blob: ed48de17f4baaca307e93a5225bc42e18c86ea6c [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
Denis Vlasenko0beaff82007-09-21 13:16:32 +00008 * Copyright (C) 2003 by Glenn McGrath
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 *
Rob Landley2b26fd52006-02-24 02:30:39 +000011 * MAINTAINER: Rob Landley <rob@landley.net>
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +000012 *
Denys Vlasenko0ef64bd2010-08-16 20:14:46 +020013 * Licensed under GPLv2, see file LICENSE in this source tree.
Eric Andersen6b6b3f61999-10-28 16:06:25 +000014 */
15
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000016/* Code overview.
Denys Vlasenko52d83702011-05-03 00:51:43 +020017 *
18 * Files are laid out to avoid unnecessary function declarations. So for
19 * example, every function add_cmd calls occurs before add_cmd in this file.
20 *
21 * add_cmd() is called on each line of sed command text (from a file or from
22 * the command line). It calls get_address() and parse_cmd_args(). The
23 * resulting sed_cmd_t structures are appended to a linked list
24 * (G.sed_cmd_head/G.sed_cmd_tail).
25 *
Denys Vlasenkoa221bc52011-09-13 18:40:22 +020026 * process_files() does actual sedding, reading data lines from each input FILE*
Denys Vlasenko52d83702011-05-03 00:51:43 +020027 * (which could be stdin) and applying the sed command list (sed_cmd_head) to
28 * each of the resulting lines.
29 *
30 * sed_main() is where external code calls into this, with a command line.
31 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000032
Denys Vlasenko52d83702011-05-03 00:51:43 +020033/* Supported features and commands in this version of sed:
34 *
35 * - comments ('#')
36 * - address matching: num|/matchstr/[,num|/matchstr/|$]command
37 * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
38 * - edit commands: (a)ppend, (i)nsert, (c)hange
39 * - file commands: (r)ead
40 * - backreferences in substitution expressions (\0, \1, \2...\9)
41 * - grouped commands: {cmd1;cmd2}
42 * - transliteration (y/source-chars/dest-chars/)
43 * - pattern space hold space storing / swapping (g, h, x)
44 * - labels / branching (: label, b, t, T)
45 *
46 * (Note: Specifying an address (range) to match is *optional*; commands
47 * default to the whole pattern space if no specific address match was
48 * requested.)
49 *
50 * Todo:
51 * - Create a wrapper around regex to make libc's regex conform with sed
52 *
Mimi Li37a79c02012-07-24 13:20:12 +020053 * Reference
54 * http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
55 * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html
Denys Vlasenko63f4d322015-04-17 14:24:55 +020056 * http://sed.sourceforge.net/sedfaq3.html
Denys Vlasenko52d83702011-05-03 00:51:43 +020057 */
Mark Whitley6315ce62000-07-10 22:55:51 +000058
Denys Vlasenko73225b62013-11-13 12:45:33 +010059//config:config SED
60//config: bool "sed"
61//config: default y
62//config: help
63//config: sed is used to perform text transformations on a file
64//config: or input from a pipeline.
65
66//kbuild:lib-$(CONFIG_SED) += sed.o
67
68//applet:IF_SED(APPLET(sed, BB_DIR_BIN, BB_SUID_DROP))
69
Pere Orga6a3e01d2011-04-01 22:56:30 +020070//usage:#define sed_trivial_usage
Denys Vlasenkoa82e32d2013-10-30 13:00:00 +010071//usage: "[-inrE] [-f FILE]... [-e CMD]... [FILE]...\n"
72//usage: "or: sed [-inrE] CMD [FILE]..."
Pere Orga6a3e01d2011-04-01 22:56:30 +020073//usage:#define sed_full_usage "\n\n"
Denys Vlasenko66426762011-06-05 03:58:28 +020074//usage: " -e CMD Add CMD to sed commands to be executed"
Pere Orga6a3e01d2011-04-01 22:56:30 +020075//usage: "\n -f FILE Add FILE contents to sed commands to be executed"
Simon B8c343952012-05-06 13:59:15 +020076//usage: "\n -i[SFX] Edit files in-place (otherwise sends to stdout)"
Denys Vlasenko1d7ad7a2012-06-21 09:45:11 +020077//usage: "\n Optionally back files up, appending SFX"
Pere Orga6a3e01d2011-04-01 22:56:30 +020078//usage: "\n -n Suppress automatic printing of pattern space"
Denys Vlasenkoa82e32d2013-10-30 13:00:00 +010079//usage: "\n -r,-E Use extended regex syntax"
Pere Orga6a3e01d2011-04-01 22:56:30 +020080//usage: "\n"
81//usage: "\nIf no -e or -f, the first non-option argument is the sed command string."
82//usage: "\nRemaining arguments are input files (stdin if none)."
83//usage:
84//usage:#define sed_example_usage
85//usage: "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
86//usage: "bar\n"
87
Denis Vlasenkob6adbf12007-05-26 19:00:18 +000088#include "libbb.h"
Denys Vlasenkoe6a2f4c2016-04-21 16:26:30 +020089#include "common_bufsiz.h"
"Vladimir N. Oleynik"23f62fc2005-09-14 16:59:11 +000090#include "xregex.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000091
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +020092#if 0
93# define dbg(...) bb_error_msg(__VA_ARGS__)
94#else
95# define dbg(...) ((void)0)
96#endif
97
98
Denys Vlasenko2e284a42010-08-01 04:14:46 +020099enum {
100 OPT_in_place = 1 << 0,
101};
102
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000103/* Each sed command turns into one of these structures. */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +0000104typedef struct sed_cmd_s {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000105 /* Ordered by alignment requirements: currently 36 bytes on x86 */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000106 struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
Mark Whitley2dc192f2000-11-03 19:47:00 +0000107
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000108 /* address storage */
109 regex_t *beg_match; /* sed -e '/match/cmd' */
110 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
111 regex_t *sub_match; /* For 's/sub_match/string/' */
112 int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200113 int beg_line_orig; /* copy of the above, needed for -i */
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200114 int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($). -2-N = +N */
115 int end_line_orig;
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +0000116
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100117 FILE *sw_file; /* File (sw) command writes to, NULL for none. */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000118 char *string; /* Data string for (saicytb) commands. */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000119
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000120 unsigned which_match; /* (s) Which match to replace (0 for all) */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000121
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000122 /* Bitfields (gcc won't group them if we don't) */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000123 unsigned invert:1; /* the '!' after the address */
124 unsigned in_match:1; /* Next line also included in match? */
125 unsigned sub_p:1; /* (s) print option */
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000126
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000127 char sw_last_char; /* Last line written by (sw) had no '\n' */
Glenn L McGrathc18ce372003-09-13 06:57:39 +0000128
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000129 /* GENERAL FIELDS */
130 char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +0000131} sed_cmd_t;
Mark Whitley6315ce62000-07-10 22:55:51 +0000132
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000133static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000134
Denis Vlasenko17a15262007-03-26 20:48:46 +0000135struct globals {
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000136 /* options */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000137 int be_quiet, regex_type;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100138
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000139 FILE *nonstdout;
140 char *outname, *hold_space;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100141 smallint exitcode;
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000142
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100143 /* list of input files */
Denys Vlasenkodeb07692013-11-28 12:08:51 +0100144 int current_input_file, last_input_file;
145 char **input_file_list;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100146 FILE *current_fp;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000147
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000148 regmatch_t regmatch[10];
149 regex_t *previous_regex_ptr;
Denis Vlasenko9213a9e2006-09-17 16:28:10 +0000150
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000151 /* linked list of sed commands */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200152 sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000153
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100154 /* linked list of append lines */
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000155 llist_t *append_head;
156
157 char *add_cmd_line;
158
159 struct pipeline {
Denys Vlasenkofb132e42010-10-29 11:46:52 +0200160 char *buf; /* Space to hold string */
161 int idx; /* Space used */
162 int len; /* Space allocated */
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000163 } pipeline;
Denys Vlasenko98a4c7c2010-02-04 15:00:15 +0100164} FIX_ALIASING;
Denys Vlasenkoe6a2f4c2016-04-21 16:26:30 +0200165#define G (*(struct globals*)bb_common_bufsiz1)
Denis Vlasenko74324c82007-06-04 10:16:52 +0000166#define INIT_G() do { \
Denys Vlasenko47cfbf32016-04-21 18:18:48 +0200167 setup_common_bufsiz(); \
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200168 BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
Denis Vlasenko74324c82007-06-04 10:16:52 +0000169 G.sed_cmd_tail = &G.sed_cmd_head; \
170} while (0)
171
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000172
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +0000173#if ENABLE_FEATURE_CLEAN_UP
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000174static void sed_free_and_close_stuff(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000175{
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200176 sed_cmd_t *sed_cmd = G.sed_cmd_head;
Mark Whitley6315ce62000-07-10 22:55:51 +0000177
Denis Vlasenko17a15262007-03-26 20:48:46 +0000178 llist_free(G.append_head, free);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000179
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000180 while (sed_cmd) {
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000181 sed_cmd_t *sed_cmd_next = sed_cmd->next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000182
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000183 if (sed_cmd->sw_file)
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100184 fclose(sed_cmd->sw_file);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000185
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000186 if (sed_cmd->beg_match) {
187 regfree(sed_cmd->beg_match);
188 free(sed_cmd->beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000189 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000190 if (sed_cmd->end_match) {
191 regfree(sed_cmd->end_match);
192 free(sed_cmd->end_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000193 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000194 if (sed_cmd->sub_match) {
195 regfree(sed_cmd->sub_match);
196 free(sed_cmd->sub_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000197 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000198 free(sed_cmd->string);
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000199 free(sed_cmd);
200 sed_cmd = sed_cmd_next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000201 }
Rob Landleyce4f0e92004-10-30 06:54:19 +0000202
Denis Vlasenko60818682007-09-28 22:07:23 +0000203 free(G.hold_space);
Rob Landleydcc28662004-11-25 07:21:47 +0000204
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100205 if (G.current_fp)
206 fclose(G.current_fp);
Mark Whitley6315ce62000-07-10 22:55:51 +0000207}
Denis Vlasenko666da5e2006-12-26 18:17:42 +0000208#else
209void sed_free_and_close_stuff(void);
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000210#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000211
Rob Landley53302f82004-02-18 09:54:15 +0000212/* If something bad happens during -i operation, delete temp file */
213
214static void cleanup_outname(void)
215{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000216 if (G.outname) unlink(G.outname);
Rob Landley53302f82004-02-18 09:54:15 +0000217}
218
Denis Vlasenko40276642007-11-13 16:48:10 +0000219/* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000220
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000221static void parse_escapes(char *dest, const char *string, int len, char from, char to)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000222{
Denis Vlasenko80778502006-10-25 12:46:46 +0000223 int i = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000224
Denis Vlasenko80778502006-10-25 12:46:46 +0000225 while (i < len) {
226 if (string[i] == '\\') {
227 if (!to || string[i+1] == from) {
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000228 *dest++ = to ? to : string[i+1];
Denis Vlasenko80778502006-10-25 12:46:46 +0000229 i += 2;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000230 continue;
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000231 }
232 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000233 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000234 /* TODO: is it safe wrt a string with trailing '\\' ? */
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000235 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000236 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000237 *dest = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000238}
239
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000240static char *copy_parsing_escapes(const char *string, int len)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000241{
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200242 const char *s;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000243 char *dest = xmalloc(len + 1);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000244
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200245 /* sed recognizes \n */
Denys Vlasenko6a0abcc2011-05-03 00:52:22 +0200246 /* GNU sed also recognizes \t and \r */
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200247 for (s = "\nn\tt\rr"; *s; s += 2) {
248 parse_escapes(dest, string, len, s[1], s[0]);
249 string = dest;
250 len = strlen(dest);
251 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000252 return dest;
253}
254
255
Mark Whitley6315ce62000-07-10 22:55:51 +0000256/*
Eric Andersen28b3c532001-01-02 11:01:31 +0000257 * index_of_next_unescaped_regexp_delim - walks left to right through a string
258 * beginning at a specified index and returns the index of the next regular
Denis Vlasenko40276642007-11-13 16:48:10 +0000259 * expression delimiter (typically a forward slash ('/')) not preceded by
Rob Landley4795e4e2006-07-26 17:25:08 +0000260 * a backslash ('\'). A negative delimiter disables square bracket checking.
Mark Whitley6315ce62000-07-10 22:55:51 +0000261 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000262static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
Mark Whitley6315ce62000-07-10 22:55:51 +0000263{
Matt Kraaia0065d52001-08-24 14:45:50 +0000264 int bracket = -1;
265 int escaped = 0;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000266 int idx = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000267 char ch;
Matt Kraaia0065d52001-08-24 14:45:50 +0000268
Rob Landley4795e4e2006-07-26 17:25:08 +0000269 if (delimiter < 0) {
270 bracket--;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000271 delimiter = -delimiter;
Rob Landley4795e4e2006-07-26 17:25:08 +0000272 }
273
Denys Vlasenko52d83702011-05-03 00:51:43 +0200274 for (; (ch = str[idx]) != '\0'; idx++) {
Rob Landley4795e4e2006-07-26 17:25:08 +0000275 if (bracket >= 0) {
Denys Vlasenko52d83702011-05-03 00:51:43 +0200276 if (ch == ']'
277 && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
278 ) {
Matt Kraaia0065d52001-08-24 14:45:50 +0000279 bracket = -1;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200280 }
Matt Kraaia0065d52001-08-24 14:45:50 +0000281 } else if (escaped)
282 escaped = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000283 else if (ch == '\\')
Matt Kraaia0065d52001-08-24 14:45:50 +0000284 escaped = 1;
Rob Landley4795e4e2006-07-26 17:25:08 +0000285 else if (bracket == -1 && ch == '[')
Matt Kraaia0065d52001-08-24 14:45:50 +0000286 bracket = idx;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000287 else if (ch == delimiter)
Mark Whitley97562bd2000-07-17 20:06:42 +0000288 return idx;
289 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000290
Mark Whitley97562bd2000-07-17 20:06:42 +0000291 /* if we make it to here, we've hit the end of the string */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000292 bb_error_msg_and_die("unmatched '%c'", delimiter);
Mark Whitley6315ce62000-07-10 22:55:51 +0000293}
294
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000295/*
296 * Returns the index of the third delimiter
297 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000298static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000299{
Denis Vlasenko9356b502007-01-29 23:44:38 +0000300 const char *cmdstr_ptr = cmdstr;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100301 unsigned char delimiter;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000302 int idx = 0;
303
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000304 /* verify that the 's' or 'y' is followed by something. That something
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000305 * (typically a 'slash') is now our regexp delimiter... */
Rob Landley4795e4e2006-07-26 17:25:08 +0000306 if (*cmdstr == '\0')
307 bb_error_msg_and_die("bad format in substitution expression");
Denis Vlasenko80778502006-10-25 12:46:46 +0000308 delimiter = *cmdstr_ptr++;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000309
310 /* save the match string */
311 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000312 *match = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000313
314 /* save the replacement string */
315 cmdstr_ptr += idx + 1;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100316 idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000317 *replace = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000318
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000319 return ((cmdstr_ptr - cmdstr) + idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000320}
321
Mark Whitley6315ce62000-07-10 22:55:51 +0000322/*
323 * returns the index in the string just past where the address ends.
324 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000325static int get_address(const char *my_str, int *linenum, regex_t ** regex)
Mark Whitley6315ce62000-07-10 22:55:51 +0000326{
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000327 const char *pos = my_str;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000328
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000329 if (isdigit(*my_str)) {
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000330 *linenum = strtol(my_str, (char**)&pos, 10);
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000331 /* endstr shouldnt ever equal NULL */
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000332 } else if (*my_str == '$') {
Mark Whitley0915c4b2001-06-11 23:50:06 +0000333 *linenum = -1;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000334 pos++;
335 } else if (*my_str == '/' || *my_str == '\\') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000336 int next;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000337 char delimiter;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000338 char *temp;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000339
Denis Vlasenko80778502006-10-25 12:46:46 +0000340 delimiter = '/';
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100341 if (*my_str == '\\')
342 delimiter = *++pos;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000343 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000344 temp = copy_parsing_escapes(pos, next);
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100345 *regex = xzalloc(sizeof(regex_t));
Denys Vlasenkob0e9b722013-07-21 22:09:44 +0200346 xregcomp(*regex, temp, G.regex_type);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000347 free(temp);
348 /* Move position to next character after last delimiter */
Rob Landley4795e4e2006-07-26 17:25:08 +0000349 pos += (next+1);
Mark Whitley6315ce62000-07-10 22:55:51 +0000350 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000351 return pos - my_str;
Mark Whitley6315ce62000-07-10 22:55:51 +0000352}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000353
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000354/* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
Denis Vlasenko68404f12008-03-17 09:00:54 +0000355static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000356{
Denis Vlasenko80778502006-10-25 12:46:46 +0000357 int start = 0, idx, hack = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000358
359 /* Skip whitespace, then grab filename to end of line */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000360 while (isspace(filecmdstr[start]))
361 start++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000362 idx = start;
Denis Vlasenko9356b502007-01-29 23:44:38 +0000363 while (filecmdstr[idx] && filecmdstr[idx] != '\n')
364 idx++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000365
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000366 /* If lines glued together, put backslash back. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000367 if (filecmdstr[idx] == '\n')
368 hack = 1;
Denis Vlasenko80778502006-10-25 12:46:46 +0000369 if (idx == start)
370 bb_error_msg_and_die("empty filename");
Rob Landleyd921b2e2006-08-03 15:41:12 +0000371 *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
Denis Vlasenko9356b502007-01-29 23:44:38 +0000372 if (hack)
373 (*retval)[idx] = '\\';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000374
375 return idx;
376}
377
Denis Vlasenko9356b502007-01-29 23:44:38 +0000378static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000379{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000380 int cflags = G.regex_type;
Mark Whitley06f35292000-07-13 19:58:04 +0000381 char *match;
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000382 int idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000383
384 /*
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000385 * A substitution command should look something like this:
David A. Wheeler80a068d2013-12-05 20:42:17 -0500386 * s/match/replace/ #giIpw
Mark Whitley0e4cec02000-08-21 21:29:20 +0000387 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000388 * mandatory optional
Mark Whitley06f35292000-07-13 19:58:04 +0000389 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000390 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
Mark Whitley06f35292000-07-13 19:58:04 +0000391
Mark Whitley97562bd2000-07-17 20:06:42 +0000392 /* determine the number of back references in the match string */
393 /* Note: we compute this here rather than in the do_subst_command()
394 * function to save processor time, at the expense of a little more memory
395 * (4 bits) per sed_cmd */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000396
Mark Whitley06f35292000-07-13 19:58:04 +0000397 /* process the flags */
Mark Whitley70705d72000-07-14 19:06:30 +0000398
Denis Vlasenko80778502006-10-25 12:46:46 +0000399 sed_cmd->which_match = 1;
Denys Vlasenko9caea242014-09-16 01:11:13 +0200400 dbg("s flags:'%s'", substr + idx + 1);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000401 while (substr[++idx]) {
Denys Vlasenko9caea242014-09-16 01:11:13 +0200402 dbg("s flag:'%c'", substr[idx]);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000403 /* Parse match number */
Denis Vlasenko80778502006-10-25 12:46:46 +0000404 if (isdigit(substr[idx])) {
405 if (match[0] != '^') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000406 /* Match 0 treated as all, multiple matches we take the last one. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000407 const char *pos = substr + idx;
408/* FIXME: error check? */
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000409 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
Denys Vlasenko9caea242014-09-16 01:11:13 +0200410 idx = pos - substr - 1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000411 }
412 continue;
413 }
Rob Landley40ec4ae2004-01-04 06:42:14 +0000414 /* Skip spaces */
Denys Vlasenko6935ec92009-10-22 19:42:26 +0200415 if (isspace(substr[idx]))
416 continue;
Rob Landley40ec4ae2004-01-04 06:42:14 +0000417
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000418 switch (substr[idx]) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000419 /* Replace all occurrences */
420 case 'g':
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000421 if (match[0] != '^')
422 sed_cmd->which_match = 0;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000423 break;
424 /* Print pattern space */
425 case 'p':
426 sed_cmd->sub_p = 1;
427 break;
428 /* Write to file */
429 case 'w':
430 {
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100431 char *fname;
432 idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname);
433 sed_cmd->sw_file = xfopen_for_write(fname);
434 sed_cmd->sw_last_char = '\n';
435 free(fname);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000436 break;
437 }
438 /* Ignore case (gnu exension) */
David A. Wheeler80a068d2013-12-05 20:42:17 -0500439 case 'i':
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000440 case 'I':
441 cflags |= REG_ICASE;
442 break;
443 /* Comment */
444 case '#':
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +0200445 // while (substr[++idx]) continue;
446 idx += strlen(substr + idx); // same
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000447 /* Fall through */
448 /* End of command */
449 case ';':
450 case '}':
451 goto out;
452 default:
Denys Vlasenko9caea242014-09-16 01:11:13 +0200453 dbg("s bad flags:'%s'", substr + idx);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000454 bb_error_msg_and_die("bad option in substitution expression");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000455 }
456 }
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +0200457 out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000458 /* compile the match string into a regex */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000459 if (*match != '\0') {
460 /* If match is empty, we use last regex used at runtime */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100461 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
462 dbg("xregcomp('%s',%x)", match, cflags);
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000463 xregcomp(sed_cmd->sub_match, match, cflags);
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100464 dbg("regcomp ok");
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000465 }
Mark Whitley06f35292000-07-13 19:58:04 +0000466 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000467
468 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000469}
470
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000471/*
472 * Process the commands arguments
473 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000474static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000475{
Denys Vlasenko52d83702011-05-03 00:51:43 +0200476 static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
477 enum {
478 IDX_s = 0,
479 IDX_a,
480 IDX_i,
481 IDX_c,
482 IDX_r,
483 IDX_w,
484 IDX_colon,
485 IDX_b,
486 IDX_t,
487 IDX_T,
488 IDX_y,
489 IDX_d,
490 IDX_D,
491 IDX_g,
492 IDX_G,
493 IDX_h,
494 IDX_H,
495 IDX_l,
496 IDX_n,
497 IDX_N,
498 IDX_p,
499 IDX_P,
500 IDX_q,
501 IDX_x,
502 IDX_equal,
503 IDX_lbrace,
504 IDX_rbrace,
505 IDX_nul
506 };
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200507 unsigned idx;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200508
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200509 BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul);
510
511 idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200512
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000513 /* handle (s)ubstitution command */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200514 if (idx == IDX_s) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000515 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
Denys Vlasenko52d83702011-05-03 00:51:43 +0200516 }
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000517 /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200518 else if (idx <= IDX_c) { /* a,i,c */
Mimi Li37a79c02012-07-24 13:20:12 +0200519 if (idx < IDX_c) { /* a,i */
520 if (sed_cmd->end_line || sed_cmd->end_match)
521 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
522 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000523 for (;;) {
524 if (*cmdstr == '\n' || *cmdstr == '\\') {
Rob Landley25d82392004-04-01 09:23:30 +0000525 cmdstr++;
526 break;
Denys Vlasenkoc0dab372009-10-22 22:28:08 +0200527 }
528 if (!isspace(*cmdstr))
Denis Vlasenko80778502006-10-25 12:46:46 +0000529 break;
Denys Vlasenkoc0dab372009-10-22 22:28:08 +0200530 cmdstr++;
Rob Landley25d82392004-04-01 09:23:30 +0000531 }
Rob Landleyd921b2e2006-08-03 15:41:12 +0000532 sed_cmd->string = xstrdup(cmdstr);
Denis Vlasenko40276642007-11-13 16:48:10 +0000533 /* "\anychar" -> "anychar" */
534 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000535 cmdstr += strlen(cmdstr);
Denys Vlasenko52d83702011-05-03 00:51:43 +0200536 }
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000537 /* handle file cmds: (r)ead */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200538 else if (idx <= IDX_w) { /* r,w */
Mimi Li37a79c02012-07-24 13:20:12 +0200539 if (idx < IDX_w) { /* r */
540 if (sed_cmd->end_line || sed_cmd->end_match)
541 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
542 }
Denis Vlasenko68404f12008-03-17 09:00:54 +0000543 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000544 if (sed_cmd->cmd == 'w') {
Denis Vlasenko5415c852008-07-21 23:05:26 +0000545 sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000546 sed_cmd->sw_last_char = '\n';
547 }
Denys Vlasenko52d83702011-05-03 00:51:43 +0200548 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000549 /* handle branch commands */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200550 else if (idx <= IDX_T) { /* :,b,t,T */
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000551 int length;
552
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000553 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000554 length = strcspn(cmdstr, semicolon_whitespace);
555 if (length) {
Rob Landleyd921b2e2006-08-03 15:41:12 +0000556 sed_cmd->string = xstrndup(cmdstr, length);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000557 cmdstr += length;
558 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000559 }
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000560 /* translation command */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200561 else if (idx == IDX_y) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000562 char *match, *replace;
Denis Vlasenko80778502006-10-25 12:46:46 +0000563 int i = cmdstr[0];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000564
Denis Vlasenko80778502006-10-25 12:46:46 +0000565 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000566 /* \n already parsed, but \delimiter needs unescaping. */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000567 parse_escapes(match, match, strlen(match), i, i);
568 parse_escapes(replace, replace, strlen(replace), i, i);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000569
Rob Landley9ffd4232006-05-21 18:30:35 +0000570 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000571 for (i = 0; match[i] && replace[i]; i++) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000572 sed_cmd->string[i*2] = match[i];
573 sed_cmd->string[i*2+1] = replace[i];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000574 }
575 free(match);
576 free(replace);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000577 }
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000578 /* if it wasnt a single-letter command that takes no arguments
579 * then it must be an invalid command.
580 */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200581 else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
Denis Vlasenko80778502006-10-25 12:46:46 +0000582 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
Mark Whitley1f3b9f22001-05-11 22:27:13 +0000583 }
Mark Whitley70705d72000-07-14 19:06:30 +0000584
585 /* give back whatever's left over */
Denis Vlasenko80778502006-10-25 12:46:46 +0000586 return cmdstr;
Eric Andersen50d63601999-11-09 01:47:36 +0000587}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000588
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000589
590/* Parse address+command sets, skipping comment lines. */
591
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000592static void add_cmd(const char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000593{
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000594 sed_cmd_t *sed_cmd;
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200595 unsigned len, n;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000596
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000597 /* Append this line to any unfinished line from last time. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000598 if (G.add_cmd_line) {
599 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
600 free(G.add_cmd_line);
Denis Vlasenko945bd3d2007-04-12 21:20:25 +0000601 cmdstr = G.add_cmd_line = tp;
Rob Landley12d87552006-06-05 17:32:44 +0000602 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000603
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200604 /* If this line ends with unescaped backslash, request next line. */
605 n = len = strlen(cmdstr);
606 while (n && cmdstr[n-1] == '\\')
607 n--;
608 if ((len - n) & 1) { /* if odd number of trailing backslashes */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000609 if (!G.add_cmd_line)
610 G.add_cmd_line = xstrdup(cmdstr);
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200611 G.add_cmd_line[len-1] = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000612 return;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000613 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000614
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000615 /* Loop parsing all commands in this line. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000616 while (*cmdstr) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000617 /* Skip leading whitespace and semicolons */
618 cmdstr += strspn(cmdstr, semicolon_whitespace);
619
620 /* If no more commands, exit. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000621 if (!*cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000622
623 /* if this is a comment, jump past it and keep going */
624 if (*cmdstr == '#') {
625 /* "#n" is the same as using -n on the command line */
Denis Vlasenko80778502006-10-25 12:46:46 +0000626 if (cmdstr[1] == 'n')
Denis Vlasenko17a15262007-03-26 20:48:46 +0000627 G.be_quiet++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000628 cmdstr = strpbrk(cmdstr, "\n\r");
629 if (!cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000630 continue;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000631 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000632
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000633 /* parse the command
634 * format is: [addr][,addr][!]cmd
635 * |----||-----||-|
636 * part1 part2 part3
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000637 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000638
Rob Landley9ffd4232006-05-21 18:30:35 +0000639 sed_cmd = xzalloc(sizeof(sed_cmd_t));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000640
641 /* first part (if present) is an address: either a '$', a number or a /regex/ */
642 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200643 sed_cmd->beg_line_orig = sed_cmd->beg_line;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000644
645 /* second part (if present) will begin with a comma */
646 if (*cmdstr == ',') {
647 int idx;
648
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000649 cmdstr++;
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200650 if (*cmdstr == '+' && isdigit(cmdstr[1])) {
651 /* http://sed.sourceforge.net/sedfaq3.html#s3.3
652 * Under GNU sed 3.02+, ssed, and sed15+, <address2>
653 * may also be a notation of the form +num,
654 * indicating the next num lines after <address1> is
655 * matched.
656 * GNU sed 4.2.1 accepts even "+" (meaning "+0").
657 * We don't (we check for isdigit, see above), think
658 * about the "+-3" case.
659 */
660 char *end;
661 /* code is smaller compared to using &cmdstr here: */
662 idx = strtol(cmdstr+1, &end, 10);
663 sed_cmd->end_line = -2 - idx;
664 cmdstr = end;
665 } else {
666 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
667 cmdstr += idx;
668 idx--; /* if 0, trigger error check below */
669 }
670 if (idx < 0)
Denis Vlasenko80778502006-10-25 12:46:46 +0000671 bb_error_msg_and_die("no address after comma");
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200672 sed_cmd->end_line_orig = sed_cmd->end_line;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000673 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000674
675 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000676 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000677
678 /* Check for inversion flag */
679 if (*cmdstr == '!') {
680 sed_cmd->invert = 1;
681 cmdstr++;
682
683 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000684 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000685 }
686
687 /* last part (mandatory) will be a command */
Denis Vlasenko80778502006-10-25 12:46:46 +0000688 if (!*cmdstr)
689 bb_error_msg_and_die("missing command");
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200690 sed_cmd->cmd = *cmdstr++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000691 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
692
Denys Vlasenkoe93d1562013-07-08 01:43:40 +0200693 /* cmdstr now points past args.
694 * GNU sed requires a separator, if there are more commands,
695 * else it complains "char N: extra characters after command".
696 * Example: "sed 'p;d'". We also allow "sed 'pd'".
697 */
698
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000699 /* Add the command to the command array */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200700 *G.sed_cmd_tail = sed_cmd;
701 G.sed_cmd_tail = &sed_cmd->next;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000702 }
703
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000704 /* If we glued multiple lines together, free the memory. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000705 free(G.add_cmd_line);
706 G.add_cmd_line = NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +0000707}
708
Rob Landleydcc28662004-11-25 07:21:47 +0000709/* Append to a string, reallocating memory as necessary. */
710
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000711#define PIPE_GROW 64
Eric Andersenc52a6b02001-11-10 10:49:42 +0000712
Eric Andersen14f5c8d2005-04-16 19:39:00 +0000713static void pipe_putc(char c)
Eric Andersenc52a6b02001-11-10 10:49:42 +0000714{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000715 if (G.pipeline.idx == G.pipeline.len) {
716 G.pipeline.buf = xrealloc(G.pipeline.buf,
717 G.pipeline.len + PIPE_GROW);
718 G.pipeline.len += PIPE_GROW;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000719 }
Denis Vlasenko17a15262007-03-26 20:48:46 +0000720 G.pipeline.buf[G.pipeline.idx++] = c;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000721}
722
Rob Landley4795e4e2006-07-26 17:25:08 +0000723static void do_subst_w_backrefs(char *line, char *replace)
Mark Whitley97562bd2000-07-17 20:06:42 +0000724{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200725 int i, j;
Mark Whitley97562bd2000-07-17 20:06:42 +0000726
727 /* go through the replacement string */
728 for (i = 0; replace[i]; i++) {
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200729 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000730 if (replace[i] == '\\') {
731 unsigned backref = replace[++i] - '0';
732 if (backref <= 9) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000733 /* print out the text held in G.regmatch[backref] */
734 if (G.regmatch[backref].rm_so != -1) {
735 j = G.regmatch[backref].rm_so;
736 while (j < G.regmatch[backref].rm_eo)
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000737 pipe_putc(line[j++]);
738 }
739 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000740 }
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000741 /* I _think_ it is impossible to get '\' to be
742 * the last char in replace string. Thus we dont check
743 * for replace[i] == NUL. (counterexample anyone?) */
744 /* if we find a backslash escaped character, print the character */
745 pipe_putc(replace[i]);
746 continue;
Mark Whitley97562bd2000-07-17 20:06:42 +0000747 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000748 /* if we find an unescaped '&' print out the whole matched text. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000749 if (replace[i] == '&') {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000750 j = G.regmatch[0].rm_so;
751 while (j < G.regmatch[0].rm_eo)
Denis Vlasenko80778502006-10-25 12:46:46 +0000752 pipe_putc(line[j++]);
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000753 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000754 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000755 /* Otherwise just output the character. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000756 pipe_putc(replace[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000757 }
758}
759
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200760static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
Mark Whitley4f7fe772000-07-13 20:01:58 +0000761{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200762 char *line = *line_p;
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000763 unsigned match_count = 0;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200764 bool altered = 0;
765 bool prev_match_empty = 1;
766 bool tried_at_eol = 0;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000767 regex_t *current_regex;
768
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200769 current_regex = sed_cmd->sub_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000770 /* Handle empty regex. */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200771 if (!current_regex) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000772 current_regex = G.previous_regex_ptr;
Denis Vlasenko80778502006-10-25 12:46:46 +0000773 if (!current_regex)
Denis Vlasenkod3d004d2006-10-27 09:02:31 +0000774 bb_error_msg_and_die("no previous regexp");
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200775 }
776 G.previous_regex_ptr = current_regex;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000777
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000778 /* Find the first match */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100779 dbg("matching '%s'", line);
780 if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
781 dbg("no match");
Mark Whitley2dc192f2000-11-03 19:47:00 +0000782 return 0;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100783 }
784 dbg("match");
Mark Whitley2dc192f2000-11-03 19:47:00 +0000785
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000786 /* Initialize temporary output buffer. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000787 G.pipeline.buf = xmalloc(PIPE_GROW);
788 G.pipeline.len = PIPE_GROW;
789 G.pipeline.idx = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000790
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000791 /* Now loop through, substituting for matches */
792 do {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200793 int start = G.regmatch[0].rm_so;
794 int end = G.regmatch[0].rm_eo;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000795 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000796
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000797 match_count++;
798
799 /* If we aren't interested in this match, output old line to
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200800 * end of match and continue */
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000801 if (sed_cmd->which_match
802 && (sed_cmd->which_match != match_count)
803 ) {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200804 for (i = 0; i < end; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200805 pipe_putc(*line++);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200806 /* Null match? Print one more char */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200807 if (start == end && *line)
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200808 pipe_putc(*line++);
Denys Vlasenkob84dafb2012-04-24 19:27:34 +0200809 goto next;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000810 }
811
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200812 /* Print everything before the match */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200813 for (i = 0; i < start; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200814 pipe_putc(line[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000815
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200816 /* Then print the substitution string,
817 * unless we just matched empty string after non-empty one.
818 * Example: string "cccd", pattern "c*", repl "R":
819 * result is "RdR", not "RRdR": first match "ccc",
820 * second is "" before "d", third is "" after "d".
821 * Second match is NOT replaced!
822 */
Denys Vlasenko84406e42012-06-07 16:34:57 +0200823 if (prev_match_empty || start != 0 || start != end) {
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200824 //dbg("%d %d %d", prev_match_empty, start, end);
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200825 dbg("inserting replacement at %d in '%s'", start, line);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200826 do_subst_w_backrefs(line, sed_cmd->string);
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200827 /* Flag that something has changed */
828 altered = 1;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200829 } else {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200830 dbg("NOT inserting replacement at %d in '%s'", start, line);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200831 }
Mark Whitley97562bd2000-07-17 20:06:42 +0000832
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200833 /* If matched string is empty (f.e. "c*" pattern),
834 * copy verbatim one char after it before attempting more matches
835 */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200836 prev_match_empty = (start == end);
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200837 if (prev_match_empty) {
838 if (!line[end]) {
839 tried_at_eol = 1;
840 } else {
841 pipe_putc(line[end]);
842 end++;
843 }
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200844 }
845
846 /* Advance past the match */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200847 dbg("line += %d", end);
848 line += end;
Mark Whitley97562bd2000-07-17 20:06:42 +0000849
Mark Whitley2dc192f2000-11-03 19:47:00 +0000850 /* if we're not doing this globally, get out now */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100851 if (sed_cmd->which_match != 0)
852 break;
Denys Vlasenkob84dafb2012-04-24 19:27:34 +0200853 next:
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200854 /* Exit if we are at EOL and already tried matching at it */
855 if (*line == '\0') {
856 if (tried_at_eol)
857 break;
858 tried_at_eol = 1;
859 }
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200860
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200861//maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100862 } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
Mark Whitley2dc192f2000-11-03 19:47:00 +0000863
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000864 /* Copy rest of string into output pipeline */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200865 while (1) {
866 char c = *line++;
867 pipe_putc(c);
868 if (c == '\0')
869 break;
870 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000871
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200872 free(*line_p);
873 *line_p = G.pipeline.buf;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000874 return altered;
875}
Mark Whitley6315ce62000-07-10 22:55:51 +0000876
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000877/* Set command pointer to point to this label. (Does not handle null label.) */
Rob Landley4795e4e2006-07-26 17:25:08 +0000878static sed_cmd_t *branch_to(char *label)
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000879{
880 sed_cmd_t *sed_cmd;
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000881
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200882 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000883 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
884 return sed_cmd;
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000885 }
886 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000887 bb_error_msg_and_die("can't find label for jump to '%s'", label);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000888}
Mark Whitley6315ce62000-07-10 22:55:51 +0000889
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000890static void append(char *s)
891{
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +0100892 llist_add_to_end(&G.append_head, s);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000893}
894
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100895/* Output line of text. */
896/* Note:
897 * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
898 * Without them, we had this:
899 * echo -n thingy >z1
900 * echo -n again >z2
901 * >znull
902 * sed "s/i/z/" z1 z2 znull | hexdump -vC
903 * output:
904 * gnu sed 4.1.5:
905 * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
906 * bbox:
907 * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
Rob Landleydcc28662004-11-25 07:21:47 +0000908 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000909enum {
910 NO_EOL_CHAR = 1,
Denis Vlasenko86811802007-01-29 17:10:19 +0000911 LAST_IS_NUL = 2,
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000912};
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100913static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
914{
915 char lpc = *last_puts_char;
916
917 /* Need to insert a '\n' between two files because first file's
918 * last line wasn't terminated? */
919 if (lpc != '\n' && lpc != '\0') {
920 fputc('\n', file);
921 lpc = '\n';
922 }
923 fputs(s, file);
924
925 /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
926 if (s[0])
927 lpc = 'x';
928
929 /* had trailing '\0' and it was last char of file? */
930 if (last_gets_char == LAST_IS_NUL) {
931 fputc('\0', file);
932 lpc = 'x'; /* */
933 } else
934 /* had trailing '\n' or '\0'? */
935 if (last_gets_char != NO_EOL_CHAR) {
936 fputc(last_gets_char, file);
937 lpc = last_gets_char;
938 }
939
940 if (ferror(file)) {
941 xfunc_error_retval = 4; /* It's what gnu sed exits with... */
942 bb_error_msg_and_die(bb_msg_write_error);
943 }
944 *last_puts_char = lpc;
945}
946
947static void flush_append(char *last_puts_char, char last_gets_char)
948{
949 char *data;
950
951 /* Output appended lines. */
952 while ((data = (char *)llist_pop(&G.append_head))) {
953 puts_maybe_newline(data, G.nonstdout, last_puts_char, last_gets_char);
954 free(data);
955 }
956}
957
958/* Get next line of input from G.input_file_list, flushing append buffer and
959 * noting if we ran out of files without a newline on the last line we read.
960 */
961static char *get_next_line(char *gets_char, char *last_puts_char, char last_gets_char)
Rob Landleydcc28662004-11-25 07:21:47 +0000962{
Denis Vlasenko80778502006-10-25 12:46:46 +0000963 char *temp = NULL;
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000964 int len;
965 char gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000966
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100967 flush_append(last_puts_char, last_gets_char);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000968
969 /* will be returned if last line in the file
970 * doesn't end with either '\n' or '\0' */
971 gc = NO_EOL_CHAR;
Denys Vlasenkodeb07692013-11-28 12:08:51 +0100972 for (; G.current_input_file <= G.last_input_file; G.current_input_file++) {
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100973 FILE *fp = G.current_fp;
974 if (!fp) {
975 const char *path = G.input_file_list[G.current_input_file];
976 fp = stdin;
977 if (path != bb_msg_standard_input) {
978 fp = fopen_or_warn(path, "r");
979 if (!fp) {
980 G.exitcode = EXIT_FAILURE;
981 continue;
982 }
983 }
984 G.current_fp = fp;
985 }
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000986 /* Read line up to a newline or NUL byte, inclusive,
987 * return malloc'ed char[]. length of the chunk read
988 * is stored in len. NULL if EOF/error */
Denis Vlasenko86811802007-01-29 17:10:19 +0000989 temp = bb_get_chunk_from_file(fp, &len);
Rob Landley2b26fd52006-02-24 02:30:39 +0000990 if (temp) {
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000991 /* len > 0 here, it's ok to do temp[len-1] */
992 char c = temp[len-1];
993 if (c == '\n' || c == '\0') {
994 temp[len-1] = '\0';
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000995 gc = c;
Denis Vlasenko86811802007-01-29 17:10:19 +0000996 if (c == '\0') {
997 int ch = fgetc(fp);
998 if (ch != EOF)
999 ungetc(ch, fp);
1000 else
1001 gc = LAST_IS_NUL;
1002 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +00001003 }
Denis Vlasenkoc562bb72007-01-29 17:08:51 +00001004 /* else we put NO_EOL_CHAR into *gets_char */
1005 break;
1006
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001007 /* NB: I had the idea of peeking next file(s) and returning
1008 * NO_EOL_CHAR only if it is the *last* non-empty
1009 * input file. But there is a case where this won't work:
1010 * file1: "a woo\nb woo"
1011 * file2: "c no\nd no"
1012 * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
1013 * (note: *no* newline after "b bang"!) */
Denis Vlasenko8b22b072006-12-02 17:58:10 +00001014 }
1015 /* Close this file and advance to next one */
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001016 fclose_if_not_stdin(fp);
1017 G.current_fp = NULL;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001018 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001019 *gets_char = gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001020 return temp;
1021}
1022
Denis Vlasenko17a15262007-03-26 20:48:46 +00001023#define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001024
Denis Vlasenko8274e062007-08-06 03:41:08 +00001025static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
1026{
1027 int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
1028 if (retval)
1029 G.previous_regex_ptr = sed_cmd->beg_match;
1030 return retval;
1031}
1032
Rob Landley2b26fd52006-02-24 02:30:39 +00001033/* Process all the lines in all the files */
1034
Rob Landleydcc28662004-11-25 07:21:47 +00001035static void process_files(void)
Mark Whitley6315ce62000-07-10 22:55:51 +00001036{
Rob Landleyce4f0e92004-10-30 06:54:19 +00001037 char *pattern_space, *next_line;
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001038 int linenum = 0;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001039 char last_puts_char = '\n';
1040 char last_gets_char, next_gets_char;
Denis Vlasenko2ea630f2006-12-10 02:52:19 +00001041 sed_cmd_t *sed_cmd;
1042 int substituted;
Mark Whitley6315ce62000-07-10 22:55:51 +00001043
Rob Landley2b26fd52006-02-24 02:30:39 +00001044 /* Prime the pump */
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001045 next_line = get_next_line(&next_gets_char, &last_puts_char, '\n' /*last_gets_char*/);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001046
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001047 /* Go through every line in each file */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +00001048 again:
Denis Vlasenko2ea630f2006-12-10 02:52:19 +00001049 substituted = 0;
Glenn L McGrath505bd0f2003-03-08 05:21:02 +00001050
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001051 /* Advance to next line. Stop if out of lines. */
1052 pattern_space = next_line;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001053 if (!pattern_space)
1054 return;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001055 last_gets_char = next_gets_char;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001056
1057 /* Read one line in advance so we can act on the last line,
1058 * the '$' address */
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001059 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001060 linenum++;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001061
1062 /* For every line, go through all the commands */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +00001063 restart:
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001064 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001065 int old_matched, matched;
Eric Andersen52a3c272003-12-23 08:53:51 +00001066
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001067 old_matched = sed_cmd->in_match;
Mark Whitley0915c4b2001-06-11 23:50:06 +00001068
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001069 /* Determine if this command matches this line: */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001070
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001071 dbg("match1:%d", sed_cmd->in_match);
1072 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1073 && !sed_cmd->beg_match && !sed_cmd->end_match));
1074 dbg("match3:%d", (sed_cmd->beg_line > 0
1075 && (sed_cmd->end_line || sed_cmd->end_match
1076 ? (sed_cmd->beg_line <= linenum)
1077 : (sed_cmd->beg_line == linenum)
1078 )
1079 ));
1080 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1081 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001082
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001083 /* Are we continuing a previous multi-line match? */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001084 sed_cmd->in_match = sed_cmd->in_match
1085 /* Or is no range necessary? */
1086 || (!sed_cmd->beg_line && !sed_cmd->end_line
1087 && !sed_cmd->beg_match && !sed_cmd->end_match)
1088 /* Or did we match the start of a numerical range? */
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001089 || (sed_cmd->beg_line > 0
1090 && (sed_cmd->end_line || sed_cmd->end_match
Denys Vlasenkoc4679242010-06-04 01:31:48 +02001091 /* note: even if end is numeric and is < linenum too,
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001092 * GNU sed matches! We match too, therefore we don't
1093 * check here that linenum <= end.
1094 * Example:
1095 * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1096 * first three input lines are deleted;
1097 * 4th line is matched and printed
1098 * by "2,3" (!) and by "4" ranges
1099 */
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001100 ? (sed_cmd->beg_line <= linenum) /* N,end */
1101 : (sed_cmd->beg_line == linenum) /* N */
1102 )
1103 )
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001104 /* Or does this line match our begin address regex? */
Denis Vlasenko8274e062007-08-06 03:41:08 +00001105 || (beg_match(sed_cmd, pattern_space))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001106 /* Or did we match last line of input? */
1107 || (sed_cmd->beg_line == -1 && next_line == NULL);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001108
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001109 /* Snapshot the value */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001110 matched = sed_cmd->in_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001111
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001112 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1113 sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001114
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001115 /* Is this line the end of the current match? */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001116
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001117 if (matched) {
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001118 if (sed_cmd->end_line <= -2) {
1119 /* address2 is +N, i.e. N lines from beg_line */
1120 sed_cmd->end_line = linenum + (-sed_cmd->end_line - 2);
1121 }
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001122 /* once matched, "n,xxx" range is dead, disabling it */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001123 if (sed_cmd->beg_line > 0) {
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001124 sed_cmd->beg_line = -2;
Denys Vlasenko2e284a42010-08-01 04:14:46 +02001125 }
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001126 dbg("end1:%d", sed_cmd->end_line ? sed_cmd->end_line == -1
1127 ? !next_line : (sed_cmd->end_line <= linenum)
1128 : !sed_cmd->end_match);
1129 dbg("end2:%d", sed_cmd->end_match && old_matched
1130 && !regexec(sed_cmd->end_match,pattern_space, 0, NULL, 0));
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001131 sed_cmd->in_match = !(
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001132 /* has the ending line come, or is this a single address command? */
Denys Vlasenko52d83702011-05-03 00:51:43 +02001133 (sed_cmd->end_line
1134 ? sed_cmd->end_line == -1
1135 ? !next_line
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001136 : (sed_cmd->end_line <= linenum)
1137 : !sed_cmd->end_match
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001138 )
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001139 /* or does this line matches our last address regex */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001140 || (sed_cmd->end_match && old_matched
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001141 && (regexec(sed_cmd->end_match,
Denys Vlasenko6830ade2013-01-15 13:58:01 +01001142 pattern_space, 0, NULL, 0) == 0)
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001143 )
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001144 );
Erik Andersene49d5ec2000-02-08 19:58:47 +00001145 }
Erik Andersen1266a131999-12-29 22:19:46 +00001146
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001147 /* Skip blocks of commands we didn't match */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001148 if (sed_cmd->cmd == '{') {
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001149 if (sed_cmd->invert ? matched : !matched) {
Denys Vlasenkof2c16ed2010-04-20 04:00:03 -04001150 unsigned nest_cnt = 0;
1151 while (1) {
1152 if (sed_cmd->cmd == '{')
1153 nest_cnt++;
1154 if (sed_cmd->cmd == '}') {
1155 nest_cnt--;
1156 if (nest_cnt == 0)
1157 break;
1158 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001159 sed_cmd = sed_cmd->next;
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001160 if (!sed_cmd)
1161 bb_error_msg_and_die("unterminated {");
1162 }
1163 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001164 continue;
1165 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001166
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001167 /* Okay, so did this line match? */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001168 if (sed_cmd->invert ? matched : !matched)
1169 continue; /* no */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001170
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001171 /* Update last used regex in case a blank substitute BRE is found */
1172 if (sed_cmd->beg_match) {
1173 G.previous_regex_ptr = sed_cmd->beg_match;
1174 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001175
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001176 /* actual sedding */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +01001177 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1178 pattern_space, next_line, sed_cmd->cmd);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001179 switch (sed_cmd->cmd) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001180
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001181 /* Print line number */
1182 case '=':
1183 fprintf(G.nonstdout, "%d\n", linenum);
1184 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001185
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001186 /* Write the current pattern space up to the first newline */
1187 case 'P':
1188 {
1189 char *tmp = strchr(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001190 if (tmp) {
1191 *tmp = '\0';
1192 /* TODO: explain why '\n' below */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001193 sed_puts(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001194 *tmp = '\n';
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001195 break;
1196 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001197 /* Fall Through */
1198 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001199
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001200 /* Write the current pattern space to output */
1201 case 'p':
1202 /* NB: we print this _before_ the last line
1203 * (of current file) is printed. Even if
1204 * that line is nonterminated, we print
1205 * '\n' here (gnu sed does the same) */
1206 sed_puts(pattern_space, '\n');
1207 break;
1208 /* Delete up through first newline */
1209 case 'D':
1210 {
1211 char *tmp = strchr(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001212 if (tmp) {
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001213 overlapping_strcpy(pattern_space, tmp + 1);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001214 goto restart;
1215 }
1216 }
1217 /* discard this line. */
1218 case 'd':
1219 goto discard_line;
1220
1221 /* Substitute with regex */
1222 case 's':
1223 if (!do_subst_command(sed_cmd, &pattern_space))
1224 break;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +02001225 dbg("do_subst_command succeeded:'%s'", pattern_space);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001226 substituted |= 1;
1227
1228 /* handle p option */
1229 if (sed_cmd->sub_p)
1230 sed_puts(pattern_space, last_gets_char);
1231 /* handle w option */
1232 if (sed_cmd->sw_file)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001233 puts_maybe_newline(
1234 pattern_space, sed_cmd->sw_file,
1235 &sed_cmd->sw_last_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001236 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001237
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001238 /* Append line to linked list to be printed later */
1239 case 'a':
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +01001240 append(xstrdup(sed_cmd->string));
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001241 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001242
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001243 /* Insert text before this line */
1244 case 'i':
1245 sed_puts(sed_cmd->string, '\n');
1246 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001247
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001248 /* Cut and paste text (replace) */
1249 case 'c':
1250 /* Only triggers on last line of a matching range. */
1251 if (!sed_cmd->in_match)
Denys Vlasenko96a18332010-04-19 22:36:07 -04001252 sed_puts(sed_cmd->string, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001253 goto discard_line;
1254
1255 /* Read file, append contents to output */
1256 case 'r':
1257 {
1258 FILE *rfile;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001259 rfile = fopen_for_read(sed_cmd->string);
1260 if (rfile) {
1261 char *line;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001262 while ((line = xmalloc_fgetline(rfile))
1263 != NULL)
1264 append(line);
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +01001265 fclose(rfile);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001266 }
1267
1268 break;
1269 }
1270
1271 /* Write pattern space to file. */
1272 case 'w':
1273 puts_maybe_newline(
1274 pattern_space, sed_cmd->sw_file,
1275 &sed_cmd->sw_last_char, last_gets_char);
1276 break;
1277
1278 /* Read next line from input */
1279 case 'n':
1280 if (!G.be_quiet)
1281 sed_puts(pattern_space, last_gets_char);
1282 if (next_line) {
1283 free(pattern_space);
1284 pattern_space = next_line;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001285 last_gets_char = next_gets_char;
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001286 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001287 substituted = 0;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001288 linenum++;
1289 break;
1290 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001291 /* fall through */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001292
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001293 /* Quit. End of script, end of input. */
1294 case 'q':
1295 /* Exit the outer while loop */
1296 free(next_line);
1297 next_line = NULL;
1298 goto discard_commands;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001299
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001300 /* Append the next line to the current line */
1301 case 'N':
1302 {
1303 int len;
1304 /* If no next line, jump to end of script and exit. */
Denys Vlasenko0d555fc2010-08-16 16:26:33 +02001305 /* http://www.gnu.org/software/sed/manual/sed.html:
1306 * "Most versions of sed exit without printing anything
1307 * when the N command is issued on the last line of
1308 * a file. GNU sed prints pattern space before exiting
1309 * unless of course the -n command switch has been
1310 * specified. This choice is by design."
1311 */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001312 if (next_line == NULL) {
Denys Vlasenko0d555fc2010-08-16 16:26:33 +02001313 //goto discard_line;
1314 goto discard_commands; /* GNU behavior */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001315 }
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001316 /* Append next_line, read new next_line. */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001317 len = strlen(pattern_space);
Denys Vlasenko90377872010-01-08 09:07:50 +01001318 pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001319 pattern_space[len] = '\n';
1320 strcpy(pattern_space + len+1, next_line);
1321 last_gets_char = next_gets_char;
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001322 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001323 linenum++;
1324 break;
1325 }
1326
1327 /* Test/branch if substitution occurred */
1328 case 't':
1329 if (!substituted) break;
1330 substituted = 0;
1331 /* Fall through */
1332 /* Test/branch if substitution didn't occur */
1333 case 'T':
1334 if (substituted) break;
1335 /* Fall through */
1336 /* Branch to label */
1337 case 'b':
1338 if (!sed_cmd->string) goto discard_commands;
1339 else sed_cmd = branch_to(sed_cmd->string);
1340 break;
1341 /* Transliterate characters */
1342 case 'y':
1343 {
1344 int i, j;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001345 for (i = 0; pattern_space[i]; i++) {
1346 for (j = 0; sed_cmd->string[j]; j += 2) {
1347 if (pattern_space[i] == sed_cmd->string[j]) {
1348 pattern_space[i] = sed_cmd->string[j + 1];
1349 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001350 }
1351 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001352 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001353
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001354 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001355 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001356 case 'g': /* Replace pattern space with hold space */
1357 free(pattern_space);
1358 pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1359 break;
1360 case 'G': /* Append newline and hold space to pattern space */
1361 {
1362 int pattern_space_size = 2;
1363 int hold_space_size = 0;
1364
1365 if (pattern_space)
1366 pattern_space_size += strlen(pattern_space);
1367 if (G.hold_space)
1368 hold_space_size = strlen(G.hold_space);
1369 pattern_space = xrealloc(pattern_space,
1370 pattern_space_size + hold_space_size);
1371 if (pattern_space_size == 2)
1372 pattern_space[0] = 0;
1373 strcat(pattern_space, "\n");
1374 if (G.hold_space)
1375 strcat(pattern_space, G.hold_space);
1376 last_gets_char = '\n';
1377
1378 break;
1379 }
1380 case 'h': /* Replace hold space with pattern space */
1381 free(G.hold_space);
1382 G.hold_space = xstrdup(pattern_space);
1383 break;
1384 case 'H': /* Append newline and pattern space to hold space */
1385 {
1386 int hold_space_size = 2;
1387 int pattern_space_size = 0;
1388
1389 if (G.hold_space)
1390 hold_space_size += strlen(G.hold_space);
1391 if (pattern_space)
1392 pattern_space_size = strlen(pattern_space);
1393 G.hold_space = xrealloc(G.hold_space,
1394 hold_space_size + pattern_space_size);
1395
1396 if (hold_space_size == 2)
1397 *G.hold_space = 0;
1398 strcat(G.hold_space, "\n");
1399 if (pattern_space)
1400 strcat(G.hold_space, pattern_space);
1401
1402 break;
1403 }
1404 case 'x': /* Exchange hold and pattern space */
1405 {
1406 char *tmp = pattern_space;
Denys Vlasenko90a99042009-09-06 02:36:23 +02001407 pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001408 last_gets_char = '\n';
1409 G.hold_space = tmp;
1410 break;
1411 }
1412 } /* switch */
1413 } /* for each cmd */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001414
1415 /*
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001416 * Exit point from sedding...
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001417 */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001418 discard_commands:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001419 /* we will print the line unless we were told to be quiet ('-n')
1420 or if the line was suppressed (ala 'd'elete) */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001421 if (!G.be_quiet)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001422 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001423
1424 /* Delete and such jump here. */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001425 discard_line:
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001426 flush_append(&last_puts_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001427 free(pattern_space);
1428
1429 goto again;
Erik Andersen1266a131999-12-29 22:19:46 +00001430}
1431
Glenn L McGrath42c25732003-10-04 05:27:56 +00001432/* It is possible to have a command line argument with embedded
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001433 * newlines. This counts as multiple command lines.
1434 * However, newline can be escaped: 's/e/z\<newline>z/'
Denys Vlasenko40ab27a2013-07-08 02:04:44 +02001435 * add_cmd() handles this.
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001436 */
Glenn L McGrath42c25732003-10-04 05:27:56 +00001437
1438static void add_cmd_block(char *cmdstr)
1439{
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001440 char *sv, *eol;
Glenn L McGrath42c25732003-10-04 05:27:56 +00001441
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001442 cmdstr = sv = xstrdup(cmdstr);
1443 do {
1444 eol = strchr(cmdstr, '\n');
Denys Vlasenko40ab27a2013-07-08 02:04:44 +02001445 if (eol)
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001446 *eol = '\0';
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001447 add_cmd(cmdstr);
1448 cmdstr = eol + 1;
1449 } while (eol);
1450 free(sv);
Glenn L McGrath42c25732003-10-04 05:27:56 +00001451}
1452
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +00001453int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +00001454int sed_main(int argc UNUSED_PARAM, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001455{
Denis Vlasenko67b23e62006-10-03 21:00:06 +00001456 unsigned opt;
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001457 llist_t *opt_e, *opt_f;
Simon B8c343952012-05-06 13:59:15 +02001458 char *opt_i;
1459
1460#if ENABLE_LONG_OPTS
1461 static const char sed_longopts[] ALIGN1 =
1462 /* name has_arg short */
1463 "in-place\0" Optional_argument "i"
1464 "regexp-extended\0" No_argument "r"
1465 "quiet\0" No_argument "n"
1466 "silent\0" No_argument "n"
1467 "expression\0" Required_argument "e"
1468 "file\0" Required_argument "f";
1469#endif
1470
Denis Vlasenko74324c82007-06-04 10:16:52 +00001471 INIT_G();
Rob Landleye3f5a3f2006-05-09 03:53:55 +00001472
Mark Whitley858c1ad2000-07-11 21:38:47 +00001473 /* destroy command strings on exit */
Rob Landley0b656282006-05-08 22:17:23 +00001474 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
Mark Whitley858c1ad2000-07-11 21:38:47 +00001475
Rob Landleyfae1dc82005-11-20 07:44:35 +00001476 /* Lie to autoconf when it starts asking stupid questions. */
Simon B8c343952012-05-06 13:59:15 +02001477 if (argv[1] && strcmp(argv[1], "--version") == 0) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001478 puts("This is not GNU sed version 4.0");
1479 return 0;
Eric Andersenc06f5682004-02-04 10:57:46 +00001480 }
Eric Andersenc06f5682004-02-04 10:57:46 +00001481
Mark Whitley6315ce62000-07-10 22:55:51 +00001482 /* do normal option parsing */
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001483 opt_e = opt_f = NULL;
Simon B8c343952012-05-06 13:59:15 +02001484 opt_i = NULL;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001485 opt_complementary = "e::f::" /* can occur multiple times */
1486 "nn"; /* count -n */
Simon B8c343952012-05-06 13:59:15 +02001487
1488 IF_LONG_OPTS(applet_long_options = sed_longopts);
1489
Denys Vlasenko2e284a42010-08-01 04:14:46 +02001490 /* -i must be first, to match OPT_in_place definition */
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001491 /* -E is a synonym of -r:
1492 * GNU sed 4.2.1 mentions it in neither --help
1493 * nor manpage, but does recognize it.
1494 */
1495 opt = getopt32(argv, "i::rEne:f:", &opt_i, &opt_e, &opt_f,
Denis Vlasenko17a15262007-03-26 20:48:46 +00001496 &G.be_quiet); /* counter for -n */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001497 //argc -= optind;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001498 argv += optind;
1499 if (opt & OPT_in_place) { // -i
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001500 atexit(cleanup_outname);
1501 }
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001502 if (opt & (2|4))
1503 G.regex_type |= REG_EXTENDED; // -r or -E
1504 //if (opt & 8)
1505 // G.be_quiet++; // -n (implemented with a counter instead)
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001506 while (opt_e) { // -e
Denis Vlasenkod50dda82008-06-15 05:40:56 +00001507 add_cmd_block(llist_pop(&opt_e));
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001508 }
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001509 while (opt_f) { // -f
1510 char *line;
1511 FILE *cmdfile;
Denys Vlasenkoe4d925b2016-04-08 00:20:36 +02001512 cmdfile = xfopen_stdin(llist_pop(&opt_f));
Denis Vlasenko8ee649a2008-03-26 20:04:27 +00001513 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001514 add_cmd(line);
1515 free(line);
1516 }
Denys Vlasenkoe4d925b2016-04-08 00:20:36 +02001517 fclose_if_not_stdin(cmdfile);
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001518 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001519 /* if we didn't get a pattern from -e or -f, use argv[0] */
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001520 if (!(opt & 0x30)) {
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001521 if (!*argv)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001522 bb_show_usage();
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001523 add_cmd_block(*argv++);
Mark Whitley6315ce62000-07-10 22:55:51 +00001524 }
Glenn L McGrath42c25732003-10-04 05:27:56 +00001525 /* Flush any unfinished commands. */
1526 add_cmd("");
Mark Whitley6315ce62000-07-10 22:55:51 +00001527
Rob Landley53302f82004-02-18 09:54:15 +00001528 /* By default, we write to stdout */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001529 G.nonstdout = stdout;
Rob Landley53302f82004-02-18 09:54:15 +00001530
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001531 /* argv[0..(argc-1)] should be names of file to process. If no
Mark Whitley6315ce62000-07-10 22:55:51 +00001532 * files were specified or '-' was specified, take input from stdin.
1533 * Otherwise, we process all the files specified. */
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001534 G.input_file_list = argv;
1535 if (!argv[0]) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001536 if (opt & OPT_in_place)
Denis Vlasenko80778502006-10-25 12:46:46 +00001537 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001538 argv[0] = (char*)bb_msg_standard_input;
1539 /* G.last_input_file = 0; - already is */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001540 } else {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001541 goto start;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001542
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001543 for (; *argv; argv++) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001544 struct stat statbuf;
1545 int nonstdoutfd;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001546 sed_cmd_t *sed_cmd;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001547
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001548 G.last_input_file++;
1549 start:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001550 if (!(opt & OPT_in_place)) {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001551 if (LONE_DASH(*argv)) {
1552 *argv = (char*)bb_msg_standard_input;
1553 process_files();
1554 }
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001555 continue;
1556 }
1557
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001558 /* -i: process each FILE separately: */
1559
Denys Vlasenkocd738712014-10-05 02:44:34 +02001560 if (stat(*argv, &statbuf) != 0) {
1561 bb_simple_perror_msg(*argv);
1562 G.exitcode = EXIT_FAILURE;
1563 G.current_input_file++;
1564 continue;
1565 }
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001566 G.outname = xasprintf("%sXXXXXX", *argv);
Alexander Shishkin67227372010-10-22 13:27:16 +02001567 nonstdoutfd = xmkstemp(G.outname);
Denys Vlasenkoa7ccdee2009-11-15 23:28:11 +01001568 G.nonstdout = xfdopen_for_write(nonstdoutfd);
Denys Vlasenko57992482009-11-13 09:09:07 +01001569 /* Set permissions/owner of output file */
Denys Vlasenko3b727cc2010-06-18 02:12:56 +02001570 /* chmod'ing AFTER chown would preserve suid/sgid bits,
1571 * but GNU sed 4.2.1 does not preserve them either */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001572 fchmod(nonstdoutfd, statbuf.st_mode);
Denys Vlasenko57992482009-11-13 09:09:07 +01001573 fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001574
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001575 process_files();
Denis Vlasenko17a15262007-03-26 20:48:46 +00001576 fclose(G.nonstdout);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001577 G.nonstdout = stdout;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001578
Simon B8c343952012-05-06 13:59:15 +02001579 if (opt_i) {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001580 char *backupname = xasprintf("%s%s", *argv, opt_i);
1581 xrename(*argv, backupname);
Simon B8c343952012-05-06 13:59:15 +02001582 free(backupname);
1583 }
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001584 /* else unlink(*argv); - rename below does this */
1585 xrename(G.outname, *argv); //TODO: rollback backup on error?
Denis Vlasenko17a15262007-03-26 20:48:46 +00001586 free(G.outname);
Denis Vlasenko40276642007-11-13 16:48:10 +00001587 G.outname = NULL;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001588
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001589 /* Fix disabled range matches and mangled ",+N" ranges */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001590 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1591 sed_cmd->beg_line = sed_cmd->beg_line_orig;
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001592 sed_cmd->end_line = sed_cmd->end_line_orig;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001593 }
Mark Whitley6315ce62000-07-10 22:55:51 +00001594 }
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001595 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001596 * if (G.current_input_file[G.current_input_file] == NULL)
1597 * return G.exitcode;
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001598 * but it's not needed since process_files() works correctly
1599 * in this case too. */
Mark Whitley6315ce62000-07-10 22:55:51 +00001600 }
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001601
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001602 process_files();
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001603
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001604 return G.exitcode;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001605}