n,
OPTBIT_q,
OPTBIT_v,
OPTBIT_s,
OPTBIT_c,
OPTBIT_F,
OPTBIT_i,
OPTBIT_H,
OPTBIT_h,
OPTBIT_e,
OPTBIT_f,
OPTBIT_L,
OPTBIT_o,
OPTBIT_r,
USE_FEATURE_GREP_CONTEXT( OPTBIT_A ,)
USE_FEATURE_GREP_CONTEXT( OPTBIT_B ,)
USE_FEATURE_GREP_CONTEXT( OPTBIT_C ,)
USE_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,)
USE_DESKTOP( OPTBIT_w ,)
2007-02-25 08:08:20 +05:30
OPT_l = 1 << OPTBIT_l,
OPT_n = 1 << OPTBIT_n,
OPT_q = 1 << OPTBIT_q,
OPT_v = 1 << OPTBIT_v,
OPT_s = 1 << OPTBIT_s,
OPT_c = 1 << OPTBIT_c,
OPT_F = 1 << OPTBIT_F,
OPT_i = 1 << OPTBIT_i,
OPT_H = 1 << OPTBIT_H,
OPT_h = 1 << OPTBIT_h,
OPT_e = 1 << OPTBIT_e,
OPT_f = 1 << OPTBIT_f,
OPT_L = 1 << OPTBIT_L,
OPT_o = 1 << OPTBIT_o,
OPT_r = 1 << OPTBIT_r,
OPT_A = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_A)) + 0,
OPT_B = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_B)) + 0,
OPT_C = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_C)) + 0,
OPT_E = USE_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
OPT_w = USE_DESKTOP( (1 << OPTBIT_w)) + 0,
2007-02-25 08:08:20 +05:30
};
#define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
#define PRINT_LINE_NUM (option_mask32 & OPT_n)
#define BE_QUIET (option_mask32 & OPT_q)
#define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
#define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
#define FGREP_FLAG (option_mask32 & OPT_F)
2007-02-25 08:08:20 +05:30
#define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
2006-10-14 19:54:30 +05:30
typedef unsigned char byte_t;
static int reflags;
2006-10-14 19:54:30 +05:30
static byte_t invert_search;
static byte_t print_filename;
static byte_t open_errors;
#if ENABLE_FEATURE_GREP_CONTEXT
static int lines_before;
static int lines_after;
static char **before_buf;
static int last_line_printed;
#endif /* ENABLE_FEATURE_GREP_CONTEXT */
/* globals used internally */
static llist_t *pattern_head; /* growable list of patterns to match */
2006-10-14 19:54:30 +05:30
static const char *cur_file; /* the current file we are reading */
typedef struct grep_list_data_t {
char *pattern;
regex_t preg;
#define PATTERN_MEM_A 1
#define COMPILED 2
int flg_mem_alocated_compiled;
} grep_list_data_t;
static void print_line(const char *line, int linenum, char decoration)
{
#if ENABLE_FEATURE_GREP_CONTEXT
/* possibly print the little '--' separator */
if ((lines_before || lines_after) && last_line_printed &&
last_line_printed < linenum - 1) {
puts("--");
}
last_line_printed = linenum;
#endif
2006-10-14 19:54:30 +05:30
if (print_filename)
printf("%s%c", cur_file, decoration);
if (PRINT_LINE_NUM)
printf("%i%c", linenum, decoration);
/* Emulate weird GNU grep behavior with -ov */
2007-02-25 08:08:20 +05:30
if ((option_mask32 & (OPT_v+OPT_o)) != (OPT_v+OPT_o))
puts(line);
}
1999-11-08 22:30:52 +05:30
static int grep_file(FILE *file)
{
char *line;
2006-10-14 19:54:30 +05:30
byte_t ret;
int linenum = 0;
int nmatches = 0;
regmatch_t regmatch;
#if ENABLE_FEATURE_GREP_CONTEXT
int print_n_lines_after = 0;
int curpos = 0; /* track where we are in the circular 'before' buffer */
int idx = 0; /* used for iteration through the circular buffer */
#endif /* ENABLE_FEATURE_GREP_CONTEXT */
while ((line = xmalloc_getline(file)) != NULL) {
llist_t *pattern_ptr = pattern_head;
grep_list_data_t * gl;
linenum++;
ret = 0;
while (pattern_ptr) {
gl = (grep_list_data_t *)pattern_ptr->data;
if (FGREP_FLAG) {
ret = strstr(line, gl->pattern) != NULL;
} else {
/*
* test for a postitive-assertion match (regexec returns success (0)
* and the user did not specify invert search), or a negative-assertion
* match (regexec returns failure (REG_NOMATCH) and the user specified
* invert search)
*/
2006-09-30 02:34:12 +05:30
if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
gl->flg_mem_alocated_compiled |= COMPILED;
xregcomp(&(gl->preg), gl->pattern, reflags);
}
regmatch.rm_so = 0;
regmatch.rm_eo = 0;
2007-02-25 08:08:20 +05:30
if (regexec(&(gl->preg), line, 1, &regmatch, 0) == 0) {
if (!(option_mask32 & OPT_w))
ret = 1;
else {
char c = ' ';
if (regmatch.rm_so)
c = line[regmatch.rm_so - 1];
if (!isalnum(c) && c != '_') {
c = line[regmatch.rm_eo];
if (!c || (!isalnum(c) && c != '_'))
ret = 1;
}
}
}
}
pattern_ptr = pattern_ptr->link;
} /* while (pattern_ptr) */
if (ret ^ invert_search) {
if (PRINT_FILES_WITH_MATCHES || BE_QUIET)
free(line);
/* if we found a match but were told to be quiet, stop here */
if (BE_QUIET || PRINT_FILES_WITHOUT_MATCHES)
return -1;
/* keep track of matches */
nmatches++;
/* if we're just printing filenames, we stop after the first match */
if (PRINT_FILES_WITH_MATCHES)
break;
/* print the matched line */
if (PRINT_MATCH_COUNTS == 0) {
#if ENABLE_FEATURE_GREP_CONTEXT
int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
/* if we were told to print 'before' lines and there is at least
* one line in the circular buffer, print them */
if (lines_before && before_buf[prevpos] != NULL) {
int first_buf_entry_line_num = linenum - lines_before;
/* advance to the first entry in the circular buffer, and
* figure out the line number is of the first line in the
* buffer */
idx = curpos;
while (before_buf[idx] == NULL) {
idx = (idx + 1) % lines_before;
first_buf_entry_line_num++;
}
/* now print each line in the buffer, clearing them as we go */
while (before_buf[idx] != NULL) {
print_line(before_buf[idx], first_buf_entry_line_num, '-');
free(before_buf[idx]);
before_buf[idx] = NULL;
idx = (idx + 1) % lines_before;
first_buf_entry_line_num++;
}
}
/* make a note that we need to print 'after' lines */
print_n_lines_after = lines_after;
#endif
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_o) {
line[regmatch.rm_eo] = '\0';
print_line(line + regmatch.rm_so, linenum, ':');
} else {
print_line(line, linenum, ':');
}
}
}
#if ENABLE_FEATURE_GREP_CONTEXT
else { /* no match */
/* Add the line to the circular 'before' buffer */
if (lines_before) {
free(before_buf[curpos]);
before_buf[curpos] = xstrdup(line);
curpos = (curpos + 1) % lines_before;
}
}
/* if we need to print some context lines after the last match, do so */
if (print_n_lines_after && (last_line_printed != linenum)) {
print_line(line, linenum, '-');
print_n_lines_after--;
}
#endif /* ENABLE_FEATURE_GREP_CONTEXT */
free(line);
1999-11-08 22:30:52 +05:30
}
/* special-case file post-processing for options where we don't print line
* matches, just filenames and possibly match counts */
/* grep -c: print [filename:]count, even if count is zero */
if (PRINT_MATCH_COUNTS) {
2006-10-14 19:54:30 +05:30
if (print_filename)
printf("%s:", cur_file);
2006-10-04 01:26:34 +05:30
printf("%d\n", nmatches);
}
/* grep -l: print just the filename, but only if we grepped the line in the file */
if (PRINT_FILES_WITH_MATCHES && nmatches > 0) {
puts(cur_file);
}
/* grep -L: print just the filename, but only if we didn't grep the line in the file */
if (PRINT_FILES_WITHOUT_MATCHES && nmatches == 0) {
puts(cur_file);
}
return nmatches;
}
#if ENABLE_FEATURE_CLEAN_UP
#define new_grep_list_data(p, m) add_grep_list_data(p, m)
static char * add_grep_list_data(char *pattern, int flg_used_mem)
#else
#define new_grep_list_data(p, m) add_grep_list_data(p)
static char * add_grep_list_data(char *pattern)
#endif
{
grep_list_data_t *gl = xmalloc(sizeof(grep_list_data_t));
gl->pattern = pattern;
#if ENABLE_FEATURE_CLEAN_UP
gl->flg_mem_alocated_compiled = flg_used_mem;
#else
gl->flg_mem_alocated_compiled = 0;
#endif
return (char *)gl;
}
static void load_regexes_from_file(llist_t *fopt)
{
char *line;
FILE *f;
2006-09-30 02:34:12 +05:30
while (fopt) {
llist_t *cur = fopt;
char *ffile = cur->data;
fopt = cur->link;
free(cur);
f = xfopen(ffile, "r");
while ((line = xmalloc_getline(f)) != NULL) {
llist_add_to(&pattern_head,
new_grep_list_data(line, PATTERN_MEM_A));
}
}
}
static int file_action_grep(const char *filename, struct stat *statbuf, void* matched, int depth)
2006-10-14 19:54:30 +05:30
{
FILE *file = fopen(filename, "r");
if (file == NULL) {
if (!SUPPRESS_ERR_MSGS)
bb_perror_msg("%s", cur_file);
open_errors = 1;
return 0;
}
cur_file = filename;
*(int*)matched += grep_file(file);
fclose(file);
2006-10-14 19:54:30 +05:30
return 1;
}
static int grep_dir(const char *dir)
{
int matched = 0;
recursive_action(dir,
/* recurse=yes */ ACTION_RECURSE |
/* followLinks=no */
/* depthFirst=yes */ ACTION_DEPTHFIRST,
2006-10-14 19:54:30 +05:30
/* fileAction= */ file_action_grep,
/* dirAction= */ NULL,
/* userData= */ &matched,
/* depth= */ 0);
2006-10-14 19:54:30 +05:30
return matched;
}
int grep_main(int argc, char **argv);
int grep_main(int argc, char **argv)
1999-10-05 21:54:54 +05:30
{
FILE *file;
int matched;
llist_t *fopt = NULL;
/* do normal option parsing */
#if ENABLE_FEATURE_GREP_CONTEXT
char *slines_after;
char *slines_before;
char *Copt;
opt_complementary = "H-h:e::f::C-AB";
2006-10-14 19:54:30 +05:30
getopt32(argc, argv,
2007-02-25 08:08:20 +05:30
OPTSTR_GREP,
&pattern_head, &fopt,
&slines_after, &slines_before, &Copt);
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_C) {
/* -C unsets prev -A and -B, but following -A or -B
may override it */
2007-02-25 08:08:20 +05:30
if (!(option_mask32 & OPT_A)) /* not overridden */
slines_after = Copt;
2007-02-25 08:08:20 +05:30
if (!(option_mask32 & OPT_B)) /* not overridden */
slines_before = Copt;
2007-02-25 08:08:20 +05:30
option_mask32 |= OPT_A|OPT_B; /* for parser */
}
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_A) {
2006-10-14 19:54:30 +05:30
lines_after = xatoi_u(slines_after);
}
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_B) {
2006-10-14 19:54:30 +05:30
lines_before = xatoi_u(slines_before);
}
/* sanity checks */
2007-02-25 08:08:20 +05:30
if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
option_mask32 &= ~OPT_n;
lines_before = 0;
lines_after = 0;
2006-09-30 02:34:12 +05:30
} else if (lines_before > 0)
before_buf = xzalloc(lines_before * sizeof(char *));
#else
/* with auto sanity checks */
opt_complementary = "H-h:e::f::c-n:q-n:l-n";
2007-02-25 08:08:20 +05:30
getopt32(argc, argv, OPTSTR_GREP,
&pattern_head, &fopt);
#endif
2007-02-25 08:08:20 +05:30
invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
if (pattern_head != NULL) {
/* convert char **argv to grep_list_data_t */
llist_t *cur;
2006-09-30 02:34:12 +05:30
for (cur = pattern_head; cur; cur = cur->link)
cur->data = new_grep_list_data(cur->data, 0);
}
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_f)
load_regexes_from_file(fopt);
2006-10-04 02:30:43 +05:30
if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
2007-02-25 08:08:20 +05:30
option_mask32 |= OPT_F;
2007-02-25 08:08:20 +05:30
if (!(option_mask32 & (OPT_o | OPT_w)))
reflags = REG_NOSUB;
2006-09-30 02:34:12 +05:30
if (ENABLE_FEATURE_GREP_EGREP_ALIAS &&
2007-02-25 08:08:20 +05:30
(applet_name[0] == 'e' || (option_mask32 & OPT_E)))
reflags |= REG_EXTENDED;
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_i)
reflags |= REG_ICASE;
argv += optind;
argc -= optind;
/* if we didn't get a pattern from a -e and no command file was specified,
* argv[optind] should be the pattern. no pattern, no worky */
if (pattern_head == NULL) {
2007-02-25 08:08:20 +05:30
char *pattern;
if (*argv == NULL)
2003-03-19 14:43:01 +05:30
bb_show_usage();
2007-02-25 08:08:20 +05:30
pattern = new_grep_list_data(*argv++, 0);
llist_add_to(&pattern_head, pattern);
argc--;
}
/* argv[(optind)..(argc-1)] should be names of file to grep through. If
2006-10-14 19:54:30 +05:30
* there is more than one file to grep, we will print the filenames. */
if (argc > 1)
2006-10-14 19:54:30 +05:30
print_filename = 1;
/* -H / -h of course override */
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_H)
print_filename = 1;
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_h)
print_filename = 0;
1999-10-05 21:54:54 +05:30
2000-06-29 04:29:30 +05:30
/* If no files were specified, or '-' was specified, take input from
* stdin. Otherwise, we grep through all the files specified. */
if (argc == 0)
argc++;
matched = 0;
while (argc--) {
cur_file = *argv++;
2006-10-14 19:54:30 +05:30
file = stdin;
2006-09-30 02:34:12 +05:30
if (!cur_file || (*cur_file == '-' && !cur_file[1])) {
cur_file = "(standard input)";
} else {
2007-02-25 08:08:20 +05:30
if (option_mask32 & OPT_r) {
2006-10-14 19:54:30 +05:30
struct stat st;
if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
2007-02-25 08:08:20 +05:30
if (!(option_mask32 & OPT_h))
2006-10-14 20:21:59 +05:30
print_filename = 1;
2006-10-14 19:54:30 +05:30
matched += grep_dir(cur_file);
goto grep_done;
}
}
/* else: fopen(dir) will succeed, but reading won't */
file = fopen(cur_file, "r");
2006-10-14 19:54:30 +05:30
if (file == NULL) {
if (!SUPPRESS_ERR_MSGS)
bb_perror_msg("%s", cur_file);
open_errors = 1;
continue;
}
}
2006-10-14 19:54:30 +05:30
matched += grep_file(file);
fclose_if_not_stdin(file);
2006-10-14 19:54:30 +05:30
grep_done:
if (matched < 0) {
/* we found a match but were told to be quiet, stop here and
* return success */
break;
}
}
/* destroy all the elments in the pattern list */
if (ENABLE_FEATURE_CLEAN_UP) {
while (pattern_head) {
llist_t *pattern_head_ptr = pattern_head;
grep_list_data_t *gl =
(grep_list_data_t *)pattern_head_ptr->data;
pattern_head = pattern_head->link;
2006-09-30 02:34:12 +05:30
if ((gl->flg_mem_alocated_compiled & PATTERN_MEM_A))
free(gl->pattern);
2006-09-30 02:34:12 +05:30
if ((gl->flg_mem_alocated_compiled & COMPILED))
regfree(&(gl->preg));
free(gl);
free(pattern_head_ptr);
}
}
/* 0 = success, 1 = failed, 2 = error */
/* If the -q option is specified, the exit status shall be zero
* if an input line is selected, even if an error was detected. */
2006-09-30 02:34:12 +05:30
if (BE_QUIET && matched)
return 0;
2006-10-14 19:54:30 +05:30
if (open_errors)
return 2;
return !matched; /* invert return value 0 = success, 1 = failed */
}