blob: 9c4c8e1486f78a525b0162c1edf9c2de931318a5 [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"
"Vladimir N. Oleynik"23f62fc2005-09-14 16:59:11 +000089#include "xregex.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000090
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +020091#if 0
92# define dbg(...) bb_error_msg(__VA_ARGS__)
93#else
94# define dbg(...) ((void)0)
95#endif
96
97
Denys Vlasenko2e284a42010-08-01 04:14:46 +020098enum {
99 OPT_in_place = 1 << 0,
100};
101
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000102/* Each sed command turns into one of these structures. */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +0000103typedef struct sed_cmd_s {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000104 /* Ordered by alignment requirements: currently 36 bytes on x86 */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000105 struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
Mark Whitley2dc192f2000-11-03 19:47:00 +0000106
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000107 /* address storage */
108 regex_t *beg_match; /* sed -e '/match/cmd' */
109 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
110 regex_t *sub_match; /* For 's/sub_match/string/' */
111 int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200112 int beg_line_orig; /* copy of the above, needed for -i */
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200113 int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($). -2-N = +N */
114 int end_line_orig;
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +0000115
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100116 FILE *sw_file; /* File (sw) command writes to, NULL for none. */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000117 char *string; /* Data string for (saicytb) commands. */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000118
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000119 unsigned which_match; /* (s) Which match to replace (0 for all) */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +0000120
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000121 /* Bitfields (gcc won't group them if we don't) */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000122 unsigned invert:1; /* the '!' after the address */
123 unsigned in_match:1; /* Next line also included in match? */
124 unsigned sub_p:1; /* (s) print option */
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000125
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +0000126 char sw_last_char; /* Last line written by (sw) had no '\n' */
Glenn L McGrathc18ce372003-09-13 06:57:39 +0000127
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000128 /* GENERAL FIELDS */
129 char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +0000130} sed_cmd_t;
Mark Whitley6315ce62000-07-10 22:55:51 +0000131
Denis Vlasenko6ca409e2007-08-12 20:58:27 +0000132static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000133
Denis Vlasenko17a15262007-03-26 20:48:46 +0000134struct globals {
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000135 /* options */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000136 int be_quiet, regex_type;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100137
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000138 FILE *nonstdout;
139 char *outname, *hold_space;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100140 smallint exitcode;
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000141
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100142 /* list of input files */
Denys Vlasenkodeb07692013-11-28 12:08:51 +0100143 int current_input_file, last_input_file;
144 char **input_file_list;
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100145 FILE *current_fp;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000146
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000147 regmatch_t regmatch[10];
148 regex_t *previous_regex_ptr;
Denis Vlasenko9213a9e2006-09-17 16:28:10 +0000149
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000150 /* linked list of sed commands */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200151 sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000152
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100153 /* linked list of append lines */
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000154 llist_t *append_head;
155
156 char *add_cmd_line;
157
158 struct pipeline {
Denys Vlasenkofb132e42010-10-29 11:46:52 +0200159 char *buf; /* Space to hold string */
160 int idx; /* Space used */
161 int len; /* Space allocated */
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000162 } pipeline;
Denys Vlasenko98a4c7c2010-02-04 15:00:15 +0100163} FIX_ALIASING;
Denis Vlasenko17a15262007-03-26 20:48:46 +0000164#define G (*(struct globals*)&bb_common_bufsiz1)
Denis Vlasenko74324c82007-06-04 10:16:52 +0000165#define INIT_G() do { \
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200166 BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
Denis Vlasenko74324c82007-06-04 10:16:52 +0000167 G.sed_cmd_tail = &G.sed_cmd_head; \
168} while (0)
169
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000170
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +0000171#if ENABLE_FEATURE_CLEAN_UP
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000172static void sed_free_and_close_stuff(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000173{
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200174 sed_cmd_t *sed_cmd = G.sed_cmd_head;
Mark Whitley6315ce62000-07-10 22:55:51 +0000175
Denis Vlasenko17a15262007-03-26 20:48:46 +0000176 llist_free(G.append_head, free);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000177
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000178 while (sed_cmd) {
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000179 sed_cmd_t *sed_cmd_next = sed_cmd->next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000180
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000181 if (sed_cmd->sw_file)
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100182 fclose(sed_cmd->sw_file);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000183
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000184 if (sed_cmd->beg_match) {
185 regfree(sed_cmd->beg_match);
186 free(sed_cmd->beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000187 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000188 if (sed_cmd->end_match) {
189 regfree(sed_cmd->end_match);
190 free(sed_cmd->end_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000191 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000192 if (sed_cmd->sub_match) {
193 regfree(sed_cmd->sub_match);
194 free(sed_cmd->sub_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000195 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000196 free(sed_cmd->string);
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000197 free(sed_cmd);
198 sed_cmd = sed_cmd_next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000199 }
Rob Landleyce4f0e92004-10-30 06:54:19 +0000200
Denis Vlasenko60818682007-09-28 22:07:23 +0000201 free(G.hold_space);
Rob Landleydcc28662004-11-25 07:21:47 +0000202
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100203 if (G.current_fp)
204 fclose(G.current_fp);
Mark Whitley6315ce62000-07-10 22:55:51 +0000205}
Denis Vlasenko666da5e2006-12-26 18:17:42 +0000206#else
207void sed_free_and_close_stuff(void);
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000208#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000209
Rob Landley53302f82004-02-18 09:54:15 +0000210/* If something bad happens during -i operation, delete temp file */
211
212static void cleanup_outname(void)
213{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000214 if (G.outname) unlink(G.outname);
Rob Landley53302f82004-02-18 09:54:15 +0000215}
216
Denis Vlasenko40276642007-11-13 16:48:10 +0000217/* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000218
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000219static void parse_escapes(char *dest, const char *string, int len, char from, char to)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000220{
Denis Vlasenko80778502006-10-25 12:46:46 +0000221 int i = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000222
Denis Vlasenko80778502006-10-25 12:46:46 +0000223 while (i < len) {
224 if (string[i] == '\\') {
225 if (!to || string[i+1] == from) {
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000226 *dest++ = to ? to : string[i+1];
Denis Vlasenko80778502006-10-25 12:46:46 +0000227 i += 2;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000228 continue;
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000229 }
230 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000231 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000232 /* TODO: is it safe wrt a string with trailing '\\' ? */
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000233 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000234 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000235 *dest = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000236}
237
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000238static char *copy_parsing_escapes(const char *string, int len)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000239{
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200240 const char *s;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000241 char *dest = xmalloc(len + 1);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000242
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200243 /* sed recognizes \n */
Denys Vlasenko6a0abcc2011-05-03 00:52:22 +0200244 /* GNU sed also recognizes \t and \r */
Pascal Bellardd3e4be32011-05-05 00:26:37 +0200245 for (s = "\nn\tt\rr"; *s; s += 2) {
246 parse_escapes(dest, string, len, s[1], s[0]);
247 string = dest;
248 len = strlen(dest);
249 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000250 return dest;
251}
252
253
Mark Whitley6315ce62000-07-10 22:55:51 +0000254/*
Eric Andersen28b3c532001-01-02 11:01:31 +0000255 * index_of_next_unescaped_regexp_delim - walks left to right through a string
256 * beginning at a specified index and returns the index of the next regular
Denis Vlasenko40276642007-11-13 16:48:10 +0000257 * expression delimiter (typically a forward slash ('/')) not preceded by
Rob Landley4795e4e2006-07-26 17:25:08 +0000258 * a backslash ('\'). A negative delimiter disables square bracket checking.
Mark Whitley6315ce62000-07-10 22:55:51 +0000259 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000260static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
Mark Whitley6315ce62000-07-10 22:55:51 +0000261{
Matt Kraaia0065d52001-08-24 14:45:50 +0000262 int bracket = -1;
263 int escaped = 0;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000264 int idx = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000265 char ch;
Matt Kraaia0065d52001-08-24 14:45:50 +0000266
Rob Landley4795e4e2006-07-26 17:25:08 +0000267 if (delimiter < 0) {
268 bracket--;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000269 delimiter = -delimiter;
Rob Landley4795e4e2006-07-26 17:25:08 +0000270 }
271
Denys Vlasenko52d83702011-05-03 00:51:43 +0200272 for (; (ch = str[idx]) != '\0'; idx++) {
Rob Landley4795e4e2006-07-26 17:25:08 +0000273 if (bracket >= 0) {
Denys Vlasenko52d83702011-05-03 00:51:43 +0200274 if (ch == ']'
275 && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
276 ) {
Matt Kraaia0065d52001-08-24 14:45:50 +0000277 bracket = -1;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200278 }
Matt Kraaia0065d52001-08-24 14:45:50 +0000279 } else if (escaped)
280 escaped = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000281 else if (ch == '\\')
Matt Kraaia0065d52001-08-24 14:45:50 +0000282 escaped = 1;
Rob Landley4795e4e2006-07-26 17:25:08 +0000283 else if (bracket == -1 && ch == '[')
Matt Kraaia0065d52001-08-24 14:45:50 +0000284 bracket = idx;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000285 else if (ch == delimiter)
Mark Whitley97562bd2000-07-17 20:06:42 +0000286 return idx;
287 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000288
Mark Whitley97562bd2000-07-17 20:06:42 +0000289 /* if we make it to here, we've hit the end of the string */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000290 bb_error_msg_and_die("unmatched '%c'", delimiter);
Mark Whitley6315ce62000-07-10 22:55:51 +0000291}
292
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000293/*
294 * Returns the index of the third delimiter
295 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000296static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000297{
Denis Vlasenko9356b502007-01-29 23:44:38 +0000298 const char *cmdstr_ptr = cmdstr;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100299 unsigned char delimiter;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000300 int idx = 0;
301
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000302 /* verify that the 's' or 'y' is followed by something. That something
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000303 * (typically a 'slash') is now our regexp delimiter... */
Rob Landley4795e4e2006-07-26 17:25:08 +0000304 if (*cmdstr == '\0')
305 bb_error_msg_and_die("bad format in substitution expression");
Denis Vlasenko80778502006-10-25 12:46:46 +0000306 delimiter = *cmdstr_ptr++;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000307
308 /* save the match string */
309 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000310 *match = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000311
312 /* save the replacement string */
313 cmdstr_ptr += idx + 1;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100314 idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000315 *replace = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000316
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000317 return ((cmdstr_ptr - cmdstr) + idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000318}
319
Mark Whitley6315ce62000-07-10 22:55:51 +0000320/*
321 * returns the index in the string just past where the address ends.
322 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000323static int get_address(const char *my_str, int *linenum, regex_t ** regex)
Mark Whitley6315ce62000-07-10 22:55:51 +0000324{
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000325 const char *pos = my_str;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000326
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000327 if (isdigit(*my_str)) {
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000328 *linenum = strtol(my_str, (char**)&pos, 10);
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000329 /* endstr shouldnt ever equal NULL */
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000330 } else if (*my_str == '$') {
Mark Whitley0915c4b2001-06-11 23:50:06 +0000331 *linenum = -1;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000332 pos++;
333 } else if (*my_str == '/' || *my_str == '\\') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000334 int next;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000335 char delimiter;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000336 char *temp;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000337
Denis Vlasenko80778502006-10-25 12:46:46 +0000338 delimiter = '/';
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100339 if (*my_str == '\\')
340 delimiter = *++pos;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000341 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000342 temp = copy_parsing_escapes(pos, next);
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100343 *regex = xzalloc(sizeof(regex_t));
Denys Vlasenkob0e9b722013-07-21 22:09:44 +0200344 xregcomp(*regex, temp, G.regex_type);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000345 free(temp);
346 /* Move position to next character after last delimiter */
Rob Landley4795e4e2006-07-26 17:25:08 +0000347 pos += (next+1);
Mark Whitley6315ce62000-07-10 22:55:51 +0000348 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000349 return pos - my_str;
Mark Whitley6315ce62000-07-10 22:55:51 +0000350}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000351
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000352/* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
Denis Vlasenko68404f12008-03-17 09:00:54 +0000353static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000354{
Denis Vlasenko80778502006-10-25 12:46:46 +0000355 int start = 0, idx, hack = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000356
357 /* Skip whitespace, then grab filename to end of line */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000358 while (isspace(filecmdstr[start]))
359 start++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000360 idx = start;
Denis Vlasenko9356b502007-01-29 23:44:38 +0000361 while (filecmdstr[idx] && filecmdstr[idx] != '\n')
362 idx++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000363
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000364 /* If lines glued together, put backslash back. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000365 if (filecmdstr[idx] == '\n')
366 hack = 1;
Denis Vlasenko80778502006-10-25 12:46:46 +0000367 if (idx == start)
368 bb_error_msg_and_die("empty filename");
Rob Landleyd921b2e2006-08-03 15:41:12 +0000369 *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
Denis Vlasenko9356b502007-01-29 23:44:38 +0000370 if (hack)
371 (*retval)[idx] = '\\';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000372
373 return idx;
374}
375
Denis Vlasenko9356b502007-01-29 23:44:38 +0000376static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000377{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000378 int cflags = G.regex_type;
Mark Whitley06f35292000-07-13 19:58:04 +0000379 char *match;
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000380 int idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000381
382 /*
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000383 * A substitution command should look something like this:
David A. Wheeler80a068d2013-12-05 20:42:17 -0500384 * s/match/replace/ #giIpw
Mark Whitley0e4cec02000-08-21 21:29:20 +0000385 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000386 * mandatory optional
Mark Whitley06f35292000-07-13 19:58:04 +0000387 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000388 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
Mark Whitley06f35292000-07-13 19:58:04 +0000389
Mark Whitley97562bd2000-07-17 20:06:42 +0000390 /* determine the number of back references in the match string */
391 /* Note: we compute this here rather than in the do_subst_command()
392 * function to save processor time, at the expense of a little more memory
393 * (4 bits) per sed_cmd */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000394
Mark Whitley06f35292000-07-13 19:58:04 +0000395 /* process the flags */
Mark Whitley70705d72000-07-14 19:06:30 +0000396
Denis Vlasenko80778502006-10-25 12:46:46 +0000397 sed_cmd->which_match = 1;
Denys Vlasenko9caea242014-09-16 01:11:13 +0200398 dbg("s flags:'%s'", substr + idx + 1);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000399 while (substr[++idx]) {
Denys Vlasenko9caea242014-09-16 01:11:13 +0200400 dbg("s flag:'%c'", substr[idx]);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000401 /* Parse match number */
Denis Vlasenko80778502006-10-25 12:46:46 +0000402 if (isdigit(substr[idx])) {
403 if (match[0] != '^') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000404 /* Match 0 treated as all, multiple matches we take the last one. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000405 const char *pos = substr + idx;
406/* FIXME: error check? */
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000407 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
Denys Vlasenko9caea242014-09-16 01:11:13 +0200408 idx = pos - substr - 1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000409 }
410 continue;
411 }
Rob Landley40ec4ae2004-01-04 06:42:14 +0000412 /* Skip spaces */
Denys Vlasenko6935ec92009-10-22 19:42:26 +0200413 if (isspace(substr[idx]))
414 continue;
Rob Landley40ec4ae2004-01-04 06:42:14 +0000415
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000416 switch (substr[idx]) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000417 /* Replace all occurrences */
418 case 'g':
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000419 if (match[0] != '^')
420 sed_cmd->which_match = 0;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000421 break;
422 /* Print pattern space */
423 case 'p':
424 sed_cmd->sub_p = 1;
425 break;
426 /* Write to file */
427 case 'w':
428 {
Denys Vlasenko2a4bba32016-01-24 15:52:16 +0100429 char *fname;
430 idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname);
431 sed_cmd->sw_file = xfopen_for_write(fname);
432 sed_cmd->sw_last_char = '\n';
433 free(fname);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000434 break;
435 }
436 /* Ignore case (gnu exension) */
David A. Wheeler80a068d2013-12-05 20:42:17 -0500437 case 'i':
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000438 case 'I':
439 cflags |= REG_ICASE;
440 break;
441 /* Comment */
442 case '#':
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +0200443 // while (substr[++idx]) continue;
444 idx += strlen(substr + idx); // same
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000445 /* Fall through */
446 /* End of command */
447 case ';':
448 case '}':
449 goto out;
450 default:
Denys Vlasenko9caea242014-09-16 01:11:13 +0200451 dbg("s bad flags:'%s'", substr + idx);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000452 bb_error_msg_and_die("bad option in substitution expression");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000453 }
454 }
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +0200455 out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000456 /* compile the match string into a regex */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000457 if (*match != '\0') {
458 /* If match is empty, we use last regex used at runtime */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100459 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
460 dbg("xregcomp('%s',%x)", match, cflags);
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000461 xregcomp(sed_cmd->sub_match, match, cflags);
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100462 dbg("regcomp ok");
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000463 }
Mark Whitley06f35292000-07-13 19:58:04 +0000464 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000465
466 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000467}
468
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000469/*
470 * Process the commands arguments
471 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000472static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000473{
Denys Vlasenko52d83702011-05-03 00:51:43 +0200474 static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
475 enum {
476 IDX_s = 0,
477 IDX_a,
478 IDX_i,
479 IDX_c,
480 IDX_r,
481 IDX_w,
482 IDX_colon,
483 IDX_b,
484 IDX_t,
485 IDX_T,
486 IDX_y,
487 IDX_d,
488 IDX_D,
489 IDX_g,
490 IDX_G,
491 IDX_h,
492 IDX_H,
493 IDX_l,
494 IDX_n,
495 IDX_N,
496 IDX_p,
497 IDX_P,
498 IDX_q,
499 IDX_x,
500 IDX_equal,
501 IDX_lbrace,
502 IDX_rbrace,
503 IDX_nul
504 };
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200505 unsigned idx;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200506
Denys Vlasenko7b85ec32015-10-13 17:17:34 +0200507 BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul);
508
509 idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
Denys Vlasenko52d83702011-05-03 00:51:43 +0200510
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000511 /* handle (s)ubstitution command */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200512 if (idx == IDX_s) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000513 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
Denys Vlasenko52d83702011-05-03 00:51:43 +0200514 }
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000515 /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200516 else if (idx <= IDX_c) { /* a,i,c */
Mimi Li37a79c02012-07-24 13:20:12 +0200517 if (idx < IDX_c) { /* a,i */
518 if (sed_cmd->end_line || sed_cmd->end_match)
519 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
520 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000521 for (;;) {
522 if (*cmdstr == '\n' || *cmdstr == '\\') {
Rob Landley25d82392004-04-01 09:23:30 +0000523 cmdstr++;
524 break;
Denys Vlasenkoc0dab372009-10-22 22:28:08 +0200525 }
526 if (!isspace(*cmdstr))
Denis Vlasenko80778502006-10-25 12:46:46 +0000527 break;
Denys Vlasenkoc0dab372009-10-22 22:28:08 +0200528 cmdstr++;
Rob Landley25d82392004-04-01 09:23:30 +0000529 }
Rob Landleyd921b2e2006-08-03 15:41:12 +0000530 sed_cmd->string = xstrdup(cmdstr);
Denis Vlasenko40276642007-11-13 16:48:10 +0000531 /* "\anychar" -> "anychar" */
532 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000533 cmdstr += strlen(cmdstr);
Denys Vlasenko52d83702011-05-03 00:51:43 +0200534 }
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000535 /* handle file cmds: (r)ead */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200536 else if (idx <= IDX_w) { /* r,w */
Mimi Li37a79c02012-07-24 13:20:12 +0200537 if (idx < IDX_w) { /* r */
538 if (sed_cmd->end_line || sed_cmd->end_match)
539 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
540 }
Denis Vlasenko68404f12008-03-17 09:00:54 +0000541 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000542 if (sed_cmd->cmd == 'w') {
Denis Vlasenko5415c852008-07-21 23:05:26 +0000543 sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000544 sed_cmd->sw_last_char = '\n';
545 }
Denys Vlasenko52d83702011-05-03 00:51:43 +0200546 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000547 /* handle branch commands */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200548 else if (idx <= IDX_T) { /* :,b,t,T */
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000549 int length;
550
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000551 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000552 length = strcspn(cmdstr, semicolon_whitespace);
553 if (length) {
Rob Landleyd921b2e2006-08-03 15:41:12 +0000554 sed_cmd->string = xstrndup(cmdstr, length);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000555 cmdstr += length;
556 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000557 }
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000558 /* translation command */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200559 else if (idx == IDX_y) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000560 char *match, *replace;
Denis Vlasenko80778502006-10-25 12:46:46 +0000561 int i = cmdstr[0];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000562
Denis Vlasenko80778502006-10-25 12:46:46 +0000563 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000564 /* \n already parsed, but \delimiter needs unescaping. */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000565 parse_escapes(match, match, strlen(match), i, i);
566 parse_escapes(replace, replace, strlen(replace), i, i);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000567
Rob Landley9ffd4232006-05-21 18:30:35 +0000568 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000569 for (i = 0; match[i] && replace[i]; i++) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000570 sed_cmd->string[i*2] = match[i];
571 sed_cmd->string[i*2+1] = replace[i];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000572 }
573 free(match);
574 free(replace);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000575 }
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000576 /* if it wasnt a single-letter command that takes no arguments
577 * then it must be an invalid command.
578 */
Denys Vlasenko52d83702011-05-03 00:51:43 +0200579 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 +0000580 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
Mark Whitley1f3b9f22001-05-11 22:27:13 +0000581 }
Mark Whitley70705d72000-07-14 19:06:30 +0000582
583 /* give back whatever's left over */
Denis Vlasenko80778502006-10-25 12:46:46 +0000584 return cmdstr;
Eric Andersen50d63601999-11-09 01:47:36 +0000585}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000586
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000587
588/* Parse address+command sets, skipping comment lines. */
589
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000590static void add_cmd(const char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000591{
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000592 sed_cmd_t *sed_cmd;
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200593 unsigned len, n;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000594
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000595 /* Append this line to any unfinished line from last time. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000596 if (G.add_cmd_line) {
597 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
598 free(G.add_cmd_line);
Denis Vlasenko945bd3d2007-04-12 21:20:25 +0000599 cmdstr = G.add_cmd_line = tp;
Rob Landley12d87552006-06-05 17:32:44 +0000600 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000601
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200602 /* If this line ends with unescaped backslash, request next line. */
603 n = len = strlen(cmdstr);
604 while (n && cmdstr[n-1] == '\\')
605 n--;
606 if ((len - n) & 1) { /* if odd number of trailing backslashes */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000607 if (!G.add_cmd_line)
608 G.add_cmd_line = xstrdup(cmdstr);
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200609 G.add_cmd_line[len-1] = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000610 return;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000611 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000612
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000613 /* Loop parsing all commands in this line. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000614 while (*cmdstr) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000615 /* Skip leading whitespace and semicolons */
616 cmdstr += strspn(cmdstr, semicolon_whitespace);
617
618 /* If no more commands, exit. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000619 if (!*cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000620
621 /* if this is a comment, jump past it and keep going */
622 if (*cmdstr == '#') {
623 /* "#n" is the same as using -n on the command line */
Denis Vlasenko80778502006-10-25 12:46:46 +0000624 if (cmdstr[1] == 'n')
Denis Vlasenko17a15262007-03-26 20:48:46 +0000625 G.be_quiet++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000626 cmdstr = strpbrk(cmdstr, "\n\r");
627 if (!cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000628 continue;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000629 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000630
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000631 /* parse the command
632 * format is: [addr][,addr][!]cmd
633 * |----||-----||-|
634 * part1 part2 part3
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000635 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000636
Rob Landley9ffd4232006-05-21 18:30:35 +0000637 sed_cmd = xzalloc(sizeof(sed_cmd_t));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000638
639 /* first part (if present) is an address: either a '$', a number or a /regex/ */
640 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200641 sed_cmd->beg_line_orig = sed_cmd->beg_line;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000642
643 /* second part (if present) will begin with a comma */
644 if (*cmdstr == ',') {
645 int idx;
646
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000647 cmdstr++;
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200648 if (*cmdstr == '+' && isdigit(cmdstr[1])) {
649 /* http://sed.sourceforge.net/sedfaq3.html#s3.3
650 * Under GNU sed 3.02+, ssed, and sed15+, <address2>
651 * may also be a notation of the form +num,
652 * indicating the next num lines after <address1> is
653 * matched.
654 * GNU sed 4.2.1 accepts even "+" (meaning "+0").
655 * We don't (we check for isdigit, see above), think
656 * about the "+-3" case.
657 */
658 char *end;
659 /* code is smaller compared to using &cmdstr here: */
660 idx = strtol(cmdstr+1, &end, 10);
661 sed_cmd->end_line = -2 - idx;
662 cmdstr = end;
663 } else {
664 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
665 cmdstr += idx;
666 idx--; /* if 0, trigger error check below */
667 }
668 if (idx < 0)
Denis Vlasenko80778502006-10-25 12:46:46 +0000669 bb_error_msg_and_die("no address after comma");
Denys Vlasenko63f4d322015-04-17 14:24:55 +0200670 sed_cmd->end_line_orig = sed_cmd->end_line;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000671 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000672
673 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000674 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000675
676 /* Check for inversion flag */
677 if (*cmdstr == '!') {
678 sed_cmd->invert = 1;
679 cmdstr++;
680
681 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000682 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000683 }
684
685 /* last part (mandatory) will be a command */
Denis Vlasenko80778502006-10-25 12:46:46 +0000686 if (!*cmdstr)
687 bb_error_msg_and_die("missing command");
Denys Vlasenkoa2215b92010-05-12 01:49:04 +0200688 sed_cmd->cmd = *cmdstr++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000689 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
690
Denys Vlasenkoe93d1562013-07-08 01:43:40 +0200691 /* cmdstr now points past args.
692 * GNU sed requires a separator, if there are more commands,
693 * else it complains "char N: extra characters after command".
694 * Example: "sed 'p;d'". We also allow "sed 'pd'".
695 */
696
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000697 /* Add the command to the command array */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200698 *G.sed_cmd_tail = sed_cmd;
699 G.sed_cmd_tail = &sed_cmd->next;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000700 }
701
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000702 /* If we glued multiple lines together, free the memory. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000703 free(G.add_cmd_line);
704 G.add_cmd_line = NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +0000705}
706
Rob Landleydcc28662004-11-25 07:21:47 +0000707/* Append to a string, reallocating memory as necessary. */
708
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000709#define PIPE_GROW 64
Eric Andersenc52a6b02001-11-10 10:49:42 +0000710
Eric Andersen14f5c8d2005-04-16 19:39:00 +0000711static void pipe_putc(char c)
Eric Andersenc52a6b02001-11-10 10:49:42 +0000712{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000713 if (G.pipeline.idx == G.pipeline.len) {
714 G.pipeline.buf = xrealloc(G.pipeline.buf,
715 G.pipeline.len + PIPE_GROW);
716 G.pipeline.len += PIPE_GROW;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000717 }
Denis Vlasenko17a15262007-03-26 20:48:46 +0000718 G.pipeline.buf[G.pipeline.idx++] = c;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000719}
720
Rob Landley4795e4e2006-07-26 17:25:08 +0000721static void do_subst_w_backrefs(char *line, char *replace)
Mark Whitley97562bd2000-07-17 20:06:42 +0000722{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200723 int i, j;
Mark Whitley97562bd2000-07-17 20:06:42 +0000724
725 /* go through the replacement string */
726 for (i = 0; replace[i]; i++) {
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200727 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000728 if (replace[i] == '\\') {
729 unsigned backref = replace[++i] - '0';
730 if (backref <= 9) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000731 /* print out the text held in G.regmatch[backref] */
732 if (G.regmatch[backref].rm_so != -1) {
733 j = G.regmatch[backref].rm_so;
734 while (j < G.regmatch[backref].rm_eo)
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000735 pipe_putc(line[j++]);
736 }
737 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000738 }
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000739 /* I _think_ it is impossible to get '\' to be
740 * the last char in replace string. Thus we dont check
741 * for replace[i] == NUL. (counterexample anyone?) */
742 /* if we find a backslash escaped character, print the character */
743 pipe_putc(replace[i]);
744 continue;
Mark Whitley97562bd2000-07-17 20:06:42 +0000745 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000746 /* if we find an unescaped '&' print out the whole matched text. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000747 if (replace[i] == '&') {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000748 j = G.regmatch[0].rm_so;
749 while (j < G.regmatch[0].rm_eo)
Denis Vlasenko80778502006-10-25 12:46:46 +0000750 pipe_putc(line[j++]);
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000751 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000752 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000753 /* Otherwise just output the character. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000754 pipe_putc(replace[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000755 }
756}
757
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200758static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
Mark Whitley4f7fe772000-07-13 20:01:58 +0000759{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200760 char *line = *line_p;
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000761 unsigned match_count = 0;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200762 bool altered = 0;
763 bool prev_match_empty = 1;
764 bool tried_at_eol = 0;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000765 regex_t *current_regex;
766
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200767 current_regex = sed_cmd->sub_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000768 /* Handle empty regex. */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200769 if (!current_regex) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000770 current_regex = G.previous_regex_ptr;
Denis Vlasenko80778502006-10-25 12:46:46 +0000771 if (!current_regex)
Denis Vlasenkod3d004d2006-10-27 09:02:31 +0000772 bb_error_msg_and_die("no previous regexp");
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200773 }
774 G.previous_regex_ptr = current_regex;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000775
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000776 /* Find the first match */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100777 dbg("matching '%s'", line);
778 if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
779 dbg("no match");
Mark Whitley2dc192f2000-11-03 19:47:00 +0000780 return 0;
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100781 }
782 dbg("match");
Mark Whitley2dc192f2000-11-03 19:47:00 +0000783
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000784 /* Initialize temporary output buffer. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000785 G.pipeline.buf = xmalloc(PIPE_GROW);
786 G.pipeline.len = PIPE_GROW;
787 G.pipeline.idx = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000788
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000789 /* Now loop through, substituting for matches */
790 do {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200791 int start = G.regmatch[0].rm_so;
792 int end = G.regmatch[0].rm_eo;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000793 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000794
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000795 match_count++;
796
797 /* If we aren't interested in this match, output old line to
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200798 * end of match and continue */
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000799 if (sed_cmd->which_match
800 && (sed_cmd->which_match != match_count)
801 ) {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200802 for (i = 0; i < end; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200803 pipe_putc(*line++);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200804 /* Null match? Print one more char */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200805 if (start == end && *line)
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200806 pipe_putc(*line++);
Denys Vlasenkob84dafb2012-04-24 19:27:34 +0200807 goto next;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000808 }
809
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200810 /* Print everything before the match */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200811 for (i = 0; i < start; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200812 pipe_putc(line[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000813
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200814 /* Then print the substitution string,
815 * unless we just matched empty string after non-empty one.
816 * Example: string "cccd", pattern "c*", repl "R":
817 * result is "RdR", not "RRdR": first match "ccc",
818 * second is "" before "d", third is "" after "d".
819 * Second match is NOT replaced!
820 */
Denys Vlasenko84406e42012-06-07 16:34:57 +0200821 if (prev_match_empty || start != 0 || start != end) {
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200822 //dbg("%d %d %d", prev_match_empty, start, end);
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200823 dbg("inserting replacement at %d in '%s'", start, line);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200824 do_subst_w_backrefs(line, sed_cmd->string);
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200825 /* Flag that something has changed */
826 altered = 1;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200827 } else {
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200828 dbg("NOT inserting replacement at %d in '%s'", start, line);
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200829 }
Mark Whitley97562bd2000-07-17 20:06:42 +0000830
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200831 /* If matched string is empty (f.e. "c*" pattern),
832 * copy verbatim one char after it before attempting more matches
833 */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200834 prev_match_empty = (start == end);
Denys Vlasenko37ca36a2012-06-08 10:25:31 +0200835 if (prev_match_empty) {
836 if (!line[end]) {
837 tried_at_eol = 1;
838 } else {
839 pipe_putc(line[end]);
840 end++;
841 }
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200842 }
843
844 /* Advance past the match */
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200845 dbg("line += %d", end);
846 line += end;
Mark Whitley97562bd2000-07-17 20:06:42 +0000847
Mark Whitley2dc192f2000-11-03 19:47:00 +0000848 /* if we're not doing this globally, get out now */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100849 if (sed_cmd->which_match != 0)
850 break;
Denys Vlasenkob84dafb2012-04-24 19:27:34 +0200851 next:
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +0200852 /* Exit if we are at EOL and already tried matching at it */
853 if (*line == '\0') {
854 if (tried_at_eol)
855 break;
856 tried_at_eol = 1;
857 }
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200858
Denys Vlasenkoc35545a2012-06-04 14:45:09 +0200859//maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +0100860 } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
Mark Whitley2dc192f2000-11-03 19:47:00 +0000861
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000862 /* Copy rest of string into output pipeline */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200863 while (1) {
864 char c = *line++;
865 pipe_putc(c);
866 if (c == '\0')
867 break;
868 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000869
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200870 free(*line_p);
871 *line_p = G.pipeline.buf;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000872 return altered;
873}
Mark Whitley6315ce62000-07-10 22:55:51 +0000874
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000875/* Set command pointer to point to this label. (Does not handle null label.) */
Rob Landley4795e4e2006-07-26 17:25:08 +0000876static sed_cmd_t *branch_to(char *label)
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000877{
878 sed_cmd_t *sed_cmd;
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000879
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +0200880 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000881 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
882 return sed_cmd;
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000883 }
884 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000885 bb_error_msg_and_die("can't find label for jump to '%s'", label);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000886}
Mark Whitley6315ce62000-07-10 22:55:51 +0000887
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000888static void append(char *s)
889{
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +0100890 llist_add_to_end(&G.append_head, s);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000891}
892
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100893/* Output line of text. */
894/* Note:
895 * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
896 * Without them, we had this:
897 * echo -n thingy >z1
898 * echo -n again >z2
899 * >znull
900 * sed "s/i/z/" z1 z2 znull | hexdump -vC
901 * output:
902 * gnu sed 4.1.5:
903 * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
904 * bbox:
905 * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
Rob Landleydcc28662004-11-25 07:21:47 +0000906 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000907enum {
908 NO_EOL_CHAR = 1,
Denis Vlasenko86811802007-01-29 17:10:19 +0000909 LAST_IS_NUL = 2,
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000910};
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100911static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
912{
913 char lpc = *last_puts_char;
914
915 /* Need to insert a '\n' between two files because first file's
916 * last line wasn't terminated? */
917 if (lpc != '\n' && lpc != '\0') {
918 fputc('\n', file);
919 lpc = '\n';
920 }
921 fputs(s, file);
922
923 /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
924 if (s[0])
925 lpc = 'x';
926
927 /* had trailing '\0' and it was last char of file? */
928 if (last_gets_char == LAST_IS_NUL) {
929 fputc('\0', file);
930 lpc = 'x'; /* */
931 } else
932 /* had trailing '\n' or '\0'? */
933 if (last_gets_char != NO_EOL_CHAR) {
934 fputc(last_gets_char, file);
935 lpc = last_gets_char;
936 }
937
938 if (ferror(file)) {
939 xfunc_error_retval = 4; /* It's what gnu sed exits with... */
940 bb_error_msg_and_die(bb_msg_write_error);
941 }
942 *last_puts_char = lpc;
943}
944
945static void flush_append(char *last_puts_char, char last_gets_char)
946{
947 char *data;
948
949 /* Output appended lines. */
950 while ((data = (char *)llist_pop(&G.append_head))) {
951 puts_maybe_newline(data, G.nonstdout, last_puts_char, last_gets_char);
952 free(data);
953 }
954}
955
956/* Get next line of input from G.input_file_list, flushing append buffer and
957 * noting if we ran out of files without a newline on the last line we read.
958 */
959static char *get_next_line(char *gets_char, char *last_puts_char, char last_gets_char)
Rob Landleydcc28662004-11-25 07:21:47 +0000960{
Denis Vlasenko80778502006-10-25 12:46:46 +0000961 char *temp = NULL;
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000962 int len;
963 char gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000964
Denys Vlasenkoc44539f2013-10-30 14:25:22 +0100965 flush_append(last_puts_char, last_gets_char);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000966
967 /* will be returned if last line in the file
968 * doesn't end with either '\n' or '\0' */
969 gc = NO_EOL_CHAR;
Denys Vlasenkodeb07692013-11-28 12:08:51 +0100970 for (; G.current_input_file <= G.last_input_file; G.current_input_file++) {
Denys Vlasenko259b3c02013-11-28 03:14:16 +0100971 FILE *fp = G.current_fp;
972 if (!fp) {
973 const char *path = G.input_file_list[G.current_input_file];
974 fp = stdin;
975 if (path != bb_msg_standard_input) {
976 fp = fopen_or_warn(path, "r");
977 if (!fp) {
978 G.exitcode = EXIT_FAILURE;
979 continue;
980 }
981 }
982 G.current_fp = fp;
983 }
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000984 /* Read line up to a newline or NUL byte, inclusive,
985 * return malloc'ed char[]. length of the chunk read
986 * is stored in len. NULL if EOF/error */
Denis Vlasenko86811802007-01-29 17:10:19 +0000987 temp = bb_get_chunk_from_file(fp, &len);
Rob Landley2b26fd52006-02-24 02:30:39 +0000988 if (temp) {
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000989 /* len > 0 here, it's ok to do temp[len-1] */
990 char c = temp[len-1];
991 if (c == '\n' || c == '\0') {
992 temp[len-1] = '\0';
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000993 gc = c;
Denis Vlasenko86811802007-01-29 17:10:19 +0000994 if (c == '\0') {
995 int ch = fgetc(fp);
996 if (ch != EOF)
997 ungetc(ch, fp);
998 else
999 gc = LAST_IS_NUL;
1000 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +00001001 }
Denis Vlasenkoc562bb72007-01-29 17:08:51 +00001002 /* else we put NO_EOL_CHAR into *gets_char */
1003 break;
1004
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001005 /* NB: I had the idea of peeking next file(s) and returning
1006 * NO_EOL_CHAR only if it is the *last* non-empty
1007 * input file. But there is a case where this won't work:
1008 * file1: "a woo\nb woo"
1009 * file2: "c no\nd no"
1010 * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
1011 * (note: *no* newline after "b bang"!) */
Denis Vlasenko8b22b072006-12-02 17:58:10 +00001012 }
1013 /* Close this file and advance to next one */
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001014 fclose_if_not_stdin(fp);
1015 G.current_fp = NULL;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001016 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001017 *gets_char = gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001018 return temp;
1019}
1020
Denis Vlasenko17a15262007-03-26 20:48:46 +00001021#define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001022
Denis Vlasenko8274e062007-08-06 03:41:08 +00001023static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
1024{
1025 int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
1026 if (retval)
1027 G.previous_regex_ptr = sed_cmd->beg_match;
1028 return retval;
1029}
1030
Rob Landley2b26fd52006-02-24 02:30:39 +00001031/* Process all the lines in all the files */
1032
Rob Landleydcc28662004-11-25 07:21:47 +00001033static void process_files(void)
Mark Whitley6315ce62000-07-10 22:55:51 +00001034{
Rob Landleyce4f0e92004-10-30 06:54:19 +00001035 char *pattern_space, *next_line;
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001036 int linenum = 0;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001037 char last_puts_char = '\n';
1038 char last_gets_char, next_gets_char;
Denis Vlasenko2ea630f2006-12-10 02:52:19 +00001039 sed_cmd_t *sed_cmd;
1040 int substituted;
Mark Whitley6315ce62000-07-10 22:55:51 +00001041
Rob Landley2b26fd52006-02-24 02:30:39 +00001042 /* Prime the pump */
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001043 next_line = get_next_line(&next_gets_char, &last_puts_char, '\n' /*last_gets_char*/);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001044
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001045 /* Go through every line in each file */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +00001046 again:
Denis Vlasenko2ea630f2006-12-10 02:52:19 +00001047 substituted = 0;
Glenn L McGrath505bd0f2003-03-08 05:21:02 +00001048
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001049 /* Advance to next line. Stop if out of lines. */
1050 pattern_space = next_line;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001051 if (!pattern_space)
1052 return;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001053 last_gets_char = next_gets_char;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001054
1055 /* Read one line in advance so we can act on the last line,
1056 * the '$' address */
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001057 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001058 linenum++;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001059
1060 /* For every line, go through all the commands */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +00001061 restart:
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001062 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001063 int old_matched, matched;
Eric Andersen52a3c272003-12-23 08:53:51 +00001064
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001065 old_matched = sed_cmd->in_match;
Mark Whitley0915c4b2001-06-11 23:50:06 +00001066
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001067 /* Determine if this command matches this line: */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001068
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001069 dbg("match1:%d", sed_cmd->in_match);
1070 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1071 && !sed_cmd->beg_match && !sed_cmd->end_match));
1072 dbg("match3:%d", (sed_cmd->beg_line > 0
1073 && (sed_cmd->end_line || sed_cmd->end_match
1074 ? (sed_cmd->beg_line <= linenum)
1075 : (sed_cmd->beg_line == linenum)
1076 )
1077 ));
1078 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1079 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001080
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001081 /* Are we continuing a previous multi-line match? */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001082 sed_cmd->in_match = sed_cmd->in_match
1083 /* Or is no range necessary? */
1084 || (!sed_cmd->beg_line && !sed_cmd->end_line
1085 && !sed_cmd->beg_match && !sed_cmd->end_match)
1086 /* Or did we match the start of a numerical range? */
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001087 || (sed_cmd->beg_line > 0
1088 && (sed_cmd->end_line || sed_cmd->end_match
Denys Vlasenkoc4679242010-06-04 01:31:48 +02001089 /* note: even if end is numeric and is < linenum too,
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001090 * GNU sed matches! We match too, therefore we don't
1091 * check here that linenum <= end.
1092 * Example:
1093 * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1094 * first three input lines are deleted;
1095 * 4th line is matched and printed
1096 * by "2,3" (!) and by "4" ranges
1097 */
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001098 ? (sed_cmd->beg_line <= linenum) /* N,end */
1099 : (sed_cmd->beg_line == linenum) /* N */
1100 )
1101 )
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001102 /* Or does this line match our begin address regex? */
Denis Vlasenko8274e062007-08-06 03:41:08 +00001103 || (beg_match(sed_cmd, pattern_space))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001104 /* Or did we match last line of input? */
1105 || (sed_cmd->beg_line == -1 && next_line == NULL);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001106
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001107 /* Snapshot the value */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001108 matched = sed_cmd->in_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001109
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001110 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1111 sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001112
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001113 /* Is this line the end of the current match? */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001114
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001115 if (matched) {
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001116 if (sed_cmd->end_line <= -2) {
1117 /* address2 is +N, i.e. N lines from beg_line */
1118 sed_cmd->end_line = linenum + (-sed_cmd->end_line - 2);
1119 }
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001120 /* once matched, "n,xxx" range is dead, disabling it */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001121 if (sed_cmd->beg_line > 0) {
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001122 sed_cmd->beg_line = -2;
Denys Vlasenko2e284a42010-08-01 04:14:46 +02001123 }
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001124 dbg("end1:%d", sed_cmd->end_line ? sed_cmd->end_line == -1
1125 ? !next_line : (sed_cmd->end_line <= linenum)
1126 : !sed_cmd->end_match);
1127 dbg("end2:%d", sed_cmd->end_match && old_matched
1128 && !regexec(sed_cmd->end_match,pattern_space, 0, NULL, 0));
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001129 sed_cmd->in_match = !(
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001130 /* has the ending line come, or is this a single address command? */
Denys Vlasenko52d83702011-05-03 00:51:43 +02001131 (sed_cmd->end_line
1132 ? sed_cmd->end_line == -1
1133 ? !next_line
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001134 : (sed_cmd->end_line <= linenum)
1135 : !sed_cmd->end_match
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001136 )
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001137 /* or does this line matches our last address regex */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001138 || (sed_cmd->end_match && old_matched
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001139 && (regexec(sed_cmd->end_match,
Denys Vlasenko6830ade2013-01-15 13:58:01 +01001140 pattern_space, 0, NULL, 0) == 0)
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001141 )
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +02001142 );
Erik Andersene49d5ec2000-02-08 19:58:47 +00001143 }
Erik Andersen1266a131999-12-29 22:19:46 +00001144
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001145 /* Skip blocks of commands we didn't match */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001146 if (sed_cmd->cmd == '{') {
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001147 if (sed_cmd->invert ? matched : !matched) {
Denys Vlasenkof2c16ed2010-04-20 04:00:03 -04001148 unsigned nest_cnt = 0;
1149 while (1) {
1150 if (sed_cmd->cmd == '{')
1151 nest_cnt++;
1152 if (sed_cmd->cmd == '}') {
1153 nest_cnt--;
1154 if (nest_cnt == 0)
1155 break;
1156 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001157 sed_cmd = sed_cmd->next;
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001158 if (!sed_cmd)
1159 bb_error_msg_and_die("unterminated {");
1160 }
1161 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001162 continue;
1163 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001164
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001165 /* Okay, so did this line match? */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001166 if (sed_cmd->invert ? matched : !matched)
1167 continue; /* no */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001168
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001169 /* Update last used regex in case a blank substitute BRE is found */
1170 if (sed_cmd->beg_match) {
1171 G.previous_regex_ptr = sed_cmd->beg_match;
1172 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001173
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001174 /* actual sedding */
Denys Vlasenkod29ae7e2012-01-15 20:06:03 +01001175 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1176 pattern_space, next_line, sed_cmd->cmd);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001177 switch (sed_cmd->cmd) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001178
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001179 /* Print line number */
1180 case '=':
1181 fprintf(G.nonstdout, "%d\n", linenum);
1182 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001183
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001184 /* Write the current pattern space up to the first newline */
1185 case 'P':
1186 {
1187 char *tmp = strchr(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001188 if (tmp) {
1189 *tmp = '\0';
1190 /* TODO: explain why '\n' below */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001191 sed_puts(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001192 *tmp = '\n';
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001193 break;
1194 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001195 /* Fall Through */
1196 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001197
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001198 /* Write the current pattern space to output */
1199 case 'p':
1200 /* NB: we print this _before_ the last line
1201 * (of current file) is printed. Even if
1202 * that line is nonterminated, we print
1203 * '\n' here (gnu sed does the same) */
1204 sed_puts(pattern_space, '\n');
1205 break;
1206 /* Delete up through first newline */
1207 case 'D':
1208 {
1209 char *tmp = strchr(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001210 if (tmp) {
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001211 overlapping_strcpy(pattern_space, tmp + 1);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001212 goto restart;
1213 }
1214 }
1215 /* discard this line. */
1216 case 'd':
1217 goto discard_line;
1218
1219 /* Substitute with regex */
1220 case 's':
1221 if (!do_subst_command(sed_cmd, &pattern_space))
1222 break;
Denys Vlasenko21f6fbf2012-06-04 14:44:47 +02001223 dbg("do_subst_command succeeded:'%s'", pattern_space);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001224 substituted |= 1;
1225
1226 /* handle p option */
1227 if (sed_cmd->sub_p)
1228 sed_puts(pattern_space, last_gets_char);
1229 /* handle w option */
1230 if (sed_cmd->sw_file)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001231 puts_maybe_newline(
1232 pattern_space, sed_cmd->sw_file,
1233 &sed_cmd->sw_last_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001234 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001235
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001236 /* Append line to linked list to be printed later */
1237 case 'a':
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +01001238 append(xstrdup(sed_cmd->string));
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001239 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001240
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001241 /* Insert text before this line */
1242 case 'i':
1243 sed_puts(sed_cmd->string, '\n');
1244 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001245
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001246 /* Cut and paste text (replace) */
1247 case 'c':
1248 /* Only triggers on last line of a matching range. */
1249 if (!sed_cmd->in_match)
Denys Vlasenko96a18332010-04-19 22:36:07 -04001250 sed_puts(sed_cmd->string, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001251 goto discard_line;
1252
1253 /* Read file, append contents to output */
1254 case 'r':
1255 {
1256 FILE *rfile;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001257 rfile = fopen_for_read(sed_cmd->string);
1258 if (rfile) {
1259 char *line;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001260 while ((line = xmalloc_fgetline(rfile))
1261 != NULL)
1262 append(line);
Denys Vlasenko9d46a7a2013-10-30 10:22:47 +01001263 fclose(rfile);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001264 }
1265
1266 break;
1267 }
1268
1269 /* Write pattern space to file. */
1270 case 'w':
1271 puts_maybe_newline(
1272 pattern_space, sed_cmd->sw_file,
1273 &sed_cmd->sw_last_char, last_gets_char);
1274 break;
1275
1276 /* Read next line from input */
1277 case 'n':
1278 if (!G.be_quiet)
1279 sed_puts(pattern_space, last_gets_char);
1280 if (next_line) {
1281 free(pattern_space);
1282 pattern_space = next_line;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001283 last_gets_char = next_gets_char;
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001284 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001285 substituted = 0;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001286 linenum++;
1287 break;
1288 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001289 /* fall through */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001290
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001291 /* Quit. End of script, end of input. */
1292 case 'q':
1293 /* Exit the outer while loop */
1294 free(next_line);
1295 next_line = NULL;
1296 goto discard_commands;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001297
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001298 /* Append the next line to the current line */
1299 case 'N':
1300 {
1301 int len;
1302 /* If no next line, jump to end of script and exit. */
Denys Vlasenko0d555fc2010-08-16 16:26:33 +02001303 /* http://www.gnu.org/software/sed/manual/sed.html:
1304 * "Most versions of sed exit without printing anything
1305 * when the N command is issued on the last line of
1306 * a file. GNU sed prints pattern space before exiting
1307 * unless of course the -n command switch has been
1308 * specified. This choice is by design."
1309 */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001310 if (next_line == NULL) {
Denys Vlasenko0d555fc2010-08-16 16:26:33 +02001311 //goto discard_line;
1312 goto discard_commands; /* GNU behavior */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001313 }
Denys Vlasenkobf5f99f2010-06-04 01:29:52 +02001314 /* Append next_line, read new next_line. */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001315 len = strlen(pattern_space);
Denys Vlasenko90377872010-01-08 09:07:50 +01001316 pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001317 pattern_space[len] = '\n';
1318 strcpy(pattern_space + len+1, next_line);
1319 last_gets_char = next_gets_char;
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001320 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001321 linenum++;
1322 break;
1323 }
1324
1325 /* Test/branch if substitution occurred */
1326 case 't':
1327 if (!substituted) break;
1328 substituted = 0;
1329 /* Fall through */
1330 /* Test/branch if substitution didn't occur */
1331 case 'T':
1332 if (substituted) break;
1333 /* Fall through */
1334 /* Branch to label */
1335 case 'b':
1336 if (!sed_cmd->string) goto discard_commands;
1337 else sed_cmd = branch_to(sed_cmd->string);
1338 break;
1339 /* Transliterate characters */
1340 case 'y':
1341 {
1342 int i, j;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001343 for (i = 0; pattern_space[i]; i++) {
1344 for (j = 0; sed_cmd->string[j]; j += 2) {
1345 if (pattern_space[i] == sed_cmd->string[j]) {
1346 pattern_space[i] = sed_cmd->string[j + 1];
1347 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001348 }
1349 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001350 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001351
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001352 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001353 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001354 case 'g': /* Replace pattern space with hold space */
1355 free(pattern_space);
1356 pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1357 break;
1358 case 'G': /* Append newline and hold space to pattern space */
1359 {
1360 int pattern_space_size = 2;
1361 int hold_space_size = 0;
1362
1363 if (pattern_space)
1364 pattern_space_size += strlen(pattern_space);
1365 if (G.hold_space)
1366 hold_space_size = strlen(G.hold_space);
1367 pattern_space = xrealloc(pattern_space,
1368 pattern_space_size + hold_space_size);
1369 if (pattern_space_size == 2)
1370 pattern_space[0] = 0;
1371 strcat(pattern_space, "\n");
1372 if (G.hold_space)
1373 strcat(pattern_space, G.hold_space);
1374 last_gets_char = '\n';
1375
1376 break;
1377 }
1378 case 'h': /* Replace hold space with pattern space */
1379 free(G.hold_space);
1380 G.hold_space = xstrdup(pattern_space);
1381 break;
1382 case 'H': /* Append newline and pattern space to hold space */
1383 {
1384 int hold_space_size = 2;
1385 int pattern_space_size = 0;
1386
1387 if (G.hold_space)
1388 hold_space_size += strlen(G.hold_space);
1389 if (pattern_space)
1390 pattern_space_size = strlen(pattern_space);
1391 G.hold_space = xrealloc(G.hold_space,
1392 hold_space_size + pattern_space_size);
1393
1394 if (hold_space_size == 2)
1395 *G.hold_space = 0;
1396 strcat(G.hold_space, "\n");
1397 if (pattern_space)
1398 strcat(G.hold_space, pattern_space);
1399
1400 break;
1401 }
1402 case 'x': /* Exchange hold and pattern space */
1403 {
1404 char *tmp = pattern_space;
Denys Vlasenko90a99042009-09-06 02:36:23 +02001405 pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001406 last_gets_char = '\n';
1407 G.hold_space = tmp;
1408 break;
1409 }
1410 } /* switch */
1411 } /* for each cmd */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001412
1413 /*
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001414 * Exit point from sedding...
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001415 */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001416 discard_commands:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001417 /* we will print the line unless we were told to be quiet ('-n')
1418 or if the line was suppressed (ala 'd'elete) */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001419 if (!G.be_quiet)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001420 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001421
1422 /* Delete and such jump here. */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001423 discard_line:
Denys Vlasenkoc44539f2013-10-30 14:25:22 +01001424 flush_append(&last_puts_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001425 free(pattern_space);
1426
1427 goto again;
Erik Andersen1266a131999-12-29 22:19:46 +00001428}
1429
Glenn L McGrath42c25732003-10-04 05:27:56 +00001430/* It is possible to have a command line argument with embedded
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001431 * newlines. This counts as multiple command lines.
1432 * However, newline can be escaped: 's/e/z\<newline>z/'
Denys Vlasenko40ab27a2013-07-08 02:04:44 +02001433 * add_cmd() handles this.
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001434 */
Glenn L McGrath42c25732003-10-04 05:27:56 +00001435
1436static void add_cmd_block(char *cmdstr)
1437{
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001438 char *sv, *eol;
Glenn L McGrath42c25732003-10-04 05:27:56 +00001439
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001440 cmdstr = sv = xstrdup(cmdstr);
1441 do {
1442 eol = strchr(cmdstr, '\n');
Denys Vlasenko40ab27a2013-07-08 02:04:44 +02001443 if (eol)
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001444 *eol = '\0';
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001445 add_cmd(cmdstr);
1446 cmdstr = eol + 1;
1447 } while (eol);
1448 free(sv);
Glenn L McGrath42c25732003-10-04 05:27:56 +00001449}
1450
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +00001451int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +00001452int sed_main(int argc UNUSED_PARAM, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001453{
Denis Vlasenko67b23e62006-10-03 21:00:06 +00001454 unsigned opt;
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001455 llist_t *opt_e, *opt_f;
Simon B8c343952012-05-06 13:59:15 +02001456 char *opt_i;
1457
1458#if ENABLE_LONG_OPTS
1459 static const char sed_longopts[] ALIGN1 =
1460 /* name has_arg short */
1461 "in-place\0" Optional_argument "i"
1462 "regexp-extended\0" No_argument "r"
1463 "quiet\0" No_argument "n"
1464 "silent\0" No_argument "n"
1465 "expression\0" Required_argument "e"
1466 "file\0" Required_argument "f";
1467#endif
1468
Denis Vlasenko74324c82007-06-04 10:16:52 +00001469 INIT_G();
Rob Landleye3f5a3f2006-05-09 03:53:55 +00001470
Mark Whitley858c1ad2000-07-11 21:38:47 +00001471 /* destroy command strings on exit */
Rob Landley0b656282006-05-08 22:17:23 +00001472 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
Mark Whitley858c1ad2000-07-11 21:38:47 +00001473
Rob Landleyfae1dc82005-11-20 07:44:35 +00001474 /* Lie to autoconf when it starts asking stupid questions. */
Simon B8c343952012-05-06 13:59:15 +02001475 if (argv[1] && strcmp(argv[1], "--version") == 0) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001476 puts("This is not GNU sed version 4.0");
1477 return 0;
Eric Andersenc06f5682004-02-04 10:57:46 +00001478 }
Eric Andersenc06f5682004-02-04 10:57:46 +00001479
Mark Whitley6315ce62000-07-10 22:55:51 +00001480 /* do normal option parsing */
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001481 opt_e = opt_f = NULL;
Simon B8c343952012-05-06 13:59:15 +02001482 opt_i = NULL;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001483 opt_complementary = "e::f::" /* can occur multiple times */
1484 "nn"; /* count -n */
Simon B8c343952012-05-06 13:59:15 +02001485
1486 IF_LONG_OPTS(applet_long_options = sed_longopts);
1487
Denys Vlasenko2e284a42010-08-01 04:14:46 +02001488 /* -i must be first, to match OPT_in_place definition */
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001489 /* -E is a synonym of -r:
1490 * GNU sed 4.2.1 mentions it in neither --help
1491 * nor manpage, but does recognize it.
1492 */
1493 opt = getopt32(argv, "i::rEne:f:", &opt_i, &opt_e, &opt_f,
Denis Vlasenko17a15262007-03-26 20:48:46 +00001494 &G.be_quiet); /* counter for -n */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001495 //argc -= optind;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001496 argv += optind;
1497 if (opt & OPT_in_place) { // -i
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001498 atexit(cleanup_outname);
1499 }
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001500 if (opt & (2|4))
1501 G.regex_type |= REG_EXTENDED; // -r or -E
1502 //if (opt & 8)
1503 // G.be_quiet++; // -n (implemented with a counter instead)
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001504 while (opt_e) { // -e
Denis Vlasenkod50dda82008-06-15 05:40:56 +00001505 add_cmd_block(llist_pop(&opt_e));
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001506 }
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001507 while (opt_f) { // -f
1508 char *line;
1509 FILE *cmdfile;
Denys Vlasenkoe4d925b2016-04-08 00:20:36 +02001510 cmdfile = xfopen_stdin(llist_pop(&opt_f));
Denis Vlasenko8ee649a2008-03-26 20:04:27 +00001511 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001512 add_cmd(line);
1513 free(line);
1514 }
Denys Vlasenkoe4d925b2016-04-08 00:20:36 +02001515 fclose_if_not_stdin(cmdfile);
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001516 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001517 /* if we didn't get a pattern from -e or -f, use argv[0] */
David A. Wheeleraf0cdee2013-10-29 00:52:48 +01001518 if (!(opt & 0x30)) {
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001519 if (!*argv)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001520 bb_show_usage();
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001521 add_cmd_block(*argv++);
Mark Whitley6315ce62000-07-10 22:55:51 +00001522 }
Glenn L McGrath42c25732003-10-04 05:27:56 +00001523 /* Flush any unfinished commands. */
1524 add_cmd("");
Mark Whitley6315ce62000-07-10 22:55:51 +00001525
Rob Landley53302f82004-02-18 09:54:15 +00001526 /* By default, we write to stdout */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001527 G.nonstdout = stdout;
Rob Landley53302f82004-02-18 09:54:15 +00001528
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001529 /* argv[0..(argc-1)] should be names of file to process. If no
Mark Whitley6315ce62000-07-10 22:55:51 +00001530 * files were specified or '-' was specified, take input from stdin.
1531 * Otherwise, we process all the files specified. */
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001532 G.input_file_list = argv;
1533 if (!argv[0]) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001534 if (opt & OPT_in_place)
Denis Vlasenko80778502006-10-25 12:46:46 +00001535 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001536 argv[0] = (char*)bb_msg_standard_input;
1537 /* G.last_input_file = 0; - already is */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001538 } else {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001539 goto start;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001540
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001541 for (; *argv; argv++) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001542 struct stat statbuf;
1543 int nonstdoutfd;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001544 sed_cmd_t *sed_cmd;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001545
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001546 G.last_input_file++;
1547 start:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001548 if (!(opt & OPT_in_place)) {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001549 if (LONE_DASH(*argv)) {
1550 *argv = (char*)bb_msg_standard_input;
1551 process_files();
1552 }
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001553 continue;
1554 }
1555
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001556 /* -i: process each FILE separately: */
1557
Denys Vlasenkocd738712014-10-05 02:44:34 +02001558 if (stat(*argv, &statbuf) != 0) {
1559 bb_simple_perror_msg(*argv);
1560 G.exitcode = EXIT_FAILURE;
1561 G.current_input_file++;
1562 continue;
1563 }
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001564 G.outname = xasprintf("%sXXXXXX", *argv);
Alexander Shishkin67227372010-10-22 13:27:16 +02001565 nonstdoutfd = xmkstemp(G.outname);
Denys Vlasenkoa7ccdee2009-11-15 23:28:11 +01001566 G.nonstdout = xfdopen_for_write(nonstdoutfd);
Denys Vlasenko57992482009-11-13 09:09:07 +01001567 /* Set permissions/owner of output file */
Denys Vlasenko3b727cc2010-06-18 02:12:56 +02001568 /* chmod'ing AFTER chown would preserve suid/sgid bits,
1569 * but GNU sed 4.2.1 does not preserve them either */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001570 fchmod(nonstdoutfd, statbuf.st_mode);
Denys Vlasenko57992482009-11-13 09:09:07 +01001571 fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001572
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001573 process_files();
Denis Vlasenko17a15262007-03-26 20:48:46 +00001574 fclose(G.nonstdout);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001575 G.nonstdout = stdout;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001576
Simon B8c343952012-05-06 13:59:15 +02001577 if (opt_i) {
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001578 char *backupname = xasprintf("%s%s", *argv, opt_i);
1579 xrename(*argv, backupname);
Simon B8c343952012-05-06 13:59:15 +02001580 free(backupname);
1581 }
Denys Vlasenkodeb07692013-11-28 12:08:51 +01001582 /* else unlink(*argv); - rename below does this */
1583 xrename(G.outname, *argv); //TODO: rollback backup on error?
Denis Vlasenko17a15262007-03-26 20:48:46 +00001584 free(G.outname);
Denis Vlasenko40276642007-11-13 16:48:10 +00001585 G.outname = NULL;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001586
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001587 /* Fix disabled range matches and mangled ",+N" ranges */
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001588 for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1589 sed_cmd->beg_line = sed_cmd->beg_line_orig;
Denys Vlasenko63f4d322015-04-17 14:24:55 +02001590 sed_cmd->end_line = sed_cmd->end_line_orig;
Denys Vlasenkoa7d6bb32011-08-16 13:29:34 +02001591 }
Mark Whitley6315ce62000-07-10 22:55:51 +00001592 }
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001593 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001594 * if (G.current_input_file[G.current_input_file] == NULL)
1595 * return G.exitcode;
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001596 * but it's not needed since process_files() works correctly
1597 * in this case too. */
Mark Whitley6315ce62000-07-10 22:55:51 +00001598 }
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001599
Denys Vlasenkoeb08b6e2010-06-19 17:51:06 +02001600 process_files();
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001601
Denys Vlasenko259b3c02013-11-28 03:14:16 +01001602 return G.exitcode;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001603}