blob: 168a69c5b8c8f9ca8a4b91291e7379f73e465d75 [file] [log] [blame]
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001/* vi: set sw=4 ts=4: */
Eric Andersen6b6b3f61999-10-28 16:06:25 +00002/*
Mark Whitley6315ce62000-07-10 22:55:51 +00003 * sed.c - very minimalist version of sed
Eric Andersen6b6b3f61999-10-28 16:06:25 +00004 *
Eric Andersenbdfd0d72001-10-24 05:00:29 +00005 * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
6 * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
Matt Kraai5ed78ad2002-01-03 21:12:34 +00007 * Copyright (C) 2002 Matt Kraai
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +00008 * Copyright (C) 2003 by Glenn McGrath <bug1@optushome.com.au>
Rob Landley25d82392004-04-01 09:23:30 +00009 * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
Erik Andersen1266a131999-12-29 22:19:46 +000010 *
Eric Andersen6b6b3f61999-10-28 16:06:25 +000011 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License as published by
13 * the Free Software Foundation; either version 2 of the License, or
14 * (at your option) any later version.
15 *
16 * This program is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * General Public License for more details.
20 *
21 * You should have received a copy of the GNU General Public License
22 * along with this program; if not, write to the Free Software
23 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
24 *
25 */
26
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000027/* Code overview.
28
29 Files are laid out to avoid unnecessary function declarations. So for
30 example, every function add_cmd calls occurs before add_cmd in this file.
31
32 add_cmd() is called on each line of sed command text (from a file or from
33 the command line). It calls get_address() and parse_cmd_args(). The
34 resulting sed_cmd_t structures are appended to a linked list
35 (sed_cmd_head/sed_cmd_tail).
36
37 process_file() does actual sedding, reading data lines from an input FILE *
38 (which could be stdin) and applying the sed command list (sed_cmd_head) to
39 each of the resulting lines.
40
41 sed_main() is where external code calls into this, with a command line.
42*/
43
44
Mark Whitley6315ce62000-07-10 22:55:51 +000045/*
46 Supported features and commands in this version of sed:
47
48 - comments ('#')
Mark Whitley94074a92000-07-14 00:00:15 +000049 - address matching: num|/matchstr/[,num|/matchstr/|$]command
50 - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
51 - edit commands: (a)ppend, (i)nsert, (c)hange
Mark Whitley1f3b9f22001-05-11 22:27:13 +000052 - file commands: (r)ead
Mark Whitley97562bd2000-07-17 20:06:42 +000053 - backreferences in substitution expressions (\1, \2...\9)
Glenn L McGrathf50ce312003-03-09 15:12:24 +000054 - grouped commands: {cmd1;cmd2}
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000055 - transliteration (y/source-chars/dest-chars/)
56 - pattern space hold space storing / swapping (g, h, x)
57 - labels / branching (: label, b, t)
Glenn L McGrathf50ce312003-03-09 15:12:24 +000058
Mark Whitley6315ce62000-07-10 22:55:51 +000059 (Note: Specifying an address (range) to match is *optional*; commands
60 default to the whole pattern space if no specific address match was
61 requested.)
62
63 Unsupported features:
64
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000065 - GNU extensions
Glenn L McGrathf36635c2003-09-13 15:12:22 +000066 - and more.
Glenn L McGrathd5eadea2003-03-09 02:39:29 +000067
Glenn L McGrath2570b432003-09-16 05:25:43 +000068 Todo:
69
70 - Create a wrapper around regex to make libc's regex conform with sed
71 - Fix bugs
72
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000073
Glenn L McGrathd5eadea2003-03-09 02:39:29 +000074 Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
Mark Whitley6315ce62000-07-10 22:55:51 +000075*/
76
Eric Andersen6b6b3f61999-10-28 16:06:25 +000077#include <stdio.h>
Glenn L McGrath8d6395d2003-04-08 11:56:11 +000078#include <unistd.h> /* for getopt() */
Mark Whitley6315ce62000-07-10 22:55:51 +000079#include <regex.h>
Glenn L McGrath8d6395d2003-04-08 11:56:11 +000080#include <string.h> /* for strdup() */
Eric Andersen6b6b3f61999-10-28 16:06:25 +000081#include <errno.h>
Glenn L McGrath8d6395d2003-04-08 11:56:11 +000082#include <ctype.h> /* for isspace() */
Eric Andersened3ef502001-01-27 08:24:39 +000083#include <stdlib.h>
Eric Andersen3570a342000-09-25 21:45:58 +000084#include "busybox.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000085
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +000086typedef struct sed_cmd_s {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000087 /* Ordered by alignment requirements: currently 36 bytes on x86 */
Mark Whitley2dc192f2000-11-03 19:47:00 +000088
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000089 /* address storage */
90 regex_t *beg_match; /* sed -e '/match/cmd' */
91 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
92 regex_t *sub_match; /* For 's/sub_match/string/' */
93 int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
94 int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
Mark Whitley6315ce62000-07-10 22:55:51 +000095
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000096 FILE *file; /* File (sr) command writes to, -1 for none. */
97 char *string; /* Data string for (saicytb) commands. */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +000098
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000099 unsigned short which_match; /* (s) Which match to replace (0 for all) */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000100
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000101 /* Bitfields (gcc won't group them if we don't) */
102 unsigned int invert:1; /* the '!' after the address */
103 unsigned int in_match:1; /* Next line also included in match? */
104 unsigned int no_newline:1; /* Last line written by (sr) had no '\n' */
105 unsigned int sub_p:1; /* (s) print option */
Glenn L McGrathc18ce372003-09-13 06:57:39 +0000106
Mark Whitley2dc192f2000-11-03 19:47:00 +0000107
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000108 /* GENERAL FIELDS */
109 char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
110 struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +0000111} sed_cmd_t;
Mark Whitley6315ce62000-07-10 22:55:51 +0000112
113/* globals */
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000114/* options */
Rob Landley53302f82004-02-18 09:54:15 +0000115static int be_quiet = 0, in_place=0;
116FILE *nonstdout;
117char *outname;
118
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000119
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000120static const char bad_format_in_subst[] =
121 "bad format in substitution expression";
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000122const char *const semicolon_whitespace = "; \n\r\t\v";
123
124regmatch_t regmatch[10];
125static regex_t *previous_regex_ptr = NULL;
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000126
Glenn L McGrathc949bfa2003-03-28 03:53:31 +0000127/* linked list of sed commands */
128static sed_cmd_t sed_cmd_head;
129static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000130
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000131/* Linked list of append lines */
132struct append_list {
133 char *string;
134 struct append_list *next;
135};
136struct append_list *append_head=NULL, *append_tail=NULL;
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000137
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000138#ifdef CONFIG_FEATURE_CLEAN_UP
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000139static void free_and_close_stuff(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000140{
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000141 sed_cmd_t *sed_cmd = sed_cmd_head.next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000142
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000143 while(append_head) {
144 append_tail=append_head->next;
145 free(append_head->string);
146 free(append_head);
147 append_head=append_tail;
148 }
149
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000150 while (sed_cmd) {
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000151 sed_cmd_t *sed_cmd_next = sed_cmd->next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000152
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000153 if(sed_cmd->file)
154 bb_xprint_and_close_file(sed_cmd->file);
155
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000156 if (sed_cmd->beg_match) {
157 regfree(sed_cmd->beg_match);
158 free(sed_cmd->beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000159 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000160 if (sed_cmd->end_match) {
161 regfree(sed_cmd->end_match);
162 free(sed_cmd->end_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000163 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000164 if (sed_cmd->sub_match) {
165 regfree(sed_cmd->sub_match);
166 free(sed_cmd->sub_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000167 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000168 free(sed_cmd->string);
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000169 free(sed_cmd);
170 sed_cmd = sed_cmd_next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000171 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000172}
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000173#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000174
Rob Landley53302f82004-02-18 09:54:15 +0000175/* If something bad happens during -i operation, delete temp file */
176
177static void cleanup_outname(void)
178{
179 if(outname) unlink(outname);
180}
181
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000182/* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
183
184static void parse_escapes(char *dest, const char *string, int len, char from, char to)
185{
186 int i=0;
187
188 while(i<len) {
189 if(string[i] == '\\') {
Glenn L McGrath738fb332003-10-01 06:45:11 +0000190 if(!to || string[i+1] == from) {
191 *(dest++) = to ? to : string[i+1];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000192 i+=2;
193 continue;
194 } else *(dest++)=string[i++];
195 }
196 *(dest++) = string[i++];
197 }
198 *dest=0;
199}
200
201static char *copy_parsing_slashn(const char *string, int len)
202{
203 char *dest=xmalloc(len+1);
204
205 parse_escapes(dest,string,len,'n','\n');
206 return dest;
207}
208
209
Mark Whitley6315ce62000-07-10 22:55:51 +0000210/*
Eric Andersen28b3c532001-01-02 11:01:31 +0000211 * index_of_next_unescaped_regexp_delim - walks left to right through a string
212 * beginning at a specified index and returns the index of the next regular
Eric Andersenc7bda1c2004-03-15 08:29:22 +0000213 * expression delimiter (typically a forward * slash ('/')) not preceeded by
Eric Andersen28b3c532001-01-02 11:01:31 +0000214 * a backslash ('\').
Mark Whitley6315ce62000-07-10 22:55:51 +0000215 */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000216static int index_of_next_unescaped_regexp_delim(const char delimiter,
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000217 const char *str)
Mark Whitley6315ce62000-07-10 22:55:51 +0000218{
Matt Kraaia0065d52001-08-24 14:45:50 +0000219 int bracket = -1;
220 int escaped = 0;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000221 int idx = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000222 char ch;
Matt Kraaia0065d52001-08-24 14:45:50 +0000223
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000224 for (; (ch = str[idx]); idx++) {
Matt Kraaia0065d52001-08-24 14:45:50 +0000225 if (bracket != -1) {
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000226 if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000227 && str[idx - 1] == '^')))
Matt Kraaia0065d52001-08-24 14:45:50 +0000228 bracket = -1;
229 } else if (escaped)
230 escaped = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000231 else if (ch == '\\')
Matt Kraaia0065d52001-08-24 14:45:50 +0000232 escaped = 1;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000233 else if (ch == '[')
Matt Kraaia0065d52001-08-24 14:45:50 +0000234 bracket = idx;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000235 else if (ch == delimiter)
Mark Whitley97562bd2000-07-17 20:06:42 +0000236 return idx;
237 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000238
Mark Whitley97562bd2000-07-17 20:06:42 +0000239 /* if we make it to here, we've hit the end of the string */
240 return -1;
Mark Whitley6315ce62000-07-10 22:55:51 +0000241}
242
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000243/*
244 * Returns the index of the third delimiter
245 */
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000246static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
247{
248 const char *cmdstr_ptr = cmdstr;
249 char delimiter;
250 int idx = 0;
251
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000252 /* verify that the 's' or 'y' is followed by something. That something
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000253 * (typically a 'slash') is now our regexp delimiter... */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000254 if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
255 delimiter = *(cmdstr_ptr++);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000256
257 /* save the match string */
258 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
259 if (idx == -1) {
Glenn L McGrath9a52bb62003-03-30 09:38:40 +0000260 bb_error_msg_and_die(bad_format_in_subst);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000261 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000262 *match = copy_parsing_slashn(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000263
264 /* save the replacement string */
265 cmdstr_ptr += idx + 1;
266 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
267 if (idx == -1) {
Glenn L McGrath9a52bb62003-03-30 09:38:40 +0000268 bb_error_msg_and_die(bad_format_in_subst);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000269 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000270 *replace = copy_parsing_slashn(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000271
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000272 return ((cmdstr_ptr - cmdstr) + idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000273}
274
Mark Whitley6315ce62000-07-10 22:55:51 +0000275/*
276 * returns the index in the string just past where the address ends.
277 */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000278static int get_address(char *my_str, int *linenum, regex_t ** regex)
Mark Whitley6315ce62000-07-10 22:55:51 +0000279{
Glenn L McGrathe3e28d32003-09-15 09:22:04 +0000280 char *pos = my_str;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000281
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000282 if (isdigit(*my_str)) {
283 *linenum = strtol(my_str, &pos, 10);
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000284 /* endstr shouldnt ever equal NULL */
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000285 } else if (*my_str == '$') {
Mark Whitley0915c4b2001-06-11 23:50:06 +0000286 *linenum = -1;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000287 pos++;
288 } else if (*my_str == '/' || *my_str == '\\') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000289 int next;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000290 char delimiter;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000291 char *temp;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000292
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000293 if (*my_str == '\\') delimiter = *(++pos);
294 else delimiter = '/';
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000295 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000296 if (next == -1)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000297 bb_error_msg_and_die("unterminated match expression");
Eric Andersenc7bda1c2004-03-15 08:29:22 +0000298
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000299 temp=copy_parsing_slashn(pos,next);
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000300 *regex = (regex_t *) xmalloc(sizeof(regex_t));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000301 xregcomp(*regex, temp, REG_NEWLINE);
302 free(temp);
303 /* Move position to next character after last delimiter */
304 pos+=(next+1);
Mark Whitley6315ce62000-07-10 22:55:51 +0000305 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000306 return pos - my_str;
Mark Whitley6315ce62000-07-10 22:55:51 +0000307}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000308
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000309/* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
310static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
311{
312 int start = 0, idx, hack=0;
313
314 /* Skip whitespace, then grab filename to end of line */
315 while (isspace(filecmdstr[start])) start++;
316 idx=start;
317 while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
318 /* If lines glued together, put backslash back. */
319 if(filecmdstr[idx]=='\n') hack=1;
320 if(idx==start) bb_error_msg_and_die("Empty filename");
321 *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
322 if(hack) *(idx+*retval)='\\';
323
324 return idx;
325}
326
Eric Andersen638da752003-10-09 08:18:36 +0000327static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000328{
Glenn L McGrathe01f9662003-03-18 08:37:57 +0000329 int cflags = 0;
Mark Whitley06f35292000-07-13 19:58:04 +0000330 char *match;
331 int idx = 0;
332
333 /*
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000334 * A substitution command should look something like this:
335 * s/match/replace/ #gIpw
Mark Whitley0e4cec02000-08-21 21:29:20 +0000336 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000337 * mandatory optional
Mark Whitley06f35292000-07-13 19:58:04 +0000338 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000339 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
Mark Whitley06f35292000-07-13 19:58:04 +0000340
Mark Whitley97562bd2000-07-17 20:06:42 +0000341 /* determine the number of back references in the match string */
342 /* Note: we compute this here rather than in the do_subst_command()
343 * function to save processor time, at the expense of a little more memory
344 * (4 bits) per sed_cmd */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000345
Mark Whitley06f35292000-07-13 19:58:04 +0000346 /* process the flags */
Mark Whitley70705d72000-07-14 19:06:30 +0000347
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000348 sed_cmd->which_match=1;
349 while (substr[++idx]) {
350 /* Parse match number */
351 if(isdigit(substr[idx])) {
352 if(match[0]!='^') {
353 /* Match 0 treated as all, multiple matches we take the last one. */
354 char *pos=substr+idx;
355 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
356 idx=pos-substr;
357 }
358 continue;
359 }
Rob Landley40ec4ae2004-01-04 06:42:14 +0000360 /* Skip spaces */
361 if(isspace(substr[idx])) continue;
362
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000363 switch (substr[idx]) {
364 /* Replace all occurrences */
365 case 'g':
366 if (match[0] != '^') sed_cmd->which_match = 0;
367 break;
368 /* Print pattern space */
369 case 'p':
370 sed_cmd->sub_p = 1;
371 break;
372 case 'w':
373 {
374 char *temp;
375 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
Eric Andersenc7bda1c2004-03-15 08:29:22 +0000376
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000377 break;
378 }
379 /* Ignore case (gnu exension) */
380 case 'I':
381 cflags |= REG_ICASE;
382 break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000383 case ';':
384 case '}':
385 goto out;
386 default:
387 bb_error_msg_and_die("bad option in substitution expression");
388 }
389 }
Glenn L McGrath2570b432003-09-16 05:25:43 +0000390out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000391 /* compile the match string into a regex */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000392 if (*match != '\0') {
393 /* If match is empty, we use last regex used at runtime */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000394 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000395 xregcomp(sed_cmd->sub_match, match, cflags);
396 }
Mark Whitley06f35292000-07-13 19:58:04 +0000397 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000398
399 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000400}
401
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000402/*
403 * Process the commands arguments
404 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000405static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000406{
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000407 /* handle (s)ubstitution command */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000408 if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000409 /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
410 else if (strchr("aic", sed_cmd->cmd)) {
411 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000412 bb_error_msg_and_die
413 ("only a beginning address can be specified for edit commands");
Rob Landley25d82392004-04-01 09:23:30 +0000414 for(;;) {
415 if(*cmdstr=='\n' || *cmdstr=='\\') {
416 cmdstr++;
417 break;
418 } else if(isspace(*cmdstr)) cmdstr++;
419 else break;
420 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000421 sed_cmd->string = bb_xstrdup(cmdstr);
Glenn L McGrath738fb332003-10-01 06:45:11 +0000422 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000423 cmdstr += strlen(cmdstr);
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000424 /* handle file cmds: (r)ead */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000425 } else if(strchr("rw", sed_cmd->cmd)) {
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000426 if (sed_cmd->end_line || sed_cmd->end_match)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000427 bb_error_msg_and_die("Command only uses one address");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000428 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
429 if(sed_cmd->cmd=='w')
430 sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000431 /* handle branch commands */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000432 } else if (strchr(":bt", sed_cmd->cmd)) {
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000433 int length;
434
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000435 while(isspace(*cmdstr)) cmdstr++;
Glenn L McGrathf4523562003-09-14 06:01:14 +0000436 length = strcspn(cmdstr, semicolon_whitespace);
437 if (length) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000438 sed_cmd->string = strndup(cmdstr, length);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000439 cmdstr += length;
440 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000441 }
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000442 /* translation command */
443 else if (sed_cmd->cmd == 'y') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000444 char *match, *replace;
445 int i=cmdstr[0];
446
447 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
448 /* \n already parsed, but \delimiter needs unescaping. */
449 parse_escapes(match,match,strlen(match),i,i);
450 parse_escapes(replace,replace,strlen(replace),i,i);
451
452 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
453 for (i = 0; match[i] && replace[i]; i++) {
454 sed_cmd->string[i * 2] = match[i];
455 sed_cmd->string[(i * 2) + 1] = replace[i];
456 }
457 free(match);
458 free(replace);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000459 }
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000460 /* if it wasnt a single-letter command that takes no arguments
461 * then it must be an invalid command.
462 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000463 else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000464 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
Mark Whitley1f3b9f22001-05-11 22:27:13 +0000465 }
Mark Whitley70705d72000-07-14 19:06:30 +0000466
467 /* give back whatever's left over */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000468 return (cmdstr);
Eric Andersen50d63601999-11-09 01:47:36 +0000469}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000470
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000471
472/* Parse address+command sets, skipping comment lines. */
473
474void add_cmd(char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000475{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000476 static char *add_cmd_line=NULL;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000477 sed_cmd_t *sed_cmd;
Glenn L McGrath586d86c2003-10-09 07:22:59 +0000478 int temp;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000479
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000480 /* Append this line to any unfinished line from last time. */
481 if(add_cmd_line) {
482 int lastlen=strlen(add_cmd_line);
Eric Andersen638da752003-10-09 08:18:36 +0000483 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
Erik Andersen1266a131999-12-29 22:19:46 +0000484
Eric Andersen638da752003-10-09 08:18:36 +0000485 memcpy(tmp,add_cmd_line,lastlen);
486 tmp[lastlen]='\n';
487 strcpy(tmp+lastlen+1,cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000488 free(add_cmd_line);
Eric Andersen638da752003-10-09 08:18:36 +0000489 cmdstr=add_cmd_line=tmp;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000490 } else add_cmd_line=NULL;
491
492 /* If this line ends with backslash, request next line. */
Glenn L McGrath586d86c2003-10-09 07:22:59 +0000493 temp=strlen(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000494 if(temp && cmdstr[temp-1]=='\\') {
495 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
496 add_cmd_line[temp-1]=0;
497 return;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000498 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000499
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000500 /* Loop parsing all commands in this line. */
501 while(*cmdstr) {
502 /* Skip leading whitespace and semicolons */
503 cmdstr += strspn(cmdstr, semicolon_whitespace);
504
505 /* If no more commands, exit. */
506 if(!*cmdstr) break;
507
508 /* if this is a comment, jump past it and keep going */
509 if (*cmdstr == '#') {
510 /* "#n" is the same as using -n on the command line */
511 if (cmdstr[1] == 'n') be_quiet++;
512 if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
513 continue;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000514 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000515
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000516 /* parse the command
517 * format is: [addr][,addr][!]cmd
518 * |----||-----||-|
519 * part1 part2 part3
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000520 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000521
522 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
523
524 /* first part (if present) is an address: either a '$', a number or a /regex/ */
525 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
526
527 /* second part (if present) will begin with a comma */
528 if (*cmdstr == ',') {
529 int idx;
530
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000531 cmdstr++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000532 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
533 if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
534 cmdstr += idx;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000535 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000536
537 /* skip whitespace before the command */
538 while (isspace(*cmdstr)) cmdstr++;
539
540 /* Check for inversion flag */
541 if (*cmdstr == '!') {
542 sed_cmd->invert = 1;
543 cmdstr++;
544
545 /* skip whitespace before the command */
546 while (isspace(*cmdstr)) cmdstr++;
547 }
548
549 /* last part (mandatory) will be a command */
550 if (!*cmdstr) bb_error_msg_and_die("missing command");
551 sed_cmd->cmd = *(cmdstr++);
552 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
553
554 /* Add the command to the command array */
555 sed_cmd_tail->next = sed_cmd;
556 sed_cmd_tail = sed_cmd_tail->next;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000557 }
558
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000559 /* If we glued multiple lines together, free the memory. */
560 if(add_cmd_line) {
561 free(add_cmd_line);
562 add_cmd_line=NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +0000563 }
564}
565
Eric Andersenc52a6b02001-11-10 10:49:42 +0000566struct pipeline {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000567 char *buf; /* Space to hold string */
568 int idx; /* Space used */
569 int len; /* Space allocated */
570} pipeline;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000571
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000572#define PIPE_GROW 64
Eric Andersenc52a6b02001-11-10 10:49:42 +0000573
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000574void pipe_putc(char c)
Eric Andersenc52a6b02001-11-10 10:49:42 +0000575{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000576 if(pipeline.idx==pipeline.len) {
577 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
578 pipeline.len+=PIPE_GROW;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000579 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000580 pipeline.buf[pipeline.idx++] = (c);
Eric Andersenc52a6b02001-11-10 10:49:42 +0000581}
582
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000583static void do_subst_w_backrefs(const char *line, const char *replace)
Mark Whitley97562bd2000-07-17 20:06:42 +0000584{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000585 int i,j;
Mark Whitley97562bd2000-07-17 20:06:42 +0000586
587 /* go through the replacement string */
588 for (i = 0; replace[i]; i++) {
589 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
Glenn L McGrath0ad4daa2003-10-01 10:26:23 +0000590 if (replace[i] == '\\' && replace[i+1]>'0' && replace[i+1]<='9') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000591 int backref=replace[++i]-'0';
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000592
Mark Whitley97562bd2000-07-17 20:06:42 +0000593 /* print out the text held in regmatch[backref] */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000594 if(regmatch[backref].rm_so != -1)
595 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
596 pipe_putc(line[j]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000597 }
598
Mark Whitley83e85f62000-07-25 20:48:44 +0000599 /* if we find a backslash escaped character, print the character */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000600 else if (replace[i] == '\\') pipe_putc(replace[++i]);
Mark Whitley83e85f62000-07-25 20:48:44 +0000601
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000602 /* if we find an unescaped '&' print out the whole matched text. */
603 else if (replace[i] == '&')
Mark Whitley97562bd2000-07-17 20:06:42 +0000604 for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000605 pipe_putc(line[j]);
606 /* Otherwise just output the character. */
607 else pipe_putc(replace[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000608 }
609}
610
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000611static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
Mark Whitley4f7fe772000-07-13 20:01:58 +0000612{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000613 char *oldline = *line;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000614 int altered = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000615 int match_count=0;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000616 regex_t *current_regex;
617
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000618 /* Handle empty regex. */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000619 if (sed_cmd->sub_match == NULL) {
620 current_regex = previous_regex_ptr;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000621 if(!current_regex)
622 bb_error_msg_and_die("No previous regexp.");
623 } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000624
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000625 /* Find the first match */
626 if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
Mark Whitley2dc192f2000-11-03 19:47:00 +0000627 return 0;
628
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000629 /* Initialize temporary output buffer. */
630 pipeline.buf=xmalloc(PIPE_GROW);
631 pipeline.len=PIPE_GROW;
632 pipeline.idx=0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000633
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000634 /* Now loop through, substituting for matches */
635 do {
Mark Whitley4f7fe772000-07-13 20:01:58 +0000636 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000637
Eric Andersenc06f5682004-02-04 10:57:46 +0000638 /* Work around bug in glibc regexec, demonstrated by:
639 echo " a.b" | busybox sed 's [^ .]* x g'
640 The match_count check is so not to break
641 echo "hi" | busybox sed 's/^/!/g' */
642 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
643 pipe_putc(*(oldline++));
644 continue;
645 }
646
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000647 match_count++;
648
649 /* If we aren't interested in this match, output old line to
650 end of match and continue */
651 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
652 for(i=0;i<regmatch[0].rm_eo;i++)
653 pipe_putc(oldline[i]);
654 continue;
655 }
656
Mark Whitley2dc192f2000-11-03 19:47:00 +0000657 /* print everything before the match */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000658 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000659
Mark Whitley2dc192f2000-11-03 19:47:00 +0000660 /* then print the substitution string */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000661 do_subst_w_backrefs(oldline, sed_cmd->string);
Mark Whitley97562bd2000-07-17 20:06:42 +0000662
Mark Whitley2dc192f2000-11-03 19:47:00 +0000663 /* advance past the match */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000664 oldline += regmatch[0].rm_eo;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000665 /* flag that something has changed */
666 altered++;
Mark Whitley97562bd2000-07-17 20:06:42 +0000667
Mark Whitley2dc192f2000-11-03 19:47:00 +0000668 /* if we're not doing this globally, get out now */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000669 if (sed_cmd->which_match) break;
670 } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
Mark Whitley2dc192f2000-11-03 19:47:00 +0000671
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000672 /* Copy rest of string into output pipeline */
673
674 while(*oldline) pipe_putc(*(oldline++));
675 pipe_putc(0);
Mark Whitley2dc192f2000-11-03 19:47:00 +0000676
Eric Andersenb76cb682001-08-22 05:58:16 +0000677 free(*line);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000678 *line = pipeline.buf;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000679 return altered;
680}
Mark Whitley6315ce62000-07-10 22:55:51 +0000681
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000682/* Set command pointer to point to this label. (Does not handle null label.) */
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000683static sed_cmd_t *branch_to(const char *label)
684{
685 sed_cmd_t *sed_cmd;
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000686
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000687 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000688 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
Glenn L McGrathf4523562003-09-14 06:01:14 +0000689 return (sed_cmd);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000690 }
691 }
Glenn L McGrathf4523562003-09-14 06:01:14 +0000692 bb_error_msg_and_die("Can't find label for jump to `%s'", label);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000693}
Mark Whitley6315ce62000-07-10 22:55:51 +0000694
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000695/* Append copy of string to append buffer */
696static void append(char *s)
697{
698 struct append_list *temp=calloc(1,sizeof(struct append_list));
699
700 if(append_head)
701 append_tail=(append_tail->next=temp);
702 else append_head=append_tail=temp;
703 temp->string=strdup(s);
704}
705
706static void flush_append(void)
707{
708 /* Output appended lines. */
709 while(append_head) {
Rob Landley53302f82004-02-18 09:54:15 +0000710 fprintf(nonstdout,"%s\n",append_head->string);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000711 append_tail=append_head->next;
712 free(append_head->string);
713 free(append_head);
714 append_head=append_tail;
715 }
716 append_head=append_tail=NULL;
717}
718
719/* Get next line of input, flushing append buffer and noting if we hit EOF
720 * without a newline on the last line.
721 */
722static char *get_next_line(FILE * file, int *no_newline)
723{
724 char *temp;
725 int len;
726
727 flush_append();
728 temp=bb_get_line_from_file(file);
729 if(temp) {
730 len=strlen(temp);
731 if(len && temp[len-1]=='\n') temp[len-1]=0;
732 else *no_newline=1;
733 }
734
735 return temp;
736}
737
738/* Output line of text. missing_newline means the last line output did not
739 end with a newline. no_newline means this line does not end with a
740 newline. */
741
742static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
743{
744 if(missing_newline) fputc('\n',file);
745 fputs(s,file);
746 if(!no_newline) fputc('\n',file);
747
Rob Landley53302f82004-02-18 09:54:15 +0000748 if(ferror(file)) {
749 fprintf(stderr,"Write failed.\n");
750 exit(4); /* It's what gnu sed exits with... */
751 }
752
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000753 return no_newline;
754}
755
Rob Landley53302f82004-02-18 09:54:15 +0000756#define sed_puts(s,n) missing_newline=puts_maybe_newline(s,nonstdout,missing_newline,n)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000757
Rob Landley53302f82004-02-18 09:54:15 +0000758static void process_file(FILE *file)
Mark Whitley6315ce62000-07-10 22:55:51 +0000759{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000760 char *pattern_space, *next_line, *hold_space=NULL;
761 static int linenum = 0, missing_newline=0;
762 int no_newline,next_no_newline=0;
Mark Whitley6315ce62000-07-10 22:55:51 +0000763
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000764 next_line = get_next_line(file,&next_no_newline);
765
Mark Whitley6315ce62000-07-10 22:55:51 +0000766 /* go through every line in the file */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000767 for(;;) {
Glenn L McGrathc949bfa2003-03-28 03:53:31 +0000768 sed_cmd_t *sed_cmd;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000769 int substituted=0;
770
771 /* Advance to next line. Stop if out of lines. */
772 if(!(pattern_space=next_line)) break;
773 no_newline=next_no_newline;
Glenn L McGrath505bd0f2003-03-08 05:21:02 +0000774
775 /* Read one line in advance so we can act on the last line, the '$' address */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000776 next_line = get_next_line(file,&next_no_newline);
Mark Whitley6315ce62000-07-10 22:55:51 +0000777 linenum++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000778restart:
Mark Whitley6315ce62000-07-10 22:55:51 +0000779 /* for every line, go through all the commands */
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000780 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Eric Andersen52a3c272003-12-23 08:53:51 +0000781 int old_matched, matched;
782
783 old_matched = sed_cmd->in_match;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000784
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000785 /* Determine if this command matches this line: */
786
787 /* Are we continuing a previous multi-line match? */
788
789 sed_cmd->in_match = sed_cmd->in_match
790
791 /* Or is no range necessary? */
792 || (!sed_cmd->beg_line && !sed_cmd->end_line
793 && !sed_cmd->beg_match && !sed_cmd->end_match)
794
795 /* Or did we match the start of a numerical range? */
796 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
797
798 /* Or does this line match our begin address regex? */
799 || (sed_cmd->beg_match &&
800 !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
801
802 /* Or did we match last line of input? */
803 || (sed_cmd->beg_line == -1 && next_line == NULL);
804
805 /* Snapshot the value */
806
807 matched = sed_cmd->in_match;
808
809 /* Is this line the end of the current match? */
810
811 if(matched) {
812 sed_cmd->in_match = !(
813 /* has the ending line come, or is this a single address command? */
814 (sed_cmd->end_line ?
815 sed_cmd->end_line==-1 ?
816 !next_line
817 : sed_cmd->end_line<=linenum
818 : !sed_cmd->end_match)
819 /* or does this line matches our last address regex */
Eric Andersen52a3c272003-12-23 08:53:51 +0000820 || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000821 );
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000822 }
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000823
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000824 /* Skip blocks of commands we didn't match. */
825 if (sed_cmd->cmd == '{') {
826 if(sed_cmd->invert ? matched : !matched)
827 while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
828 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
829 continue;
830 }
831
832 /* Okay, so did this line match? */
833 if (sed_cmd->invert ? !matched : matched) {
834 /* Update last used regex in case a blank substitute BRE is found */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000835 if (sed_cmd->beg_match) {
836 previous_regex_ptr = sed_cmd->beg_match;
837 }
Mark Whitley0915c4b2001-06-11 23:50:06 +0000838
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000839 /* actual sedding */
Eric Andersenc52a6b02001-11-10 10:49:42 +0000840 switch (sed_cmd->cmd) {
Mark Whitley0915c4b2001-06-11 23:50:06 +0000841
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000842 /* Print line number */
843 case '=':
Rob Landley53302f82004-02-18 09:54:15 +0000844 fprintf(nonstdout,"%d\n", linenum);
Glenn L McGrath2eed0e22003-09-15 06:28:45 +0000845 break;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000846
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000847 /* Write the current pattern space up to the first newline */
848 case 'P':
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000849 {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000850 char *tmp = strchr(pattern_space, '\n');
Glenn L McGrath0c518322003-03-30 03:41:53 +0000851
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000852 if (tmp) {
853 *tmp = '\0';
854 sed_puts(pattern_space,1);
855 *tmp = '\n';
856 break;
857 }
858 /* Fall Through */
859 }
860
861 /* Write the current pattern space to output */
862 case 'p':
863 sed_puts(pattern_space,no_newline);
864 break;
865 /* Delete up through first newline */
866 case 'D':
867 {
868 char *tmp = strchr(pattern_space,'\n');
869
870 if(tmp) {
871 tmp=bb_xstrdup(tmp+1);
872 free(pattern_space);
873 pattern_space=tmp;
874 goto restart;
Glenn L McGrath0c518322003-03-30 03:41:53 +0000875 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000876 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000877 /* discard this line. */
878 case 'd':
879 goto discard_line;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000880
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000881 /* Substitute with regex */
882 case 's':
883 if(do_subst_command(sed_cmd, &pattern_space)) {
884 substituted|=1;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000885
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000886 /* handle p option */
887 if(sed_cmd->sub_p)
888 sed_puts(pattern_space,no_newline);
889 /* handle w option */
890 if(sed_cmd->file)
891 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
Mark Whitley0915c4b2001-06-11 23:50:06 +0000892
Glenn L McGrathd7fe39b2003-04-09 15:52:32 +0000893 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000894 break;
895
896 /* Append line to linked list to be printed later */
897 case 'a':
898 {
899 append(sed_cmd->string);
900 break;
Glenn L McGrathd87a7ac2003-04-09 15:26:14 +0000901 }
Glenn L McGrathd7fe39b2003-04-09 15:52:32 +0000902
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000903 /* Insert text before this line */
904 case 'i':
905 sed_puts(sed_cmd->string,1);
Glenn L McGrath2570b432003-09-16 05:25:43 +0000906 break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000907
908 /* Cut and paste text (replace) */
909 case 'c':
910 /* Only triggers on last line of a matching range. */
911 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,1);
912 goto discard_line;
913
914 /* Read file, append contents to output */
915 case 'r':
Glenn L McGrathf4523562003-09-14 06:01:14 +0000916 {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000917 FILE *outfile;
918
919 outfile = fopen(sed_cmd->string, "r");
920 if (outfile) {
921 char *line;
922
923 while ((line = bb_get_chomped_line_from_file(outfile))
924 != NULL)
925 append(line);
926 bb_xprint_and_close_file(outfile);
927 }
928
929 break;
930 }
931
932 /* Write pattern space to file. */
933 case 'w':
934 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
935 break;
936
937 /* Read next line from input */
938 case 'n':
939 if (!be_quiet)
940 sed_puts(pattern_space,no_newline);
941 if (next_line) {
942 free(pattern_space);
943 pattern_space = next_line;
944 no_newline=next_no_newline;
945 next_line = get_next_line(file,&next_no_newline);
946 linenum++;
947 break;
948 }
949 /* fall through */
950
951 /* Quit. End of script, end of input. */
952 case 'q':
953 /* Exit the outer while loop */
954 free(next_line);
955 next_line = NULL;
956 goto discard_commands;
957
958 /* Append the next line to the current line */
959 case 'N':
960 {
961 /* If no next line, jump to end of script and exit. */
962 if (next_line == NULL) {
963 /* Jump to end of script and exit */
964 free(next_line);
965 next_line = NULL;
966 goto discard_line;
967 /* append next_line, read new next_line. */
Glenn L McGrathf4523562003-09-14 06:01:14 +0000968 } else {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000969 int len=strlen(pattern_space);
970
971 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
972 pattern_space[len]='\n';
973 strcpy(pattern_space+len+1, next_line);
974 no_newline=next_no_newline;
975 next_line = get_next_line(file,&next_no_newline);
976 linenum++;
Glenn L McGrathf4523562003-09-14 06:01:14 +0000977 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000978 break;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000979 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000980
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000981 /* Test if substition worked, branch if so. */
982 case 't':
983 if (!substituted) break;
984 substituted=0;
985 /* Fall through */
986 /* Branch to label */
987 case 'b':
988 if (!sed_cmd->string) goto discard_commands;
989 else sed_cmd = branch_to(sed_cmd->string);
990 break;
991 /* Transliterate characters */
992 case 'y':
993 {
994 int i;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000995
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000996 for (i = 0; pattern_space[i]; i++) {
997 int j;
998
999 for (j = 0; sed_cmd->string[j]; j += 2) {
1000 if (pattern_space[i] == sed_cmd->string[j]) {
1001 pattern_space[i] = sed_cmd->string[j + 1];
1002 }
Glenn L McGrathf01b46d2003-03-30 08:02:18 +00001003 }
1004 }
Glenn L McGrath294d1132003-09-14 16:28:08 +00001005
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001006 break;
Glenn L McGrath91e19782003-04-26 07:40:07 +00001007 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001008 case 'g': /* Replace pattern space with hold space */
1009 free(pattern_space);
1010 if (hold_space) {
1011 pattern_space = strdup(hold_space);
1012 no_newline=0;
1013 }
1014 break;
1015 case 'G': /* Append newline and hold space to pattern space */
1016 {
1017 int pattern_space_size = 2;
1018 int hold_space_size = 0;
1019
1020 if (pattern_space)
1021 pattern_space_size += strlen(pattern_space);
1022 if (hold_space) hold_space_size = strlen(hold_space);
1023 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1024 if (pattern_space_size == 2) pattern_space[0]=0;
Glenn L McGrath977451e2003-09-15 12:07:48 +00001025 strcat(pattern_space, "\n");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001026 if (hold_space) strcat(pattern_space, hold_space);
1027 no_newline=0;
Glenn L McGrath8417c8c2003-09-14 15:24:18 +00001028
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001029 break;
Glenn L McGrath91e19782003-04-26 07:40:07 +00001030 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001031 case 'h': /* Replace hold space with pattern space */
1032 free(hold_space);
1033 hold_space = strdup(pattern_space);
1034 break;
1035 case 'H': /* Append newline and pattern space to hold space */
1036 {
1037 int hold_space_size = 2;
1038 int pattern_space_size = 0;
Glenn L McGrath8417c8c2003-09-14 15:24:18 +00001039
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001040 if (hold_space) hold_space_size += strlen(hold_space);
1041 if (pattern_space)
1042 pattern_space_size = strlen(pattern_space);
1043 hold_space = xrealloc(hold_space,
1044 hold_space_size + pattern_space_size);
1045
1046 if (hold_space_size == 2) hold_space[0]=0;
Glenn L McGrath8417c8c2003-09-14 15:24:18 +00001047 strcat(hold_space, "\n");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001048 if (pattern_space) strcat(hold_space, pattern_space);
1049
1050 break;
Glenn L McGrath4dc1d252003-09-14 01:25:31 +00001051 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001052 case 'x': /* Exchange hold and pattern space */
1053 {
1054 char *tmp = pattern_space;
1055 pattern_space = hold_space;
1056 no_newline=0;
1057 hold_space = tmp;
1058 break;
Glenn L McGrath8417c8c2003-09-14 15:24:18 +00001059 }
Mark Whitley0915c4b2001-06-11 23:50:06 +00001060 }
Robert Griebl47abc492002-06-11 23:43:27 +00001061 }
Erik Andersene49d5ec2000-02-08 19:58:47 +00001062 }
Erik Andersen1266a131999-12-29 22:19:46 +00001063
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001064 /*
1065 * exit point from sedding...
1066 */
1067discard_commands:
1068 /* we will print the line unless we were told to be quiet ('-n')
1069 or if the line was suppressed (ala 'd'elete) */
1070 if (!be_quiet) sed_puts(pattern_space,no_newline);
1071
1072 /* Delete and such jump here. */
1073discard_line:
1074 flush_append();
Glenn L McGrathc6adada2003-04-07 12:24:44 +00001075 free(pattern_space);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001076 }
Erik Andersen1266a131999-12-29 22:19:46 +00001077}
1078
Glenn L McGrath42c25732003-10-04 05:27:56 +00001079/* It is possible to have a command line argument with embedded
1080 newlines. This counts as multiple command lines. */
1081
1082static void add_cmd_block(char *cmdstr)
1083{
1084 int go=1;
1085 char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1086
1087 while(go) {
1088 int len=strcspn(temp2,"\n");
1089 if(!temp2[len]) go=0;
1090 else temp2[len]=0;
1091 add_cmd(temp2);
1092 temp2+=len+1;
1093 }
1094 free(temp);
1095}
1096
Erik Andersen1266a131999-12-29 22:19:46 +00001097extern int sed_main(int argc, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001098{
Matt Kraaia5f09c62001-11-12 16:44:55 +00001099 int opt, status = EXIT_SUCCESS;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001100
Eric Andersenbdfd0d72001-10-24 05:00:29 +00001101#ifdef CONFIG_FEATURE_CLEAN_UP
Mark Whitley858c1ad2000-07-11 21:38:47 +00001102 /* destroy command strings on exit */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001103 if (atexit(free_and_close_stuff) == -1)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001104 bb_perror_msg_and_die("atexit");
Mark Whitleyc41e8c82000-07-12 23:35:21 +00001105#endif
Mark Whitley858c1ad2000-07-11 21:38:47 +00001106
Eric Andersenc06f5682004-02-04 10:57:46 +00001107#define LIE_TO_AUTOCONF
1108#ifdef LIE_TO_AUTOCONF
1109 if(argc==2 && !strcmp(argv[1],"--version")) {
1110 printf("This is not GNU sed version 4.0\n");
1111 exit(0);
1112 }
1113#endif
1114
Mark Whitley6315ce62000-07-10 22:55:51 +00001115 /* do normal option parsing */
Rob Landley53302f82004-02-18 09:54:15 +00001116 while ((opt = getopt(argc, argv, "ine:f:")) > 0) {
Mark Whitley6315ce62000-07-10 22:55:51 +00001117 switch (opt) {
Rob Landley53302f82004-02-18 09:54:15 +00001118 case 'i':
1119 in_place++;
1120 atexit(cleanup_outname);
1121 break;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001122 case 'n':
1123 be_quiet++;
1124 break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001125 case 'e':
Glenn L McGrath42c25732003-10-04 05:27:56 +00001126 add_cmd_block(optarg);
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001127 break;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001128 case 'f':
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001129 {
1130 FILE *cmdfile;
1131 char *line;
1132
1133 cmdfile = bb_xfopen(optarg, "r");
1134
1135 while ((line = bb_get_chomped_line_from_file(cmdfile))
1136 != NULL) {
1137 add_cmd(line);
1138 free(line);
1139 }
1140 bb_xprint_and_close_file(cmdfile);
1141
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001142 break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001143 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001144 default:
1145 bb_show_usage();
Erik Andersene49d5ec2000-02-08 19:58:47 +00001146 }
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001147 }
Mark Whitley6315ce62000-07-10 22:55:51 +00001148
1149 /* if we didn't get a pattern from a -e and no command file was specified,
1150 * argv[optind] should be the pattern. no pattern, no worky */
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +00001151 if (sed_cmd_head.next == NULL) {
Mark Whitley6315ce62000-07-10 22:55:51 +00001152 if (argv[optind] == NULL)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001153 bb_show_usage();
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001154 else
Glenn L McGrath42c25732003-10-04 05:27:56 +00001155 add_cmd_block(argv[optind++]);
Mark Whitley6315ce62000-07-10 22:55:51 +00001156 }
Glenn L McGrath42c25732003-10-04 05:27:56 +00001157 /* Flush any unfinished commands. */
1158 add_cmd("");
Mark Whitley6315ce62000-07-10 22:55:51 +00001159
Rob Landley53302f82004-02-18 09:54:15 +00001160 /* By default, we write to stdout */
1161 nonstdout=stdout;
1162
Mark Whitley6315ce62000-07-10 22:55:51 +00001163 /* argv[(optind)..(argc-1)] should be names of file to process. If no
1164 * files were specified or '-' was specified, take input from stdin.
1165 * Otherwise, we process all the files specified. */
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001166 if (argv[optind] == NULL) {
Rob Landley53302f82004-02-18 09:54:15 +00001167 if(in_place) {
1168 fprintf(stderr,"sed: Filename required for -i\n");
1169 exit(1);
1170 }
Mark Whitley6315ce62000-07-10 22:55:51 +00001171 process_file(stdin);
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001172 } else {
Mark Whitley6315ce62000-07-10 22:55:51 +00001173 int i;
1174 FILE *file;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001175
Mark Whitley6315ce62000-07-10 22:55:51 +00001176 for (i = optind; i < argc; i++) {
Rob Landley53302f82004-02-18 09:54:15 +00001177 if(!strcmp(argv[i], "-") && !in_place) {
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001178 process_file(stdin);
1179 } else {
1180 file = bb_wfopen(argv[i], "r");
1181 if (file) {
Rob Landley53302f82004-02-18 09:54:15 +00001182 if(in_place) {
1183 struct stat statbuf;
1184 outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1185 strcat(outname,"XXXXXX");
1186 /* Set permissions of output file */
1187 fstat(fileno(file),&statbuf);
1188 mkstemp(outname);
1189 nonstdout=bb_wfopen(outname,"w");
1190 /* Set permissions of output file */
1191 fstat(fileno(file),&statbuf);
1192 fchmod(fileno(file),statbuf.st_mode);
1193 atexit(cleanup_outname);
1194 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001195 process_file(file);
1196 fclose(file);
Rob Landley53302f82004-02-18 09:54:15 +00001197 if(in_place) {
1198 fclose(nonstdout);
1199 nonstdout=stdout;
1200 unlink(argv[i]);
1201 rename(outname,argv[i]);
1202 free(outname);
1203 outname=0;
1204 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001205 } else {
1206 status = EXIT_FAILURE;
1207 }
1208 }
Mark Whitley6315ce62000-07-10 22:55:51 +00001209 }
1210 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001211
Matt Kraaia5f09c62001-11-12 16:44:55 +00001212 return status;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001213}