2003-10-01 08:36:16 +05:30
|
|
|
/* vi: set sw=4 ts=4: */
|
1999-10-28 21:36:25 +05:30
|
|
|
/*
|
2000-07-11 04:25:51 +05:30
|
|
|
* sed.c - very minimalist version of sed
|
1999-10-28 21:36:25 +05:30
|
|
|
*
|
2001-10-24 10:30:29 +05:30
|
|
|
* Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
|
|
|
|
* Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
|
2002-01-04 02:42:34 +05:30
|
|
|
* Copyright (C) 2002 Matt Kraai
|
2007-09-21 18:46:32 +05:30
|
|
|
* Copyright (C) 2003 by Glenn McGrath
|
2004-04-01 14:53:30 +05:30
|
|
|
* Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
|
1999-12-30 03:49:46 +05:30
|
|
|
*
|
2006-02-24 08:00:39 +05:30
|
|
|
* MAINTAINER: Rob Landley <rob@landley.net>
|
2006-03-02 01:44:16 +05:30
|
|
|
*
|
2010-08-16 23:44:46 +05:30
|
|
|
* Licensed under GPLv2, see file LICENSE in this source tree.
|
1999-10-28 21:36:25 +05:30
|
|
|
*/
|
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Code overview.
|
2011-05-03 04:21:43 +05:30
|
|
|
*
|
|
|
|
* Files are laid out to avoid unnecessary function declarations. So for
|
|
|
|
* example, every function add_cmd calls occurs before add_cmd in this file.
|
|
|
|
*
|
|
|
|
* add_cmd() is called on each line of sed command text (from a file or from
|
|
|
|
* the command line). It calls get_address() and parse_cmd_args(). The
|
|
|
|
* resulting sed_cmd_t structures are appended to a linked list
|
|
|
|
* (G.sed_cmd_head/G.sed_cmd_tail).
|
|
|
|
*
|
2011-09-13 22:10:22 +05:30
|
|
|
* process_files() does actual sedding, reading data lines from each input FILE*
|
2011-05-03 04:21:43 +05:30
|
|
|
* (which could be stdin) and applying the sed command list (sed_cmd_head) to
|
|
|
|
* each of the resulting lines.
|
|
|
|
*
|
|
|
|
* sed_main() is where external code calls into this, with a command line.
|
|
|
|
*/
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2011-05-03 04:21:43 +05:30
|
|
|
/* Supported features and commands in this version of sed:
|
|
|
|
*
|
|
|
|
* - comments ('#')
|
|
|
|
* - address matching: num|/matchstr/[,num|/matchstr/|$]command
|
|
|
|
* - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
|
|
|
|
* - edit commands: (a)ppend, (i)nsert, (c)hange
|
|
|
|
* - file commands: (r)ead
|
|
|
|
* - backreferences in substitution expressions (\0, \1, \2...\9)
|
|
|
|
* - grouped commands: {cmd1;cmd2}
|
|
|
|
* - transliteration (y/source-chars/dest-chars/)
|
|
|
|
* - pattern space hold space storing / swapping (g, h, x)
|
|
|
|
* - labels / branching (: label, b, t, T)
|
|
|
|
*
|
|
|
|
* (Note: Specifying an address (range) to match is *optional*; commands
|
|
|
|
* default to the whole pattern space if no specific address match was
|
|
|
|
* requested.)
|
|
|
|
*
|
|
|
|
* Todo:
|
|
|
|
* - Create a wrapper around regex to make libc's regex conform with sed
|
|
|
|
*
|
2012-07-24 16:50:12 +05:30
|
|
|
* Reference
|
|
|
|
* http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
|
|
|
|
* http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html
|
2011-05-03 04:21:43 +05:30
|
|
|
*/
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2013-11-13 17:15:33 +05:30
|
|
|
//config:config SED
|
|
|
|
//config: bool "sed"
|
|
|
|
//config: default y
|
|
|
|
//config: help
|
|
|
|
//config: sed is used to perform text transformations on a file
|
|
|
|
//config: or input from a pipeline.
|
|
|
|
|
|
|
|
//kbuild:lib-$(CONFIG_SED) += sed.o
|
|
|
|
|
|
|
|
//applet:IF_SED(APPLET(sed, BB_DIR_BIN, BB_SUID_DROP))
|
|
|
|
|
2011-04-02 02:26:30 +05:30
|
|
|
//usage:#define sed_trivial_usage
|
2013-10-30 17:30:00 +05:30
|
|
|
//usage: "[-inrE] [-f FILE]... [-e CMD]... [FILE]...\n"
|
|
|
|
//usage: "or: sed [-inrE] CMD [FILE]..."
|
2011-04-02 02:26:30 +05:30
|
|
|
//usage:#define sed_full_usage "\n\n"
|
2011-06-05 07:28:28 +05:30
|
|
|
//usage: " -e CMD Add CMD to sed commands to be executed"
|
2011-04-02 02:26:30 +05:30
|
|
|
//usage: "\n -f FILE Add FILE contents to sed commands to be executed"
|
2012-05-06 17:29:15 +05:30
|
|
|
//usage: "\n -i[SFX] Edit files in-place (otherwise sends to stdout)"
|
2012-06-21 13:15:11 +05:30
|
|
|
//usage: "\n Optionally back files up, appending SFX"
|
2011-04-02 02:26:30 +05:30
|
|
|
//usage: "\n -n Suppress automatic printing of pattern space"
|
2013-10-30 17:30:00 +05:30
|
|
|
//usage: "\n -r,-E Use extended regex syntax"
|
2011-04-02 02:26:30 +05:30
|
|
|
//usage: "\n"
|
|
|
|
//usage: "\nIf no -e or -f, the first non-option argument is the sed command string."
|
|
|
|
//usage: "\nRemaining arguments are input files (stdin if none)."
|
|
|
|
//usage:
|
|
|
|
//usage:#define sed_example_usage
|
|
|
|
//usage: "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
|
|
|
|
//usage: "bar\n"
|
|
|
|
|
2007-05-27 00:30:18 +05:30
|
|
|
#include "libbb.h"
|
2005-09-14 22:29:11 +05:30
|
|
|
#include "xregex.h"
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2011-08-16 16:59:34 +05:30
|
|
|
#if 0
|
|
|
|
# define dbg(...) bb_error_msg(__VA_ARGS__)
|
|
|
|
#else
|
|
|
|
# define dbg(...) ((void)0)
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
2010-08-01 07:44:46 +05:30
|
|
|
enum {
|
|
|
|
OPT_in_place = 1 << 0,
|
|
|
|
};
|
|
|
|
|
2006-05-09 09:23:55 +05:30
|
|
|
/* Each sed command turns into one of these structures. */
|
2003-03-09 15:53:57 +05:30
|
|
|
typedef struct sed_cmd_s {
|
2006-10-25 18:16:03 +05:30
|
|
|
/* Ordered by alignment requirements: currently 36 bytes on x86 */
|
2007-01-31 03:55:16 +05:30
|
|
|
struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
|
2003-04-12 21:40:42 +05:30
|
|
|
|
2006-10-25 18:16:03 +05:30
|
|
|
/* address storage */
|
|
|
|
regex_t *beg_match; /* sed -e '/match/cmd' */
|
|
|
|
regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
|
|
|
|
regex_t *sub_match; /* For 's/sub_match/string/' */
|
|
|
|
int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
|
2011-08-16 16:59:34 +05:30
|
|
|
int beg_line_orig; /* copy of the above, needed for -i */
|
2006-10-25 18:16:03 +05:30
|
|
|
int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
|
2006-03-02 01:44:16 +05:30
|
|
|
|
2007-01-29 20:01:47 +05:30
|
|
|
FILE *sw_file; /* File (sw) command writes to, -1 for none. */
|
2006-10-25 18:16:03 +05:30
|
|
|
char *string; /* Data string for (saicytb) commands. */
|
2003-09-13 12:27:39 +05:30
|
|
|
|
2008-02-16 18:48:17 +05:30
|
|
|
unsigned which_match; /* (s) Which match to replace (0 for all) */
|
2000-11-04 01:17:00 +05:30
|
|
|
|
2006-10-25 18:16:03 +05:30
|
|
|
/* Bitfields (gcc won't group them if we don't) */
|
2007-01-31 03:55:16 +05:30
|
|
|
unsigned invert:1; /* the '!' after the address */
|
|
|
|
unsigned in_match:1; /* Next line also included in match? */
|
|
|
|
unsigned sub_p:1; /* (s) print option */
|
2006-12-03 01:42:12 +05:30
|
|
|
|
2007-01-31 03:55:16 +05:30
|
|
|
char sw_last_char; /* Last line written by (sw) had no '\n' */
|
2001-11-10 16:19:42 +05:30
|
|
|
|
2006-10-25 18:16:03 +05:30
|
|
|
/* GENERAL FIELDS */
|
|
|
|
char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
|
2003-03-09 15:53:57 +05:30
|
|
|
} sed_cmd_t;
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2007-08-13 02:28:27 +05:30
|
|
|
static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2007-03-27 02:18:46 +05:30
|
|
|
struct globals {
|
2006-05-09 09:23:55 +05:30
|
|
|
/* options */
|
2006-12-10 07:39:12 +05:30
|
|
|
int be_quiet, regex_type;
|
2013-11-28 07:44:16 +05:30
|
|
|
|
2006-05-09 09:23:55 +05:30
|
|
|
FILE *nonstdout;
|
|
|
|
char *outname, *hold_space;
|
2013-11-28 07:44:16 +05:30
|
|
|
smallint exitcode;
|
2006-05-09 09:23:55 +05:30
|
|
|
|
2013-11-28 07:44:16 +05:30
|
|
|
/* list of input files */
|
2013-11-28 16:38:51 +05:30
|
|
|
int current_input_file, last_input_file;
|
|
|
|
char **input_file_list;
|
2013-11-28 07:44:16 +05:30
|
|
|
FILE *current_fp;
|
2006-05-09 09:23:55 +05:30
|
|
|
|
|
|
|
regmatch_t regmatch[10];
|
|
|
|
regex_t *previous_regex_ptr;
|
2006-09-17 21:58:10 +05:30
|
|
|
|
2006-05-09 09:23:55 +05:30
|
|
|
/* linked list of sed commands */
|
2011-08-16 16:59:34 +05:30
|
|
|
sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
|
2003-04-09 07:13:54 +05:30
|
|
|
|
2013-11-28 07:44:16 +05:30
|
|
|
/* linked list of append lines */
|
2006-05-09 09:23:55 +05:30
|
|
|
llist_t *append_head;
|
|
|
|
|
|
|
|
char *add_cmd_line;
|
|
|
|
|
|
|
|
struct pipeline {
|
2010-10-29 15:16:52 +05:30
|
|
|
char *buf; /* Space to hold string */
|
|
|
|
int idx; /* Space used */
|
|
|
|
int len; /* Space allocated */
|
2006-05-09 09:23:55 +05:30
|
|
|
} pipeline;
|
2010-02-04 19:30:15 +05:30
|
|
|
} FIX_ALIASING;
|
2007-03-27 02:18:46 +05:30
|
|
|
#define G (*(struct globals*)&bb_common_bufsiz1)
|
2010-06-04 04:59:52 +05:30
|
|
|
struct BUG_G_too_big {
|
2011-02-03 04:35:48 +05:30
|
|
|
char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
|
2010-06-04 04:59:52 +05:30
|
|
|
};
|
2007-06-04 15:46:52 +05:30
|
|
|
#define INIT_G() do { \
|
|
|
|
G.sed_cmd_tail = &G.sed_cmd_head; \
|
|
|
|
} while (0)
|
|
|
|
|
2003-04-09 07:13:54 +05:30
|
|
|
|
2006-03-02 01:44:16 +05:30
|
|
|
#if ENABLE_FEATURE_CLEAN_UP
|
2006-12-10 07:39:12 +05:30
|
|
|
static void sed_free_and_close_stuff(void)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2011-08-16 16:59:34 +05:30
|
|
|
sed_cmd_t *sed_cmd = G.sed_cmd_head;
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2007-03-27 02:18:46 +05:30
|
|
|
llist_free(G.append_head, free);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2003-03-28 09:53:23 +05:30
|
|
|
while (sed_cmd) {
|
2003-04-09 07:13:54 +05:30
|
|
|
sed_cmd_t *sed_cmd_next = sed_cmd->next;
|
2000-02-09 01:28:47 +05:30
|
|
|
|
2007-01-29 20:01:47 +05:30
|
|
|
if (sed_cmd->sw_file)
|
|
|
|
xprint_and_close_file(sed_cmd->sw_file);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2003-03-28 09:53:23 +05:30
|
|
|
if (sed_cmd->beg_match) {
|
|
|
|
regfree(sed_cmd->beg_match);
|
|
|
|
free(sed_cmd->beg_match);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2003-03-28 09:53:23 +05:30
|
|
|
if (sed_cmd->end_match) {
|
|
|
|
regfree(sed_cmd->end_match);
|
|
|
|
free(sed_cmd->end_match);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2003-03-28 09:53:23 +05:30
|
|
|
if (sed_cmd->sub_match) {
|
|
|
|
regfree(sed_cmd->sub_match);
|
|
|
|
free(sed_cmd->sub_match);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
free(sed_cmd->string);
|
2003-03-28 09:53:23 +05:30
|
|
|
free(sed_cmd);
|
|
|
|
sed_cmd = sed_cmd_next;
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
2004-10-30 12:24:19 +05:30
|
|
|
|
2007-09-29 03:37:23 +05:30
|
|
|
free(G.hold_space);
|
2004-11-25 12:51:47 +05:30
|
|
|
|
2013-11-28 07:44:16 +05:30
|
|
|
if (G.current_fp)
|
|
|
|
fclose(G.current_fp);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2006-12-26 23:47:42 +05:30
|
|
|
#else
|
|
|
|
void sed_free_and_close_stuff(void);
|
2000-07-13 05:05:21 +05:30
|
|
|
#endif
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2004-02-18 15:24:15 +05:30
|
|
|
/* If something bad happens during -i operation, delete temp file */
|
|
|
|
|
|
|
|
static void cleanup_outname(void)
|
|
|
|
{
|
2007-03-27 02:18:46 +05:30
|
|
|
if (G.outname) unlink(G.outname);
|
2004-02-18 15:24:15 +05:30
|
|
|
}
|
|
|
|
|
2007-11-13 22:18:10 +05:30
|
|
|
/* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2007-01-30 04:21:25 +05:30
|
|
|
static void parse_escapes(char *dest, const char *string, int len, char from, char to)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2006-10-25 18:16:46 +05:30
|
|
|
int i = 0;
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-10-25 18:16:46 +05:30
|
|
|
while (i < len) {
|
|
|
|
if (string[i] == '\\') {
|
|
|
|
if (!to || string[i+1] == from) {
|
2007-01-20 03:03:19 +05:30
|
|
|
*dest++ = to ? to : string[i+1];
|
2006-10-25 18:16:46 +05:30
|
|
|
i += 2;
|
2003-10-01 08:36:16 +05:30
|
|
|
continue;
|
2007-01-20 03:03:19 +05:30
|
|
|
}
|
|
|
|
*dest++ = string[i++];
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
2007-11-13 22:18:10 +05:30
|
|
|
/* TODO: is it safe wrt a string with trailing '\\' ? */
|
2007-01-20 03:03:19 +05:30
|
|
|
*dest++ = string[i++];
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
2007-11-13 22:18:10 +05:30
|
|
|
*dest = '\0';
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
|
2007-01-30 04:21:25 +05:30
|
|
|
static char *copy_parsing_escapes(const char *string, int len)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2011-05-05 03:56:37 +05:30
|
|
|
const char *s;
|
2006-12-10 07:39:12 +05:30
|
|
|
char *dest = xmalloc(len + 1);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2011-05-05 03:56:37 +05:30
|
|
|
/* sed recognizes \n */
|
2011-05-03 04:22:22 +05:30
|
|
|
/* GNU sed also recognizes \t and \r */
|
2011-05-05 03:56:37 +05:30
|
|
|
for (s = "\nn\tt\rr"; *s; s += 2) {
|
|
|
|
parse_escapes(dest, string, len, s[1], s[0]);
|
|
|
|
string = dest;
|
|
|
|
len = strlen(dest);
|
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
return dest;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/*
|
2001-01-02 16:31:31 +05:30
|
|
|
* index_of_next_unescaped_regexp_delim - walks left to right through a string
|
|
|
|
* beginning at a specified index and returns the index of the next regular
|
2007-11-13 22:18:10 +05:30
|
|
|
* expression delimiter (typically a forward slash ('/')) not preceded by
|
2006-07-26 22:55:08 +05:30
|
|
|
* a backslash ('\'). A negative delimiter disables square bracket checking.
|
2000-07-11 04:25:51 +05:30
|
|
|
*/
|
2007-01-30 04:21:25 +05:30
|
|
|
static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2001-08-24 20:15:50 +05:30
|
|
|
int bracket = -1;
|
|
|
|
int escaped = 0;
|
2003-03-30 13:32:18 +05:30
|
|
|
int idx = 0;
|
2001-11-10 16:19:42 +05:30
|
|
|
char ch;
|
2001-08-24 20:15:50 +05:30
|
|
|
|
2006-07-26 22:55:08 +05:30
|
|
|
if (delimiter < 0) {
|
|
|
|
bracket--;
|
2006-12-10 07:39:12 +05:30
|
|
|
delimiter = -delimiter;
|
2006-07-26 22:55:08 +05:30
|
|
|
}
|
|
|
|
|
2011-05-03 04:21:43 +05:30
|
|
|
for (; (ch = str[idx]) != '\0'; idx++) {
|
2006-07-26 22:55:08 +05:30
|
|
|
if (bracket >= 0) {
|
2011-05-03 04:21:43 +05:30
|
|
|
if (ch == ']'
|
|
|
|
&& !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
|
|
|
|
) {
|
2001-08-24 20:15:50 +05:30
|
|
|
bracket = -1;
|
2011-05-03 04:21:43 +05:30
|
|
|
}
|
2001-08-24 20:15:50 +05:30
|
|
|
} else if (escaped)
|
|
|
|
escaped = 0;
|
2001-11-10 16:19:42 +05:30
|
|
|
else if (ch == '\\')
|
2001-08-24 20:15:50 +05:30
|
|
|
escaped = 1;
|
2006-07-26 22:55:08 +05:30
|
|
|
else if (bracket == -1 && ch == '[')
|
2001-08-24 20:15:50 +05:30
|
|
|
bracket = idx;
|
2003-03-09 14:14:49 +05:30
|
|
|
else if (ch == delimiter)
|
2000-07-18 01:36:42 +05:30
|
|
|
return idx;
|
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
/* if we make it to here, we've hit the end of the string */
|
2006-12-10 07:39:12 +05:30
|
|
|
bb_error_msg_and_die("unmatched '%c'", delimiter);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
|
|
|
|
2003-09-14 09:36:12 +05:30
|
|
|
/*
|
|
|
|
* Returns the index of the third delimiter
|
|
|
|
*/
|
2007-01-30 05:14:38 +05:30
|
|
|
static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
|
2003-03-30 13:32:18 +05:30
|
|
|
{
|
2007-01-30 05:14:38 +05:30
|
|
|
const char *cmdstr_ptr = cmdstr;
|
2012-01-16 00:36:03 +05:30
|
|
|
unsigned char delimiter;
|
2003-03-30 13:32:18 +05:30
|
|
|
int idx = 0;
|
|
|
|
|
2003-09-14 09:36:12 +05:30
|
|
|
/* verify that the 's' or 'y' is followed by something. That something
|
2003-03-30 13:32:18 +05:30
|
|
|
* (typically a 'slash') is now our regexp delimiter... */
|
2006-07-26 22:55:08 +05:30
|
|
|
if (*cmdstr == '\0')
|
|
|
|
bb_error_msg_and_die("bad format in substitution expression");
|
2006-10-25 18:16:46 +05:30
|
|
|
delimiter = *cmdstr_ptr++;
|
2003-03-30 13:32:18 +05:30
|
|
|
|
|
|
|
/* save the match string */
|
|
|
|
idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
|
2005-11-20 13:14:35 +05:30
|
|
|
*match = copy_parsing_escapes(cmdstr_ptr, idx);
|
2003-03-30 13:32:18 +05:30
|
|
|
|
|
|
|
/* save the replacement string */
|
|
|
|
cmdstr_ptr += idx + 1;
|
2012-01-16 00:36:03 +05:30
|
|
|
idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
|
2005-11-20 13:14:35 +05:30
|
|
|
*replace = copy_parsing_escapes(cmdstr_ptr, idx);
|
2003-03-30 13:32:18 +05:30
|
|
|
|
2003-04-08 17:26:11 +05:30
|
|
|
return ((cmdstr_ptr - cmdstr) + idx);
|
2003-03-30 13:32:18 +05:30
|
|
|
}
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/*
|
|
|
|
* returns the index in the string just past where the address ends.
|
|
|
|
*/
|
2007-01-30 04:21:25 +05:30
|
|
|
static int get_address(const char *my_str, int *linenum, regex_t ** regex)
|
1999-12-30 03:49:46 +05:30
|
|
|
{
|
2007-01-30 04:21:25 +05:30
|
|
|
const char *pos = my_str;
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2003-09-14 09:36:12 +05:30
|
|
|
if (isdigit(*my_str)) {
|
2007-01-30 04:21:25 +05:30
|
|
|
*linenum = strtol(my_str, (char**)&pos, 10);
|
2003-03-09 14:14:49 +05:30
|
|
|
/* endstr shouldnt ever equal NULL */
|
2003-09-14 09:36:12 +05:30
|
|
|
} else if (*my_str == '$') {
|
2001-06-12 05:20:06 +05:30
|
|
|
*linenum = -1;
|
2003-09-14 09:36:12 +05:30
|
|
|
pos++;
|
|
|
|
} else if (*my_str == '/' || *my_str == '\\') {
|
2003-10-01 08:36:16 +05:30
|
|
|
int next;
|
2003-03-30 13:32:18 +05:30
|
|
|
char delimiter;
|
2003-10-01 08:36:16 +05:30
|
|
|
char *temp;
|
2003-03-09 14:14:49 +05:30
|
|
|
|
2006-10-25 18:16:46 +05:30
|
|
|
delimiter = '/';
|
2012-01-16 00:36:03 +05:30
|
|
|
if (*my_str == '\\')
|
|
|
|
delimiter = *++pos;
|
2003-09-14 09:36:12 +05:30
|
|
|
next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
|
2006-12-10 07:39:12 +05:30
|
|
|
temp = copy_parsing_escapes(pos, next);
|
2012-01-16 00:36:03 +05:30
|
|
|
*regex = xzalloc(sizeof(regex_t));
|
2013-07-22 01:39:44 +05:30
|
|
|
xregcomp(*regex, temp, G.regex_type);
|
2003-10-01 08:36:16 +05:30
|
|
|
free(temp);
|
|
|
|
/* Move position to next character after last delimiter */
|
2006-07-26 22:55:08 +05:30
|
|
|
pos += (next+1);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2003-09-14 09:36:12 +05:30
|
|
|
return pos - my_str;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
|
2008-03-17 14:30:54 +05:30
|
|
|
static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2006-10-25 18:16:46 +05:30
|
|
|
int start = 0, idx, hack = 0;
|
2003-10-01 08:36:16 +05:30
|
|
|
|
|
|
|
/* Skip whitespace, then grab filename to end of line */
|
2007-01-30 05:14:38 +05:30
|
|
|
while (isspace(filecmdstr[start]))
|
|
|
|
start++;
|
2006-10-25 18:16:46 +05:30
|
|
|
idx = start;
|
2007-01-30 05:14:38 +05:30
|
|
|
while (filecmdstr[idx] && filecmdstr[idx] != '\n')
|
|
|
|
idx++;
|
2006-10-25 18:16:46 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* If lines glued together, put backslash back. */
|
2007-01-30 05:14:38 +05:30
|
|
|
if (filecmdstr[idx] == '\n')
|
|
|
|
hack = 1;
|
2006-10-25 18:16:46 +05:30
|
|
|
if (idx == start)
|
|
|
|
bb_error_msg_and_die("empty filename");
|
2006-08-03 21:11:12 +05:30
|
|
|
*retval = xstrndup(filecmdstr+start, idx-start+hack+1);
|
2007-01-30 05:14:38 +05:30
|
|
|
if (hack)
|
|
|
|
(*retval)[idx] = '\\';
|
2003-10-01 08:36:16 +05:30
|
|
|
|
|
|
|
return idx;
|
|
|
|
}
|
|
|
|
|
2007-01-30 05:14:38 +05:30
|
|
|
static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
|
2000-07-14 01:28:04 +05:30
|
|
|
{
|
2007-03-27 02:18:46 +05:30
|
|
|
int cflags = G.regex_type;
|
2000-07-14 01:28:04 +05:30
|
|
|
char *match;
|
2007-01-30 04:21:25 +05:30
|
|
|
int idx;
|
2000-07-14 01:28:04 +05:30
|
|
|
|
|
|
|
/*
|
2003-10-01 08:36:16 +05:30
|
|
|
* A substitution command should look something like this:
|
2013-12-06 07:12:17 +05:30
|
|
|
* s/match/replace/ #giIpw
|
2000-08-22 02:59:20 +05:30
|
|
|
* || | |||
|
2000-07-14 01:28:04 +05:30
|
|
|
* mandatory optional
|
|
|
|
*/
|
2003-10-01 08:36:16 +05:30
|
|
|
idx = parse_regex_delim(substr, &match, &sed_cmd->string);
|
2000-07-14 01:28:04 +05:30
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
/* determine the number of back references in the match string */
|
|
|
|
/* Note: we compute this here rather than in the do_subst_command()
|
|
|
|
* function to save processor time, at the expense of a little more memory
|
|
|
|
* (4 bits) per sed_cmd */
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2000-07-14 01:28:04 +05:30
|
|
|
/* process the flags */
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-10-25 18:16:46 +05:30
|
|
|
sed_cmd->which_match = 1;
|
2014-09-16 04:41:13 +05:30
|
|
|
dbg("s flags:'%s'", substr + idx + 1);
|
2003-10-01 08:36:16 +05:30
|
|
|
while (substr[++idx]) {
|
2014-09-16 04:41:13 +05:30
|
|
|
dbg("s flag:'%c'", substr[idx]);
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Parse match number */
|
2006-10-25 18:16:46 +05:30
|
|
|
if (isdigit(substr[idx])) {
|
|
|
|
if (match[0] != '^') {
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Match 0 treated as all, multiple matches we take the last one. */
|
2007-01-30 05:14:38 +05:30
|
|
|
const char *pos = substr + idx;
|
|
|
|
/* FIXME: error check? */
|
2008-02-16 18:48:17 +05:30
|
|
|
sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
|
2014-09-16 04:41:13 +05:30
|
|
|
idx = pos - substr - 1;
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
continue;
|
|
|
|
}
|
2004-01-04 12:12:14 +05:30
|
|
|
/* Skip spaces */
|
2009-10-22 23:12:26 +05:30
|
|
|
if (isspace(substr[idx]))
|
|
|
|
continue;
|
2004-01-04 12:12:14 +05:30
|
|
|
|
2000-07-14 01:28:04 +05:30
|
|
|
switch (substr[idx]) {
|
2006-10-25 18:16:03 +05:30
|
|
|
/* Replace all occurrences */
|
|
|
|
case 'g':
|
2008-02-16 18:48:17 +05:30
|
|
|
if (match[0] != '^')
|
|
|
|
sed_cmd->which_match = 0;
|
2006-10-25 18:16:03 +05:30
|
|
|
break;
|
|
|
|
/* Print pattern space */
|
|
|
|
case 'p':
|
|
|
|
sed_cmd->sub_p = 1;
|
|
|
|
break;
|
|
|
|
/* Write to file */
|
|
|
|
case 'w':
|
|
|
|
{
|
|
|
|
char *temp;
|
2008-03-17 14:30:54 +05:30
|
|
|
idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
|
2006-10-25 18:16:03 +05:30
|
|
|
break;
|
|
|
|
}
|
|
|
|
/* Ignore case (gnu exension) */
|
2013-12-06 07:12:17 +05:30
|
|
|
case 'i':
|
2006-10-25 18:16:03 +05:30
|
|
|
case 'I':
|
|
|
|
cflags |= REG_ICASE;
|
|
|
|
break;
|
|
|
|
/* Comment */
|
|
|
|
case '#':
|
2010-06-04 04:59:52 +05:30
|
|
|
// while (substr[++idx]) continue;
|
|
|
|
idx += strlen(substr + idx); // same
|
2006-10-25 18:16:03 +05:30
|
|
|
/* Fall through */
|
|
|
|
/* End of command */
|
|
|
|
case ';':
|
|
|
|
case '}':
|
|
|
|
goto out;
|
|
|
|
default:
|
2014-09-16 04:41:13 +05:30
|
|
|
dbg("s bad flags:'%s'", substr + idx);
|
2006-10-25 18:16:03 +05:30
|
|
|
bb_error_msg_and_die("bad option in substitution expression");
|
2000-07-14 01:28:04 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
2010-06-04 04:59:52 +05:30
|
|
|
out:
|
2000-07-18 01:36:42 +05:30
|
|
|
/* compile the match string into a regex */
|
2003-04-08 12:12:45 +05:30
|
|
|
if (*match != '\0') {
|
|
|
|
/* If match is empty, we use last regex used at runtime */
|
2012-01-16 00:36:03 +05:30
|
|
|
sed_cmd->sub_match = xzalloc(sizeof(regex_t));
|
|
|
|
dbg("xregcomp('%s',%x)", match, cflags);
|
2003-04-08 12:12:45 +05:30
|
|
|
xregcomp(sed_cmd->sub_match, match, cflags);
|
2012-01-16 00:36:03 +05:30
|
|
|
dbg("regcomp ok");
|
2003-04-08 12:12:45 +05:30
|
|
|
}
|
2000-07-14 01:28:04 +05:30
|
|
|
free(match);
|
2000-07-15 00:36:30 +05:30
|
|
|
|
|
|
|
return idx;
|
2000-07-14 01:28:04 +05:30
|
|
|
}
|
|
|
|
|
2003-03-18 06:49:23 +05:30
|
|
|
/*
|
|
|
|
* Process the commands arguments
|
|
|
|
*/
|
2007-01-30 05:14:38 +05:30
|
|
|
static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2011-05-03 04:21:43 +05:30
|
|
|
static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
|
|
|
|
enum {
|
|
|
|
IDX_s = 0,
|
|
|
|
IDX_a,
|
|
|
|
IDX_i,
|
|
|
|
IDX_c,
|
|
|
|
IDX_r,
|
|
|
|
IDX_w,
|
|
|
|
IDX_colon,
|
|
|
|
IDX_b,
|
|
|
|
IDX_t,
|
|
|
|
IDX_T,
|
|
|
|
IDX_y,
|
|
|
|
IDX_d,
|
|
|
|
IDX_D,
|
|
|
|
IDX_g,
|
|
|
|
IDX_G,
|
|
|
|
IDX_h,
|
|
|
|
IDX_H,
|
|
|
|
IDX_l,
|
|
|
|
IDX_n,
|
|
|
|
IDX_N,
|
|
|
|
IDX_p,
|
|
|
|
IDX_P,
|
|
|
|
IDX_q,
|
|
|
|
IDX_x,
|
|
|
|
IDX_equal,
|
|
|
|
IDX_lbrace,
|
|
|
|
IDX_rbrace,
|
|
|
|
IDX_nul
|
|
|
|
};
|
|
|
|
struct chk { char chk[sizeof(cmd_letters)-1 == IDX_nul ? 1 : -1]; };
|
|
|
|
|
|
|
|
unsigned idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
|
|
|
|
|
2003-03-09 20:42:24 +05:30
|
|
|
/* handle (s)ubstitution command */
|
2011-05-03 04:21:43 +05:30
|
|
|
if (idx == IDX_s) {
|
2006-10-25 18:16:46 +05:30
|
|
|
cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
|
2011-05-03 04:21:43 +05:30
|
|
|
}
|
2003-03-09 20:42:24 +05:30
|
|
|
/* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
|
2011-05-03 04:21:43 +05:30
|
|
|
else if (idx <= IDX_c) { /* a,i,c */
|
2012-07-24 16:50:12 +05:30
|
|
|
if (idx < IDX_c) { /* a,i */
|
|
|
|
if (sed_cmd->end_line || sed_cmd->end_match)
|
|
|
|
bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
|
|
|
|
}
|
2006-10-25 18:16:46 +05:30
|
|
|
for (;;) {
|
|
|
|
if (*cmdstr == '\n' || *cmdstr == '\\') {
|
|
|
|
cmdstr++;
|
|
|
|
break;
|
2009-10-23 01:58:08 +05:30
|
|
|
}
|
|
|
|
if (!isspace(*cmdstr))
|
2004-04-01 14:53:30 +05:30
|
|
|
break;
|
2009-10-23 01:58:08 +05:30
|
|
|
cmdstr++;
|
2004-04-01 14:53:30 +05:30
|
|
|
}
|
2006-08-03 21:11:12 +05:30
|
|
|
sed_cmd->string = xstrdup(cmdstr);
|
2007-11-13 22:18:10 +05:30
|
|
|
/* "\anychar" -> "anychar" */
|
|
|
|
parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
|
2003-10-01 08:36:16 +05:30
|
|
|
cmdstr += strlen(cmdstr);
|
2011-05-03 04:21:43 +05:30
|
|
|
}
|
2003-03-09 20:42:24 +05:30
|
|
|
/* handle file cmds: (r)ead */
|
2011-05-03 04:21:43 +05:30
|
|
|
else if (idx <= IDX_w) { /* r,w */
|
2012-07-24 16:50:12 +05:30
|
|
|
if (idx < IDX_w) { /* r */
|
|
|
|
if (sed_cmd->end_line || sed_cmd->end_match)
|
|
|
|
bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
|
|
|
|
}
|
2008-03-17 14:30:54 +05:30
|
|
|
cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
|
2007-01-29 22:38:51 +05:30
|
|
|
if (sed_cmd->cmd == 'w') {
|
2008-07-22 04:35:26 +05:30
|
|
|
sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
|
2007-01-29 22:38:51 +05:30
|
|
|
sed_cmd->sw_last_char = '\n';
|
|
|
|
}
|
2011-05-03 04:21:43 +05:30
|
|
|
}
|
2003-03-28 10:13:39 +05:30
|
|
|
/* handle branch commands */
|
2011-05-03 04:21:43 +05:30
|
|
|
else if (idx <= IDX_T) { /* :,b,t,T */
|
2003-03-28 10:13:39 +05:30
|
|
|
int length;
|
|
|
|
|
2006-10-25 18:16:03 +05:30
|
|
|
cmdstr = skip_whitespace(cmdstr);
|
2003-09-14 11:31:14 +05:30
|
|
|
length = strcspn(cmdstr, semicolon_whitespace);
|
|
|
|
if (length) {
|
2006-08-03 21:11:12 +05:30
|
|
|
sed_cmd->string = xstrndup(cmdstr, length);
|
2003-09-14 11:31:14 +05:30
|
|
|
cmdstr += length;
|
|
|
|
}
|
2003-03-28 10:13:39 +05:30
|
|
|
}
|
2003-03-30 13:32:18 +05:30
|
|
|
/* translation command */
|
2011-05-03 04:21:43 +05:30
|
|
|
else if (idx == IDX_y) {
|
2003-10-01 08:36:16 +05:30
|
|
|
char *match, *replace;
|
2006-10-25 18:16:46 +05:30
|
|
|
int i = cmdstr[0];
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-10-25 18:16:46 +05:30
|
|
|
cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
|
2003-10-01 08:36:16 +05:30
|
|
|
/* \n already parsed, but \delimiter needs unescaping. */
|
2006-12-10 07:39:12 +05:30
|
|
|
parse_escapes(match, match, strlen(match), i, i);
|
|
|
|
parse_escapes(replace, replace, strlen(replace), i, i);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-05-22 00:00:35 +05:30
|
|
|
sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
|
2003-10-01 08:36:16 +05:30
|
|
|
for (i = 0; match[i] && replace[i]; i++) {
|
2006-10-25 18:16:46 +05:30
|
|
|
sed_cmd->string[i*2] = match[i];
|
|
|
|
sed_cmd->string[i*2+1] = replace[i];
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
free(match);
|
|
|
|
free(replace);
|
2003-03-30 13:32:18 +05:30
|
|
|
}
|
2003-03-18 06:49:23 +05:30
|
|
|
/* if it wasnt a single-letter command that takes no arguments
|
|
|
|
* then it must be an invalid command.
|
|
|
|
*/
|
2011-05-03 04:21:43 +05:30
|
|
|
else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
|
2006-10-25 18:16:46 +05:30
|
|
|
bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
|
2003-03-09 20:42:24 +05:30
|
|
|
}
|
|
|
|
|
|
|
|
/* give back whatever's left over */
|
2006-10-25 18:16:46 +05:30
|
|
|
return cmdstr;
|
2003-03-09 20:42:24 +05:30
|
|
|
}
|
|
|
|
|
2003-09-14 09:36:12 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Parse address+command sets, skipping comment lines. */
|
2003-03-09 20:42:24 +05:30
|
|
|
|
2007-01-30 04:21:25 +05:30
|
|
|
static void add_cmd(const char *cmdstr)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
|
|
|
sed_cmd_t *sed_cmd;
|
2010-05-12 05:19:04 +05:30
|
|
|
unsigned len, n;
|
2003-03-09 20:42:24 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Append this line to any unfinished line from last time. */
|
2007-03-27 02:18:46 +05:30
|
|
|
if (G.add_cmd_line) {
|
|
|
|
char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
|
|
|
|
free(G.add_cmd_line);
|
2007-04-13 02:50:25 +05:30
|
|
|
cmdstr = G.add_cmd_line = tp;
|
2006-06-05 23:02:44 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2010-05-12 05:19:04 +05:30
|
|
|
/* If this line ends with unescaped backslash, request next line. */
|
|
|
|
n = len = strlen(cmdstr);
|
|
|
|
while (n && cmdstr[n-1] == '\\')
|
|
|
|
n--;
|
|
|
|
if ((len - n) & 1) { /* if odd number of trailing backslashes */
|
2007-03-27 02:18:46 +05:30
|
|
|
if (!G.add_cmd_line)
|
|
|
|
G.add_cmd_line = xstrdup(cmdstr);
|
2010-05-12 05:19:04 +05:30
|
|
|
G.add_cmd_line[len-1] = '\0';
|
2003-10-01 08:36:16 +05:30
|
|
|
return;
|
2003-03-09 20:42:24 +05:30
|
|
|
}
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Loop parsing all commands in this line. */
|
2006-10-25 18:16:46 +05:30
|
|
|
while (*cmdstr) {
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Skip leading whitespace and semicolons */
|
|
|
|
cmdstr += strspn(cmdstr, semicolon_whitespace);
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* If no more commands, exit. */
|
2006-10-25 18:16:46 +05:30
|
|
|
if (!*cmdstr) break;
|
2003-09-14 09:36:12 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* if this is a comment, jump past it and keep going */
|
|
|
|
if (*cmdstr == '#') {
|
|
|
|
/* "#n" is the same as using -n on the command line */
|
2006-10-25 18:16:46 +05:30
|
|
|
if (cmdstr[1] == 'n')
|
2007-03-27 02:18:46 +05:30
|
|
|
G.be_quiet++;
|
2006-10-25 18:16:46 +05:30
|
|
|
cmdstr = strpbrk(cmdstr, "\n\r");
|
|
|
|
if (!cmdstr) break;
|
2003-10-01 08:36:16 +05:30
|
|
|
continue;
|
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* parse the command
|
|
|
|
* format is: [addr][,addr][!]cmd
|
|
|
|
* |----||-----||-|
|
|
|
|
* part1 part2 part3
|
|
|
|
*/
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2006-05-22 00:00:35 +05:30
|
|
|
sed_cmd = xzalloc(sizeof(sed_cmd_t));
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* first part (if present) is an address: either a '$', a number or a /regex/ */
|
|
|
|
cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
|
2011-08-16 16:59:34 +05:30
|
|
|
sed_cmd->beg_line_orig = sed_cmd->beg_line;
|
2001-12-21 21:34:12 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* second part (if present) will begin with a comma */
|
|
|
|
if (*cmdstr == ',') {
|
|
|
|
int idx;
|
2002-06-12 05:13:27 +05:30
|
|
|
|
2003-03-09 16:36:38 +05:30
|
|
|
cmdstr++;
|
2003-10-01 08:36:16 +05:30
|
|
|
idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
|
2006-10-25 18:16:46 +05:30
|
|
|
if (!idx)
|
|
|
|
bb_error_msg_and_die("no address after comma");
|
2003-10-01 08:36:16 +05:30
|
|
|
cmdstr += idx;
|
2003-03-09 20:42:24 +05:30
|
|
|
}
|
2003-03-09 16:36:38 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* skip whitespace before the command */
|
2006-10-25 18:16:03 +05:30
|
|
|
cmdstr = skip_whitespace(cmdstr);
|
2003-03-09 20:42:24 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Check for inversion flag */
|
|
|
|
if (*cmdstr == '!') {
|
|
|
|
sed_cmd->invert = 1;
|
|
|
|
cmdstr++;
|
2003-04-08 12:12:45 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* skip whitespace before the command */
|
2006-10-25 18:16:03 +05:30
|
|
|
cmdstr = skip_whitespace(cmdstr);
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
2003-04-08 12:12:45 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* last part (mandatory) will be a command */
|
2006-10-25 18:16:46 +05:30
|
|
|
if (!*cmdstr)
|
|
|
|
bb_error_msg_and_die("missing command");
|
2010-05-12 05:19:04 +05:30
|
|
|
sed_cmd->cmd = *cmdstr++;
|
2003-10-01 08:36:16 +05:30
|
|
|
cmdstr = parse_cmd_args(sed_cmd, cmdstr);
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2013-07-08 05:13:40 +05:30
|
|
|
/* cmdstr now points past args.
|
|
|
|
* GNU sed requires a separator, if there are more commands,
|
|
|
|
* else it complains "char N: extra characters after command".
|
|
|
|
* Example: "sed 'p;d'". We also allow "sed 'pd'".
|
|
|
|
*/
|
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Add the command to the command array */
|
2011-08-16 16:59:34 +05:30
|
|
|
*G.sed_cmd_tail = sed_cmd;
|
|
|
|
G.sed_cmd_tail = &sed_cmd->next;
|
2003-03-30 09:11:53 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* If we glued multiple lines together, free the memory. */
|
2007-03-27 02:18:46 +05:30
|
|
|
free(G.add_cmd_line);
|
|
|
|
G.add_cmd_line = NULL;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2004-11-25 12:51:47 +05:30
|
|
|
/* Append to a string, reallocating memory as necessary. */
|
|
|
|
|
2003-04-08 17:26:11 +05:30
|
|
|
#define PIPE_GROW 64
|
2001-11-10 16:19:42 +05:30
|
|
|
|
2005-04-17 01:09:00 +05:30
|
|
|
static void pipe_putc(char c)
|
2001-11-10 16:19:42 +05:30
|
|
|
{
|
2007-03-27 02:18:46 +05:30
|
|
|
if (G.pipeline.idx == G.pipeline.len) {
|
|
|
|
G.pipeline.buf = xrealloc(G.pipeline.buf,
|
|
|
|
G.pipeline.len + PIPE_GROW);
|
|
|
|
G.pipeline.len += PIPE_GROW;
|
2001-11-10 16:19:42 +05:30
|
|
|
}
|
2007-03-27 02:18:46 +05:30
|
|
|
G.pipeline.buf[G.pipeline.idx++] = c;
|
2001-11-10 16:19:42 +05:30
|
|
|
}
|
|
|
|
|
2006-07-26 22:55:08 +05:30
|
|
|
static void do_subst_w_backrefs(char *line, char *replace)
|
2000-07-18 01:36:42 +05:30
|
|
|
{
|
2009-08-17 05:05:04 +05:30
|
|
|
int i, j;
|
2000-07-18 01:36:42 +05:30
|
|
|
|
|
|
|
/* go through the replacement string */
|
|
|
|
for (i = 0; replace[i]; i++) {
|
2012-06-04 18:14:47 +05:30
|
|
|
/* if we find a backreference (\1, \2, etc.) print the backref'ed text */
|
2007-01-18 04:46:16 +05:30
|
|
|
if (replace[i] == '\\') {
|
|
|
|
unsigned backref = replace[++i] - '0';
|
|
|
|
if (backref <= 9) {
|
2007-03-27 02:18:46 +05:30
|
|
|
/* print out the text held in G.regmatch[backref] */
|
|
|
|
if (G.regmatch[backref].rm_so != -1) {
|
|
|
|
j = G.regmatch[backref].rm_so;
|
|
|
|
while (j < G.regmatch[backref].rm_eo)
|
2007-01-18 04:46:16 +05:30
|
|
|
pipe_putc(line[j++]);
|
|
|
|
}
|
|
|
|
continue;
|
2006-10-25 18:16:46 +05:30
|
|
|
}
|
2007-01-18 04:46:16 +05:30
|
|
|
/* I _think_ it is impossible to get '\' to be
|
|
|
|
* the last char in replace string. Thus we dont check
|
|
|
|
* for replace[i] == NUL. (counterexample anyone?) */
|
|
|
|
/* if we find a backslash escaped character, print the character */
|
|
|
|
pipe_putc(replace[i]);
|
|
|
|
continue;
|
2000-07-18 01:36:42 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
/* if we find an unescaped '&' print out the whole matched text. */
|
2007-01-18 04:46:16 +05:30
|
|
|
if (replace[i] == '&') {
|
2007-03-27 02:18:46 +05:30
|
|
|
j = G.regmatch[0].rm_so;
|
|
|
|
while (j < G.regmatch[0].rm_eo)
|
2006-10-25 18:16:46 +05:30
|
|
|
pipe_putc(line[j++]);
|
2007-01-18 04:46:16 +05:30
|
|
|
continue;
|
2006-10-25 18:16:46 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Otherwise just output the character. */
|
2007-01-18 04:46:16 +05:30
|
|
|
pipe_putc(replace[i]);
|
2000-07-18 01:36:42 +05:30
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-08-17 05:05:04 +05:30
|
|
|
static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
|
2000-07-14 01:31:58 +05:30
|
|
|
{
|
2009-08-17 05:05:04 +05:30
|
|
|
char *line = *line_p;
|
2008-05-13 07:57:31 +05:30
|
|
|
unsigned match_count = 0;
|
2012-06-04 18:14:47 +05:30
|
|
|
bool altered = 0;
|
|
|
|
bool prev_match_empty = 1;
|
|
|
|
bool tried_at_eol = 0;
|
2003-04-08 12:12:45 +05:30
|
|
|
regex_t *current_regex;
|
|
|
|
|
2009-08-17 05:05:04 +05:30
|
|
|
current_regex = sed_cmd->sub_match;
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Handle empty regex. */
|
2009-08-17 05:05:04 +05:30
|
|
|
if (!current_regex) {
|
2007-03-27 02:18:46 +05:30
|
|
|
current_regex = G.previous_regex_ptr;
|
2006-10-25 18:16:46 +05:30
|
|
|
if (!current_regex)
|
2006-10-27 14:32:31 +05:30
|
|
|
bb_error_msg_and_die("no previous regexp");
|
2009-08-17 05:05:04 +05:30
|
|
|
}
|
|
|
|
G.previous_regex_ptr = current_regex;
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Find the first match */
|
2012-01-16 00:36:03 +05:30
|
|
|
dbg("matching '%s'", line);
|
|
|
|
if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
|
|
|
|
dbg("no match");
|
2000-11-04 01:17:00 +05:30
|
|
|
return 0;
|
2012-01-16 00:36:03 +05:30
|
|
|
}
|
|
|
|
dbg("match");
|
2000-07-18 01:36:42 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Initialize temporary output buffer. */
|
2007-03-27 02:18:46 +05:30
|
|
|
G.pipeline.buf = xmalloc(PIPE_GROW);
|
|
|
|
G.pipeline.len = PIPE_GROW;
|
|
|
|
G.pipeline.idx = 0;
|
2003-10-01 08:36:16 +05:30
|
|
|
|
|
|
|
/* Now loop through, substituting for matches */
|
|
|
|
do {
|
2012-06-04 18:15:09 +05:30
|
|
|
int start = G.regmatch[0].rm_so;
|
|
|
|
int end = G.regmatch[0].rm_eo;
|
2000-11-04 01:17:00 +05:30
|
|
|
int i;
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
match_count++;
|
|
|
|
|
|
|
|
/* If we aren't interested in this match, output old line to
|
2012-06-04 18:14:47 +05:30
|
|
|
* end of match and continue */
|
2008-05-13 07:57:31 +05:30
|
|
|
if (sed_cmd->which_match
|
|
|
|
&& (sed_cmd->which_match != match_count)
|
|
|
|
) {
|
2012-06-04 18:15:09 +05:30
|
|
|
for (i = 0; i < end; i++)
|
2009-08-17 05:05:04 +05:30
|
|
|
pipe_putc(*line++);
|
2012-06-04 18:14:47 +05:30
|
|
|
/* Null match? Print one more char */
|
2012-06-04 18:15:09 +05:30
|
|
|
if (start == end && *line)
|
2012-06-04 18:14:47 +05:30
|
|
|
pipe_putc(*line++);
|
2012-04-24 22:57:34 +05:30
|
|
|
goto next;
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
|
2012-06-04 18:14:47 +05:30
|
|
|
/* Print everything before the match */
|
2012-06-04 18:15:09 +05:30
|
|
|
for (i = 0; i < start; i++)
|
2009-08-17 05:05:04 +05:30
|
|
|
pipe_putc(line[i]);
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2012-06-04 18:14:47 +05:30
|
|
|
/* Then print the substitution string,
|
|
|
|
* unless we just matched empty string after non-empty one.
|
|
|
|
* Example: string "cccd", pattern "c*", repl "R":
|
|
|
|
* result is "RdR", not "RRdR": first match "ccc",
|
|
|
|
* second is "" before "d", third is "" after "d".
|
|
|
|
* Second match is NOT replaced!
|
|
|
|
*/
|
2012-06-07 20:04:57 +05:30
|
|
|
if (prev_match_empty || start != 0 || start != end) {
|
2012-06-08 13:55:31 +05:30
|
|
|
//dbg("%d %d %d", prev_match_empty, start, end);
|
2012-06-04 18:15:09 +05:30
|
|
|
dbg("inserting replacement at %d in '%s'", start, line);
|
2012-06-04 18:14:47 +05:30
|
|
|
do_subst_w_backrefs(line, sed_cmd->string);
|
2012-06-08 13:55:31 +05:30
|
|
|
/* Flag that something has changed */
|
|
|
|
altered = 1;
|
2012-06-04 18:14:47 +05:30
|
|
|
} else {
|
2012-06-04 18:15:09 +05:30
|
|
|
dbg("NOT inserting replacement at %d in '%s'", start, line);
|
2012-06-04 18:14:47 +05:30
|
|
|
}
|
|
|
|
|
|
|
|
/* If matched string is empty (f.e. "c*" pattern),
|
|
|
|
* copy verbatim one char after it before attempting more matches
|
|
|
|
*/
|
2012-06-04 18:15:09 +05:30
|
|
|
prev_match_empty = (start == end);
|
2012-06-08 13:55:31 +05:30
|
|
|
if (prev_match_empty) {
|
|
|
|
if (!line[end]) {
|
|
|
|
tried_at_eol = 1;
|
|
|
|
} else {
|
|
|
|
pipe_putc(line[end]);
|
|
|
|
end++;
|
|
|
|
}
|
2012-06-04 18:14:47 +05:30
|
|
|
}
|
2000-11-04 01:17:00 +05:30
|
|
|
|
2012-06-04 18:14:47 +05:30
|
|
|
/* Advance past the match */
|
2012-06-04 18:15:09 +05:30
|
|
|
dbg("line += %d", end);
|
|
|
|
line += end;
|
2000-07-18 01:36:42 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* if we're not doing this globally, get out now */
|
2012-01-16 00:36:03 +05:30
|
|
|
if (sed_cmd->which_match != 0)
|
|
|
|
break;
|
2012-04-24 22:57:34 +05:30
|
|
|
next:
|
2012-06-04 18:14:47 +05:30
|
|
|
/* Exit if we are at EOL and already tried matching at it */
|
|
|
|
if (*line == '\0') {
|
|
|
|
if (tried_at_eol)
|
|
|
|
break;
|
|
|
|
tried_at_eol = 1;
|
|
|
|
}
|
2000-11-04 01:17:00 +05:30
|
|
|
|
2012-06-04 18:15:09 +05:30
|
|
|
//maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
|
2012-01-16 00:36:03 +05:30
|
|
|
} while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-08-17 05:05:04 +05:30
|
|
|
/* Copy rest of string into output pipeline */
|
|
|
|
while (1) {
|
|
|
|
char c = *line++;
|
|
|
|
pipe_putc(c);
|
|
|
|
if (c == '\0')
|
|
|
|
break;
|
|
|
|
}
|
2000-11-04 01:17:00 +05:30
|
|
|
|
2009-08-17 05:05:04 +05:30
|
|
|
free(*line_p);
|
|
|
|
*line_p = G.pipeline.buf;
|
2000-07-14 01:31:58 +05:30
|
|
|
return altered;
|
|
|
|
}
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Set command pointer to point to this label. (Does not handle null label.) */
|
2006-07-26 22:55:08 +05:30
|
|
|
static sed_cmd_t *branch_to(char *label)
|
2003-03-28 10:13:39 +05:30
|
|
|
{
|
|
|
|
sed_cmd_t *sed_cmd;
|
2003-04-11 22:40:23 +05:30
|
|
|
|
2011-08-16 16:59:34 +05:30
|
|
|
for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
|
2006-10-25 18:16:46 +05:30
|
|
|
if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
|
|
|
|
return sed_cmd;
|
2003-03-28 10:13:39 +05:30
|
|
|
}
|
|
|
|
}
|
2006-10-25 18:16:46 +05:30
|
|
|
bb_error_msg_and_die("can't find label for jump to '%s'", label);
|
2003-03-28 10:13:39 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
static void append(char *s)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2013-10-30 14:52:47 +05:30
|
|
|
llist_add_to_end(&G.append_head, s);
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
|
2013-10-30 18:55:22 +05:30
|
|
|
/* Output line of text. */
|
|
|
|
/* Note:
|
|
|
|
* The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
|
|
|
|
* Without them, we had this:
|
|
|
|
* echo -n thingy >z1
|
|
|
|
* echo -n again >z2
|
|
|
|
* >znull
|
|
|
|
* sed "s/i/z/" z1 z2 znull | hexdump -vC
|
|
|
|
* output:
|
|
|
|
* gnu sed 4.1.5:
|
|
|
|
* 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
|
|
|
|
* bbox:
|
|
|
|
* 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
|
|
|
|
*/
|
|
|
|
enum {
|
|
|
|
NO_EOL_CHAR = 1,
|
|
|
|
LAST_IS_NUL = 2,
|
|
|
|
};
|
|
|
|
static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
|
|
|
|
{
|
|
|
|
char lpc = *last_puts_char;
|
|
|
|
|
|
|
|
/* Need to insert a '\n' between two files because first file's
|
|
|
|
* last line wasn't terminated? */
|
|
|
|
if (lpc != '\n' && lpc != '\0') {
|
|
|
|
fputc('\n', file);
|
|
|
|
lpc = '\n';
|
|
|
|
}
|
|
|
|
fputs(s, file);
|
|
|
|
|
|
|
|
/* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
|
|
|
|
if (s[0])
|
|
|
|
lpc = 'x';
|
|
|
|
|
|
|
|
/* had trailing '\0' and it was last char of file? */
|
|
|
|
if (last_gets_char == LAST_IS_NUL) {
|
|
|
|
fputc('\0', file);
|
|
|
|
lpc = 'x'; /* */
|
|
|
|
} else
|
|
|
|
/* had trailing '\n' or '\0'? */
|
|
|
|
if (last_gets_char != NO_EOL_CHAR) {
|
|
|
|
fputc(last_gets_char, file);
|
|
|
|
lpc = last_gets_char;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ferror(file)) {
|
|
|
|
xfunc_error_retval = 4; /* It's what gnu sed exits with... */
|
|
|
|
bb_error_msg_and_die(bb_msg_write_error);
|
|
|
|
}
|
|
|
|
*last_puts_char = lpc;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void flush_append(char *last_puts_char, char last_gets_char)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2006-05-09 03:47:23 +05:30
|
|
|
char *data;
|
|
|
|
|
2003-10-01 08:36:16 +05:30
|
|
|
/* Output appended lines. */
|
2007-03-27 02:18:46 +05:30
|
|
|
while ((data = (char *)llist_pop(&G.append_head))) {
|
2013-10-30 18:55:22 +05:30
|
|
|
puts_maybe_newline(data, G.nonstdout, last_puts_char, last_gets_char);
|
2006-05-09 03:47:23 +05:30
|
|
|
free(data);
|
2003-10-01 08:36:16 +05:30
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-03-27 02:18:46 +05:30
|
|
|
/* Get next line of input from G.input_file_list, flushing append buffer and
|
2004-11-25 12:51:47 +05:30
|
|
|
* noting if we ran out of files without a newline on the last line we read.
|
2003-10-01 08:36:16 +05:30
|
|
|
*/
|
2013-10-30 18:55:22 +05:30
|
|
|
static char *get_next_line(char *gets_char, char *last_puts_char, char last_gets_char)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2006-10-25 18:16:46 +05:30
|
|
|
char *temp = NULL;
|
2007-01-29 22:38:51 +05:30
|
|
|
int len;
|
|
|
|
char gc;
|
2006-12-02 23:28:10 +05:30
|
|
|
|
2013-10-30 18:55:22 +05:30
|
|
|
flush_append(last_puts_char, last_gets_char);
|
2007-01-29 20:01:47 +05:30
|
|
|
|
|
|
|
/* will be returned if last line in the file
|
|
|
|
* doesn't end with either '\n' or '\0' */
|
|
|
|
gc = NO_EOL_CHAR;
|
2013-11-28 16:38:51 +05:30
|
|
|
for (; G.current_input_file <= G.last_input_file; G.current_input_file++) {
|
2013-11-28 07:44:16 +05:30
|
|
|
FILE *fp = G.current_fp;
|
|
|
|
if (!fp) {
|
|
|
|
const char *path = G.input_file_list[G.current_input_file];
|
|
|
|
fp = stdin;
|
|
|
|
if (path != bb_msg_standard_input) {
|
|
|
|
fp = fopen_or_warn(path, "r");
|
|
|
|
if (!fp) {
|
|
|
|
G.exitcode = EXIT_FAILURE;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
G.current_fp = fp;
|
|
|
|
}
|
2007-01-18 04:46:16 +05:30
|
|
|
/* Read line up to a newline or NUL byte, inclusive,
|
|
|
|
* return malloc'ed char[]. length of the chunk read
|
|
|
|
* is stored in len. NULL if EOF/error */
|
2007-01-29 22:40:19 +05:30
|
|
|
temp = bb_get_chunk_from_file(fp, &len);
|
2006-02-24 08:00:39 +05:30
|
|
|
if (temp) {
|
2006-12-02 23:28:10 +05:30
|
|
|
/* len > 0 here, it's ok to do temp[len-1] */
|
|
|
|
char c = temp[len-1];
|
|
|
|
if (c == '\n' || c == '\0') {
|
|
|
|
temp[len-1] = '\0';
|
2007-01-29 22:38:51 +05:30
|
|
|
gc = c;
|
2007-01-29 22:40:19 +05:30
|
|
|
if (c == '\0') {
|
|
|
|
int ch = fgetc(fp);
|
|
|
|
if (ch != EOF)
|
|
|
|
ungetc(ch, fp);
|
|
|
|
else
|
|
|
|
gc = LAST_IS_NUL;
|
|
|
|
}
|
2006-12-02 23:28:10 +05:30
|
|
|
}
|
2007-01-29 22:38:51 +05:30
|
|
|
/* else we put NO_EOL_CHAR into *gets_char */
|
|
|
|
break;
|
|
|
|
|
2007-01-29 20:01:47 +05:30
|
|
|
/* NB: I had the idea of peeking next file(s) and returning
|
|
|
|
* NO_EOL_CHAR only if it is the *last* non-empty
|
|
|
|
* input file. But there is a case where this won't work:
|
|
|
|
* file1: "a woo\nb woo"
|
|
|
|
* file2: "c no\nd no"
|
|
|
|
* sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
|
|
|
|
* (note: *no* newline after "b bang"!) */
|
2006-12-02 23:28:10 +05:30
|
|
|
}
|
|
|
|
/* Close this file and advance to next one */
|
2013-11-28 07:44:16 +05:30
|
|
|
fclose_if_not_stdin(fp);
|
|
|
|
G.current_fp = NULL;
|
2003-03-08 10:51:02 +05:30
|
|
|
}
|
2007-01-29 20:01:47 +05:30
|
|
|
*gets_char = gc;
|
2003-10-01 08:36:16 +05:30
|
|
|
return temp;
|
|
|
|
}
|
|
|
|
|
2007-03-27 02:18:46 +05:30
|
|
|
#define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2007-08-06 09:11:08 +05:30
|
|
|
static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
|
|
|
|
{
|
|
|
|
int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
|
|
|
|
if (retval)
|
|
|
|
G.previous_regex_ptr = sed_cmd->beg_match;
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2006-02-24 08:00:39 +05:30
|
|
|
/* Process all the lines in all the files */
|
|
|
|
|
2004-11-25 12:51:47 +05:30
|
|
|
static void process_files(void)
|
2003-10-01 08:36:16 +05:30
|
|
|
{
|
2004-10-30 12:24:19 +05:30
|
|
|
char *pattern_space, *next_line;
|
2007-01-29 04:56:15 +05:30
|
|
|
int linenum = 0;
|
2007-01-29 20:01:47 +05:30
|
|
|
char last_puts_char = '\n';
|
|
|
|
char last_gets_char, next_gets_char;
|
2006-12-10 08:22:19 +05:30
|
|
|
sed_cmd_t *sed_cmd;
|
|
|
|
int substituted;
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-02-24 08:00:39 +05:30
|
|
|
/* Prime the pump */
|
2013-10-30 18:55:22 +05:30
|
|
|
next_line = get_next_line(&next_gets_char, &last_puts_char, '\n' /*last_gets_char*/);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Go through every line in each file */
|
2008-02-28 23:29:01 +05:30
|
|
|
again:
|
2006-12-10 08:22:19 +05:30
|
|
|
substituted = 0;
|
2003-04-12 21:40:42 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Advance to next line. Stop if out of lines. */
|
|
|
|
pattern_space = next_line;
|
2009-06-30 22:49:37 +05:30
|
|
|
if (!pattern_space)
|
|
|
|
return;
|
2007-01-29 20:01:47 +05:30
|
|
|
last_gets_char = next_gets_char;
|
2001-06-12 05:20:06 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Read one line in advance so we can act on the last line,
|
|
|
|
* the '$' address */
|
2013-10-30 18:55:22 +05:30
|
|
|
next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
|
2006-12-10 07:39:12 +05:30
|
|
|
linenum++;
|
2009-06-30 22:49:37 +05:30
|
|
|
|
|
|
|
/* For every line, go through all the commands */
|
2008-02-28 23:29:01 +05:30
|
|
|
restart:
|
2011-08-16 16:59:34 +05:30
|
|
|
for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
|
2006-12-10 07:39:12 +05:30
|
|
|
int old_matched, matched;
|
|
|
|
|
|
|
|
old_matched = sed_cmd->in_match;
|
|
|
|
|
|
|
|
/* Determine if this command matches this line: */
|
2009-07-14 04:32:57 +05:30
|
|
|
|
2011-08-16 16:59:34 +05:30
|
|
|
dbg("match1:%d", sed_cmd->in_match);
|
|
|
|
dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
|
|
|
|
&& !sed_cmd->beg_match && !sed_cmd->end_match));
|
|
|
|
dbg("match3:%d", (sed_cmd->beg_line > 0
|
|
|
|
&& (sed_cmd->end_line || sed_cmd->end_match
|
|
|
|
? (sed_cmd->beg_line <= linenum)
|
|
|
|
: (sed_cmd->beg_line == linenum)
|
|
|
|
)
|
|
|
|
));
|
|
|
|
dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
|
|
|
|
dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
|
2010-06-04 04:59:52 +05:30
|
|
|
|
2009-07-14 04:32:57 +05:30
|
|
|
/* Are we continuing a previous multi-line match? */
|
2006-12-10 07:39:12 +05:30
|
|
|
sed_cmd->in_match = sed_cmd->in_match
|
|
|
|
/* Or is no range necessary? */
|
|
|
|
|| (!sed_cmd->beg_line && !sed_cmd->end_line
|
|
|
|
&& !sed_cmd->beg_match && !sed_cmd->end_match)
|
|
|
|
/* Or did we match the start of a numerical range? */
|
2010-06-04 04:59:52 +05:30
|
|
|
|| (sed_cmd->beg_line > 0
|
|
|
|
&& (sed_cmd->end_line || sed_cmd->end_match
|
2010-06-04 05:01:48 +05:30
|
|
|
/* note: even if end is numeric and is < linenum too,
|
2011-08-16 16:59:34 +05:30
|
|
|
* GNU sed matches! We match too, therefore we don't
|
|
|
|
* check here that linenum <= end.
|
|
|
|
* Example:
|
|
|
|
* printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
|
|
|
|
* first three input lines are deleted;
|
|
|
|
* 4th line is matched and printed
|
|
|
|
* by "2,3" (!) and by "4" ranges
|
|
|
|
*/
|
2010-06-04 04:59:52 +05:30
|
|
|
? (sed_cmd->beg_line <= linenum) /* N,end */
|
|
|
|
: (sed_cmd->beg_line == linenum) /* N */
|
|
|
|
)
|
|
|
|
)
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Or does this line match our begin address regex? */
|
2007-08-06 09:11:08 +05:30
|
|
|
|| (beg_match(sed_cmd, pattern_space))
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Or did we match last line of input? */
|
|
|
|
|| (sed_cmd->beg_line == -1 && next_line == NULL);
|
|
|
|
|
2009-07-14 04:32:57 +05:30
|
|
|
/* Snapshot the value */
|
2006-12-10 07:39:12 +05:30
|
|
|
matched = sed_cmd->in_match;
|
|
|
|
|
2011-08-16 16:59:34 +05:30
|
|
|
dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
|
|
|
|
sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Is this line the end of the current match? */
|
2009-07-14 04:32:57 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
if (matched) {
|
2009-07-14 04:32:57 +05:30
|
|
|
/* once matched, "n,xxx" range is dead, disabling it */
|
2011-08-16 16:59:34 +05:30
|
|
|
if (sed_cmd->beg_line > 0) {
|
2009-07-14 04:32:57 +05:30
|
|
|
sed_cmd->beg_line = -2;
|
2010-08-01 07:44:46 +05:30
|
|
|
}
|
2009-07-14 04:32:57 +05:30
|
|
|
sed_cmd->in_match = !(
|
2006-12-10 07:39:12 +05:30
|
|
|
/* has the ending line come, or is this a single address command? */
|
2011-05-03 04:21:43 +05:30
|
|
|
(sed_cmd->end_line
|
|
|
|
? sed_cmd->end_line == -1
|
|
|
|
? !next_line
|
2006-12-10 07:39:12 +05:30
|
|
|
: (sed_cmd->end_line <= linenum)
|
|
|
|
: !sed_cmd->end_match
|
2009-07-14 04:32:57 +05:30
|
|
|
)
|
2006-12-10 07:39:12 +05:30
|
|
|
/* or does this line matches our last address regex */
|
2009-07-14 04:32:57 +05:30
|
|
|
|| (sed_cmd->end_match && old_matched
|
2006-12-10 07:39:12 +05:30
|
|
|
&& (regexec(sed_cmd->end_match,
|
2013-01-15 18:28:01 +05:30
|
|
|
pattern_space, 0, NULL, 0) == 0)
|
2011-08-16 16:59:34 +05:30
|
|
|
)
|
2009-07-14 04:32:57 +05:30
|
|
|
);
|
2006-12-10 07:39:12 +05:30
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Skip blocks of commands we didn't match */
|
2006-12-10 07:39:12 +05:30
|
|
|
if (sed_cmd->cmd == '{') {
|
2007-01-29 04:56:15 +05:30
|
|
|
if (sed_cmd->invert ? matched : !matched) {
|
2010-04-20 13:30:03 +05:30
|
|
|
unsigned nest_cnt = 0;
|
|
|
|
while (1) {
|
|
|
|
if (sed_cmd->cmd == '{')
|
|
|
|
nest_cnt++;
|
|
|
|
if (sed_cmd->cmd == '}') {
|
|
|
|
nest_cnt--;
|
|
|
|
if (nest_cnt == 0)
|
|
|
|
break;
|
|
|
|
}
|
2006-12-10 07:39:12 +05:30
|
|
|
sed_cmd = sed_cmd->next;
|
2007-01-29 04:56:15 +05:30
|
|
|
if (!sed_cmd)
|
|
|
|
bb_error_msg_and_die("unterminated {");
|
|
|
|
}
|
|
|
|
}
|
2006-12-10 07:39:12 +05:30
|
|
|
continue;
|
|
|
|
}
|
2003-03-30 09:11:53 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Okay, so did this line match? */
|
2009-06-30 22:49:37 +05:30
|
|
|
if (sed_cmd->invert ? matched : !matched)
|
|
|
|
continue; /* no */
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Update last used regex in case a blank substitute BRE is found */
|
|
|
|
if (sed_cmd->beg_match) {
|
|
|
|
G.previous_regex_ptr = sed_cmd->beg_match;
|
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* actual sedding */
|
2012-01-16 00:36:03 +05:30
|
|
|
dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
|
|
|
|
pattern_space, next_line, sed_cmd->cmd);
|
2009-06-30 22:49:37 +05:30
|
|
|
switch (sed_cmd->cmd) {
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Print line number */
|
|
|
|
case '=':
|
|
|
|
fprintf(G.nonstdout, "%d\n", linenum);
|
|
|
|
break;
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Write the current pattern space up to the first newline */
|
|
|
|
case 'P':
|
|
|
|
{
|
|
|
|
char *tmp = strchr(pattern_space, '\n');
|
|
|
|
if (tmp) {
|
|
|
|
*tmp = '\0';
|
|
|
|
/* TODO: explain why '\n' below */
|
2007-01-29 20:01:47 +05:30
|
|
|
sed_puts(pattern_space, '\n');
|
2009-06-30 22:49:37 +05:30
|
|
|
*tmp = '\n';
|
2006-12-10 07:39:12 +05:30
|
|
|
break;
|
|
|
|
}
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Fall Through */
|
|
|
|
}
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Write the current pattern space to output */
|
|
|
|
case 'p':
|
|
|
|
/* NB: we print this _before_ the last line
|
|
|
|
* (of current file) is printed. Even if
|
|
|
|
* that line is nonterminated, we print
|
|
|
|
* '\n' here (gnu sed does the same) */
|
|
|
|
sed_puts(pattern_space, '\n');
|
|
|
|
break;
|
|
|
|
/* Delete up through first newline */
|
|
|
|
case 'D':
|
|
|
|
{
|
|
|
|
char *tmp = strchr(pattern_space, '\n');
|
|
|
|
if (tmp) {
|
2010-06-04 04:59:52 +05:30
|
|
|
overlapping_strcpy(pattern_space, tmp + 1);
|
2009-06-30 22:49:37 +05:30
|
|
|
goto restart;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/* discard this line. */
|
|
|
|
case 'd':
|
|
|
|
goto discard_line;
|
2006-10-25 18:16:46 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Substitute with regex */
|
|
|
|
case 's':
|
|
|
|
if (!do_subst_command(sed_cmd, &pattern_space))
|
2006-12-10 07:39:12 +05:30
|
|
|
break;
|
2012-06-04 18:14:47 +05:30
|
|
|
dbg("do_subst_command succeeded:'%s'", pattern_space);
|
2009-06-30 22:49:37 +05:30
|
|
|
substituted |= 1;
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* handle p option */
|
|
|
|
if (sed_cmd->sub_p)
|
|
|
|
sed_puts(pattern_space, last_gets_char);
|
|
|
|
/* handle w option */
|
|
|
|
if (sed_cmd->sw_file)
|
|
|
|
puts_maybe_newline(
|
|
|
|
pattern_space, sed_cmd->sw_file,
|
|
|
|
&sed_cmd->sw_last_char, last_gets_char);
|
|
|
|
break;
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Append line to linked list to be printed later */
|
|
|
|
case 'a':
|
2013-10-30 14:52:47 +05:30
|
|
|
append(xstrdup(sed_cmd->string));
|
2009-06-30 22:49:37 +05:30
|
|
|
break;
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Insert text before this line */
|
|
|
|
case 'i':
|
|
|
|
sed_puts(sed_cmd->string, '\n');
|
|
|
|
break;
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Cut and paste text (replace) */
|
|
|
|
case 'c':
|
|
|
|
/* Only triggers on last line of a matching range. */
|
|
|
|
if (!sed_cmd->in_match)
|
2010-04-20 08:06:07 +05:30
|
|
|
sed_puts(sed_cmd->string, '\n');
|
2009-06-30 22:49:37 +05:30
|
|
|
goto discard_line;
|
2003-04-11 22:40:23 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Read file, append contents to output */
|
|
|
|
case 'r':
|
|
|
|
{
|
|
|
|
FILE *rfile;
|
|
|
|
rfile = fopen_for_read(sed_cmd->string);
|
|
|
|
if (rfile) {
|
|
|
|
char *line;
|
|
|
|
while ((line = xmalloc_fgetline(rfile))
|
|
|
|
!= NULL)
|
|
|
|
append(line);
|
2013-10-30 14:52:47 +05:30
|
|
|
fclose(rfile);
|
2006-12-10 07:39:12 +05:30
|
|
|
}
|
2003-04-09 21:22:32 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
break;
|
|
|
|
}
|
2006-10-25 18:16:46 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Write pattern space to file. */
|
|
|
|
case 'w':
|
|
|
|
puts_maybe_newline(
|
|
|
|
pattern_space, sed_cmd->sw_file,
|
|
|
|
&sed_cmd->sw_last_char, last_gets_char);
|
|
|
|
break;
|
2006-12-10 07:39:12 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Read next line from input */
|
|
|
|
case 'n':
|
|
|
|
if (!G.be_quiet)
|
|
|
|
sed_puts(pattern_space, last_gets_char);
|
|
|
|
if (next_line) {
|
|
|
|
free(pattern_space);
|
|
|
|
pattern_space = next_line;
|
2007-01-29 20:01:47 +05:30
|
|
|
last_gets_char = next_gets_char;
|
2013-10-30 18:55:22 +05:30
|
|
|
next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
|
2009-06-30 22:49:37 +05:30
|
|
|
substituted = 0;
|
2006-12-10 07:39:12 +05:30
|
|
|
linenum++;
|
|
|
|
break;
|
|
|
|
}
|
2009-06-30 22:49:37 +05:30
|
|
|
/* fall through */
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Quit. End of script, end of input. */
|
|
|
|
case 'q':
|
|
|
|
/* Exit the outer while loop */
|
|
|
|
free(next_line);
|
|
|
|
next_line = NULL;
|
|
|
|
goto discard_commands;
|
2006-10-25 18:16:46 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Append the next line to the current line */
|
|
|
|
case 'N':
|
|
|
|
{
|
|
|
|
int len;
|
|
|
|
/* If no next line, jump to end of script and exit. */
|
2010-08-16 19:56:33 +05:30
|
|
|
/* http://www.gnu.org/software/sed/manual/sed.html:
|
|
|
|
* "Most versions of sed exit without printing anything
|
|
|
|
* when the N command is issued on the last line of
|
|
|
|
* a file. GNU sed prints pattern space before exiting
|
|
|
|
* unless of course the -n command switch has been
|
|
|
|
* specified. This choice is by design."
|
|
|
|
*/
|
2009-06-30 22:49:37 +05:30
|
|
|
if (next_line == NULL) {
|
2010-08-16 19:56:33 +05:30
|
|
|
//goto discard_line;
|
|
|
|
goto discard_commands; /* GNU behavior */
|
2006-12-10 07:39:12 +05:30
|
|
|
}
|
2010-06-04 04:59:52 +05:30
|
|
|
/* Append next_line, read new next_line. */
|
2009-06-30 22:49:37 +05:30
|
|
|
len = strlen(pattern_space);
|
2010-01-08 13:37:50 +05:30
|
|
|
pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
|
2009-06-30 22:49:37 +05:30
|
|
|
pattern_space[len] = '\n';
|
|
|
|
strcpy(pattern_space + len+1, next_line);
|
|
|
|
last_gets_char = next_gets_char;
|
2013-10-30 18:55:22 +05:30
|
|
|
next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
|
2009-06-30 22:49:37 +05:30
|
|
|
linenum++;
|
|
|
|
break;
|
|
|
|
}
|
2006-10-25 18:16:46 +05:30
|
|
|
|
2009-06-30 22:49:37 +05:30
|
|
|
/* Test/branch if substitution occurred */
|
|
|
|
case 't':
|
|
|
|
if (!substituted) break;
|
|
|
|
substituted = 0;
|
|
|
|
/* Fall through */
|
|
|
|
/* Test/branch if substitution didn't occur */
|
|
|
|
case 'T':
|
|
|
|
if (substituted) break;
|
|
|
|
/* Fall through */
|
|
|
|
/* Branch to label */
|
|
|
|
case 'b':
|
|
|
|
if (!sed_cmd->string) goto discard_commands;
|
|
|
|
else sed_cmd = branch_to(sed_cmd->string);
|
|
|
|
break;
|
|
|
|
/* Transliterate characters */
|
|
|
|
case 'y':
|
|
|
|
{
|
|
|
|
int i, j;
|
|
|
|
for (i = 0; pattern_space[i]; i++) {
|
|
|
|
for (j = 0; sed_cmd->string[j]; j += 2) {
|
|
|
|
if (pattern_space[i] == sed_cmd->string[j]) {
|
|
|
|
pattern_space[i] = sed_cmd->string[j + 1];
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2002-06-12 05:13:27 +05:30
|
|
|
}
|
2009-06-30 22:49:37 +05:30
|
|
|
|
|
|
|
break;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2009-06-30 22:49:37 +05:30
|
|
|
case 'g': /* Replace pattern space with hold space */
|
|
|
|
free(pattern_space);
|
|
|
|
pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
|
|
|
|
break;
|
|
|
|
case 'G': /* Append newline and hold space to pattern space */
|
|
|
|
{
|
|
|
|
int pattern_space_size = 2;
|
|
|
|
int hold_space_size = 0;
|
|
|
|
|
|
|
|
if (pattern_space)
|
|
|
|
pattern_space_size += strlen(pattern_space);
|
|
|
|
if (G.hold_space)
|
|
|
|
hold_space_size = strlen(G.hold_space);
|
|
|
|
pattern_space = xrealloc(pattern_space,
|
|
|
|
pattern_space_size + hold_space_size);
|
|
|
|
if (pattern_space_size == 2)
|
|
|
|
pattern_space[0] = 0;
|
|
|
|
strcat(pattern_space, "\n");
|
|
|
|
if (G.hold_space)
|
|
|
|
strcat(pattern_space, G.hold_space);
|
|
|
|
last_gets_char = '\n';
|
|
|
|
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
case 'h': /* Replace hold space with pattern space */
|
|
|
|
free(G.hold_space);
|
|
|
|
G.hold_space = xstrdup(pattern_space);
|
|
|
|
break;
|
|
|
|
case 'H': /* Append newline and pattern space to hold space */
|
|
|
|
{
|
|
|
|
int hold_space_size = 2;
|
|
|
|
int pattern_space_size = 0;
|
|
|
|
|
|
|
|
if (G.hold_space)
|
|
|
|
hold_space_size += strlen(G.hold_space);
|
|
|
|
if (pattern_space)
|
|
|
|
pattern_space_size = strlen(pattern_space);
|
|
|
|
G.hold_space = xrealloc(G.hold_space,
|
|
|
|
hold_space_size + pattern_space_size);
|
|
|
|
|
|
|
|
if (hold_space_size == 2)
|
|
|
|
*G.hold_space = 0;
|
|
|
|
strcat(G.hold_space, "\n");
|
|
|
|
if (pattern_space)
|
|
|
|
strcat(G.hold_space, pattern_space);
|
|
|
|
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
case 'x': /* Exchange hold and pattern space */
|
|
|
|
{
|
|
|
|
char *tmp = pattern_space;
|
2009-09-06 06:06:23 +05:30
|
|
|
pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
|
2009-06-30 22:49:37 +05:30
|
|
|
last_gets_char = '\n';
|
|
|
|
G.hold_space = tmp;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} /* switch */
|
|
|
|
} /* for each cmd */
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/*
|
2009-06-30 22:49:37 +05:30
|
|
|
* Exit point from sedding...
|
2006-12-10 07:39:12 +05:30
|
|
|
*/
|
2007-01-29 04:56:15 +05:30
|
|
|
discard_commands:
|
2006-12-10 07:39:12 +05:30
|
|
|
/* we will print the line unless we were told to be quiet ('-n')
|
|
|
|
or if the line was suppressed (ala 'd'elete) */
|
2007-03-27 02:18:46 +05:30
|
|
|
if (!G.be_quiet)
|
2007-01-29 20:01:47 +05:30
|
|
|
sed_puts(pattern_space, last_gets_char);
|
2003-10-01 08:36:16 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/* Delete and such jump here. */
|
2007-01-29 04:56:15 +05:30
|
|
|
discard_line:
|
2013-10-30 18:55:22 +05:30
|
|
|
flush_append(&last_puts_char, last_gets_char);
|
2006-12-10 07:39:12 +05:30
|
|
|
free(pattern_space);
|
|
|
|
|
|
|
|
goto again;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
|
|
|
|
2003-10-04 10:57:56 +05:30
|
|
|
/* It is possible to have a command line argument with embedded
|
2007-03-17 05:06:58 +05:30
|
|
|
* newlines. This counts as multiple command lines.
|
|
|
|
* However, newline can be escaped: 's/e/z\<newline>z/'
|
2013-07-08 05:34:44 +05:30
|
|
|
* add_cmd() handles this.
|
2007-03-17 05:06:58 +05:30
|
|
|
*/
|
2003-10-04 10:57:56 +05:30
|
|
|
|
|
|
|
static void add_cmd_block(char *cmdstr)
|
|
|
|
{
|
2007-03-17 05:06:58 +05:30
|
|
|
char *sv, *eol;
|
|
|
|
|
|
|
|
cmdstr = sv = xstrdup(cmdstr);
|
|
|
|
do {
|
|
|
|
eol = strchr(cmdstr, '\n');
|
2013-07-08 05:34:44 +05:30
|
|
|
if (eol)
|
2007-03-17 05:06:58 +05:30
|
|
|
*eol = '\0';
|
|
|
|
add_cmd(cmdstr);
|
|
|
|
cmdstr = eol + 1;
|
|
|
|
} while (eol);
|
|
|
|
free(sv);
|
2003-10-04 10:57:56 +05:30
|
|
|
}
|
|
|
|
|
2007-10-11 15:35:36 +05:30
|
|
|
int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
2008-07-05 14:48:54 +05:30
|
|
|
int sed_main(int argc UNUSED_PARAM, char **argv)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2006-10-04 02:30:06 +05:30
|
|
|
unsigned opt;
|
2006-10-02 02:35:12 +05:30
|
|
|
llist_t *opt_e, *opt_f;
|
2012-05-06 17:29:15 +05:30
|
|
|
char *opt_i;
|
|
|
|
|
|
|
|
#if ENABLE_LONG_OPTS
|
|
|
|
static const char sed_longopts[] ALIGN1 =
|
|
|
|
/* name has_arg short */
|
|
|
|
"in-place\0" Optional_argument "i"
|
|
|
|
"regexp-extended\0" No_argument "r"
|
|
|
|
"quiet\0" No_argument "n"
|
|
|
|
"silent\0" No_argument "n"
|
|
|
|
"expression\0" Required_argument "e"
|
|
|
|
"file\0" Required_argument "f";
|
|
|
|
#endif
|
|
|
|
|
2007-06-04 15:46:52 +05:30
|
|
|
INIT_G();
|
2006-05-09 09:23:55 +05:30
|
|
|
|
2000-07-12 03:08:47 +05:30
|
|
|
/* destroy command strings on exit */
|
2006-05-09 03:47:23 +05:30
|
|
|
if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
|
2000-07-12 03:08:47 +05:30
|
|
|
|
2005-11-20 13:14:35 +05:30
|
|
|
/* Lie to autoconf when it starts asking stupid questions. */
|
2012-05-06 17:29:15 +05:30
|
|
|
if (argv[1] && strcmp(argv[1], "--version") == 0) {
|
2006-12-10 07:39:12 +05:30
|
|
|
puts("This is not GNU sed version 4.0");
|
|
|
|
return 0;
|
2004-02-04 16:27:46 +05:30
|
|
|
}
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* do normal option parsing */
|
2006-10-02 02:35:12 +05:30
|
|
|
opt_e = opt_f = NULL;
|
2012-05-06 17:29:15 +05:30
|
|
|
opt_i = NULL;
|
2006-12-10 07:39:12 +05:30
|
|
|
opt_complementary = "e::f::" /* can occur multiple times */
|
|
|
|
"nn"; /* count -n */
|
2012-05-06 17:29:15 +05:30
|
|
|
|
|
|
|
IF_LONG_OPTS(applet_long_options = sed_longopts);
|
|
|
|
|
2010-08-01 07:44:46 +05:30
|
|
|
/* -i must be first, to match OPT_in_place definition */
|
2013-10-29 05:22:48 +05:30
|
|
|
/* -E is a synonym of -r:
|
|
|
|
* GNU sed 4.2.1 mentions it in neither --help
|
|
|
|
* nor manpage, but does recognize it.
|
|
|
|
*/
|
|
|
|
opt = getopt32(argv, "i::rEne:f:", &opt_i, &opt_e, &opt_f,
|
2007-03-27 02:18:46 +05:30
|
|
|
&G.be_quiet); /* counter for -n */
|
2008-03-17 14:37:36 +05:30
|
|
|
//argc -= optind;
|
2006-12-10 07:39:12 +05:30
|
|
|
argv += optind;
|
|
|
|
if (opt & OPT_in_place) { // -i
|
2006-09-22 14:26:03 +05:30
|
|
|
atexit(cleanup_outname);
|
|
|
|
}
|
2013-10-29 05:22:48 +05:30
|
|
|
if (opt & (2|4))
|
|
|
|
G.regex_type |= REG_EXTENDED; // -r or -E
|
|
|
|
//if (opt & 8)
|
|
|
|
// G.be_quiet++; // -n (implemented with a counter instead)
|
2007-04-13 02:50:25 +05:30
|
|
|
while (opt_e) { // -e
|
2008-06-15 11:10:56 +05:30
|
|
|
add_cmd_block(llist_pop(&opt_e));
|
2006-09-22 14:26:03 +05:30
|
|
|
}
|
2007-04-13 02:50:25 +05:30
|
|
|
while (opt_f) { // -f
|
|
|
|
char *line;
|
|
|
|
FILE *cmdfile;
|
2008-07-22 04:35:26 +05:30
|
|
|
cmdfile = xfopen_for_read(llist_pop(&opt_f));
|
2008-03-27 01:34:27 +05:30
|
|
|
while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
|
2007-04-13 02:50:25 +05:30
|
|
|
add_cmd(line);
|
|
|
|
free(line);
|
|
|
|
}
|
|
|
|
fclose(cmdfile);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2006-12-10 07:39:12 +05:30
|
|
|
/* if we didn't get a pattern from -e or -f, use argv[0] */
|
2013-10-29 05:22:48 +05:30
|
|
|
if (!(opt & 0x30)) {
|
2008-03-17 14:37:36 +05:30
|
|
|
if (!*argv)
|
2003-03-19 14:43:01 +05:30
|
|
|
bb_show_usage();
|
2006-12-10 07:39:12 +05:30
|
|
|
add_cmd_block(*argv++);
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2003-10-04 10:57:56 +05:30
|
|
|
/* Flush any unfinished commands. */
|
|
|
|
add_cmd("");
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2004-02-18 15:24:15 +05:30
|
|
|
/* By default, we write to stdout */
|
2007-03-27 02:18:46 +05:30
|
|
|
G.nonstdout = stdout;
|
2004-02-18 15:24:15 +05:30
|
|
|
|
2006-12-10 07:39:12 +05:30
|
|
|
/* argv[0..(argc-1)] should be names of file to process. If no
|
2000-07-11 04:25:51 +05:30
|
|
|
* files were specified or '-' was specified, take input from stdin.
|
|
|
|
* Otherwise, we process all the files specified. */
|
2013-11-28 16:38:51 +05:30
|
|
|
G.input_file_list = argv;
|
|
|
|
if (!argv[0]) {
|
2006-12-10 07:39:12 +05:30
|
|
|
if (opt & OPT_in_place)
|
2006-10-25 18:16:46 +05:30
|
|
|
bb_error_msg_and_die(bb_msg_requires_arg, "-i");
|
2013-11-28 16:38:51 +05:30
|
|
|
argv[0] = (char*)bb_msg_standard_input;
|
|
|
|
/* G.last_input_file = 0; - already is */
|
2003-04-08 17:26:11 +05:30
|
|
|
} else {
|
2013-11-28 16:38:51 +05:30
|
|
|
goto start;
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2013-11-28 16:38:51 +05:30
|
|
|
for (; *argv; argv++) {
|
2006-10-25 18:16:03 +05:30
|
|
|
struct stat statbuf;
|
|
|
|
int nonstdoutfd;
|
2011-08-16 16:59:34 +05:30
|
|
|
sed_cmd_t *sed_cmd;
|
2006-10-25 18:16:03 +05:30
|
|
|
|
2013-11-28 16:38:51 +05:30
|
|
|
G.last_input_file++;
|
|
|
|
start:
|
2006-12-10 07:39:12 +05:30
|
|
|
if (!(opt & OPT_in_place)) {
|
2013-11-28 16:38:51 +05:30
|
|
|
if (LONE_DASH(*argv)) {
|
|
|
|
*argv = (char*)bb_msg_standard_input;
|
|
|
|
process_files();
|
|
|
|
}
|
2006-10-25 18:16:03 +05:30
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2011-08-16 16:59:34 +05:30
|
|
|
/* -i: process each FILE separately: */
|
|
|
|
|
2014-10-05 06:14:34 +05:30
|
|
|
if (stat(*argv, &statbuf) != 0) {
|
|
|
|
bb_simple_perror_msg(*argv);
|
|
|
|
G.exitcode = EXIT_FAILURE;
|
|
|
|
G.current_input_file++;
|
|
|
|
continue;
|
|
|
|
}
|
2013-11-28 16:38:51 +05:30
|
|
|
G.outname = xasprintf("%sXXXXXX", *argv);
|
2010-10-22 16:57:16 +05:30
|
|
|
nonstdoutfd = xmkstemp(G.outname);
|
2009-11-16 03:58:11 +05:30
|
|
|
G.nonstdout = xfdopen_for_write(nonstdoutfd);
|
2009-11-13 13:39:07 +05:30
|
|
|
/* Set permissions/owner of output file */
|
2010-06-18 05:42:56 +05:30
|
|
|
/* chmod'ing AFTER chown would preserve suid/sgid bits,
|
|
|
|
* but GNU sed 4.2.1 does not preserve them either */
|
2006-12-10 07:39:12 +05:30
|
|
|
fchmod(nonstdoutfd, statbuf.st_mode);
|
2009-11-13 13:39:07 +05:30
|
|
|
fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
|
2011-08-16 16:59:34 +05:30
|
|
|
|
2006-10-25 18:16:03 +05:30
|
|
|
process_files();
|
2007-03-27 02:18:46 +05:30
|
|
|
fclose(G.nonstdout);
|
|
|
|
G.nonstdout = stdout;
|
2011-08-16 16:59:34 +05:30
|
|
|
|
2012-05-06 17:29:15 +05:30
|
|
|
if (opt_i) {
|
2013-11-28 16:38:51 +05:30
|
|
|
char *backupname = xasprintf("%s%s", *argv, opt_i);
|
|
|
|
xrename(*argv, backupname);
|
2012-05-06 17:29:15 +05:30
|
|
|
free(backupname);
|
|
|
|
}
|
2013-11-28 16:38:51 +05:30
|
|
|
/* else unlink(*argv); - rename below does this */
|
|
|
|
xrename(G.outname, *argv); //TODO: rollback backup on error?
|
2007-03-27 02:18:46 +05:30
|
|
|
free(G.outname);
|
2007-11-13 22:18:10 +05:30
|
|
|
G.outname = NULL;
|
2011-08-16 16:59:34 +05:30
|
|
|
|
|
|
|
/* Re-enable disabled range matches */
|
|
|
|
for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
|
|
|
|
sed_cmd->beg_line = sed_cmd->beg_line_orig;
|
|
|
|
}
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
2010-06-19 21:21:06 +05:30
|
|
|
/* Here, to handle "sed 'cmds' nonexistent_file" case we did:
|
2013-11-28 07:44:16 +05:30
|
|
|
* if (G.current_input_file[G.current_input_file] == NULL)
|
|
|
|
* return G.exitcode;
|
2010-06-19 21:21:06 +05:30
|
|
|
* but it's not needed since process_files() works correctly
|
|
|
|
* in this case too. */
|
1999-11-09 07:17:36 +05:30
|
|
|
}
|
2013-11-28 07:44:16 +05:30
|
|
|
|
2010-06-19 21:21:06 +05:30
|
|
|
process_files();
|
2003-04-08 17:26:11 +05:30
|
|
|
|
2013-11-28 07:44:16 +05:30
|
|
|
return G.exitcode;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|