1999-10-28 16:06:25 +00:00
|
|
|
/*
|
2000-07-10 22:55:51 +00:00
|
|
|
* sed.c - very minimalist version of sed
|
1999-10-28 16:06:25 +00:00
|
|
|
*
|
2001-10-24 05:00:29 +00:00
|
|
|
* Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
|
|
|
|
* Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
|
2002-01-03 21:12:34 +00:00
|
|
|
* Copyright (C) 2002 Matt Kraai
|
1999-12-29 22:19:46 +00:00
|
|
|
*
|
1999-10-28 16:06:25 +00:00
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/*
|
|
|
|
Supported features and commands in this version of sed:
|
|
|
|
|
|
|
|
- comments ('#')
|
2000-07-14 00:00:15 +00:00
|
|
|
- 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
|
2001-05-11 22:27:13 +00:00
|
|
|
- file commands: (r)ead
|
2000-07-17 20:06:42 +00:00
|
|
|
- backreferences in substitution expressions (\1, \2...\9)
|
2003-03-09 15:12:24 +00:00
|
|
|
- grouped commands: {cmd1;cmd2}
|
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
(Note: Specifying an address (range) to match is *optional*; commands
|
|
|
|
default to the whole pattern space if no specific address match was
|
|
|
|
requested.)
|
|
|
|
|
|
|
|
Unsupported features:
|
|
|
|
|
|
|
|
- transliteration (y/source-chars/dest-chars/) (use 'tr')
|
|
|
|
- no pattern space hold space storing / swapping (x, etc.)
|
|
|
|
- no labels / branching (: label, b, t, and friends)
|
|
|
|
- and lots, lots more.
|
2003-03-09 02:39:29 +00:00
|
|
|
|
|
|
|
Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
|
2000-07-10 22:55:51 +00:00
|
|
|
*/
|
|
|
|
|
1999-10-28 16:06:25 +00:00
|
|
|
#include <stdio.h>
|
2000-07-10 22:55:51 +00:00
|
|
|
#include <unistd.h> /* for getopt() */
|
|
|
|
#include <regex.h>
|
|
|
|
#include <string.h> /* for strdup() */
|
1999-10-28 16:06:25 +00:00
|
|
|
#include <errno.h>
|
2000-07-10 22:55:51 +00:00
|
|
|
#include <ctype.h> /* for isspace() */
|
2001-01-27 08:24:39 +00:00
|
|
|
#include <stdlib.h>
|
2000-09-25 21:45:58 +00:00
|
|
|
#include "busybox.h"
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
/* externs */
|
2001-01-27 08:24:39 +00:00
|
|
|
extern void xregcomp(regex_t *preg, const char *regex, int cflags);
|
2000-07-10 22:55:51 +00:00
|
|
|
extern int optind; /* in unistd.h */
|
|
|
|
extern char *optarg; /* ditto */
|
|
|
|
|
|
|
|
/* options */
|
|
|
|
static int be_quiet = 0;
|
|
|
|
|
2000-08-21 21:29:20 +00:00
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
typedef struct sed_cmd_s {
|
2001-11-10 10:49:42 +00:00
|
|
|
/* Order by alignment requirements */
|
2000-11-03 19:47:00 +00:00
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* address storage */
|
|
|
|
regex_t *beg_match; /* sed -e '/match/cmd' */
|
|
|
|
regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
|
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* SUBSTITUTION COMMAND SPECIFIC FIELDS */
|
|
|
|
|
|
|
|
/* sed -e 's/sub_match/replace/' */
|
|
|
|
regex_t *sub_match;
|
|
|
|
char *replace;
|
2001-11-10 10:49:42 +00:00
|
|
|
|
|
|
|
/* EDIT COMMAND (a,i,c) SPECIFIC FIELDS */
|
|
|
|
char *editline;
|
|
|
|
|
|
|
|
/* FILE COMMAND (r) SPECIFIC FIELDS */
|
|
|
|
char *filename;
|
|
|
|
|
|
|
|
/* address storage */
|
|
|
|
int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
|
|
|
|
int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
|
|
|
|
/* SUBSTITUTION COMMAND SPECIFIC FIELDS */
|
|
|
|
|
2000-07-17 20:06:42 +00:00
|
|
|
unsigned int num_backrefs:4; /* how many back references (\1..\9) */
|
|
|
|
/* Note: GNU/POSIX sed does not save more than nine backrefs, so
|
|
|
|
* we only use 4 bits to hold the number */
|
2000-11-03 19:47:00 +00:00
|
|
|
unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
|
|
|
|
unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
|
|
|
|
|
2001-11-10 10:49:42 +00:00
|
|
|
/* GENERAL FIELDS */
|
|
|
|
char delimiter; /* The delimiter used to separate regexps */
|
2000-11-03 19:47:00 +00:00
|
|
|
|
2001-11-10 10:49:42 +00:00
|
|
|
/* the command */
|
|
|
|
char cmd; /* p,d,s (add more at your leisure :-) */
|
2002-06-11 23:43:27 +00:00
|
|
|
|
|
|
|
/* inversion flag */
|
|
|
|
int invert; /* the '!' after the address */
|
2003-03-09 10:23:57 +00:00
|
|
|
} sed_cmd_t;
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
/* globals */
|
2003-03-09 11:06:38 +00:00
|
|
|
static sed_cmd_t **sed_cmds = NULL; /* growable arrary holding a sequence of sed cmds */
|
2000-07-10 22:55:51 +00:00
|
|
|
static int ncmds = 0; /* number of sed commands */
|
|
|
|
|
|
|
|
/*static char *cur_file = NULL;*/ /* file currently being processed XXX: do I need this? */
|
1999-10-28 16:06:25 +00:00
|
|
|
|
2001-11-10 10:49:42 +00:00
|
|
|
const char * const semicolon_whitespace = "; \n\r\t\v\0";
|
|
|
|
|
2001-10-24 05:00:29 +00:00
|
|
|
#ifdef CONFIG_FEATURE_CLEAN_UP
|
2001-11-20 16:00:19 +00:00
|
|
|
static void destroy_cmd_strs(void)
|
2000-07-10 22:55:51 +00:00
|
|
|
{
|
|
|
|
if (sed_cmds == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* destroy all the elements in the array */
|
|
|
|
while (--ncmds >= 0) {
|
2000-02-08 19:58:47 +00:00
|
|
|
|
2003-03-09 11:06:38 +00:00
|
|
|
if (sed_cmds[ncmds]->beg_match) {
|
|
|
|
regfree(sed_cmds[ncmds]->beg_match);
|
|
|
|
free(sed_cmds[ncmds]->beg_match);
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
2003-03-09 11:06:38 +00:00
|
|
|
if (sed_cmds[ncmds]->end_match) {
|
|
|
|
regfree(sed_cmds[ncmds]->end_match);
|
|
|
|
free(sed_cmds[ncmds]->end_match);
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
2003-03-09 11:06:38 +00:00
|
|
|
if (sed_cmds[ncmds]->sub_match) {
|
|
|
|
regfree(sed_cmds[ncmds]->sub_match);
|
|
|
|
free(sed_cmds[ncmds]->sub_match);
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
2003-03-09 11:06:38 +00:00
|
|
|
free(sed_cmds[ncmds]->replace);
|
2000-02-08 19:58:47 +00:00
|
|
|
}
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
/* destroy the array */
|
|
|
|
free(sed_cmds);
|
|
|
|
sed_cmds = NULL;
|
|
|
|
}
|
2000-07-12 23:35:21 +00:00
|
|
|
#endif
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
|
|
|
|
/*
|
2001-01-02 11:01:31 +00:00
|
|
|
* 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
|
|
|
|
* expression delimiter (typically a forward * slash ('/')) not preceeded by
|
|
|
|
* a backslash ('\').
|
2000-07-10 22:55:51 +00:00
|
|
|
*/
|
2003-03-09 08:44:49 +00:00
|
|
|
static int index_of_next_unescaped_regexp_delim(const char delimiter, const char *str, int idx)
|
2000-07-10 22:55:51 +00:00
|
|
|
{
|
2001-08-24 14:45:50 +00:00
|
|
|
int bracket = -1;
|
|
|
|
int escaped = 0;
|
2001-11-10 10:49:42 +00:00
|
|
|
char ch;
|
2001-08-24 14:45:50 +00:00
|
|
|
|
2001-11-10 10:49:42 +00:00
|
|
|
for ( ; (ch = str[idx]); idx++) {
|
2001-08-24 14:45:50 +00:00
|
|
|
if (bracket != -1) {
|
2001-11-10 10:49:42 +00:00
|
|
|
if (ch == ']' && !(bracket == idx - 1 ||
|
2001-08-24 14:45:50 +00:00
|
|
|
(bracket == idx - 2 && str[idx-1] == '^')))
|
|
|
|
bracket = -1;
|
|
|
|
} else if (escaped)
|
|
|
|
escaped = 0;
|
2001-11-10 10:49:42 +00:00
|
|
|
else if (ch == '\\')
|
2001-08-24 14:45:50 +00:00
|
|
|
escaped = 1;
|
2001-11-10 10:49:42 +00:00
|
|
|
else if (ch == '[')
|
2001-08-24 14:45:50 +00:00
|
|
|
bracket = idx;
|
2003-03-09 08:44:49 +00:00
|
|
|
else if (ch == delimiter)
|
2000-07-17 20:06:42 +00:00
|
|
|
return idx;
|
|
|
|
}
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2000-07-17 20:06:42 +00:00
|
|
|
/* if we make it to here, we've hit the end of the string */
|
|
|
|
return -1;
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* returns the index in the string just past where the address ends.
|
|
|
|
*/
|
2003-03-09 08:44:49 +00:00
|
|
|
static int get_address(char *delimiter, char *my_str, int *linenum, regex_t **regex)
|
1999-12-29 22:19:46 +00:00
|
|
|
{
|
2003-03-09 08:44:49 +00:00
|
|
|
int idx = 0;
|
2000-07-10 22:55:51 +00:00
|
|
|
if (isdigit(my_str[idx])) {
|
2003-03-09 08:44:49 +00:00
|
|
|
char *endstr;
|
|
|
|
*linenum = strtol(my_str, &endstr, 10);
|
|
|
|
/* endstr shouldnt ever equal NULL */
|
|
|
|
idx = endstr - my_str;
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
else if (my_str[idx] == '$') {
|
2001-06-11 23:50:06 +00:00
|
|
|
*linenum = -1;
|
2000-07-10 22:55:51 +00:00
|
|
|
idx++;
|
|
|
|
}
|
2002-08-06 21:07:17 +00:00
|
|
|
else if (my_str[idx] == '/' || my_str[idx] == '\\') {
|
2003-03-09 08:44:49 +00:00
|
|
|
int idx_start = 1;
|
|
|
|
|
2003-03-09 15:40:40 +00:00
|
|
|
*delimiter = '/';
|
2002-08-06 21:07:17 +00:00
|
|
|
if (my_str[idx] == '\\') {
|
2002-08-06 23:13:31 +00:00
|
|
|
idx_start++;
|
2003-03-09 08:44:49 +00:00
|
|
|
*delimiter = my_str[++idx];
|
2002-08-06 21:07:17 +00:00
|
|
|
}
|
2003-03-09 08:44:49 +00:00
|
|
|
idx = index_of_next_unescaped_regexp_delim(*delimiter, my_str, ++idx);
|
|
|
|
if (idx == -1) {
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("unterminated match expression");
|
2003-03-09 08:44:49 +00:00
|
|
|
}
|
2000-07-11 16:53:56 +00:00
|
|
|
my_str[idx] = '\0';
|
|
|
|
*regex = (regex_t *)xmalloc(sizeof(regex_t));
|
2002-08-06 23:13:31 +00:00
|
|
|
xregcomp(*regex, my_str+idx_start, REG_NEWLINE);
|
2000-07-13 22:52:02 +00:00
|
|
|
idx++; /* so it points to the next character after the last '/' */
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
return idx;
|
|
|
|
}
|
1999-12-29 22:19:46 +00:00
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
static int parse_subst_cmd(sed_cmd_t * const sed_cmd, const char *substr)
|
2000-07-13 19:58:04 +00:00
|
|
|
{
|
|
|
|
int oldidx, cflags = REG_NEWLINE;
|
|
|
|
char *match;
|
|
|
|
int idx = 0;
|
2000-07-17 20:06:42 +00:00
|
|
|
int j;
|
2000-07-13 19:58:04 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* the string that gets passed to this function should look like this:
|
2000-08-21 21:29:20 +00:00
|
|
|
* s/match/replace/gIp
|
|
|
|
* || | |||
|
2000-07-13 19:58:04 +00:00
|
|
|
* mandatory optional
|
|
|
|
*
|
|
|
|
* (all three of the '/' slashes are mandatory)
|
|
|
|
*/
|
|
|
|
|
2001-01-02 11:01:31 +00:00
|
|
|
/* verify that the 's' is followed by something. That something
|
|
|
|
* (typically a 'slash') is now our regexp delimiter... */
|
2003-03-09 15:12:24 +00:00
|
|
|
if (substr[idx] == '\0')
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("bad format in substitution expression");
|
2001-01-02 11:01:31 +00:00
|
|
|
else
|
|
|
|
sed_cmd->delimiter=substr[idx];
|
2000-07-13 19:58:04 +00:00
|
|
|
|
|
|
|
/* save the match string */
|
|
|
|
oldidx = idx+1;
|
2003-03-09 08:44:49 +00:00
|
|
|
idx = index_of_next_unescaped_regexp_delim(sed_cmd->delimiter, substr, ++idx);
|
2000-07-13 19:58:04 +00:00
|
|
|
if (idx == -1)
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("bad format in substitution expression");
|
2001-07-05 19:00:47 +00:00
|
|
|
match = xstrndup(substr + oldidx, idx - oldidx);
|
2000-07-13 19:58:04 +00:00
|
|
|
|
2000-07-17 20:06:42 +00:00
|
|
|
/* 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 */
|
|
|
|
|
|
|
|
/* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
|
|
|
|
for (j = 0; match[j]; j++) {
|
|
|
|
/* GNU/POSIX sed does not save more than nine backrefs */
|
2000-11-03 19:47:00 +00:00
|
|
|
if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
|
2000-07-17 20:06:42 +00:00
|
|
|
sed_cmd->num_backrefs++;
|
|
|
|
}
|
|
|
|
|
2000-07-13 19:58:04 +00:00
|
|
|
/* save the replacement string */
|
|
|
|
oldidx = idx+1;
|
2003-03-09 08:44:49 +00:00
|
|
|
idx = index_of_next_unescaped_regexp_delim(sed_cmd->delimiter, substr, ++idx);
|
2000-07-13 19:58:04 +00:00
|
|
|
if (idx == -1)
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("bad format in substitution expression");
|
2001-07-05 19:00:47 +00:00
|
|
|
sed_cmd->replace = xstrndup(substr + oldidx, idx - oldidx);
|
2000-07-13 19:58:04 +00:00
|
|
|
|
|
|
|
/* process the flags */
|
|
|
|
while (substr[++idx]) {
|
|
|
|
switch (substr[idx]) {
|
2000-07-14 19:06:30 +00:00
|
|
|
case 'g':
|
2000-11-03 19:47:00 +00:00
|
|
|
sed_cmd->sub_g = 1;
|
2000-07-14 19:06:30 +00:00
|
|
|
break;
|
|
|
|
case 'I':
|
|
|
|
cflags |= REG_ICASE;
|
|
|
|
break;
|
2000-08-21 21:29:20 +00:00
|
|
|
case 'p':
|
2000-11-03 19:47:00 +00:00
|
|
|
sed_cmd->sub_p = 1;
|
2000-08-21 21:29:20 +00:00
|
|
|
break;
|
2000-07-14 19:06:30 +00:00
|
|
|
default:
|
|
|
|
/* any whitespace or semicolon trailing after a s/// is ok */
|
2001-11-10 10:49:42 +00:00
|
|
|
if (strchr(semicolon_whitespace, substr[idx]))
|
2000-07-14 19:06:30 +00:00
|
|
|
goto out;
|
|
|
|
/* else */
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("bad option in substitution expression");
|
2000-07-13 19:58:04 +00:00
|
|
|
}
|
|
|
|
}
|
2000-07-14 19:06:30 +00:00
|
|
|
|
|
|
|
out:
|
2000-07-17 20:06:42 +00:00
|
|
|
/* compile the match string into a regex */
|
2000-07-13 19:58:04 +00:00
|
|
|
sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
|
|
|
|
xregcomp(sed_cmd->sub_match, match, cflags);
|
|
|
|
free(match);
|
2000-07-14 19:06:30 +00:00
|
|
|
|
|
|
|
return idx;
|
2000-07-13 19:58:04 +00:00
|
|
|
}
|
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
static int parse_edit_cmd(sed_cmd_t *sed_cmd, const char *editstr)
|
2000-07-14 00:00:15 +00:00
|
|
|
{
|
2002-01-03 21:12:34 +00:00
|
|
|
int i, j;
|
2000-07-14 00:00:15 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* the string that gets passed to this function should look like this:
|
|
|
|
*
|
|
|
|
* need one of these
|
|
|
|
* |
|
|
|
|
* | this backslash (immediately following the edit command) is mandatory
|
|
|
|
* | |
|
|
|
|
* [aic]\
|
|
|
|
* TEXT1\
|
|
|
|
* TEXT2\
|
|
|
|
* TEXTN
|
|
|
|
*
|
|
|
|
* as soon as we hit a TEXT line that has no trailing '\', we're done.
|
|
|
|
* this means a command like:
|
|
|
|
*
|
|
|
|
* i\
|
|
|
|
* INSERTME
|
|
|
|
*
|
|
|
|
* is a-ok.
|
|
|
|
*
|
|
|
|
*/
|
2003-03-09 15:12:24 +00:00
|
|
|
if ((*editstr != '\\') || ((editstr[1] != '\n') && (editstr[1] != '\r'))) {
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("bad format in edit expression");
|
2003-03-09 15:12:24 +00:00
|
|
|
}
|
2000-07-14 00:00:15 +00:00
|
|
|
|
|
|
|
/* store the edit line text */
|
2003-03-09 15:12:24 +00:00
|
|
|
sed_cmd->editline = xmalloc(strlen(&editstr[2]) + 2);
|
|
|
|
for (i = 2, j = 0; editstr[i] != '\0' && strchr("\r\n", editstr[i]) == NULL;
|
2002-01-03 21:12:34 +00:00
|
|
|
i++, j++) {
|
2003-03-09 15:12:24 +00:00
|
|
|
if ((editstr[i] == '\\') && strchr("\n\r", editstr[i+1]) != NULL) {
|
2002-01-03 21:12:34 +00:00
|
|
|
sed_cmd->editline[j] = '\n';
|
|
|
|
i++;
|
|
|
|
} else
|
|
|
|
sed_cmd->editline[j] = editstr[i];
|
2000-07-14 00:00:15 +00:00
|
|
|
}
|
2000-07-14 19:06:30 +00:00
|
|
|
|
2001-04-20 23:41:44 +00:00
|
|
|
/* figure out if we need to add a newline */
|
2002-01-03 21:12:34 +00:00
|
|
|
if (sed_cmd->editline[j-1] != '\n')
|
|
|
|
sed_cmd->editline[j++] = '\n';
|
2001-04-20 23:41:44 +00:00
|
|
|
|
|
|
|
/* terminate string */
|
2002-01-03 21:12:34 +00:00
|
|
|
sed_cmd->editline[j] = '\0';
|
2000-07-14 23:24:00 +00:00
|
|
|
|
2002-01-03 21:12:34 +00:00
|
|
|
return i;
|
2000-07-14 00:00:15 +00:00
|
|
|
}
|
|
|
|
|
2001-05-11 22:27:13 +00:00
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
static int parse_file_cmd(sed_cmd_t *sed_cmd, const char *filecmdstr)
|
2001-05-11 22:27:13 +00:00
|
|
|
{
|
|
|
|
int idx = 0;
|
|
|
|
int filenamelen = 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* the string that gets passed to this function should look like this:
|
|
|
|
* '[ ]filename'
|
|
|
|
* | |
|
|
|
|
* | a filename
|
|
|
|
* |
|
|
|
|
* optional whitespace
|
|
|
|
|
|
|
|
* re: the file to be read, the GNU manual says the following: "Note that
|
|
|
|
* if filename cannot be read, it is treated as if it were an empty file,
|
|
|
|
* without any error indication." Thus, all of the following commands are
|
|
|
|
* perfectly leagal:
|
|
|
|
*
|
|
|
|
* sed -e '1r noexist'
|
|
|
|
* sed -e '1r ;'
|
|
|
|
* sed -e '1r'
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* the file command may be followed by whitespace; move past it. */
|
|
|
|
while (isspace(filecmdstr[++idx]))
|
|
|
|
{ ; }
|
|
|
|
|
|
|
|
/* the first non-whitespace we get is a filename. the filename ends when we
|
|
|
|
* hit a normal sed command terminator or end of string */
|
2001-11-10 10:49:42 +00:00
|
|
|
filenamelen = strcspn(&filecmdstr[idx], semicolon_whitespace);
|
2001-05-27 14:11:52 +00:00
|
|
|
sed_cmd->filename = xmalloc(filenamelen + 1);
|
|
|
|
safe_strncpy(sed_cmd->filename, &filecmdstr[idx], filenamelen + 1);
|
2001-05-11 22:27:13 +00:00
|
|
|
|
|
|
|
return idx + filenamelen;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
static char *parse_cmd_str(sed_cmd_t * const sed_cmd, char *cmdstr)
|
2000-07-10 22:55:51 +00:00
|
|
|
{
|
2003-03-09 15:12:24 +00:00
|
|
|
/* if it was a single-letter command that takes no arguments (such as 'p'
|
|
|
|
* or 'd') all we need to do is increment the index past that command */
|
|
|
|
if (strchr("pd=", sed_cmd->cmd)) {
|
|
|
|
cmdstr++;
|
|
|
|
}
|
|
|
|
/* handle (s)ubstitution command */
|
|
|
|
else if (sed_cmd->cmd == 's') {
|
|
|
|
cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
|
|
|
|
}
|
|
|
|
/* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
|
|
|
|
else if (strchr("aic", sed_cmd->cmd)) {
|
|
|
|
if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
|
|
|
|
error_msg_and_die("only a beginning address can be specified for edit commands");
|
|
|
|
cmdstr += parse_edit_cmd(sed_cmd, cmdstr);
|
|
|
|
}
|
|
|
|
/* handle file cmds: (r)ead */
|
|
|
|
else if (sed_cmd->cmd == 'r') {
|
|
|
|
if (sed_cmd->end_line || sed_cmd->end_match)
|
|
|
|
error_msg_and_die("Command only uses one address");
|
|
|
|
cmdstr += parse_file_cmd(sed_cmd, cmdstr);
|
|
|
|
}
|
|
|
|
/* handle grouped commands */
|
|
|
|
else {
|
|
|
|
error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* give back whatever's left over */
|
|
|
|
return(cmdstr);
|
|
|
|
}
|
|
|
|
|
|
|
|
static char *add_cmd(sed_cmd_t *sed_cmd, char *cmdstr)
|
|
|
|
{
|
|
|
|
|
|
|
|
/* Skip over leading whitespace and semicolons */
|
|
|
|
cmdstr += strspn(cmdstr, semicolon_whitespace);
|
|
|
|
|
|
|
|
/* if we ate the whole thing, that means there was just trailing
|
|
|
|
* whitespace or a final / no-op semicolon. either way, get out */
|
|
|
|
if (*cmdstr == '\0') {
|
|
|
|
return(NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* if this is a comment, jump past it and keep going */
|
|
|
|
if (*cmdstr == '#') {
|
|
|
|
return(strpbrk(cmdstr, "\n\r"));
|
|
|
|
}
|
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* parse the command
|
|
|
|
* format is: [addr][,addr]cmd
|
|
|
|
* |----||-----||-|
|
|
|
|
* part1 part2 part3
|
|
|
|
*/
|
|
|
|
|
2003-03-08 05:21:02 +00:00
|
|
|
/* first part (if present) is an address: either a '$', a number or a /regex/ */
|
2003-03-09 15:12:24 +00:00
|
|
|
cmdstr += get_address(&(sed_cmd->delimiter), cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
/* second part (if present) will begin with a comma */
|
2003-03-09 11:06:38 +00:00
|
|
|
if (*cmdstr == ',') {
|
2003-03-09 15:12:24 +00:00
|
|
|
int idx;
|
2003-03-09 11:06:38 +00:00
|
|
|
cmdstr++;
|
2003-03-09 15:12:24 +00:00
|
|
|
idx = get_address(&(sed_cmd->delimiter), cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
|
|
|
|
if (idx == 0) {
|
2003-03-09 08:44:49 +00:00
|
|
|
error_msg_and_die("get_address: no address found in string\n"
|
|
|
|
"\t(you probably didn't check the string you passed me)");
|
|
|
|
}
|
2003-03-09 15:12:24 +00:00
|
|
|
cmdstr += idx;
|
2002-07-11 11:11:56 +00:00
|
|
|
}
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2001-12-21 16:04:12 +00:00
|
|
|
/* skip whitespace before the command */
|
2003-03-09 15:12:24 +00:00
|
|
|
while (isspace(*cmdstr)) {
|
2003-03-09 11:06:38 +00:00
|
|
|
cmdstr++;
|
2003-03-09 15:12:24 +00:00
|
|
|
}
|
2001-12-21 16:04:12 +00:00
|
|
|
|
2002-06-11 23:43:27 +00:00
|
|
|
/* there my be the inversion flag between part2 and part3 */
|
2003-03-09 11:06:38 +00:00
|
|
|
if (*cmdstr == '!') {
|
2002-06-11 23:43:27 +00:00
|
|
|
sed_cmd->invert = 1;
|
2003-03-09 11:06:38 +00:00
|
|
|
cmdstr++;
|
2002-06-11 23:43:27 +00:00
|
|
|
|
2003-03-09 02:39:29 +00:00
|
|
|
#ifdef SED_FEATURE_STRICT_CHECKING
|
|
|
|
/* According to the spec
|
|
|
|
* It is unspecified whether <blank>s can follow a '!' character,
|
|
|
|
* and conforming applications shall not follow a '!' character
|
|
|
|
* with <blank>s.
|
|
|
|
*/
|
2003-03-09 15:12:24 +00:00
|
|
|
if (isblank(cmdstr[idx]) {
|
2003-03-09 02:39:29 +00:00
|
|
|
error_msg_and_die("blank follows '!'");
|
|
|
|
}
|
|
|
|
#else
|
2002-06-11 23:43:27 +00:00
|
|
|
/* skip whitespace before the command */
|
2003-03-09 15:12:24 +00:00
|
|
|
while (isspace(*cmdstr)) {
|
2003-03-09 11:06:38 +00:00
|
|
|
cmdstr++;
|
2003-03-09 15:12:24 +00:00
|
|
|
}
|
2003-03-09 02:39:29 +00:00
|
|
|
#endif
|
2003-03-09 15:12:24 +00:00
|
|
|
|
2002-06-11 23:43:27 +00:00
|
|
|
}
|
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* last part (mandatory) will be a command */
|
2003-03-09 11:06:38 +00:00
|
|
|
if (*cmdstr == '\0')
|
2001-01-31 19:00:21 +00:00
|
|
|
error_msg_and_die("missing command");
|
2003-03-09 11:06:38 +00:00
|
|
|
|
|
|
|
sed_cmd->cmd = *cmdstr;
|
2003-03-09 15:12:24 +00:00
|
|
|
cmdstr++;
|
|
|
|
|
|
|
|
if (sed_cmd->cmd == '{') {
|
|
|
|
do {
|
|
|
|
char *end_ptr = strpbrk(cmdstr, ";}");
|
|
|
|
*end_ptr = '\0';
|
|
|
|
add_cmd(sed_cmd, cmdstr);
|
|
|
|
cmdstr = end_ptr + 1;
|
|
|
|
} while (*cmdstr != '\0');
|
|
|
|
} else {
|
|
|
|
|
|
|
|
cmdstr = parse_cmd_str(sed_cmd, cmdstr);
|
|
|
|
|
|
|
|
/* Add the command to the command array */
|
|
|
|
sed_cmds = xrealloc(sed_cmds, sizeof(sed_cmd_t) * (++ncmds));
|
|
|
|
sed_cmds[ncmds-1] = xmalloc(sizeof(sed_cmd_t));
|
|
|
|
memcpy(sed_cmds[ncmds-1], sed_cmd, sizeof(sed_cmd_t));
|
2003-03-09 02:44:49 +00:00
|
|
|
}
|
2003-03-09 11:06:38 +00:00
|
|
|
return(cmdstr);
|
1999-12-29 22:19:46 +00:00
|
|
|
}
|
|
|
|
|
2003-03-09 15:12:24 +00:00
|
|
|
static void add_cmd_str(char *cmdstr)
|
1999-12-29 22:19:46 +00:00
|
|
|
{
|
2000-07-14 19:06:30 +00:00
|
|
|
do {
|
2003-03-09 15:12:24 +00:00
|
|
|
sed_cmd_t *sed_cmd;
|
|
|
|
sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
|
|
|
|
cmdstr = add_cmd(sed_cmd, cmdstr);
|
|
|
|
} while (cmdstr && strlen(cmdstr));
|
1999-11-09 01:47:36 +00:00
|
|
|
}
|
1999-10-28 16:06:25 +00:00
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
static void load_cmd_file(char *filename)
|
1999-10-28 16:06:25 +00:00
|
|
|
{
|
2000-07-10 22:55:51 +00:00
|
|
|
FILE *cmdfile;
|
|
|
|
char *line;
|
2000-07-14 00:00:15 +00:00
|
|
|
char *nextline;
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2000-09-27 02:43:35 +00:00
|
|
|
cmdfile = xfopen(filename, "r");
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
while ((line = get_line_from_file(cmdfile)) != NULL) {
|
2000-07-14 00:00:15 +00:00
|
|
|
/* if a line ends with '\' it needs the next line appended to it */
|
|
|
|
while (line[strlen(line)-2] == '\\' &&
|
|
|
|
(nextline = get_line_from_file(cmdfile)) != NULL) {
|
2000-09-13 02:46:14 +00:00
|
|
|
line = xrealloc(line, strlen(line) + strlen(nextline) + 1);
|
2000-07-14 00:00:15 +00:00
|
|
|
strcat(line, nextline);
|
2000-07-14 23:24:00 +00:00
|
|
|
free(nextline);
|
2000-07-14 00:00:15 +00:00
|
|
|
}
|
2000-07-14 23:24:00 +00:00
|
|
|
/* eat trailing newline (if any) --if I don't do this, edit commands
|
|
|
|
* (aic) will print an extra newline */
|
2001-02-01 16:49:30 +00:00
|
|
|
chomp(line);
|
2000-07-10 22:55:51 +00:00
|
|
|
add_cmd_str(line);
|
|
|
|
free(line);
|
2000-02-08 19:58:47 +00:00
|
|
|
}
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
1999-12-29 22:19:46 +00:00
|
|
|
|
2001-11-10 10:49:42 +00:00
|
|
|
struct pipeline {
|
|
|
|
char *buf;
|
|
|
|
int idx;
|
|
|
|
int len;
|
|
|
|
};
|
|
|
|
|
2001-08-22 05:58:16 +00:00
|
|
|
#define PIPE_MAGIC 0x7f
|
|
|
|
#define PIPE_GROW 64
|
2001-11-10 10:49:42 +00:00
|
|
|
|
|
|
|
void pipe_putc(struct pipeline *const pipeline, char c)
|
|
|
|
{
|
|
|
|
if (pipeline->buf[pipeline->idx] == PIPE_MAGIC) {
|
|
|
|
pipeline->buf =
|
|
|
|
xrealloc(pipeline->buf, pipeline->len + PIPE_GROW);
|
|
|
|
memset(pipeline->buf + pipeline->len, 0, PIPE_GROW);
|
|
|
|
pipeline->len += PIPE_GROW;
|
|
|
|
pipeline->buf[pipeline->len - 1] = PIPE_MAGIC;
|
|
|
|
}
|
|
|
|
pipeline->buf[pipeline->idx++] = (c);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define pipeputc(c) pipe_putc(pipeline, c)
|
|
|
|
|
|
|
|
#if 0
|
2001-08-22 05:58:16 +00:00
|
|
|
{ if (pipeline[pipeline_idx] == PIPE_MAGIC) { \
|
|
|
|
pipeline = xrealloc(pipeline, pipeline_len+PIPE_GROW); \
|
|
|
|
memset(pipeline+pipeline_len, 0, PIPE_GROW); \
|
|
|
|
pipeline_len += PIPE_GROW; \
|
|
|
|
pipeline[pipeline_len-1] = PIPE_MAGIC; } \
|
|
|
|
pipeline[pipeline_idx++] = (c); }
|
2001-11-10 10:49:42 +00:00
|
|
|
#endif
|
2001-08-22 05:58:16 +00:00
|
|
|
|
|
|
|
static void print_subst_w_backrefs(const char *line, const char *replace,
|
2001-11-10 10:49:42 +00:00
|
|
|
regmatch_t *regmatch, struct pipeline *const pipeline, int matches)
|
2000-07-17 20:06:42 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* go through the replacement string */
|
|
|
|
for (i = 0; replace[i]; i++) {
|
|
|
|
/* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
|
|
|
|
if (replace[i] == '\\' && isdigit(replace[i+1])) {
|
|
|
|
int j;
|
|
|
|
char tmpstr[2];
|
|
|
|
int backref;
|
|
|
|
++i; /* i now indexes the backref number, instead of the leading slash */
|
|
|
|
tmpstr[0] = replace[i];
|
|
|
|
tmpstr[1] = 0;
|
|
|
|
backref = atoi(tmpstr);
|
|
|
|
/* print out the text held in regmatch[backref] */
|
2001-08-20 21:21:06 +00:00
|
|
|
if (backref <= matches && regmatch[backref].rm_so != -1)
|
|
|
|
for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
|
2001-08-22 05:58:16 +00:00
|
|
|
pipeputc(line[j]);
|
2000-07-17 20:06:42 +00:00
|
|
|
}
|
|
|
|
|
2000-07-25 20:48:44 +00:00
|
|
|
/* if we find a backslash escaped character, print the character */
|
|
|
|
else if (replace[i] == '\\') {
|
|
|
|
++i;
|
2001-08-22 05:58:16 +00:00
|
|
|
pipeputc(replace[i]);
|
2000-07-25 20:48:44 +00:00
|
|
|
}
|
|
|
|
|
2000-07-17 20:06:42 +00:00
|
|
|
/* if we find an unescaped '&' print out the whole matched text.
|
|
|
|
* fortunately, regmatch[0] contains the indicies to the whole matched
|
|
|
|
* expression (kinda seems like it was designed for just such a
|
|
|
|
* purpose...) */
|
|
|
|
else if (replace[i] == '&' && replace[i-1] != '\\') {
|
|
|
|
int j;
|
|
|
|
for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
|
2001-08-22 05:58:16 +00:00
|
|
|
pipeputc(line[j]);
|
2000-07-17 20:06:42 +00:00
|
|
|
}
|
|
|
|
/* nothing special, just print this char of the replacement string to stdout */
|
|
|
|
else
|
2001-08-22 05:58:16 +00:00
|
|
|
pipeputc(replace[i]);
|
2000-07-17 20:06:42 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2003-03-09 10:23:57 +00:00
|
|
|
static int do_subst_command(const sed_cmd_t *sed_cmd, char **line)
|
2000-07-13 20:01:58 +00:00
|
|
|
{
|
2001-08-22 05:58:16 +00:00
|
|
|
char *hackline = *line;
|
2001-11-10 10:49:42 +00:00
|
|
|
struct pipeline thepipe = { NULL, 0 , 0};
|
|
|
|
struct pipeline *const pipeline = &thepipe;
|
2000-07-13 20:01:58 +00:00
|
|
|
int altered = 0;
|
2000-11-03 19:47:00 +00:00
|
|
|
regmatch_t *regmatch = NULL;
|
2000-07-13 20:01:58 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* we only proceed if the substitution 'search' expression matches */
|
2001-08-22 05:58:16 +00:00
|
|
|
if (regexec(sed_cmd->sub_match, hackline, 0, NULL, 0) == REG_NOMATCH)
|
2000-11-03 19:47:00 +00:00
|
|
|
return 0;
|
2000-07-17 20:06:42 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* whaddaya know, it matched. get the number of back references */
|
|
|
|
regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
|
2000-07-17 20:06:42 +00:00
|
|
|
|
2001-08-22 05:58:16 +00:00
|
|
|
/* allocate more PIPE_GROW bytes
|
|
|
|
if replaced string is larger than original */
|
2001-11-10 10:49:42 +00:00
|
|
|
thepipe.len = strlen(hackline)+PIPE_GROW;
|
|
|
|
thepipe.buf = xcalloc(1, thepipe.len);
|
2001-08-22 05:58:16 +00:00
|
|
|
/* buffer magic */
|
2001-11-10 10:49:42 +00:00
|
|
|
thepipe.buf[thepipe.len-1] = PIPE_MAGIC;
|
2001-08-22 05:58:16 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* and now, as long as we've got a line to try matching and if we can match
|
|
|
|
* the search string, we make substitutions */
|
2001-10-23 21:12:07 +00:00
|
|
|
while ((*hackline || !altered) && (regexec(sed_cmd->sub_match, hackline,
|
|
|
|
sed_cmd->num_backrefs+1, regmatch, 0) != REG_NOMATCH) ) {
|
2000-11-03 19:47:00 +00:00
|
|
|
int i;
|
2000-07-13 20:01:58 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* print everything before the match */
|
|
|
|
for (i = 0; i < regmatch[0].rm_so; i++)
|
2001-08-22 05:58:16 +00:00
|
|
|
pipeputc(hackline[i]);
|
2000-07-13 20:01:58 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* then print the substitution string */
|
2001-08-22 05:58:16 +00:00
|
|
|
print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch,
|
2001-11-10 10:49:42 +00:00
|
|
|
pipeline, sed_cmd->num_backrefs);
|
2000-11-03 19:47:00 +00:00
|
|
|
|
|
|
|
/* advance past the match */
|
|
|
|
hackline += regmatch[0].rm_eo;
|
|
|
|
/* flag that something has changed */
|
|
|
|
altered++;
|
2000-07-17 20:06:42 +00:00
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* if we're not doing this globally, get out now */
|
|
|
|
if (!sed_cmd->sub_g)
|
|
|
|
break;
|
2000-07-13 20:01:58 +00:00
|
|
|
}
|
|
|
|
|
2001-08-22 05:58:16 +00:00
|
|
|
for (; *hackline; hackline++) pipeputc(*hackline);
|
2001-11-10 10:49:42 +00:00
|
|
|
if (thepipe.buf[thepipe.idx] == PIPE_MAGIC) thepipe.buf[thepipe.idx] = 0;
|
2000-11-03 19:47:00 +00:00
|
|
|
|
|
|
|
/* cleanup */
|
|
|
|
free(regmatch);
|
|
|
|
|
2001-08-22 05:58:16 +00:00
|
|
|
free(*line);
|
2001-11-10 10:49:42 +00:00
|
|
|
*line = thepipe.buf;
|
2000-07-13 20:01:58 +00:00
|
|
|
return altered;
|
|
|
|
}
|
2000-03-06 19:20:35 +00:00
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
static void process_file(FILE *file)
|
|
|
|
{
|
2003-03-08 05:21:02 +00:00
|
|
|
char *line;
|
2000-07-10 22:55:51 +00:00
|
|
|
static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
|
|
|
|
unsigned int still_in_range = 0;
|
2001-06-11 23:50:06 +00:00
|
|
|
int altered;
|
2000-07-10 22:55:51 +00:00
|
|
|
int i;
|
|
|
|
|
2003-03-08 05:21:02 +00:00
|
|
|
line = get_line_from_file(file);
|
|
|
|
if (line == NULL) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* go through every line in the file */
|
2003-03-08 05:21:02 +00:00
|
|
|
do {
|
|
|
|
char *next_line;
|
|
|
|
|
|
|
|
/* Read one line in advance so we can act on the last line, the '$' address */
|
|
|
|
next_line = get_line_from_file(file);
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2001-05-14 19:44:44 +00:00
|
|
|
chomp(line);
|
2000-07-10 22:55:51 +00:00
|
|
|
linenum++;
|
2001-06-11 23:50:06 +00:00
|
|
|
altered = 0;
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
/* for every line, go through all the commands */
|
|
|
|
for (i = 0; i < ncmds; i++) {
|
2003-03-09 11:06:38 +00:00
|
|
|
sed_cmd_t *sed_cmd = sed_cmds[i];
|
2002-06-11 23:43:27 +00:00
|
|
|
int deleted = 0;
|
2000-03-06 19:20:35 +00:00
|
|
|
|
2001-06-11 23:50:06 +00:00
|
|
|
/*
|
|
|
|
* entry point into sedding...
|
|
|
|
*/
|
2002-06-11 23:43:27 +00:00
|
|
|
int matched = (
|
2001-06-21 13:57:51 +00:00
|
|
|
/* no range necessary */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->beg_line == 0 && sed_cmd->end_line == 0 &&
|
|
|
|
sed_cmd->beg_match == NULL &&
|
|
|
|
sed_cmd->end_match == NULL) ||
|
2001-06-11 23:50:06 +00:00
|
|
|
/* this line number is the first address we're looking for */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->beg_line && (sed_cmd->beg_line == linenum)) ||
|
2001-06-11 23:50:06 +00:00
|
|
|
/* this line matches our first address regex */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->beg_match && (regexec(sed_cmd->beg_match, line, 0, NULL, 0) == 0)) ||
|
2001-06-11 23:50:06 +00:00
|
|
|
/* we are currently within the beginning & ending address range */
|
2003-03-08 05:21:02 +00:00
|
|
|
still_in_range || ((sed_cmd->beg_line == -1) && (next_line == NULL))
|
2002-06-11 23:43:27 +00:00
|
|
|
);
|
|
|
|
|
|
|
|
if (sed_cmd->invert ^ matched) {
|
2001-06-11 23:50:06 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* actual sedding
|
|
|
|
*/
|
2001-11-10 10:49:42 +00:00
|
|
|
switch (sed_cmd->cmd) {
|
2002-12-23 10:16:12 +00:00
|
|
|
case '=':
|
|
|
|
printf("%d\n", linenum);
|
|
|
|
break;
|
2001-06-11 23:50:06 +00:00
|
|
|
case 'p':
|
|
|
|
puts(line);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'd':
|
|
|
|
altered++;
|
2002-04-01 16:17:37 +00:00
|
|
|
deleted = 1;
|
2001-06-11 23:50:06 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case 's':
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Some special cases for 's' printing to make it compliant with
|
|
|
|
* GNU sed printing behavior (aka "The -n | s///p Matrix"):
|
|
|
|
*
|
|
|
|
* -n ONLY = never print anything regardless of any successful
|
|
|
|
* substitution
|
|
|
|
*
|
|
|
|
* s///p ONLY = always print successful substitutions, even if
|
|
|
|
* the line is going to be printed anyway (line will be printed
|
|
|
|
* twice).
|
|
|
|
*
|
|
|
|
* -n AND s///p = print ONLY a successful substitution ONE TIME;
|
|
|
|
* no other lines are printed - this is the reason why the 'p'
|
|
|
|
* flag exists in the first place.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* if the user specified that they didn't want anything printed (i.e., a -n
|
|
|
|
* flag and no 'p' flag after the s///), then there's really no point doing
|
|
|
|
* anything here. */
|
2001-11-10 10:49:42 +00:00
|
|
|
if (be_quiet && !sed_cmd->sub_p)
|
2001-06-11 23:50:06 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
/* we print the line once, unless we were told to be quiet */
|
|
|
|
if (!be_quiet)
|
2001-11-10 10:49:42 +00:00
|
|
|
altered |= do_subst_command(sed_cmd, &line);
|
2001-06-11 23:50:06 +00:00
|
|
|
|
|
|
|
/* we also print the line if we were given the 'p' flag
|
|
|
|
* (this is quite possibly the second printing) */
|
2001-11-10 10:49:42 +00:00
|
|
|
if (sed_cmd->sub_p)
|
|
|
|
altered |= do_subst_command(sed_cmd, &line);
|
2003-03-09 11:06:38 +00:00
|
|
|
if (altered && (i+1 >= ncmds || sed_cmds[i+1]->cmd != 's'))
|
2001-08-22 05:58:16 +00:00
|
|
|
puts(line);
|
2001-06-11 23:50:06 +00:00
|
|
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'a':
|
|
|
|
puts(line);
|
2001-11-10 10:49:42 +00:00
|
|
|
fputs(sed_cmd->editline, stdout);
|
2001-06-11 23:50:06 +00:00
|
|
|
altered++;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'i':
|
2001-11-10 10:49:42 +00:00
|
|
|
fputs(sed_cmd->editline, stdout);
|
2001-06-11 23:50:06 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case 'c':
|
|
|
|
/* single-address case */
|
2001-11-10 10:49:42 +00:00
|
|
|
if ((sed_cmd->end_match == NULL && sed_cmd->end_line == 0)
|
2001-06-11 23:50:06 +00:00
|
|
|
/* multi-address case */
|
2001-11-10 10:49:42 +00:00
|
|
|
/* - matching text */
|
|
|
|
|| (sed_cmd->end_match && (regexec(sed_cmd->end_match, line, 0, NULL, 0) == 0))
|
|
|
|
/* - matching line numbers */
|
|
|
|
|| (sed_cmd->end_line > 0 && sed_cmd->end_line == linenum))
|
|
|
|
{
|
|
|
|
fputs(sed_cmd->editline, stdout);
|
2001-06-11 23:50:06 +00:00
|
|
|
}
|
|
|
|
altered++;
|
|
|
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'r': {
|
|
|
|
FILE *outfile;
|
|
|
|
puts(line);
|
2001-11-10 10:49:42 +00:00
|
|
|
outfile = fopen(sed_cmd->filename, "r");
|
2001-06-11 23:50:06 +00:00
|
|
|
if (outfile)
|
|
|
|
print_file(outfile);
|
|
|
|
/* else if we couldn't open the output file,
|
|
|
|
* no biggie, just don't print anything */
|
|
|
|
altered++;
|
2002-06-11 23:43:27 +00:00
|
|
|
}
|
2001-06-11 23:50:06 +00:00
|
|
|
break;
|
|
|
|
}
|
2002-06-11 23:43:27 +00:00
|
|
|
}
|
2000-03-06 19:20:35 +00:00
|
|
|
|
2002-06-11 23:43:27 +00:00
|
|
|
/*
|
|
|
|
* exit point from sedding...
|
|
|
|
*/
|
|
|
|
if (matched) {
|
2001-06-11 23:50:06 +00:00
|
|
|
if (
|
|
|
|
/* this is a single-address command or... */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->end_line == 0 && sed_cmd->end_match == NULL) || (
|
2001-06-11 23:50:06 +00:00
|
|
|
/* we were in the middle of our address range (this
|
|
|
|
* isn't the first time through) and.. */
|
|
|
|
(still_in_range == 1) && (
|
|
|
|
/* this line number is the last address we're looking for or... */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->end_line && (sed_cmd->end_line == linenum)) ||
|
2001-06-11 23:50:06 +00:00
|
|
|
/* this line matches our last address regex */
|
2001-11-10 10:49:42 +00:00
|
|
|
(sed_cmd->end_match && (regexec(sed_cmd->end_match, line, 0, NULL, 0) == 0))
|
2001-06-11 23:50:06 +00:00
|
|
|
)
|
|
|
|
)
|
|
|
|
) {
|
|
|
|
/* we're out of our address range */
|
|
|
|
still_in_range = 0;
|
|
|
|
}
|
2000-02-08 19:58:47 +00:00
|
|
|
|
2001-06-11 23:50:06 +00:00
|
|
|
/* didn't hit the exit? then we're still in the middle of an address range */
|
|
|
|
else {
|
|
|
|
still_in_range = 1;
|
|
|
|
}
|
2000-02-08 19:58:47 +00:00
|
|
|
}
|
2002-06-11 23:43:27 +00:00
|
|
|
|
|
|
|
if (deleted)
|
|
|
|
break;
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
|
2000-11-03 19:47:00 +00:00
|
|
|
/* we will print the line unless we were told to be quiet or if the
|
|
|
|
* line was altered (via a 'd'elete or 's'ubstitution), in which case
|
|
|
|
* the altered line was already printed */
|
2001-06-11 23:50:06 +00:00
|
|
|
if (!be_quiet && !altered)
|
2001-05-14 19:53:08 +00:00
|
|
|
puts(line);
|
2000-07-10 22:55:51 +00:00
|
|
|
|
|
|
|
free(line);
|
2003-03-08 05:21:02 +00:00
|
|
|
line = next_line;
|
|
|
|
} while (line);
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
extern int sed_main(int argc, char **argv)
|
|
|
|
{
|
2001-11-12 16:44:55 +00:00
|
|
|
int opt, status = EXIT_SUCCESS;
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2001-10-24 05:00:29 +00:00
|
|
|
#ifdef CONFIG_FEATURE_CLEAN_UP
|
2000-07-11 21:38:47 +00:00
|
|
|
/* destroy command strings on exit */
|
2000-12-22 01:48:07 +00:00
|
|
|
if (atexit(destroy_cmd_strs) == -1)
|
|
|
|
perror_msg_and_die("atexit");
|
2000-07-12 23:35:21 +00:00
|
|
|
#endif
|
2000-07-11 21:38:47 +00:00
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* do normal option parsing */
|
2001-02-17 16:52:35 +00:00
|
|
|
while ((opt = getopt(argc, argv, "ne:f:")) > 0) {
|
2000-07-10 22:55:51 +00:00
|
|
|
switch (opt) {
|
|
|
|
case 'n':
|
|
|
|
be_quiet++;
|
1999-12-29 22:19:46 +00:00
|
|
|
break;
|
2000-07-10 22:55:51 +00:00
|
|
|
case 'e':
|
|
|
|
add_cmd_str(optarg);
|
|
|
|
break;
|
|
|
|
case 'f':
|
|
|
|
load_cmd_file(optarg);
|
|
|
|
break;
|
2001-02-17 16:52:35 +00:00
|
|
|
default:
|
|
|
|
show_usage();
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* 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 (ncmds == 0) {
|
|
|
|
if (argv[optind] == NULL)
|
2001-02-14 21:23:06 +00:00
|
|
|
show_usage();
|
2000-07-10 22:55:51 +00:00
|
|
|
else {
|
|
|
|
add_cmd_str(argv[optind]);
|
|
|
|
optind++;
|
|
|
|
}
|
|
|
|
}
|
1999-12-29 22:19:46 +00:00
|
|
|
|
2000-07-10 22:55:51 +00:00
|
|
|
/* argv[(optind)..(argc-1)] should be names of file to process. If no
|
|
|
|
* files were specified or '-' was specified, take input from stdin.
|
|
|
|
* Otherwise, we process all the files specified. */
|
|
|
|
if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
|
|
|
|
process_file(stdin);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
int i;
|
|
|
|
FILE *file;
|
|
|
|
for (i = optind; i < argc; i++) {
|
2001-11-17 07:23:46 +00:00
|
|
|
file = wfopen(argv[i], "r");
|
|
|
|
if (file) {
|
2000-07-10 22:55:51 +00:00
|
|
|
process_file(file);
|
|
|
|
fclose(file);
|
2001-11-12 16:44:55 +00:00
|
|
|
} else
|
|
|
|
status = EXIT_FAILURE;
|
2000-02-08 19:58:47 +00:00
|
|
|
}
|
1999-11-09 01:47:36 +00:00
|
|
|
}
|
2000-07-10 22:55:51 +00:00
|
|
|
|
2001-11-12 16:44:55 +00:00
|
|
|
return status;
|
2000-07-10 22:55:51 +00:00
|
|
|
}
|