Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 2 | /* |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 3 | * sed.c - very minimalist version of sed |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 4 | * |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 5 | * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley |
| 6 | * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org> |
Matt Kraai | 5ed78ad | 2002-01-03 21:12:34 +0000 | [diff] [blame] | 7 | * Copyright (C) 2002 Matt Kraai |
Denis Vlasenko | 0beaff8 | 2007-09-21 13:16:32 +0000 | [diff] [blame] | 8 | * Copyright (C) 2003 by Glenn McGrath |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 9 | * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net> |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 10 | * |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 11 | * MAINTAINER: Rob Landley <rob@landley.net> |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 12 | * |
Denys Vlasenko | 0ef64bd | 2010-08-16 20:14:46 +0200 | [diff] [blame] | 13 | * Licensed under GPLv2, see file LICENSE in this source tree. |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 14 | */ |
| 15 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 16 | /* Code overview. |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 17 | * |
| 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 Vlasenko | a221bc5 | 2011-09-13 18:40:22 +0200 | [diff] [blame] | 26 | * process_files() does actual sedding, reading data lines from each input FILE* |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 27 | * (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 McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 32 | |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 33 | /* Supported features and commands in this version of sed: |
| 34 | * |
| 35 | * - comments ('#') |
| 36 | * - address matching: num|/matchstr/[,num|/matchstr/|$]command |
| 37 | * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags) |
| 38 | * - edit commands: (a)ppend, (i)nsert, (c)hange |
| 39 | * - file commands: (r)ead |
| 40 | * - backreferences in substitution expressions (\0, \1, \2...\9) |
| 41 | * - grouped commands: {cmd1;cmd2} |
| 42 | * - transliteration (y/source-chars/dest-chars/) |
| 43 | * - pattern space hold space storing / swapping (g, h, x) |
| 44 | * - labels / branching (: label, b, t, T) |
| 45 | * |
| 46 | * (Note: Specifying an address (range) to match is *optional*; commands |
| 47 | * default to the whole pattern space if no specific address match was |
| 48 | * requested.) |
| 49 | * |
| 50 | * Todo: |
| 51 | * - Create a wrapper around regex to make libc's regex conform with sed |
| 52 | * |
Mimi Li | 37a79c0 | 2012-07-24 13:20:12 +0200 | [diff] [blame] | 53 | * Reference |
| 54 | * http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html |
| 55 | * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 56 | * http://sed.sourceforge.net/sedfaq3.html |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 57 | */ |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 58 | |
Denys Vlasenko | 73225b6 | 2013-11-13 12:45:33 +0100 | [diff] [blame] | 59 | //config:config SED |
| 60 | //config: bool "sed" |
| 61 | //config: default y |
| 62 | //config: help |
| 63 | //config: sed is used to perform text transformations on a file |
| 64 | //config: or input from a pipeline. |
| 65 | |
| 66 | //kbuild:lib-$(CONFIG_SED) += sed.o |
| 67 | |
| 68 | //applet:IF_SED(APPLET(sed, BB_DIR_BIN, BB_SUID_DROP)) |
| 69 | |
Pere Orga | 6a3e01d | 2011-04-01 22:56:30 +0200 | [diff] [blame] | 70 | //usage:#define sed_trivial_usage |
Denys Vlasenko | a82e32d | 2013-10-30 13:00:00 +0100 | [diff] [blame] | 71 | //usage: "[-inrE] [-f FILE]... [-e CMD]... [FILE]...\n" |
| 72 | //usage: "or: sed [-inrE] CMD [FILE]..." |
Pere Orga | 6a3e01d | 2011-04-01 22:56:30 +0200 | [diff] [blame] | 73 | //usage:#define sed_full_usage "\n\n" |
Denys Vlasenko | 6642676 | 2011-06-05 03:58:28 +0200 | [diff] [blame] | 74 | //usage: " -e CMD Add CMD to sed commands to be executed" |
Pere Orga | 6a3e01d | 2011-04-01 22:56:30 +0200 | [diff] [blame] | 75 | //usage: "\n -f FILE Add FILE contents to sed commands to be executed" |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 76 | //usage: "\n -i[SFX] Edit files in-place (otherwise sends to stdout)" |
Denys Vlasenko | 1d7ad7a | 2012-06-21 09:45:11 +0200 | [diff] [blame] | 77 | //usage: "\n Optionally back files up, appending SFX" |
Pere Orga | 6a3e01d | 2011-04-01 22:56:30 +0200 | [diff] [blame] | 78 | //usage: "\n -n Suppress automatic printing of pattern space" |
Denys Vlasenko | a82e32d | 2013-10-30 13:00:00 +0100 | [diff] [blame] | 79 | //usage: "\n -r,-E Use extended regex syntax" |
Pere Orga | 6a3e01d | 2011-04-01 22:56:30 +0200 | [diff] [blame] | 80 | //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 Vlasenko | b6adbf1 | 2007-05-26 19:00:18 +0000 | [diff] [blame] | 88 | #include "libbb.h" |
Denys Vlasenko | e6a2f4c | 2016-04-21 16:26:30 +0200 | [diff] [blame] | 89 | #include "common_bufsiz.h" |
"Vladimir N. Oleynik" | 23f62fc | 2005-09-14 16:59:11 +0000 | [diff] [blame] | 90 | #include "xregex.h" |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 91 | |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 92 | #if 0 |
| 93 | # define dbg(...) bb_error_msg(__VA_ARGS__) |
| 94 | #else |
| 95 | # define dbg(...) ((void)0) |
| 96 | #endif |
| 97 | |
| 98 | |
Denys Vlasenko | 2e284a4 | 2010-08-01 04:14:46 +0200 | [diff] [blame] | 99 | enum { |
| 100 | OPT_in_place = 1 << 0, |
| 101 | }; |
| 102 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 103 | /* Each sed command turns into one of these structures. */ |
Glenn L McGrath | e7a8bc9 | 2003-03-09 10:23:57 +0000 | [diff] [blame] | 104 | typedef struct sed_cmd_s { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 105 | /* Ordered by alignment requirements: currently 36 bytes on x86 */ |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 106 | struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */ |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 107 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 108 | /* 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 Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 113 | int beg_line_orig; /* copy of the above, needed for -i */ |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 114 | int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($). -2-N = +N */ |
| 115 | int end_line_orig; |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 116 | |
Denys Vlasenko | 2a4bba3 | 2016-01-24 15:52:16 +0100 | [diff] [blame] | 117 | FILE *sw_file; /* File (sw) command writes to, NULL for none. */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 118 | char *string; /* Data string for (saicytb) commands. */ |
Glenn L McGrath | fc4cb4d | 2003-04-12 16:10:42 +0000 | [diff] [blame] | 119 | |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 120 | unsigned which_match; /* (s) Which match to replace (0 for all) */ |
Glenn L McGrath | fc4cb4d | 2003-04-12 16:10:42 +0000 | [diff] [blame] | 121 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 122 | /* Bitfields (gcc won't group them if we don't) */ |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 123 | 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 Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 126 | |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 127 | char sw_last_char; /* Last line written by (sw) had no '\n' */ |
Glenn L McGrath | c18ce37 | 2003-09-13 06:57:39 +0000 | [diff] [blame] | 128 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 129 | /* GENERAL FIELDS */ |
| 130 | char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */ |
Glenn L McGrath | e7a8bc9 | 2003-03-09 10:23:57 +0000 | [diff] [blame] | 131 | } sed_cmd_t; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 132 | |
Denis Vlasenko | 6ca409e | 2007-08-12 20:58:27 +0000 | [diff] [blame] | 133 | static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v"; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 134 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 135 | struct globals { |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 136 | /* options */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 137 | int be_quiet, regex_type; |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 138 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 139 | FILE *nonstdout; |
| 140 | char *outname, *hold_space; |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 141 | smallint exitcode; |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 142 | |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 143 | /* list of input files */ |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 144 | int current_input_file, last_input_file; |
| 145 | char **input_file_list; |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 146 | FILE *current_fp; |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 147 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 148 | regmatch_t regmatch[10]; |
| 149 | regex_t *previous_regex_ptr; |
Denis Vlasenko | 9213a9e | 2006-09-17 16:28:10 +0000 | [diff] [blame] | 150 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 151 | /* linked list of sed commands */ |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 152 | sed_cmd_t *sed_cmd_head, **sed_cmd_tail; |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 153 | |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 154 | /* linked list of append lines */ |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 155 | llist_t *append_head; |
| 156 | |
| 157 | char *add_cmd_line; |
| 158 | |
| 159 | struct pipeline { |
Denys Vlasenko | fb132e4 | 2010-10-29 11:46:52 +0200 | [diff] [blame] | 160 | char *buf; /* Space to hold string */ |
| 161 | int idx; /* Space used */ |
| 162 | int len; /* Space allocated */ |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 163 | } pipeline; |
Denys Vlasenko | 98a4c7c | 2010-02-04 15:00:15 +0100 | [diff] [blame] | 164 | } FIX_ALIASING; |
Denys Vlasenko | e6a2f4c | 2016-04-21 16:26:30 +0200 | [diff] [blame] | 165 | #define G (*(struct globals*)bb_common_bufsiz1) |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 166 | #define INIT_G() do { \ |
Denys Vlasenko | 47cfbf3 | 2016-04-21 18:18:48 +0200 | [diff] [blame] | 167 | setup_common_bufsiz(); \ |
Denys Vlasenko | 7b85ec3 | 2015-10-13 17:17:34 +0200 | [diff] [blame] | 168 | BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \ |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 169 | G.sed_cmd_tail = &G.sed_cmd_head; \ |
| 170 | } while (0) |
| 171 | |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 172 | |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 173 | #if ENABLE_FEATURE_CLEAN_UP |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 174 | static void sed_free_and_close_stuff(void) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 175 | { |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 176 | sed_cmd_t *sed_cmd = G.sed_cmd_head; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 177 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 178 | llist_free(G.append_head, free); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 179 | |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 180 | while (sed_cmd) { |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 181 | sed_cmd_t *sed_cmd_next = sed_cmd->next; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 182 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 183 | if (sed_cmd->sw_file) |
Denys Vlasenko | 2a4bba3 | 2016-01-24 15:52:16 +0100 | [diff] [blame] | 184 | fclose(sed_cmd->sw_file); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 185 | |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 186 | if (sed_cmd->beg_match) { |
| 187 | regfree(sed_cmd->beg_match); |
| 188 | free(sed_cmd->beg_match); |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 189 | } |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 190 | if (sed_cmd->end_match) { |
| 191 | regfree(sed_cmd->end_match); |
| 192 | free(sed_cmd->end_match); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 193 | } |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 194 | if (sed_cmd->sub_match) { |
| 195 | regfree(sed_cmd->sub_match); |
| 196 | free(sed_cmd->sub_match); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 197 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 198 | free(sed_cmd->string); |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 199 | free(sed_cmd); |
| 200 | sed_cmd = sed_cmd_next; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 201 | } |
Rob Landley | ce4f0e9 | 2004-10-30 06:54:19 +0000 | [diff] [blame] | 202 | |
Denis Vlasenko | 6081868 | 2007-09-28 22:07:23 +0000 | [diff] [blame] | 203 | free(G.hold_space); |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 204 | |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 205 | if (G.current_fp) |
| 206 | fclose(G.current_fp); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 207 | } |
Denis Vlasenko | 666da5e | 2006-12-26 18:17:42 +0000 | [diff] [blame] | 208 | #else |
| 209 | void sed_free_and_close_stuff(void); |
Mark Whitley | c41e8c8 | 2000-07-12 23:35:21 +0000 | [diff] [blame] | 210 | #endif |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 211 | |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 212 | /* If something bad happens during -i operation, delete temp file */ |
| 213 | |
| 214 | static void cleanup_outname(void) |
| 215 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 216 | if (G.outname) unlink(G.outname); |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 217 | } |
| 218 | |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 219 | /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 220 | |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 221 | static unsigned parse_escapes(char *dest, const char *string, int len, char from, char to) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 222 | { |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 223 | char *d = dest; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 224 | int i = 0; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 225 | |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 226 | if (len == -1) |
| 227 | len = strlen(string); |
| 228 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 229 | while (i < len) { |
| 230 | if (string[i] == '\\') { |
| 231 | if (!to || string[i+1] == from) { |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 232 | if ((*d = to ? to : string[i+1]) == '\0') |
| 233 | return d - dest; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 234 | i += 2; |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 235 | d++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 236 | continue; |
Denis Vlasenko | 4ebaf10 | 2007-01-19 21:33:19 +0000 | [diff] [blame] | 237 | } |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 238 | i++; /* skip backslash in string[] */ |
| 239 | *d++ = '\\'; |
| 240 | /* fall through: copy next char verbatim */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 241 | } |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 242 | if ((*d = string[i++]) == '\0') |
| 243 | return d - dest; |
| 244 | d++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 245 | } |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 246 | *d = '\0'; |
| 247 | return d - dest; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 248 | } |
| 249 | |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 250 | static char *copy_parsing_escapes(const char *string, int len) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 251 | { |
Pascal Bellard | d3e4be3 | 2011-05-05 00:26:37 +0200 | [diff] [blame] | 252 | const char *s; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 253 | char *dest = xmalloc(len + 1); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 254 | |
Pascal Bellard | d3e4be3 | 2011-05-05 00:26:37 +0200 | [diff] [blame] | 255 | /* sed recognizes \n */ |
Denys Vlasenko | 6a0abcc | 2011-05-03 00:52:22 +0200 | [diff] [blame] | 256 | /* GNU sed also recognizes \t and \r */ |
Pascal Bellard | d3e4be3 | 2011-05-05 00:26:37 +0200 | [diff] [blame] | 257 | for (s = "\nn\tt\rr"; *s; s += 2) { |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 258 | len = parse_escapes(dest, string, len, s[1], s[0]); |
Pascal Bellard | d3e4be3 | 2011-05-05 00:26:37 +0200 | [diff] [blame] | 259 | string = dest; |
Pascal Bellard | d3e4be3 | 2011-05-05 00:26:37 +0200 | [diff] [blame] | 260 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 261 | return dest; |
| 262 | } |
| 263 | |
| 264 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 265 | /* |
Eric Andersen | 28b3c53 | 2001-01-02 11:01:31 +0000 | [diff] [blame] | 266 | * index_of_next_unescaped_regexp_delim - walks left to right through a string |
| 267 | * beginning at a specified index and returns the index of the next regular |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 268 | * expression delimiter (typically a forward slash ('/')) not preceded by |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 269 | * a backslash ('\'). A negative delimiter disables square bracket checking. |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 270 | */ |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 271 | static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 272 | { |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 273 | int bracket = -1; |
| 274 | int escaped = 0; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 275 | int idx = 0; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 276 | char ch; |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 277 | |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 278 | if (delimiter < 0) { |
| 279 | bracket--; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 280 | delimiter = -delimiter; |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 281 | } |
| 282 | |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 283 | for (; (ch = str[idx]) != '\0'; idx++) { |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 284 | if (bracket >= 0) { |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 285 | if (ch == ']' |
| 286 | && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^')) |
| 287 | ) { |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 288 | bracket = -1; |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 289 | } |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 290 | } else if (escaped) |
| 291 | escaped = 0; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 292 | else if (ch == '\\') |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 293 | escaped = 1; |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 294 | else if (bracket == -1 && ch == '[') |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 295 | bracket = idx; |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 296 | else if (ch == delimiter) |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 297 | return idx; |
| 298 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 299 | |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 300 | /* if we make it to here, we've hit the end of the string */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 301 | bb_error_msg_and_die("unmatched '%c'", delimiter); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 302 | } |
| 303 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 304 | /* |
| 305 | * Returns the index of the third delimiter |
| 306 | */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 307 | static int parse_regex_delim(const char *cmdstr, char **match, char **replace) |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 308 | { |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 309 | const char *cmdstr_ptr = cmdstr; |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 310 | unsigned char delimiter; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 311 | int idx = 0; |
| 312 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 313 | /* verify that the 's' or 'y' is followed by something. That something |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 314 | * (typically a 'slash') is now our regexp delimiter... */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 315 | if (*cmdstr == '\0') |
| 316 | bb_error_msg_and_die("bad format in substitution expression"); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 317 | delimiter = *cmdstr_ptr++; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 318 | |
| 319 | /* save the match string */ |
| 320 | idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr); |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 321 | *match = copy_parsing_escapes(cmdstr_ptr, idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 322 | |
| 323 | /* save the replacement string */ |
| 324 | cmdstr_ptr += idx + 1; |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 325 | idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr); |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 326 | *replace = copy_parsing_escapes(cmdstr_ptr, idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 327 | |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 328 | return ((cmdstr_ptr - cmdstr) + idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 329 | } |
| 330 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 331 | /* |
| 332 | * returns the index in the string just past where the address ends. |
| 333 | */ |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 334 | static int get_address(const char *my_str, int *linenum, regex_t ** regex) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 335 | { |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 336 | const char *pos = my_str; |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 337 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 338 | if (isdigit(*my_str)) { |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 339 | *linenum = strtol(my_str, (char**)&pos, 10); |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 340 | /* endstr shouldnt ever equal NULL */ |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 341 | } else if (*my_str == '$') { |
Mark Whitley | 0915c4b | 2001-06-11 23:50:06 +0000 | [diff] [blame] | 342 | *linenum = -1; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 343 | pos++; |
| 344 | } else if (*my_str == '/' || *my_str == '\\') { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 345 | int next; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 346 | char delimiter; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 347 | char *temp; |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 348 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 349 | delimiter = '/'; |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 350 | if (*my_str == '\\') |
| 351 | delimiter = *++pos; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 352 | next = index_of_next_unescaped_regexp_delim(delimiter, ++pos); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 353 | temp = copy_parsing_escapes(pos, next); |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 354 | *regex = xzalloc(sizeof(regex_t)); |
Denys Vlasenko | b0e9b72 | 2013-07-21 22:09:44 +0200 | [diff] [blame] | 355 | xregcomp(*regex, temp, G.regex_type); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 356 | free(temp); |
| 357 | /* Move position to next character after last delimiter */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 358 | pos += (next+1); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 359 | } |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 360 | return pos - my_str; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 361 | } |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 362 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 363 | /* Grab a filename. Whitespace at start is skipped, then goes to EOL. */ |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 364 | static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 365 | { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 366 | int start = 0, idx, hack = 0; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 367 | |
| 368 | /* Skip whitespace, then grab filename to end of line */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 369 | while (isspace(filecmdstr[start])) |
| 370 | start++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 371 | idx = start; |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 372 | while (filecmdstr[idx] && filecmdstr[idx] != '\n') |
| 373 | idx++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 374 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 375 | /* If lines glued together, put backslash back. */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 376 | if (filecmdstr[idx] == '\n') |
| 377 | hack = 1; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 378 | if (idx == start) |
| 379 | bb_error_msg_and_die("empty filename"); |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 380 | *retval = xstrndup(filecmdstr+start, idx-start+hack+1); |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 381 | if (hack) |
| 382 | (*retval)[idx] = '\\'; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 383 | |
| 384 | return idx; |
| 385 | } |
| 386 | |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 387 | static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr) |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 388 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 389 | int cflags = G.regex_type; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 390 | char *match; |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 391 | int idx; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 392 | |
| 393 | /* |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 394 | * A substitution command should look something like this: |
David A. Wheeler | 80a068d | 2013-12-05 20:42:17 -0500 | [diff] [blame] | 395 | * s/match/replace/ #giIpw |
Mark Whitley | 0e4cec0 | 2000-08-21 21:29:20 +0000 | [diff] [blame] | 396 | * || | ||| |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 397 | * mandatory optional |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 398 | */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 399 | idx = parse_regex_delim(substr, &match, &sed_cmd->string); |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 400 | |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 401 | /* determine the number of back references in the match string */ |
| 402 | /* Note: we compute this here rather than in the do_subst_command() |
| 403 | * function to save processor time, at the expense of a little more memory |
| 404 | * (4 bits) per sed_cmd */ |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 405 | |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 406 | /* process the flags */ |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 407 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 408 | sed_cmd->which_match = 1; |
Denys Vlasenko | 9caea24 | 2014-09-16 01:11:13 +0200 | [diff] [blame] | 409 | dbg("s flags:'%s'", substr + idx + 1); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 410 | while (substr[++idx]) { |
Denys Vlasenko | 9caea24 | 2014-09-16 01:11:13 +0200 | [diff] [blame] | 411 | dbg("s flag:'%c'", substr[idx]); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 412 | /* Parse match number */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 413 | if (isdigit(substr[idx])) { |
| 414 | if (match[0] != '^') { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 415 | /* Match 0 treated as all, multiple matches we take the last one. */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 416 | const char *pos = substr + idx; |
| 417 | /* FIXME: error check? */ |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 418 | sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10); |
Denys Vlasenko | 9caea24 | 2014-09-16 01:11:13 +0200 | [diff] [blame] | 419 | idx = pos - substr - 1; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 420 | } |
| 421 | continue; |
| 422 | } |
Rob Landley | 40ec4ae | 2004-01-04 06:42:14 +0000 | [diff] [blame] | 423 | /* Skip spaces */ |
Denys Vlasenko | 6935ec9 | 2009-10-22 19:42:26 +0200 | [diff] [blame] | 424 | if (isspace(substr[idx])) |
| 425 | continue; |
Rob Landley | 40ec4ae | 2004-01-04 06:42:14 +0000 | [diff] [blame] | 426 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 427 | switch (substr[idx]) { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 428 | /* Replace all occurrences */ |
| 429 | case 'g': |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 430 | if (match[0] != '^') |
| 431 | sed_cmd->which_match = 0; |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 432 | break; |
| 433 | /* Print pattern space */ |
| 434 | case 'p': |
| 435 | sed_cmd->sub_p = 1; |
| 436 | break; |
| 437 | /* Write to file */ |
| 438 | case 'w': |
| 439 | { |
Denys Vlasenko | 2a4bba3 | 2016-01-24 15:52:16 +0100 | [diff] [blame] | 440 | char *fname; |
| 441 | idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname); |
| 442 | sed_cmd->sw_file = xfopen_for_write(fname); |
| 443 | sed_cmd->sw_last_char = '\n'; |
| 444 | free(fname); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 445 | break; |
| 446 | } |
| 447 | /* Ignore case (gnu exension) */ |
David A. Wheeler | 80a068d | 2013-12-05 20:42:17 -0500 | [diff] [blame] | 448 | case 'i': |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 449 | case 'I': |
| 450 | cflags |= REG_ICASE; |
| 451 | break; |
| 452 | /* Comment */ |
| 453 | case '#': |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 454 | // while (substr[++idx]) continue; |
| 455 | idx += strlen(substr + idx); // same |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 456 | /* Fall through */ |
| 457 | /* End of command */ |
| 458 | case ';': |
| 459 | case '}': |
| 460 | goto out; |
| 461 | default: |
Denys Vlasenko | 9caea24 | 2014-09-16 01:11:13 +0200 | [diff] [blame] | 462 | dbg("s bad flags:'%s'", substr + idx); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 463 | bb_error_msg_and_die("bad option in substitution expression"); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 464 | } |
| 465 | } |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 466 | out: |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 467 | /* compile the match string into a regex */ |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 468 | if (*match != '\0') { |
| 469 | /* If match is empty, we use last regex used at runtime */ |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 470 | sed_cmd->sub_match = xzalloc(sizeof(regex_t)); |
| 471 | dbg("xregcomp('%s',%x)", match, cflags); |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 472 | xregcomp(sed_cmd->sub_match, match, cflags); |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 473 | dbg("regcomp ok"); |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 474 | } |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 475 | free(match); |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 476 | |
| 477 | return idx; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 478 | } |
| 479 | |
Glenn L McGrath | 2971ef1 | 2003-03-18 01:19:23 +0000 | [diff] [blame] | 480 | /* |
| 481 | * Process the commands arguments |
| 482 | */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 483 | static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 484 | { |
Denys Vlasenko | 3e134eb | 2016-04-22 18:09:21 +0200 | [diff] [blame] | 485 | static const char cmd_letters[] ALIGN1 = "saicrw:btTydDgGhHlnNpPqx={}"; |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 486 | enum { |
| 487 | IDX_s = 0, |
| 488 | IDX_a, |
| 489 | IDX_i, |
| 490 | IDX_c, |
| 491 | IDX_r, |
| 492 | IDX_w, |
| 493 | IDX_colon, |
| 494 | IDX_b, |
| 495 | IDX_t, |
| 496 | IDX_T, |
| 497 | IDX_y, |
| 498 | IDX_d, |
| 499 | IDX_D, |
| 500 | IDX_g, |
| 501 | IDX_G, |
| 502 | IDX_h, |
| 503 | IDX_H, |
| 504 | IDX_l, |
| 505 | IDX_n, |
| 506 | IDX_N, |
| 507 | IDX_p, |
| 508 | IDX_P, |
| 509 | IDX_q, |
| 510 | IDX_x, |
| 511 | IDX_equal, |
| 512 | IDX_lbrace, |
| 513 | IDX_rbrace, |
| 514 | IDX_nul |
| 515 | }; |
Denys Vlasenko | 7b85ec3 | 2015-10-13 17:17:34 +0200 | [diff] [blame] | 516 | unsigned idx; |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 517 | |
Denys Vlasenko | 7b85ec3 | 2015-10-13 17:17:34 +0200 | [diff] [blame] | 518 | BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul); |
| 519 | |
| 520 | idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters; |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 521 | |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 522 | /* handle (s)ubstitution command */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 523 | if (idx == IDX_s) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 524 | cmdstr += parse_subst_cmd(sed_cmd, cmdstr); |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 525 | } |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 526 | /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 527 | else if (idx <= IDX_c) { /* a,i,c */ |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 528 | unsigned len; |
| 529 | |
Mimi Li | 37a79c0 | 2012-07-24 13:20:12 +0200 | [diff] [blame] | 530 | if (idx < IDX_c) { /* a,i */ |
| 531 | if (sed_cmd->end_line || sed_cmd->end_match) |
| 532 | bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd); |
| 533 | } |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 534 | for (;;) { |
| 535 | if (*cmdstr == '\n' || *cmdstr == '\\') { |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 536 | cmdstr++; |
| 537 | break; |
Denys Vlasenko | c0dab37 | 2009-10-22 22:28:08 +0200 | [diff] [blame] | 538 | } |
| 539 | if (!isspace(*cmdstr)) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 540 | break; |
Denys Vlasenko | c0dab37 | 2009-10-22 22:28:08 +0200 | [diff] [blame] | 541 | cmdstr++; |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 542 | } |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 543 | len = strlen(cmdstr); |
| 544 | sed_cmd->string = copy_parsing_escapes(cmdstr, len); |
| 545 | cmdstr += len; |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 546 | /* "\anychar" -> "anychar" */ |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 547 | parse_escapes(sed_cmd->string, sed_cmd->string, -1, '\0', '\0'); |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 548 | } |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 549 | /* handle file cmds: (r)ead */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 550 | else if (idx <= IDX_w) { /* r,w */ |
Mimi Li | 37a79c0 | 2012-07-24 13:20:12 +0200 | [diff] [blame] | 551 | if (idx < IDX_w) { /* r */ |
| 552 | if (sed_cmd->end_line || sed_cmd->end_match) |
| 553 | bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd); |
| 554 | } |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 555 | cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string); |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 556 | if (sed_cmd->cmd == 'w') { |
Denis Vlasenko | 5415c85 | 2008-07-21 23:05:26 +0000 | [diff] [blame] | 557 | sed_cmd->sw_file = xfopen_for_write(sed_cmd->string); |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 558 | sed_cmd->sw_last_char = '\n'; |
| 559 | } |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 560 | } |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 561 | /* handle branch commands */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 562 | else if (idx <= IDX_T) { /* :,b,t,T */ |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 563 | int length; |
| 564 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 565 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | f452356 | 2003-09-14 06:01:14 +0000 | [diff] [blame] | 566 | length = strcspn(cmdstr, semicolon_whitespace); |
| 567 | if (length) { |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 568 | sed_cmd->string = xstrndup(cmdstr, length); |
Glenn L McGrath | f452356 | 2003-09-14 06:01:14 +0000 | [diff] [blame] | 569 | cmdstr += length; |
| 570 | } |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 571 | } |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 572 | /* translation command */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 573 | else if (idx == IDX_y) { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 574 | char *match, *replace; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 575 | int i = cmdstr[0]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 576 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 577 | cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 578 | /* \n already parsed, but \delimiter needs unescaping. */ |
Denys Vlasenko | cbdff15 | 2016-04-24 16:18:03 +0200 | [diff] [blame] | 579 | parse_escapes(match, match, -1, i, i); |
| 580 | parse_escapes(replace, replace, -1, i, i); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 581 | |
Rob Landley | 9ffd423 | 2006-05-21 18:30:35 +0000 | [diff] [blame] | 582 | sed_cmd->string = xzalloc((strlen(match) + 1) * 2); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 583 | for (i = 0; match[i] && replace[i]; i++) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 584 | sed_cmd->string[i*2] = match[i]; |
| 585 | sed_cmd->string[i*2+1] = replace[i]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 586 | } |
| 587 | free(match); |
| 588 | free(replace); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 589 | } |
Glenn L McGrath | 2971ef1 | 2003-03-18 01:19:23 +0000 | [diff] [blame] | 590 | /* if it wasnt a single-letter command that takes no arguments |
| 591 | * then it must be an invalid command. |
| 592 | */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 593 | else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 594 | bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd); |
Mark Whitley | 1f3b9f2 | 2001-05-11 22:27:13 +0000 | [diff] [blame] | 595 | } |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 596 | |
| 597 | /* give back whatever's left over */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 598 | return cmdstr; |
Eric Andersen | 50d6360 | 1999-11-09 01:47:36 +0000 | [diff] [blame] | 599 | } |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 600 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 601 | |
| 602 | /* Parse address+command sets, skipping comment lines. */ |
| 603 | |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 604 | static void add_cmd(const char *cmdstr) |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 605 | { |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 606 | sed_cmd_t *sed_cmd; |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 607 | unsigned len, n; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 608 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 609 | /* Append this line to any unfinished line from last time. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 610 | if (G.add_cmd_line) { |
| 611 | char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr); |
| 612 | free(G.add_cmd_line); |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 613 | cmdstr = G.add_cmd_line = tp; |
Rob Landley | 12d8755 | 2006-06-05 17:32:44 +0000 | [diff] [blame] | 614 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 615 | |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 616 | /* If this line ends with unescaped backslash, request next line. */ |
| 617 | n = len = strlen(cmdstr); |
| 618 | while (n && cmdstr[n-1] == '\\') |
| 619 | n--; |
| 620 | if ((len - n) & 1) { /* if odd number of trailing backslashes */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 621 | if (!G.add_cmd_line) |
| 622 | G.add_cmd_line = xstrdup(cmdstr); |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 623 | G.add_cmd_line[len-1] = '\0'; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 624 | return; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 625 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 626 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 627 | /* Loop parsing all commands in this line. */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 628 | while (*cmdstr) { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 629 | /* Skip leading whitespace and semicolons */ |
| 630 | cmdstr += strspn(cmdstr, semicolon_whitespace); |
| 631 | |
| 632 | /* If no more commands, exit. */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 633 | if (!*cmdstr) break; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 634 | |
| 635 | /* if this is a comment, jump past it and keep going */ |
| 636 | if (*cmdstr == '#') { |
| 637 | /* "#n" is the same as using -n on the command line */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 638 | if (cmdstr[1] == 'n') |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 639 | G.be_quiet++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 640 | cmdstr = strpbrk(cmdstr, "\n\r"); |
| 641 | if (!cmdstr) break; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 642 | continue; |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 643 | } |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 644 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 645 | /* parse the command |
| 646 | * format is: [addr][,addr][!]cmd |
| 647 | * |----||-----||-| |
| 648 | * part1 part2 part3 |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 649 | */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 650 | |
Rob Landley | 9ffd423 | 2006-05-21 18:30:35 +0000 | [diff] [blame] | 651 | sed_cmd = xzalloc(sizeof(sed_cmd_t)); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 652 | |
| 653 | /* first part (if present) is an address: either a '$', a number or a /regex/ */ |
| 654 | cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match); |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 655 | sed_cmd->beg_line_orig = sed_cmd->beg_line; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 656 | |
| 657 | /* second part (if present) will begin with a comma */ |
| 658 | if (*cmdstr == ',') { |
| 659 | int idx; |
| 660 | |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 661 | cmdstr++; |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 662 | if (*cmdstr == '+' && isdigit(cmdstr[1])) { |
| 663 | /* http://sed.sourceforge.net/sedfaq3.html#s3.3 |
| 664 | * Under GNU sed 3.02+, ssed, and sed15+, <address2> |
| 665 | * may also be a notation of the form +num, |
| 666 | * indicating the next num lines after <address1> is |
| 667 | * matched. |
| 668 | * GNU sed 4.2.1 accepts even "+" (meaning "+0"). |
| 669 | * We don't (we check for isdigit, see above), think |
| 670 | * about the "+-3" case. |
| 671 | */ |
| 672 | char *end; |
| 673 | /* code is smaller compared to using &cmdstr here: */ |
| 674 | idx = strtol(cmdstr+1, &end, 10); |
| 675 | sed_cmd->end_line = -2 - idx; |
| 676 | cmdstr = end; |
| 677 | } else { |
| 678 | idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match); |
| 679 | cmdstr += idx; |
| 680 | idx--; /* if 0, trigger error check below */ |
| 681 | } |
| 682 | if (idx < 0) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 683 | bb_error_msg_and_die("no address after comma"); |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 684 | sed_cmd->end_line_orig = sed_cmd->end_line; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 685 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 686 | |
| 687 | /* skip whitespace before the command */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 688 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 689 | |
| 690 | /* Check for inversion flag */ |
| 691 | if (*cmdstr == '!') { |
| 692 | sed_cmd->invert = 1; |
| 693 | cmdstr++; |
| 694 | |
| 695 | /* skip whitespace before the command */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 696 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 697 | } |
| 698 | |
| 699 | /* last part (mandatory) will be a command */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 700 | if (!*cmdstr) |
| 701 | bb_error_msg_and_die("missing command"); |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 702 | sed_cmd->cmd = *cmdstr++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 703 | cmdstr = parse_cmd_args(sed_cmd, cmdstr); |
| 704 | |
Denys Vlasenko | e93d156 | 2013-07-08 01:43:40 +0200 | [diff] [blame] | 705 | /* cmdstr now points past args. |
| 706 | * GNU sed requires a separator, if there are more commands, |
| 707 | * else it complains "char N: extra characters after command". |
| 708 | * Example: "sed 'p;d'". We also allow "sed 'pd'". |
| 709 | */ |
| 710 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 711 | /* Add the command to the command array */ |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 712 | *G.sed_cmd_tail = sed_cmd; |
| 713 | G.sed_cmd_tail = &sed_cmd->next; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 714 | } |
| 715 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 716 | /* If we glued multiple lines together, free the memory. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 717 | free(G.add_cmd_line); |
| 718 | G.add_cmd_line = NULL; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 719 | } |
| 720 | |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 721 | /* Append to a string, reallocating memory as necessary. */ |
| 722 | |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 723 | #define PIPE_GROW 64 |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 724 | |
Eric Andersen | 14f5c8d | 2005-04-16 19:39:00 +0000 | [diff] [blame] | 725 | static void pipe_putc(char c) |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 726 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 727 | if (G.pipeline.idx == G.pipeline.len) { |
| 728 | G.pipeline.buf = xrealloc(G.pipeline.buf, |
| 729 | G.pipeline.len + PIPE_GROW); |
| 730 | G.pipeline.len += PIPE_GROW; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 731 | } |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 732 | G.pipeline.buf[G.pipeline.idx++] = c; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 733 | } |
| 734 | |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 735 | static void do_subst_w_backrefs(char *line, char *replace) |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 736 | { |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 737 | int i, j; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 738 | |
| 739 | /* go through the replacement string */ |
| 740 | for (i = 0; replace[i]; i++) { |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 741 | /* if we find a backreference (\1, \2, etc.) print the backref'ed text */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 742 | if (replace[i] == '\\') { |
| 743 | unsigned backref = replace[++i] - '0'; |
| 744 | if (backref <= 9) { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 745 | /* print out the text held in G.regmatch[backref] */ |
| 746 | if (G.regmatch[backref].rm_so != -1) { |
| 747 | j = G.regmatch[backref].rm_so; |
| 748 | while (j < G.regmatch[backref].rm_eo) |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 749 | pipe_putc(line[j++]); |
| 750 | } |
| 751 | continue; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 752 | } |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 753 | /* I _think_ it is impossible to get '\' to be |
| 754 | * the last char in replace string. Thus we dont check |
| 755 | * for replace[i] == NUL. (counterexample anyone?) */ |
| 756 | /* if we find a backslash escaped character, print the character */ |
| 757 | pipe_putc(replace[i]); |
| 758 | continue; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 759 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 760 | /* if we find an unescaped '&' print out the whole matched text. */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 761 | if (replace[i] == '&') { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 762 | j = G.regmatch[0].rm_so; |
| 763 | while (j < G.regmatch[0].rm_eo) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 764 | pipe_putc(line[j++]); |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 765 | continue; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 766 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 767 | /* Otherwise just output the character. */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 768 | pipe_putc(replace[i]); |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 769 | } |
| 770 | } |
| 771 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 772 | static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p) |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 773 | { |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 774 | char *line = *line_p; |
Denis Vlasenko | 77ad97f | 2008-05-13 02:27:31 +0000 | [diff] [blame] | 775 | unsigned match_count = 0; |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 776 | bool altered = 0; |
| 777 | bool prev_match_empty = 1; |
| 778 | bool tried_at_eol = 0; |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 779 | regex_t *current_regex; |
| 780 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 781 | current_regex = sed_cmd->sub_match; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 782 | /* Handle empty regex. */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 783 | if (!current_regex) { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 784 | current_regex = G.previous_regex_ptr; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 785 | if (!current_regex) |
Denis Vlasenko | d3d004d | 2006-10-27 09:02:31 +0000 | [diff] [blame] | 786 | bb_error_msg_and_die("no previous regexp"); |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 787 | } |
| 788 | G.previous_regex_ptr = current_regex; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 789 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 790 | /* Find the first match */ |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 791 | dbg("matching '%s'", line); |
| 792 | if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) { |
| 793 | dbg("no match"); |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 794 | return 0; |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 795 | } |
| 796 | dbg("match"); |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 797 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 798 | /* Initialize temporary output buffer. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 799 | G.pipeline.buf = xmalloc(PIPE_GROW); |
| 800 | G.pipeline.len = PIPE_GROW; |
| 801 | G.pipeline.idx = 0; |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 802 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 803 | /* Now loop through, substituting for matches */ |
| 804 | do { |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 805 | int start = G.regmatch[0].rm_so; |
| 806 | int end = G.regmatch[0].rm_eo; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 807 | int i; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 808 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 809 | match_count++; |
| 810 | |
| 811 | /* If we aren't interested in this match, output old line to |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 812 | * end of match and continue */ |
Denis Vlasenko | 77ad97f | 2008-05-13 02:27:31 +0000 | [diff] [blame] | 813 | if (sed_cmd->which_match |
| 814 | && (sed_cmd->which_match != match_count) |
| 815 | ) { |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 816 | for (i = 0; i < end; i++) |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 817 | pipe_putc(*line++); |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 818 | /* Null match? Print one more char */ |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 819 | if (start == end && *line) |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 820 | pipe_putc(*line++); |
Denys Vlasenko | b84dafb | 2012-04-24 19:27:34 +0200 | [diff] [blame] | 821 | goto next; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 822 | } |
| 823 | |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 824 | /* Print everything before the match */ |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 825 | for (i = 0; i < start; i++) |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 826 | pipe_putc(line[i]); |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 827 | |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 828 | /* Then print the substitution string, |
| 829 | * unless we just matched empty string after non-empty one. |
| 830 | * Example: string "cccd", pattern "c*", repl "R": |
| 831 | * result is "RdR", not "RRdR": first match "ccc", |
| 832 | * second is "" before "d", third is "" after "d". |
| 833 | * Second match is NOT replaced! |
| 834 | */ |
Denys Vlasenko | 84406e4 | 2012-06-07 16:34:57 +0200 | [diff] [blame] | 835 | if (prev_match_empty || start != 0 || start != end) { |
Denys Vlasenko | 37ca36a | 2012-06-08 10:25:31 +0200 | [diff] [blame] | 836 | //dbg("%d %d %d", prev_match_empty, start, end); |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 837 | dbg("inserting replacement at %d in '%s'", start, line); |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 838 | do_subst_w_backrefs(line, sed_cmd->string); |
Denys Vlasenko | 37ca36a | 2012-06-08 10:25:31 +0200 | [diff] [blame] | 839 | /* Flag that something has changed */ |
| 840 | altered = 1; |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 841 | } else { |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 842 | dbg("NOT inserting replacement at %d in '%s'", start, line); |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 843 | } |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 844 | |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 845 | /* If matched string is empty (f.e. "c*" pattern), |
| 846 | * copy verbatim one char after it before attempting more matches |
| 847 | */ |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 848 | prev_match_empty = (start == end); |
Denys Vlasenko | 37ca36a | 2012-06-08 10:25:31 +0200 | [diff] [blame] | 849 | if (prev_match_empty) { |
| 850 | if (!line[end]) { |
| 851 | tried_at_eol = 1; |
| 852 | } else { |
| 853 | pipe_putc(line[end]); |
| 854 | end++; |
| 855 | } |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 856 | } |
| 857 | |
| 858 | /* Advance past the match */ |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 859 | dbg("line += %d", end); |
| 860 | line += end; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 861 | |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 862 | /* if we're not doing this globally, get out now */ |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 863 | if (sed_cmd->which_match != 0) |
| 864 | break; |
Denys Vlasenko | b84dafb | 2012-04-24 19:27:34 +0200 | [diff] [blame] | 865 | next: |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 866 | /* Exit if we are at EOL and already tried matching at it */ |
| 867 | if (*line == '\0') { |
| 868 | if (tried_at_eol) |
| 869 | break; |
| 870 | tried_at_eol = 1; |
| 871 | } |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 872 | |
Denys Vlasenko | c35545a | 2012-06-04 14:45:09 +0200 | [diff] [blame] | 873 | //maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL? |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 874 | } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH); |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 875 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 876 | /* Copy rest of string into output pipeline */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 877 | while (1) { |
| 878 | char c = *line++; |
| 879 | pipe_putc(c); |
| 880 | if (c == '\0') |
| 881 | break; |
| 882 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 883 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 884 | free(*line_p); |
| 885 | *line_p = G.pipeline.buf; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 886 | return altered; |
| 887 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 888 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 889 | /* Set command pointer to point to this label. (Does not handle null label.) */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 890 | static sed_cmd_t *branch_to(char *label) |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 891 | { |
| 892 | sed_cmd_t *sed_cmd; |
Glenn L McGrath | d4185b0 | 2003-04-11 17:10:23 +0000 | [diff] [blame] | 893 | |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 894 | for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 895 | if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) { |
| 896 | return sed_cmd; |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 897 | } |
| 898 | } |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 899 | bb_error_msg_and_die("can't find label for jump to '%s'", label); |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 900 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 901 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 902 | static void append(char *s) |
| 903 | { |
Denys Vlasenko | 9d46a7a | 2013-10-30 10:22:47 +0100 | [diff] [blame] | 904 | llist_add_to_end(&G.append_head, s); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 905 | } |
| 906 | |
Denys Vlasenko | c44539f | 2013-10-30 14:25:22 +0100 | [diff] [blame] | 907 | /* Output line of text. */ |
| 908 | /* Note: |
| 909 | * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed. |
| 910 | * Without them, we had this: |
| 911 | * echo -n thingy >z1 |
| 912 | * echo -n again >z2 |
| 913 | * >znull |
| 914 | * sed "s/i/z/" z1 z2 znull | hexdump -vC |
| 915 | * output: |
| 916 | * gnu sed 4.1.5: |
| 917 | * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn| |
| 918 | * bbox: |
| 919 | * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn| |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 920 | */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 921 | enum { |
| 922 | NO_EOL_CHAR = 1, |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 923 | LAST_IS_NUL = 2, |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 924 | }; |
Denys Vlasenko | c44539f | 2013-10-30 14:25:22 +0100 | [diff] [blame] | 925 | static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char) |
| 926 | { |
| 927 | char lpc = *last_puts_char; |
| 928 | |
| 929 | /* Need to insert a '\n' between two files because first file's |
| 930 | * last line wasn't terminated? */ |
| 931 | if (lpc != '\n' && lpc != '\0') { |
| 932 | fputc('\n', file); |
| 933 | lpc = '\n'; |
| 934 | } |
| 935 | fputs(s, file); |
| 936 | |
| 937 | /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */ |
| 938 | if (s[0]) |
| 939 | lpc = 'x'; |
| 940 | |
| 941 | /* had trailing '\0' and it was last char of file? */ |
| 942 | if (last_gets_char == LAST_IS_NUL) { |
| 943 | fputc('\0', file); |
| 944 | lpc = 'x'; /* */ |
| 945 | } else |
| 946 | /* had trailing '\n' or '\0'? */ |
| 947 | if (last_gets_char != NO_EOL_CHAR) { |
| 948 | fputc(last_gets_char, file); |
| 949 | lpc = last_gets_char; |
| 950 | } |
| 951 | |
| 952 | if (ferror(file)) { |
| 953 | xfunc_error_retval = 4; /* It's what gnu sed exits with... */ |
| 954 | bb_error_msg_and_die(bb_msg_write_error); |
| 955 | } |
| 956 | *last_puts_char = lpc; |
| 957 | } |
| 958 | |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 959 | static void flush_append(char *last_puts_char) |
Denys Vlasenko | c44539f | 2013-10-30 14:25:22 +0100 | [diff] [blame] | 960 | { |
| 961 | char *data; |
| 962 | |
| 963 | /* Output appended lines. */ |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 964 | while ((data = (char *)llist_pop(&G.append_head)) != NULL) { |
| 965 | /* Append command does not respect "nonterminated-ness" |
| 966 | * of last line. Try this: |
| 967 | * $ echo -n "woot" | sed -e '/woot/a woo' - |
| 968 | * woot |
| 969 | * woo |
| 970 | * (both lines are terminated with \n) |
| 971 | * Therefore we do not propagate "last_gets_char" here, |
| 972 | * pass '\n' instead: |
| 973 | */ |
| 974 | puts_maybe_newline(data, G.nonstdout, last_puts_char, '\n'); |
Denys Vlasenko | c44539f | 2013-10-30 14:25:22 +0100 | [diff] [blame] | 975 | free(data); |
| 976 | } |
| 977 | } |
| 978 | |
| 979 | /* Get next line of input from G.input_file_list, flushing append buffer and |
| 980 | * noting if we ran out of files without a newline on the last line we read. |
| 981 | */ |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 982 | static char *get_next_line(char *gets_char, char *last_puts_char) |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 983 | { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 984 | char *temp = NULL; |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 985 | int len; |
| 986 | char gc; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 987 | |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 988 | flush_append(last_puts_char); |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 989 | |
| 990 | /* will be returned if last line in the file |
| 991 | * doesn't end with either '\n' or '\0' */ |
| 992 | gc = NO_EOL_CHAR; |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 993 | for (; G.current_input_file <= G.last_input_file; G.current_input_file++) { |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 994 | FILE *fp = G.current_fp; |
| 995 | if (!fp) { |
| 996 | const char *path = G.input_file_list[G.current_input_file]; |
| 997 | fp = stdin; |
| 998 | if (path != bb_msg_standard_input) { |
| 999 | fp = fopen_or_warn(path, "r"); |
| 1000 | if (!fp) { |
| 1001 | G.exitcode = EXIT_FAILURE; |
| 1002 | continue; |
| 1003 | } |
| 1004 | } |
| 1005 | G.current_fp = fp; |
| 1006 | } |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 1007 | /* Read line up to a newline or NUL byte, inclusive, |
| 1008 | * return malloc'ed char[]. length of the chunk read |
| 1009 | * is stored in len. NULL if EOF/error */ |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 1010 | temp = bb_get_chunk_from_file(fp, &len); |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 1011 | if (temp) { |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 1012 | /* len > 0 here, it's ok to do temp[len-1] */ |
| 1013 | char c = temp[len-1]; |
| 1014 | if (c == '\n' || c == '\0') { |
| 1015 | temp[len-1] = '\0'; |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 1016 | gc = c; |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 1017 | if (c == '\0') { |
| 1018 | int ch = fgetc(fp); |
| 1019 | if (ch != EOF) |
| 1020 | ungetc(ch, fp); |
| 1021 | else |
| 1022 | gc = LAST_IS_NUL; |
| 1023 | } |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 1024 | } |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 1025 | /* else we put NO_EOL_CHAR into *gets_char */ |
| 1026 | break; |
| 1027 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1028 | /* NB: I had the idea of peeking next file(s) and returning |
| 1029 | * NO_EOL_CHAR only if it is the *last* non-empty |
| 1030 | * input file. But there is a case where this won't work: |
| 1031 | * file1: "a woo\nb woo" |
| 1032 | * file2: "c no\nd no" |
| 1033 | * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang" |
| 1034 | * (note: *no* newline after "b bang"!) */ |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 1035 | } |
| 1036 | /* Close this file and advance to next one */ |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 1037 | fclose_if_not_stdin(fp); |
| 1038 | G.current_fp = NULL; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1039 | } |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1040 | *gets_char = gc; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1041 | return temp; |
| 1042 | } |
| 1043 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1044 | #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n)) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1045 | |
Denis Vlasenko | 8274e06 | 2007-08-06 03:41:08 +0000 | [diff] [blame] | 1046 | static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space) |
| 1047 | { |
| 1048 | int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0); |
| 1049 | if (retval) |
| 1050 | G.previous_regex_ptr = sed_cmd->beg_match; |
| 1051 | return retval; |
| 1052 | } |
| 1053 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 1054 | /* Process all the lines in all the files */ |
| 1055 | |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 1056 | static void process_files(void) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1057 | { |
Rob Landley | ce4f0e9 | 2004-10-30 06:54:19 +0000 | [diff] [blame] | 1058 | char *pattern_space, *next_line; |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1059 | int linenum = 0; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1060 | char last_puts_char = '\n'; |
| 1061 | char last_gets_char, next_gets_char; |
Denis Vlasenko | 2ea630f | 2006-12-10 02:52:19 +0000 | [diff] [blame] | 1062 | sed_cmd_t *sed_cmd; |
| 1063 | int substituted; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1064 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 1065 | /* Prime the pump */ |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 1066 | next_line = get_next_line(&next_gets_char, &last_puts_char); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1067 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1068 | /* Go through every line in each file */ |
Denis Vlasenko | f39c7c0 | 2008-02-28 17:59:01 +0000 | [diff] [blame] | 1069 | again: |
Denis Vlasenko | 2ea630f | 2006-12-10 02:52:19 +0000 | [diff] [blame] | 1070 | substituted = 0; |
Glenn L McGrath | 505bd0f | 2003-03-08 05:21:02 +0000 | [diff] [blame] | 1071 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1072 | /* Advance to next line. Stop if out of lines. */ |
| 1073 | pattern_space = next_line; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1074 | if (!pattern_space) |
| 1075 | return; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1076 | last_gets_char = next_gets_char; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1077 | |
| 1078 | /* Read one line in advance so we can act on the last line, |
| 1079 | * the '$' address */ |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 1080 | next_line = get_next_line(&next_gets_char, &last_puts_char); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1081 | linenum++; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1082 | |
| 1083 | /* For every line, go through all the commands */ |
Denis Vlasenko | f39c7c0 | 2008-02-28 17:59:01 +0000 | [diff] [blame] | 1084 | restart: |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1085 | for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1086 | int old_matched, matched; |
Eric Andersen | 52a3c27 | 2003-12-23 08:53:51 +0000 | [diff] [blame] | 1087 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1088 | old_matched = sed_cmd->in_match; |
Mark Whitley | 0915c4b | 2001-06-11 23:50:06 +0000 | [diff] [blame] | 1089 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1090 | /* Determine if this command matches this line: */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1091 | |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1092 | dbg("match1:%d", sed_cmd->in_match); |
| 1093 | dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line |
| 1094 | && !sed_cmd->beg_match && !sed_cmd->end_match)); |
| 1095 | dbg("match3:%d", (sed_cmd->beg_line > 0 |
| 1096 | && (sed_cmd->end_line || sed_cmd->end_match |
| 1097 | ? (sed_cmd->beg_line <= linenum) |
| 1098 | : (sed_cmd->beg_line == linenum) |
| 1099 | ) |
| 1100 | )); |
| 1101 | dbg("match4:%d", (beg_match(sed_cmd, pattern_space))); |
| 1102 | dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL)); |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1103 | |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1104 | /* Are we continuing a previous multi-line match? */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1105 | sed_cmd->in_match = sed_cmd->in_match |
| 1106 | /* Or is no range necessary? */ |
| 1107 | || (!sed_cmd->beg_line && !sed_cmd->end_line |
| 1108 | && !sed_cmd->beg_match && !sed_cmd->end_match) |
| 1109 | /* Or did we match the start of a numerical range? */ |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1110 | || (sed_cmd->beg_line > 0 |
| 1111 | && (sed_cmd->end_line || sed_cmd->end_match |
Denys Vlasenko | c467924 | 2010-06-04 01:31:48 +0200 | [diff] [blame] | 1112 | /* note: even if end is numeric and is < linenum too, |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1113 | * GNU sed matches! We match too, therefore we don't |
| 1114 | * check here that linenum <= end. |
| 1115 | * Example: |
| 1116 | * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p' |
| 1117 | * first three input lines are deleted; |
| 1118 | * 4th line is matched and printed |
| 1119 | * by "2,3" (!) and by "4" ranges |
| 1120 | */ |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1121 | ? (sed_cmd->beg_line <= linenum) /* N,end */ |
| 1122 | : (sed_cmd->beg_line == linenum) /* N */ |
| 1123 | ) |
| 1124 | ) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1125 | /* Or does this line match our begin address regex? */ |
Denis Vlasenko | 8274e06 | 2007-08-06 03:41:08 +0000 | [diff] [blame] | 1126 | || (beg_match(sed_cmd, pattern_space)) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1127 | /* Or did we match last line of input? */ |
| 1128 | || (sed_cmd->beg_line == -1 && next_line == NULL); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1129 | |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1130 | /* Snapshot the value */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1131 | matched = sed_cmd->in_match; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1132 | |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1133 | dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d", |
| 1134 | sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1135 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1136 | /* Is this line the end of the current match? */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1137 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1138 | if (matched) { |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 1139 | if (sed_cmd->end_line <= -2) { |
| 1140 | /* address2 is +N, i.e. N lines from beg_line */ |
| 1141 | sed_cmd->end_line = linenum + (-sed_cmd->end_line - 2); |
| 1142 | } |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1143 | /* once matched, "n,xxx" range is dead, disabling it */ |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1144 | if (sed_cmd->beg_line > 0) { |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1145 | sed_cmd->beg_line = -2; |
Denys Vlasenko | 2e284a4 | 2010-08-01 04:14:46 +0200 | [diff] [blame] | 1146 | } |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 1147 | dbg("end1:%d", sed_cmd->end_line ? sed_cmd->end_line == -1 |
| 1148 | ? !next_line : (sed_cmd->end_line <= linenum) |
| 1149 | : !sed_cmd->end_match); |
| 1150 | dbg("end2:%d", sed_cmd->end_match && old_matched |
| 1151 | && !regexec(sed_cmd->end_match,pattern_space, 0, NULL, 0)); |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1152 | sed_cmd->in_match = !( |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1153 | /* has the ending line come, or is this a single address command? */ |
Denys Vlasenko | 52d8370 | 2011-05-03 00:51:43 +0200 | [diff] [blame] | 1154 | (sed_cmd->end_line |
| 1155 | ? sed_cmd->end_line == -1 |
| 1156 | ? !next_line |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1157 | : (sed_cmd->end_line <= linenum) |
| 1158 | : !sed_cmd->end_match |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1159 | ) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1160 | /* or does this line matches our last address regex */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1161 | || (sed_cmd->end_match && old_matched |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1162 | && (regexec(sed_cmd->end_match, |
Denys Vlasenko | 6830ade | 2013-01-15 13:58:01 +0100 | [diff] [blame] | 1163 | pattern_space, 0, NULL, 0) == 0) |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1164 | ) |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 1165 | ); |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 1166 | } |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 1167 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1168 | /* Skip blocks of commands we didn't match */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1169 | if (sed_cmd->cmd == '{') { |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1170 | if (sed_cmd->invert ? matched : !matched) { |
Denys Vlasenko | f2c16ed | 2010-04-20 04:00:03 -0400 | [diff] [blame] | 1171 | unsigned nest_cnt = 0; |
| 1172 | while (1) { |
| 1173 | if (sed_cmd->cmd == '{') |
| 1174 | nest_cnt++; |
| 1175 | if (sed_cmd->cmd == '}') { |
| 1176 | nest_cnt--; |
| 1177 | if (nest_cnt == 0) |
| 1178 | break; |
| 1179 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1180 | sed_cmd = sed_cmd->next; |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1181 | if (!sed_cmd) |
| 1182 | bb_error_msg_and_die("unterminated {"); |
| 1183 | } |
| 1184 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1185 | continue; |
| 1186 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1187 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1188 | /* Okay, so did this line match? */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1189 | if (sed_cmd->invert ? matched : !matched) |
| 1190 | continue; /* no */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1191 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1192 | /* Update last used regex in case a blank substitute BRE is found */ |
| 1193 | if (sed_cmd->beg_match) { |
| 1194 | G.previous_regex_ptr = sed_cmd->beg_match; |
| 1195 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1196 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1197 | /* actual sedding */ |
Denys Vlasenko | d29ae7e | 2012-01-15 20:06:03 +0100 | [diff] [blame] | 1198 | dbg("pattern_space:'%s' next_line:'%s' cmd:%c", |
| 1199 | pattern_space, next_line, sed_cmd->cmd); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1200 | switch (sed_cmd->cmd) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1201 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1202 | /* Print line number */ |
| 1203 | case '=': |
| 1204 | fprintf(G.nonstdout, "%d\n", linenum); |
| 1205 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1206 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1207 | /* Write the current pattern space up to the first newline */ |
| 1208 | case 'P': |
| 1209 | { |
| 1210 | char *tmp = strchr(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1211 | if (tmp) { |
| 1212 | *tmp = '\0'; |
| 1213 | /* TODO: explain why '\n' below */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1214 | sed_puts(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1215 | *tmp = '\n'; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1216 | break; |
| 1217 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1218 | /* Fall Through */ |
| 1219 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1220 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1221 | /* Write the current pattern space to output */ |
| 1222 | case 'p': |
| 1223 | /* NB: we print this _before_ the last line |
| 1224 | * (of current file) is printed. Even if |
| 1225 | * that line is nonterminated, we print |
| 1226 | * '\n' here (gnu sed does the same) */ |
| 1227 | sed_puts(pattern_space, '\n'); |
| 1228 | break; |
| 1229 | /* Delete up through first newline */ |
| 1230 | case 'D': |
| 1231 | { |
| 1232 | char *tmp = strchr(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1233 | if (tmp) { |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1234 | overlapping_strcpy(pattern_space, tmp + 1); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1235 | goto restart; |
| 1236 | } |
| 1237 | } |
| 1238 | /* discard this line. */ |
| 1239 | case 'd': |
| 1240 | goto discard_line; |
| 1241 | |
| 1242 | /* Substitute with regex */ |
| 1243 | case 's': |
| 1244 | if (!do_subst_command(sed_cmd, &pattern_space)) |
| 1245 | break; |
Denys Vlasenko | 21f6fbf | 2012-06-04 14:44:47 +0200 | [diff] [blame] | 1246 | dbg("do_subst_command succeeded:'%s'", pattern_space); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1247 | substituted |= 1; |
| 1248 | |
| 1249 | /* handle p option */ |
| 1250 | if (sed_cmd->sub_p) |
| 1251 | sed_puts(pattern_space, last_gets_char); |
| 1252 | /* handle w option */ |
| 1253 | if (sed_cmd->sw_file) |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1254 | puts_maybe_newline( |
| 1255 | pattern_space, sed_cmd->sw_file, |
| 1256 | &sed_cmd->sw_last_char, last_gets_char); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1257 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1258 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1259 | /* Append line to linked list to be printed later */ |
| 1260 | case 'a': |
Denys Vlasenko | 9d46a7a | 2013-10-30 10:22:47 +0100 | [diff] [blame] | 1261 | append(xstrdup(sed_cmd->string)); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1262 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1263 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1264 | /* Insert text before this line */ |
| 1265 | case 'i': |
| 1266 | sed_puts(sed_cmd->string, '\n'); |
| 1267 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1268 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1269 | /* Cut and paste text (replace) */ |
| 1270 | case 'c': |
| 1271 | /* Only triggers on last line of a matching range. */ |
| 1272 | if (!sed_cmd->in_match) |
Denys Vlasenko | 96a1833 | 2010-04-19 22:36:07 -0400 | [diff] [blame] | 1273 | sed_puts(sed_cmd->string, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1274 | goto discard_line; |
| 1275 | |
| 1276 | /* Read file, append contents to output */ |
| 1277 | case 'r': |
| 1278 | { |
| 1279 | FILE *rfile; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1280 | rfile = fopen_for_read(sed_cmd->string); |
| 1281 | if (rfile) { |
| 1282 | char *line; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1283 | while ((line = xmalloc_fgetline(rfile)) |
| 1284 | != NULL) |
| 1285 | append(line); |
Denys Vlasenko | 9d46a7a | 2013-10-30 10:22:47 +0100 | [diff] [blame] | 1286 | fclose(rfile); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1287 | } |
| 1288 | |
| 1289 | break; |
| 1290 | } |
| 1291 | |
| 1292 | /* Write pattern space to file. */ |
| 1293 | case 'w': |
| 1294 | puts_maybe_newline( |
| 1295 | pattern_space, sed_cmd->sw_file, |
| 1296 | &sed_cmd->sw_last_char, last_gets_char); |
| 1297 | break; |
| 1298 | |
| 1299 | /* Read next line from input */ |
| 1300 | case 'n': |
| 1301 | if (!G.be_quiet) |
| 1302 | sed_puts(pattern_space, last_gets_char); |
| 1303 | if (next_line) { |
| 1304 | free(pattern_space); |
| 1305 | pattern_space = next_line; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1306 | last_gets_char = next_gets_char; |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 1307 | next_line = get_next_line(&next_gets_char, &last_puts_char); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1308 | substituted = 0; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1309 | linenum++; |
| 1310 | break; |
| 1311 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1312 | /* fall through */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1313 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1314 | /* Quit. End of script, end of input. */ |
| 1315 | case 'q': |
| 1316 | /* Exit the outer while loop */ |
| 1317 | free(next_line); |
| 1318 | next_line = NULL; |
| 1319 | goto discard_commands; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1320 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1321 | /* Append the next line to the current line */ |
| 1322 | case 'N': |
| 1323 | { |
| 1324 | int len; |
| 1325 | /* If no next line, jump to end of script and exit. */ |
Denys Vlasenko | 0d555fc | 2010-08-16 16:26:33 +0200 | [diff] [blame] | 1326 | /* http://www.gnu.org/software/sed/manual/sed.html: |
| 1327 | * "Most versions of sed exit without printing anything |
| 1328 | * when the N command is issued on the last line of |
| 1329 | * a file. GNU sed prints pattern space before exiting |
| 1330 | * unless of course the -n command switch has been |
| 1331 | * specified. This choice is by design." |
| 1332 | */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1333 | if (next_line == NULL) { |
Denys Vlasenko | 0d555fc | 2010-08-16 16:26:33 +0200 | [diff] [blame] | 1334 | //goto discard_line; |
| 1335 | goto discard_commands; /* GNU behavior */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1336 | } |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1337 | /* Append next_line, read new next_line. */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1338 | len = strlen(pattern_space); |
Denys Vlasenko | 9037787 | 2010-01-08 09:07:50 +0100 | [diff] [blame] | 1339 | pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1340 | pattern_space[len] = '\n'; |
| 1341 | strcpy(pattern_space + len+1, next_line); |
| 1342 | last_gets_char = next_gets_char; |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 1343 | next_line = get_next_line(&next_gets_char, &last_puts_char); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1344 | linenum++; |
| 1345 | break; |
| 1346 | } |
| 1347 | |
| 1348 | /* Test/branch if substitution occurred */ |
| 1349 | case 't': |
| 1350 | if (!substituted) break; |
| 1351 | substituted = 0; |
| 1352 | /* Fall through */ |
| 1353 | /* Test/branch if substitution didn't occur */ |
| 1354 | case 'T': |
| 1355 | if (substituted) break; |
| 1356 | /* Fall through */ |
| 1357 | /* Branch to label */ |
| 1358 | case 'b': |
| 1359 | if (!sed_cmd->string) goto discard_commands; |
| 1360 | else sed_cmd = branch_to(sed_cmd->string); |
| 1361 | break; |
| 1362 | /* Transliterate characters */ |
| 1363 | case 'y': |
| 1364 | { |
| 1365 | int i, j; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1366 | for (i = 0; pattern_space[i]; i++) { |
| 1367 | for (j = 0; sed_cmd->string[j]; j += 2) { |
| 1368 | if (pattern_space[i] == sed_cmd->string[j]) { |
| 1369 | pattern_space[i] = sed_cmd->string[j + 1]; |
| 1370 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1371 | } |
| 1372 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1373 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1374 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1375 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1376 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1377 | case 'g': /* Replace pattern space with hold space */ |
| 1378 | free(pattern_space); |
| 1379 | pattern_space = xstrdup(G.hold_space ? G.hold_space : ""); |
| 1380 | break; |
| 1381 | case 'G': /* Append newline and hold space to pattern space */ |
| 1382 | { |
| 1383 | int pattern_space_size = 2; |
| 1384 | int hold_space_size = 0; |
| 1385 | |
| 1386 | if (pattern_space) |
| 1387 | pattern_space_size += strlen(pattern_space); |
| 1388 | if (G.hold_space) |
| 1389 | hold_space_size = strlen(G.hold_space); |
| 1390 | pattern_space = xrealloc(pattern_space, |
| 1391 | pattern_space_size + hold_space_size); |
| 1392 | if (pattern_space_size == 2) |
| 1393 | pattern_space[0] = 0; |
| 1394 | strcat(pattern_space, "\n"); |
| 1395 | if (G.hold_space) |
| 1396 | strcat(pattern_space, G.hold_space); |
| 1397 | last_gets_char = '\n'; |
| 1398 | |
| 1399 | break; |
| 1400 | } |
| 1401 | case 'h': /* Replace hold space with pattern space */ |
| 1402 | free(G.hold_space); |
| 1403 | G.hold_space = xstrdup(pattern_space); |
| 1404 | break; |
| 1405 | case 'H': /* Append newline and pattern space to hold space */ |
| 1406 | { |
| 1407 | int hold_space_size = 2; |
| 1408 | int pattern_space_size = 0; |
| 1409 | |
| 1410 | if (G.hold_space) |
| 1411 | hold_space_size += strlen(G.hold_space); |
| 1412 | if (pattern_space) |
| 1413 | pattern_space_size = strlen(pattern_space); |
| 1414 | G.hold_space = xrealloc(G.hold_space, |
| 1415 | hold_space_size + pattern_space_size); |
| 1416 | |
| 1417 | if (hold_space_size == 2) |
| 1418 | *G.hold_space = 0; |
| 1419 | strcat(G.hold_space, "\n"); |
| 1420 | if (pattern_space) |
| 1421 | strcat(G.hold_space, pattern_space); |
| 1422 | |
| 1423 | break; |
| 1424 | } |
| 1425 | case 'x': /* Exchange hold and pattern space */ |
| 1426 | { |
| 1427 | char *tmp = pattern_space; |
Denys Vlasenko | 90a9904 | 2009-09-06 02:36:23 +0200 | [diff] [blame] | 1428 | pattern_space = G.hold_space ? G.hold_space : xzalloc(1); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1429 | last_gets_char = '\n'; |
| 1430 | G.hold_space = tmp; |
| 1431 | break; |
| 1432 | } |
| 1433 | } /* switch */ |
| 1434 | } /* for each cmd */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1435 | |
| 1436 | /* |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1437 | * Exit point from sedding... |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1438 | */ |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1439 | discard_commands: |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1440 | /* we will print the line unless we were told to be quiet ('-n') |
| 1441 | or if the line was suppressed (ala 'd'elete) */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1442 | if (!G.be_quiet) |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1443 | sed_puts(pattern_space, last_gets_char); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1444 | |
| 1445 | /* Delete and such jump here. */ |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1446 | discard_line: |
Denys Vlasenko | f2559e5 | 2016-05-06 18:25:56 +0200 | [diff] [blame] | 1447 | flush_append(&last_puts_char /*,last_gets_char*/); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1448 | free(pattern_space); |
| 1449 | |
| 1450 | goto again; |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 1451 | } |
| 1452 | |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1453 | /* It is possible to have a command line argument with embedded |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1454 | * newlines. This counts as multiple command lines. |
| 1455 | * However, newline can be escaped: 's/e/z\<newline>z/' |
Denys Vlasenko | 40ab27a | 2013-07-08 02:04:44 +0200 | [diff] [blame] | 1456 | * add_cmd() handles this. |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1457 | */ |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1458 | |
| 1459 | static void add_cmd_block(char *cmdstr) |
| 1460 | { |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1461 | char *sv, *eol; |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1462 | |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1463 | cmdstr = sv = xstrdup(cmdstr); |
| 1464 | do { |
| 1465 | eol = strchr(cmdstr, '\n'); |
Denys Vlasenko | 40ab27a | 2013-07-08 02:04:44 +0200 | [diff] [blame] | 1466 | if (eol) |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1467 | *eol = '\0'; |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1468 | add_cmd(cmdstr); |
| 1469 | cmdstr = eol + 1; |
| 1470 | } while (eol); |
| 1471 | free(sv); |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1472 | } |
| 1473 | |
Denis Vlasenko | 9b49a5e | 2007-10-11 10:05:36 +0000 | [diff] [blame] | 1474 | int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
Denis Vlasenko | a60f84e | 2008-07-05 09:18:54 +0000 | [diff] [blame] | 1475 | int sed_main(int argc UNUSED_PARAM, char **argv) |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1476 | { |
Denis Vlasenko | 67b23e6 | 2006-10-03 21:00:06 +0000 | [diff] [blame] | 1477 | unsigned opt; |
Denis Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 1478 | llist_t *opt_e, *opt_f; |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1479 | char *opt_i; |
| 1480 | |
| 1481 | #if ENABLE_LONG_OPTS |
| 1482 | static const char sed_longopts[] ALIGN1 = |
| 1483 | /* name has_arg short */ |
| 1484 | "in-place\0" Optional_argument "i" |
| 1485 | "regexp-extended\0" No_argument "r" |
| 1486 | "quiet\0" No_argument "n" |
| 1487 | "silent\0" No_argument "n" |
| 1488 | "expression\0" Required_argument "e" |
| 1489 | "file\0" Required_argument "f"; |
| 1490 | #endif |
| 1491 | |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 1492 | INIT_G(); |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 1493 | |
Mark Whitley | 858c1ad | 2000-07-11 21:38:47 +0000 | [diff] [blame] | 1494 | /* destroy command strings on exit */ |
Rob Landley | 0b65628 | 2006-05-08 22:17:23 +0000 | [diff] [blame] | 1495 | if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff); |
Mark Whitley | 858c1ad | 2000-07-11 21:38:47 +0000 | [diff] [blame] | 1496 | |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 1497 | /* Lie to autoconf when it starts asking stupid questions. */ |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1498 | if (argv[1] && strcmp(argv[1], "--version") == 0) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1499 | puts("This is not GNU sed version 4.0"); |
| 1500 | return 0; |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 1501 | } |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 1502 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1503 | /* do normal option parsing */ |
Denis Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 1504 | opt_e = opt_f = NULL; |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1505 | opt_i = NULL; |
Denys Vlasenko | 237bedd | 2016-07-06 21:58:02 +0200 | [diff] [blame] | 1506 | opt_complementary = "nn"; /* count -n */ |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1507 | |
| 1508 | IF_LONG_OPTS(applet_long_options = sed_longopts); |
| 1509 | |
Denys Vlasenko | 2e284a4 | 2010-08-01 04:14:46 +0200 | [diff] [blame] | 1510 | /* -i must be first, to match OPT_in_place definition */ |
David A. Wheeler | af0cdee | 2013-10-29 00:52:48 +0100 | [diff] [blame] | 1511 | /* -E is a synonym of -r: |
| 1512 | * GNU sed 4.2.1 mentions it in neither --help |
| 1513 | * nor manpage, but does recognize it. |
| 1514 | */ |
Denys Vlasenko | 237bedd | 2016-07-06 21:58:02 +0200 | [diff] [blame] | 1515 | opt = getopt32(argv, "i::rEne:*f:*", &opt_i, &opt_e, &opt_f, |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1516 | &G.be_quiet); /* counter for -n */ |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1517 | //argc -= optind; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1518 | argv += optind; |
| 1519 | if (opt & OPT_in_place) { // -i |
Denis Vlasenko | 750fc6d | 2006-09-22 08:56:03 +0000 | [diff] [blame] | 1520 | atexit(cleanup_outname); |
| 1521 | } |
David A. Wheeler | af0cdee | 2013-10-29 00:52:48 +0100 | [diff] [blame] | 1522 | if (opt & (2|4)) |
| 1523 | G.regex_type |= REG_EXTENDED; // -r or -E |
| 1524 | //if (opt & 8) |
| 1525 | // G.be_quiet++; // -n (implemented with a counter instead) |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1526 | while (opt_e) { // -e |
Denis Vlasenko | d50dda8 | 2008-06-15 05:40:56 +0000 | [diff] [blame] | 1527 | add_cmd_block(llist_pop(&opt_e)); |
Denis Vlasenko | 750fc6d | 2006-09-22 08:56:03 +0000 | [diff] [blame] | 1528 | } |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1529 | while (opt_f) { // -f |
| 1530 | char *line; |
| 1531 | FILE *cmdfile; |
Denys Vlasenko | e4d925b | 2016-04-08 00:20:36 +0200 | [diff] [blame] | 1532 | cmdfile = xfopen_stdin(llist_pop(&opt_f)); |
Denis Vlasenko | 8ee649a | 2008-03-26 20:04:27 +0000 | [diff] [blame] | 1533 | while ((line = xmalloc_fgetline(cmdfile)) != NULL) { |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1534 | add_cmd(line); |
| 1535 | free(line); |
| 1536 | } |
Denys Vlasenko | e4d925b | 2016-04-08 00:20:36 +0200 | [diff] [blame] | 1537 | fclose_if_not_stdin(cmdfile); |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1538 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1539 | /* if we didn't get a pattern from -e or -f, use argv[0] */ |
David A. Wheeler | af0cdee | 2013-10-29 00:52:48 +0100 | [diff] [blame] | 1540 | if (!(opt & 0x30)) { |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1541 | if (!*argv) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1542 | bb_show_usage(); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1543 | add_cmd_block(*argv++); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1544 | } |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1545 | /* Flush any unfinished commands. */ |
| 1546 | add_cmd(""); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1547 | |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 1548 | /* By default, we write to stdout */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1549 | G.nonstdout = stdout; |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 1550 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1551 | /* argv[0..(argc-1)] should be names of file to process. If no |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1552 | * files were specified or '-' was specified, take input from stdin. |
| 1553 | * Otherwise, we process all the files specified. */ |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1554 | G.input_file_list = argv; |
| 1555 | if (!argv[0]) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1556 | if (opt & OPT_in_place) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 1557 | bb_error_msg_and_die(bb_msg_requires_arg, "-i"); |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1558 | argv[0] = (char*)bb_msg_standard_input; |
| 1559 | /* G.last_input_file = 0; - already is */ |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1560 | } else { |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1561 | goto start; |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1562 | |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1563 | for (; *argv; argv++) { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1564 | struct stat statbuf; |
| 1565 | int nonstdoutfd; |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1566 | sed_cmd_t *sed_cmd; |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1567 | |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1568 | G.last_input_file++; |
| 1569 | start: |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1570 | if (!(opt & OPT_in_place)) { |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1571 | if (LONE_DASH(*argv)) { |
| 1572 | *argv = (char*)bb_msg_standard_input; |
| 1573 | process_files(); |
| 1574 | } |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1575 | continue; |
| 1576 | } |
| 1577 | |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1578 | /* -i: process each FILE separately: */ |
| 1579 | |
Denys Vlasenko | cd73871 | 2014-10-05 02:44:34 +0200 | [diff] [blame] | 1580 | if (stat(*argv, &statbuf) != 0) { |
| 1581 | bb_simple_perror_msg(*argv); |
| 1582 | G.exitcode = EXIT_FAILURE; |
| 1583 | G.current_input_file++; |
| 1584 | continue; |
| 1585 | } |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1586 | G.outname = xasprintf("%sXXXXXX", *argv); |
Alexander Shishkin | 6722737 | 2010-10-22 13:27:16 +0200 | [diff] [blame] | 1587 | nonstdoutfd = xmkstemp(G.outname); |
Denys Vlasenko | a7ccdee | 2009-11-15 23:28:11 +0100 | [diff] [blame] | 1588 | G.nonstdout = xfdopen_for_write(nonstdoutfd); |
Denys Vlasenko | 5799248 | 2009-11-13 09:09:07 +0100 | [diff] [blame] | 1589 | /* Set permissions/owner of output file */ |
Denys Vlasenko | 3b727cc | 2010-06-18 02:12:56 +0200 | [diff] [blame] | 1590 | /* chmod'ing AFTER chown would preserve suid/sgid bits, |
| 1591 | * but GNU sed 4.2.1 does not preserve them either */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1592 | fchmod(nonstdoutfd, statbuf.st_mode); |
Denys Vlasenko | 5799248 | 2009-11-13 09:09:07 +0100 | [diff] [blame] | 1593 | fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid); |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1594 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1595 | process_files(); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1596 | fclose(G.nonstdout); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1597 | G.nonstdout = stdout; |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1598 | |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1599 | if (opt_i) { |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1600 | char *backupname = xasprintf("%s%s", *argv, opt_i); |
| 1601 | xrename(*argv, backupname); |
Simon B | 8c34395 | 2012-05-06 13:59:15 +0200 | [diff] [blame] | 1602 | free(backupname); |
| 1603 | } |
Denys Vlasenko | deb0769 | 2013-11-28 12:08:51 +0100 | [diff] [blame] | 1604 | /* else unlink(*argv); - rename below does this */ |
| 1605 | xrename(G.outname, *argv); //TODO: rollback backup on error? |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1606 | free(G.outname); |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 1607 | G.outname = NULL; |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1608 | |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 1609 | /* Fix disabled range matches and mangled ",+N" ranges */ |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1610 | for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) { |
| 1611 | sed_cmd->beg_line = sed_cmd->beg_line_orig; |
Denys Vlasenko | 63f4d32 | 2015-04-17 14:24:55 +0200 | [diff] [blame] | 1612 | sed_cmd->end_line = sed_cmd->end_line_orig; |
Denys Vlasenko | a7d6bb3 | 2011-08-16 13:29:34 +0200 | [diff] [blame] | 1613 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1614 | } |
Denys Vlasenko | eb08b6e | 2010-06-19 17:51:06 +0200 | [diff] [blame] | 1615 | /* Here, to handle "sed 'cmds' nonexistent_file" case we did: |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 1616 | * if (G.current_input_file[G.current_input_file] == NULL) |
| 1617 | * return G.exitcode; |
Denys Vlasenko | eb08b6e | 2010-06-19 17:51:06 +0200 | [diff] [blame] | 1618 | * but it's not needed since process_files() works correctly |
| 1619 | * in this case too. */ |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1620 | } |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 1621 | |
Denys Vlasenko | eb08b6e | 2010-06-19 17:51:06 +0200 | [diff] [blame] | 1622 | process_files(); |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1623 | |
Denys Vlasenko | 259b3c0 | 2013-11-28 03:14:16 +0100 | [diff] [blame] | 1624 | return G.exitcode; |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1625 | } |