blob: e7b2c214af2e34d7a64351cfb624a68a2189b07c [file] [log] [blame]
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001/* vi: set sw=4 ts=4: */
Eric Andersen6b6b3f61999-10-28 16:06:25 +00002/*
Mark Whitley6315ce62000-07-10 22:55:51 +00003 * sed.c - very minimalist version of sed
Eric Andersen6b6b3f61999-10-28 16:06:25 +00004 *
Eric Andersenbdfd0d72001-10-24 05:00:29 +00005 * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
6 * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
Matt Kraai5ed78ad2002-01-03 21:12:34 +00007 * Copyright (C) 2002 Matt Kraai
Denis Vlasenko0beaff82007-09-21 13:16:32 +00008 * Copyright (C) 2003 by Glenn McGrath
Rob Landley25d82392004-04-01 09:23:30 +00009 * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
Erik Andersen1266a131999-12-29 22:19:46 +000010 *
Rob Landley2b26fd52006-02-24 02:30:39 +000011 * MAINTAINER: Rob Landley <rob@landley.net>
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +000012 *
Rob Landleye9a7a622006-09-22 02:52:41 +000013 * Licensed under GPL version 2, see file LICENSE in this tarball for details.
Eric Andersen6b6b3f61999-10-28 16:06:25 +000014 */
15
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000016/* Code overview.
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
Denis Vlasenko17a15262007-03-26 20:48:46 +000024 (G.sed_cmd_head/G.sed_cmd_tail).
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000025
Denis Vlasenkodefc1ea2008-06-27 02:52:20 +000026 add_input_file() adds a FILE* to the list of input files. We need to
Rob Landleyfae1dc82005-11-20 07:44:35 +000027 know all input sources ahead of time to find the last line for the $ match.
Rob Landleydcc28662004-11-25 07:21:47 +000028
29 process_files() does actual sedding, reading data lines from each input FILE *
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000030 (which could be stdin) and applying the sed command list (sed_cmd_head) to
31 each of the resulting lines.
32
33 sed_main() is where external code calls into this, with a command line.
34*/
35
36
Mark Whitley6315ce62000-07-10 22:55:51 +000037/*
38 Supported features and commands in this version of sed:
39
40 - comments ('#')
Mark Whitley94074a92000-07-14 00:00:15 +000041 - address matching: num|/matchstr/[,num|/matchstr/|$]command
42 - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
43 - edit commands: (a)ppend, (i)nsert, (c)hange
Mark Whitley1f3b9f22001-05-11 22:27:13 +000044 - file commands: (r)ead
Rob Landleyc63fe912005-10-30 10:08:13 +000045 - backreferences in substitution expressions (\0, \1, \2...\9)
Glenn L McGrathf50ce312003-03-09 15:12:24 +000046 - grouped commands: {cmd1;cmd2}
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000047 - transliteration (y/source-chars/dest-chars/)
48 - pattern space hold space storing / swapping (g, h, x)
Rob Landley93850a52005-05-18 06:34:37 +000049 - labels / branching (: label, b, t, T)
Glenn L McGrathf50ce312003-03-09 15:12:24 +000050
Mark Whitley6315ce62000-07-10 22:55:51 +000051 (Note: Specifying an address (range) to match is *optional*; commands
52 default to the whole pattern space if no specific address match was
53 requested.)
54
Glenn L McGrath2570b432003-09-16 05:25:43 +000055 Todo:
Glenn L McGrath2570b432003-09-16 05:25:43 +000056 - Create a wrapper around regex to make libc's regex conform with sed
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000057
Glenn L McGrathd5eadea2003-03-09 02:39:29 +000058 Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
Mark Whitley6315ce62000-07-10 22:55:51 +000059*/
60
Denis Vlasenkob6adbf12007-05-26 19:00:18 +000061#include "libbb.h"
"Vladimir N. Oleynik"23f62fc2005-09-14 16:59:11 +000062#include "xregex.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000063
Rob Landleye3f5a3f2006-05-09 03:53:55 +000064/* Each sed command turns into one of these structures. */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +000065typedef struct sed_cmd_s {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +000066 /* Ordered by alignment requirements: currently 36 bytes on x86 */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +000067 struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
Mark Whitley2dc192f2000-11-03 19:47:00 +000068
Denis Vlasenkod18a3a22006-10-25 12:46:03 +000069 /* address storage */
70 regex_t *beg_match; /* sed -e '/match/cmd' */
71 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
72 regex_t *sub_match; /* For 's/sub_match/string/' */
73 int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
74 int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +000075
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +000076 FILE *sw_file; /* File (sw) command writes to, -1 for none. */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +000077 char *string; /* Data string for (saicytb) commands. */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +000078
Denis Vlasenko284d0fa2008-02-16 13:18:17 +000079 unsigned which_match; /* (s) Which match to replace (0 for all) */
Glenn L McGrathfc4cb4d2003-04-12 16:10:42 +000080
Denis Vlasenkod18a3a22006-10-25 12:46:03 +000081 /* Bitfields (gcc won't group them if we don't) */
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +000082 unsigned invert:1; /* the '!' after the address */
83 unsigned in_match:1; /* Next line also included in match? */
84 unsigned sub_p:1; /* (s) print option */
Denis Vlasenko1375bc72006-12-02 20:12:12 +000085
Denis Vlasenko34c4e5f2007-01-30 22:25:16 +000086 char sw_last_char; /* Last line written by (sw) had no '\n' */
Glenn L McGrathc18ce372003-09-13 06:57:39 +000087
Denis Vlasenkod18a3a22006-10-25 12:46:03 +000088 /* GENERAL FIELDS */
89 char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
Glenn L McGrathe7a8bc92003-03-09 10:23:57 +000090} sed_cmd_t;
Mark Whitley6315ce62000-07-10 22:55:51 +000091
Denis Vlasenko6ca409e2007-08-12 20:58:27 +000092static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
Glenn L McGrathaa5a6022003-10-01 03:06:16 +000093
Denis Vlasenko17a15262007-03-26 20:48:46 +000094struct globals {
Rob Landleye3f5a3f2006-05-09 03:53:55 +000095 /* options */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +000096 int be_quiet, regex_type;
Rob Landleye3f5a3f2006-05-09 03:53:55 +000097 FILE *nonstdout;
98 char *outname, *hold_space;
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +000099
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000100 /* List of input files */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000101 int input_file_count, current_input_file;
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000102 FILE **input_file_list;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000103
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000104 regmatch_t regmatch[10];
105 regex_t *previous_regex_ptr;
Denis Vlasenko9213a9e2006-09-17 16:28:10 +0000106
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000107 /* linked list of sed commands */
108 sed_cmd_t sed_cmd_head, *sed_cmd_tail;
109
110 /* Linked list of append lines */
111 llist_t *append_head;
112
113 char *add_cmd_line;
114
115 struct pipeline {
116 char *buf; /* Space to hold string */
117 int idx; /* Space used */
118 int len; /* Space allocated */
119 } pipeline;
Denis Vlasenko17a15262007-03-26 20:48:46 +0000120};
Denis Vlasenko17a15262007-03-26 20:48:46 +0000121#define G (*(struct globals*)&bb_common_bufsiz1)
Denis Vlasenko74324c82007-06-04 10:16:52 +0000122void BUG_sed_globals_too_big(void);
123#define INIT_G() do { \
124 if (sizeof(struct globals) > COMMON_BUFSIZE) \
125 BUG_sed_globals_too_big(); \
126 G.sed_cmd_tail = &G.sed_cmd_head; \
127} while (0)
128
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000129
Bernhard Reutner-Fischerb7f39732006-03-01 20:14:16 +0000130#if ENABLE_FEATURE_CLEAN_UP
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000131static void sed_free_and_close_stuff(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000132{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000133 sed_cmd_t *sed_cmd = G.sed_cmd_head.next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000134
Denis Vlasenko17a15262007-03-26 20:48:46 +0000135 llist_free(G.append_head, free);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000136
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000137 while (sed_cmd) {
Glenn L McGrathbd9b32b2003-04-09 01:43:54 +0000138 sed_cmd_t *sed_cmd_next = sed_cmd->next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000139
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000140 if (sed_cmd->sw_file)
141 xprint_and_close_file(sed_cmd->sw_file);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000142
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000143 if (sed_cmd->beg_match) {
144 regfree(sed_cmd->beg_match);
145 free(sed_cmd->beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000146 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000147 if (sed_cmd->end_match) {
148 regfree(sed_cmd->end_match);
149 free(sed_cmd->end_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000150 }
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000151 if (sed_cmd->sub_match) {
152 regfree(sed_cmd->sub_match);
153 free(sed_cmd->sub_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000154 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000155 free(sed_cmd->string);
Glenn L McGrath56c633c2003-03-28 04:23:23 +0000156 free(sed_cmd);
157 sed_cmd = sed_cmd_next;
Mark Whitley6315ce62000-07-10 22:55:51 +0000158 }
Rob Landleyce4f0e92004-10-30 06:54:19 +0000159
Denis Vlasenko60818682007-09-28 22:07:23 +0000160 free(G.hold_space);
Rob Landleydcc28662004-11-25 07:21:47 +0000161
Denis Vlasenko17a15262007-03-26 20:48:46 +0000162 while (G.current_input_file < G.input_file_count)
163 fclose(G.input_file_list[G.current_input_file++]);
Mark Whitley6315ce62000-07-10 22:55:51 +0000164}
Denis Vlasenko666da5e2006-12-26 18:17:42 +0000165#else
166void sed_free_and_close_stuff(void);
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000167#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000168
Rob Landley53302f82004-02-18 09:54:15 +0000169/* If something bad happens during -i operation, delete temp file */
170
171static void cleanup_outname(void)
172{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000173 if (G.outname) unlink(G.outname);
Rob Landley53302f82004-02-18 09:54:15 +0000174}
175
Denis Vlasenko40276642007-11-13 16:48:10 +0000176/* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000177
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000178static void parse_escapes(char *dest, const char *string, int len, char from, char to)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000179{
Denis Vlasenko80778502006-10-25 12:46:46 +0000180 int i = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000181
Denis Vlasenko80778502006-10-25 12:46:46 +0000182 while (i < len) {
183 if (string[i] == '\\') {
184 if (!to || string[i+1] == from) {
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000185 *dest++ = to ? to : string[i+1];
Denis Vlasenko80778502006-10-25 12:46:46 +0000186 i += 2;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000187 continue;
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000188 }
189 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000190 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000191 /* TODO: is it safe wrt a string with trailing '\\' ? */
Denis Vlasenko4ebaf102007-01-19 21:33:19 +0000192 *dest++ = string[i++];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000193 }
Denis Vlasenko40276642007-11-13 16:48:10 +0000194 *dest = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000195}
196
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000197static char *copy_parsing_escapes(const char *string, int len)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000198{
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000199 char *dest = xmalloc(len + 1);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000200
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000201 parse_escapes(dest, string, len, 'n', '\n');
Denis Vlasenko40276642007-11-13 16:48:10 +0000202 /* GNU sed also recognizes \t */
203 parse_escapes(dest, dest, strlen(dest), 't', '\t');
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000204 return dest;
205}
206
207
Mark Whitley6315ce62000-07-10 22:55:51 +0000208/*
Eric Andersen28b3c532001-01-02 11:01:31 +0000209 * index_of_next_unescaped_regexp_delim - walks left to right through a string
210 * beginning at a specified index and returns the index of the next regular
Denis Vlasenko40276642007-11-13 16:48:10 +0000211 * expression delimiter (typically a forward slash ('/')) not preceded by
Rob Landley4795e4e2006-07-26 17:25:08 +0000212 * a backslash ('\'). A negative delimiter disables square bracket checking.
Mark Whitley6315ce62000-07-10 22:55:51 +0000213 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000214static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
Mark Whitley6315ce62000-07-10 22:55:51 +0000215{
Matt Kraaia0065d52001-08-24 14:45:50 +0000216 int bracket = -1;
217 int escaped = 0;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000218 int idx = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000219 char ch;
Matt Kraaia0065d52001-08-24 14:45:50 +0000220
Rob Landley4795e4e2006-07-26 17:25:08 +0000221 if (delimiter < 0) {
222 bracket--;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000223 delimiter = -delimiter;
Rob Landley4795e4e2006-07-26 17:25:08 +0000224 }
225
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000226 for (; (ch = str[idx]); idx++) {
Rob Landley4795e4e2006-07-26 17:25:08 +0000227 if (bracket >= 0) {
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000228 if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000229 && str[idx - 1] == '^')))
Matt Kraaia0065d52001-08-24 14:45:50 +0000230 bracket = -1;
231 } else if (escaped)
232 escaped = 0;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000233 else if (ch == '\\')
Matt Kraaia0065d52001-08-24 14:45:50 +0000234 escaped = 1;
Rob Landley4795e4e2006-07-26 17:25:08 +0000235 else if (bracket == -1 && ch == '[')
Matt Kraaia0065d52001-08-24 14:45:50 +0000236 bracket = idx;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000237 else if (ch == delimiter)
Mark Whitley97562bd2000-07-17 20:06:42 +0000238 return idx;
239 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000240
Mark Whitley97562bd2000-07-17 20:06:42 +0000241 /* if we make it to here, we've hit the end of the string */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000242 bb_error_msg_and_die("unmatched '%c'", delimiter);
Mark Whitley6315ce62000-07-10 22:55:51 +0000243}
244
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000245/*
246 * Returns the index of the third delimiter
247 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000248static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000249{
Denis Vlasenko9356b502007-01-29 23:44:38 +0000250 const char *cmdstr_ptr = cmdstr;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000251 char delimiter;
252 int idx = 0;
253
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000254 /* verify that the 's' or 'y' is followed by something. That something
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000255 * (typically a 'slash') is now our regexp delimiter... */
Rob Landley4795e4e2006-07-26 17:25:08 +0000256 if (*cmdstr == '\0')
257 bb_error_msg_and_die("bad format in substitution expression");
Denis Vlasenko80778502006-10-25 12:46:46 +0000258 delimiter = *cmdstr_ptr++;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000259
260 /* save the match string */
261 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000262 *match = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000263
264 /* save the replacement string */
265 cmdstr_ptr += idx + 1;
Rob Landley4795e4e2006-07-26 17:25:08 +0000266 idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
Rob Landleyfae1dc82005-11-20 07:44:35 +0000267 *replace = copy_parsing_escapes(cmdstr_ptr, idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000268
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000269 return ((cmdstr_ptr - cmdstr) + idx);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000270}
271
Mark Whitley6315ce62000-07-10 22:55:51 +0000272/*
273 * returns the index in the string just past where the address ends.
274 */
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000275static int get_address(const char *my_str, int *linenum, regex_t ** regex)
Mark Whitley6315ce62000-07-10 22:55:51 +0000276{
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000277 const char *pos = my_str;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000278
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000279 if (isdigit(*my_str)) {
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000280 *linenum = strtol(my_str, (char**)&pos, 10);
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000281 /* endstr shouldnt ever equal NULL */
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000282 } else if (*my_str == '$') {
Mark Whitley0915c4b2001-06-11 23:50:06 +0000283 *linenum = -1;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000284 pos++;
285 } else if (*my_str == '/' || *my_str == '\\') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000286 int next;
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000287 char delimiter;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000288 char *temp;
Glenn L McGrath1fb44672003-03-09 08:44:49 +0000289
Denis Vlasenko80778502006-10-25 12:46:46 +0000290 delimiter = '/';
291 if (*my_str == '\\') delimiter = *++pos;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000292 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000293 temp = copy_parsing_escapes(pos, next);
Denis Vlasenko80778502006-10-25 12:46:46 +0000294 *regex = xmalloc(sizeof(regex_t));
Denis Vlasenko17a15262007-03-26 20:48:46 +0000295 xregcomp(*regex, temp, G.regex_type|REG_NEWLINE);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000296 free(temp);
297 /* Move position to next character after last delimiter */
Rob Landley4795e4e2006-07-26 17:25:08 +0000298 pos += (next+1);
Mark Whitley6315ce62000-07-10 22:55:51 +0000299 }
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000300 return pos - my_str;
Mark Whitley6315ce62000-07-10 22:55:51 +0000301}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000302
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000303/* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
Denis Vlasenko68404f12008-03-17 09:00:54 +0000304static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000305{
Denis Vlasenko80778502006-10-25 12:46:46 +0000306 int start = 0, idx, hack = 0;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000307
308 /* Skip whitespace, then grab filename to end of line */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000309 while (isspace(filecmdstr[start]))
310 start++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000311 idx = start;
Denis Vlasenko9356b502007-01-29 23:44:38 +0000312 while (filecmdstr[idx] && filecmdstr[idx] != '\n')
313 idx++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000314
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000315 /* If lines glued together, put backslash back. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000316 if (filecmdstr[idx] == '\n')
317 hack = 1;
Denis Vlasenko80778502006-10-25 12:46:46 +0000318 if (idx == start)
319 bb_error_msg_and_die("empty filename");
Rob Landleyd921b2e2006-08-03 15:41:12 +0000320 *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
Denis Vlasenko9356b502007-01-29 23:44:38 +0000321 if (hack)
322 (*retval)[idx] = '\\';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000323
324 return idx;
325}
326
Denis Vlasenko9356b502007-01-29 23:44:38 +0000327static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000328{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000329 int cflags = G.regex_type;
Mark Whitley06f35292000-07-13 19:58:04 +0000330 char *match;
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000331 int idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000332
333 /*
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000334 * A substitution command should look something like this:
335 * s/match/replace/ #gIpw
Mark Whitley0e4cec02000-08-21 21:29:20 +0000336 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000337 * mandatory optional
Mark Whitley06f35292000-07-13 19:58:04 +0000338 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000339 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
Mark Whitley06f35292000-07-13 19:58:04 +0000340
Mark Whitley97562bd2000-07-17 20:06:42 +0000341 /* determine the number of back references in the match string */
342 /* Note: we compute this here rather than in the do_subst_command()
343 * function to save processor time, at the expense of a little more memory
344 * (4 bits) per sed_cmd */
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000345
Mark Whitley06f35292000-07-13 19:58:04 +0000346 /* process the flags */
Mark Whitley70705d72000-07-14 19:06:30 +0000347
Denis Vlasenko80778502006-10-25 12:46:46 +0000348 sed_cmd->which_match = 1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000349 while (substr[++idx]) {
350 /* Parse match number */
Denis Vlasenko80778502006-10-25 12:46:46 +0000351 if (isdigit(substr[idx])) {
352 if (match[0] != '^') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000353 /* Match 0 treated as all, multiple matches we take the last one. */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000354 const char *pos = substr + idx;
355/* FIXME: error check? */
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000356 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000357 idx = pos - substr;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000358 }
359 continue;
360 }
Rob Landley40ec4ae2004-01-04 06:42:14 +0000361 /* Skip spaces */
Denis Vlasenko80778502006-10-25 12:46:46 +0000362 if (isspace(substr[idx])) continue;
Rob Landley40ec4ae2004-01-04 06:42:14 +0000363
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000364 switch (substr[idx]) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000365 /* Replace all occurrences */
366 case 'g':
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000367 if (match[0] != '^')
368 sed_cmd->which_match = 0;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000369 break;
370 /* Print pattern space */
371 case 'p':
372 sed_cmd->sub_p = 1;
373 break;
374 /* Write to file */
375 case 'w':
376 {
377 char *temp;
Denis Vlasenko68404f12008-03-17 09:00:54 +0000378 idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000379 break;
380 }
381 /* Ignore case (gnu exension) */
382 case 'I':
383 cflags |= REG_ICASE;
384 break;
385 /* Comment */
386 case '#':
Denis Vlasenko80778502006-10-25 12:46:46 +0000387 while (substr[++idx]) /*skip all*/;
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000388 /* Fall through */
389 /* End of command */
390 case ';':
391 case '}':
392 goto out;
393 default:
394 bb_error_msg_and_die("bad option in substitution expression");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000395 }
396 }
Glenn L McGrath2570b432003-09-16 05:25:43 +0000397out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000398 /* compile the match string into a regex */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000399 if (*match != '\0') {
400 /* If match is empty, we use last regex used at runtime */
Denis Vlasenkob95636c2006-12-19 23:36:04 +0000401 sed_cmd->sub_match = xmalloc(sizeof(regex_t));
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000402 xregcomp(sed_cmd->sub_match, match, cflags);
403 }
Mark Whitley06f35292000-07-13 19:58:04 +0000404 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000405
406 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000407}
408
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000409/*
410 * Process the commands arguments
411 */
Denis Vlasenko9356b502007-01-29 23:44:38 +0000412static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000413{
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000414 /* handle (s)ubstitution command */
Denis Vlasenko80778502006-10-25 12:46:46 +0000415 if (sed_cmd->cmd == 's')
416 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000417 /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
418 else if (strchr("aic", sed_cmd->cmd)) {
419 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000420 bb_error_msg_and_die
421 ("only a beginning address can be specified for edit commands");
Denis Vlasenko80778502006-10-25 12:46:46 +0000422 for (;;) {
423 if (*cmdstr == '\n' || *cmdstr == '\\') {
Rob Landley25d82392004-04-01 09:23:30 +0000424 cmdstr++;
425 break;
Denis Vlasenko80778502006-10-25 12:46:46 +0000426 } else if (isspace(*cmdstr))
427 cmdstr++;
428 else
429 break;
Rob Landley25d82392004-04-01 09:23:30 +0000430 }
Rob Landleyd921b2e2006-08-03 15:41:12 +0000431 sed_cmd->string = xstrdup(cmdstr);
Denis Vlasenko40276642007-11-13 16:48:10 +0000432 /* "\anychar" -> "anychar" */
433 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000434 cmdstr += strlen(cmdstr);
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000435 /* handle file cmds: (r)ead */
Denis Vlasenko80778502006-10-25 12:46:46 +0000436 } else if (strchr("rw", sed_cmd->cmd)) {
Glenn L McGrath2f8a4012003-03-09 02:44:49 +0000437 if (sed_cmd->end_line || sed_cmd->end_match)
Denis Vlasenko80778502006-10-25 12:46:46 +0000438 bb_error_msg_and_die("command only uses one address");
Denis Vlasenko68404f12008-03-17 09:00:54 +0000439 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000440 if (sed_cmd->cmd == 'w') {
Denis Vlasenko5415c852008-07-21 23:05:26 +0000441 sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000442 sed_cmd->sw_last_char = '\n';
443 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000444 /* handle branch commands */
Rob Landley93850a52005-05-18 06:34:37 +0000445 } else if (strchr(":btT", sed_cmd->cmd)) {
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000446 int length;
447
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000448 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000449 length = strcspn(cmdstr, semicolon_whitespace);
450 if (length) {
Rob Landleyd921b2e2006-08-03 15:41:12 +0000451 sed_cmd->string = xstrndup(cmdstr, length);
Glenn L McGrathf4523562003-09-14 06:01:14 +0000452 cmdstr += length;
453 }
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000454 }
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000455 /* translation command */
456 else if (sed_cmd->cmd == 'y') {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000457 char *match, *replace;
Denis Vlasenko80778502006-10-25 12:46:46 +0000458 int i = cmdstr[0];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000459
Denis Vlasenko80778502006-10-25 12:46:46 +0000460 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000461 /* \n already parsed, but \delimiter needs unescaping. */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000462 parse_escapes(match, match, strlen(match), i, i);
463 parse_escapes(replace, replace, strlen(replace), i, i);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000464
Rob Landley9ffd4232006-05-21 18:30:35 +0000465 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000466 for (i = 0; match[i] && replace[i]; i++) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000467 sed_cmd->string[i*2] = match[i];
468 sed_cmd->string[i*2+1] = replace[i];
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000469 }
470 free(match);
471 free(replace);
Glenn L McGrathf01b46d2003-03-30 08:02:18 +0000472 }
Glenn L McGrath2971ef12003-03-18 01:19:23 +0000473 /* if it wasnt a single-letter command that takes no arguments
474 * then it must be an invalid command.
475 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000476 else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000477 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
Mark Whitley1f3b9f22001-05-11 22:27:13 +0000478 }
Mark Whitley70705d72000-07-14 19:06:30 +0000479
480 /* give back whatever's left over */
Denis Vlasenko80778502006-10-25 12:46:46 +0000481 return cmdstr;
Eric Andersen50d63601999-11-09 01:47:36 +0000482}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000483
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000484
485/* Parse address+command sets, skipping comment lines. */
486
Denis Vlasenkob6aae0f2007-01-29 22:51:25 +0000487static void add_cmd(const char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000488{
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000489 sed_cmd_t *sed_cmd;
Glenn L McGrath586d86c2003-10-09 07:22:59 +0000490 int temp;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +0000491
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000492 /* Append this line to any unfinished line from last time. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000493 if (G.add_cmd_line) {
494 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
495 free(G.add_cmd_line);
Denis Vlasenko945bd3d2007-04-12 21:20:25 +0000496 cmdstr = G.add_cmd_line = tp;
Rob Landley12d87552006-06-05 17:32:44 +0000497 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000498
499 /* If this line ends with backslash, request next line. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000500 temp = strlen(cmdstr);
Denis Vlasenko945bd3d2007-04-12 21:20:25 +0000501 if (temp && cmdstr[--temp] == '\\') {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000502 if (!G.add_cmd_line)
503 G.add_cmd_line = xstrdup(cmdstr);
Denis Vlasenko945bd3d2007-04-12 21:20:25 +0000504 G.add_cmd_line[temp] = '\0';
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000505 return;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000506 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000507
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000508 /* Loop parsing all commands in this line. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000509 while (*cmdstr) {
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000510 /* Skip leading whitespace and semicolons */
511 cmdstr += strspn(cmdstr, semicolon_whitespace);
512
513 /* If no more commands, exit. */
Denis Vlasenko80778502006-10-25 12:46:46 +0000514 if (!*cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000515
516 /* if this is a comment, jump past it and keep going */
517 if (*cmdstr == '#') {
518 /* "#n" is the same as using -n on the command line */
Denis Vlasenko80778502006-10-25 12:46:46 +0000519 if (cmdstr[1] == 'n')
Denis Vlasenko17a15262007-03-26 20:48:46 +0000520 G.be_quiet++;
Denis Vlasenko80778502006-10-25 12:46:46 +0000521 cmdstr = strpbrk(cmdstr, "\n\r");
522 if (!cmdstr) break;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000523 continue;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000524 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000525
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000526 /* parse the command
527 * format is: [addr][,addr][!]cmd
528 * |----||-----||-|
529 * part1 part2 part3
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000530 */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000531
Rob Landley9ffd4232006-05-21 18:30:35 +0000532 sed_cmd = xzalloc(sizeof(sed_cmd_t));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000533
534 /* first part (if present) is an address: either a '$', a number or a /regex/ */
535 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
536
537 /* second part (if present) will begin with a comma */
538 if (*cmdstr == ',') {
539 int idx;
540
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000541 cmdstr++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000542 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
Denis Vlasenko80778502006-10-25 12:46:46 +0000543 if (!idx)
544 bb_error_msg_and_die("no address after comma");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000545 cmdstr += idx;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000546 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000547
548 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000549 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000550
551 /* Check for inversion flag */
552 if (*cmdstr == '!') {
553 sed_cmd->invert = 1;
554 cmdstr++;
555
556 /* skip whitespace before the command */
Denis Vlasenkod18a3a22006-10-25 12:46:03 +0000557 cmdstr = skip_whitespace(cmdstr);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000558 }
559
560 /* last part (mandatory) will be a command */
Denis Vlasenko80778502006-10-25 12:46:46 +0000561 if (!*cmdstr)
562 bb_error_msg_and_die("missing command");
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000563 sed_cmd->cmd = *(cmdstr++);
564 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
565
566 /* Add the command to the command array */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000567 G.sed_cmd_tail->next = sed_cmd;
568 G.sed_cmd_tail = G.sed_cmd_tail->next;
Glenn L McGrathf50ce312003-03-09 15:12:24 +0000569 }
570
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000571 /* If we glued multiple lines together, free the memory. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000572 free(G.add_cmd_line);
573 G.add_cmd_line = NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +0000574}
575
Rob Landleydcc28662004-11-25 07:21:47 +0000576/* Append to a string, reallocating memory as necessary. */
577
Glenn L McGrath8d6395d2003-04-08 11:56:11 +0000578#define PIPE_GROW 64
Eric Andersenc52a6b02001-11-10 10:49:42 +0000579
Eric Andersen14f5c8d2005-04-16 19:39:00 +0000580static void pipe_putc(char c)
Eric Andersenc52a6b02001-11-10 10:49:42 +0000581{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000582 if (G.pipeline.idx == G.pipeline.len) {
583 G.pipeline.buf = xrealloc(G.pipeline.buf,
584 G.pipeline.len + PIPE_GROW);
585 G.pipeline.len += PIPE_GROW;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000586 }
Denis Vlasenko17a15262007-03-26 20:48:46 +0000587 G.pipeline.buf[G.pipeline.idx++] = c;
Eric Andersenc52a6b02001-11-10 10:49:42 +0000588}
589
Rob Landley4795e4e2006-07-26 17:25:08 +0000590static void do_subst_w_backrefs(char *line, char *replace)
Mark Whitley97562bd2000-07-17 20:06:42 +0000591{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200592 int i, j;
Mark Whitley97562bd2000-07-17 20:06:42 +0000593
594 /* go through the replacement string */
595 for (i = 0; replace[i]; i++) {
596 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000597 if (replace[i] == '\\') {
598 unsigned backref = replace[++i] - '0';
599 if (backref <= 9) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000600 /* print out the text held in G.regmatch[backref] */
601 if (G.regmatch[backref].rm_so != -1) {
602 j = G.regmatch[backref].rm_so;
603 while (j < G.regmatch[backref].rm_eo)
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000604 pipe_putc(line[j++]);
605 }
606 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000607 }
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000608 /* I _think_ it is impossible to get '\' to be
609 * the last char in replace string. Thus we dont check
610 * for replace[i] == NUL. (counterexample anyone?) */
611 /* if we find a backslash escaped character, print the character */
612 pipe_putc(replace[i]);
613 continue;
Mark Whitley97562bd2000-07-17 20:06:42 +0000614 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000615 /* if we find an unescaped '&' print out the whole matched text. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000616 if (replace[i] == '&') {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000617 j = G.regmatch[0].rm_so;
618 while (j < G.regmatch[0].rm_eo)
Denis Vlasenko80778502006-10-25 12:46:46 +0000619 pipe_putc(line[j++]);
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000620 continue;
Denis Vlasenko80778502006-10-25 12:46:46 +0000621 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000622 /* Otherwise just output the character. */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000623 pipe_putc(replace[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000624 }
625}
626
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200627static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
Mark Whitley4f7fe772000-07-13 20:01:58 +0000628{
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200629 char *line = *line_p;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000630 int altered = 0;
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000631 unsigned match_count = 0;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000632 regex_t *current_regex;
633
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200634 current_regex = sed_cmd->sub_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000635 /* Handle empty regex. */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200636 if (!current_regex) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000637 current_regex = G.previous_regex_ptr;
Denis Vlasenko80778502006-10-25 12:46:46 +0000638 if (!current_regex)
Denis Vlasenkod3d004d2006-10-27 09:02:31 +0000639 bb_error_msg_and_die("no previous regexp");
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200640 }
641 G.previous_regex_ptr = current_regex;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000642
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000643 /* Find the first match */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200644 if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0))
Mark Whitley2dc192f2000-11-03 19:47:00 +0000645 return 0;
646
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000647 /* Initialize temporary output buffer. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000648 G.pipeline.buf = xmalloc(PIPE_GROW);
649 G.pipeline.len = PIPE_GROW;
650 G.pipeline.idx = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000651
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000652 /* Now loop through, substituting for matches */
653 do {
Mark Whitley4f7fe772000-07-13 20:01:58 +0000654 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000655
Eric Andersenc06f5682004-02-04 10:57:46 +0000656 /* Work around bug in glibc regexec, demonstrated by:
657 echo " a.b" | busybox sed 's [^ .]* x g'
658 The match_count check is so not to break
659 echo "hi" | busybox sed 's/^/!/g' */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000660 if (!G.regmatch[0].rm_so && !G.regmatch[0].rm_eo && match_count) {
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200661 pipe_putc(*line++);
Eric Andersenc06f5682004-02-04 10:57:46 +0000662 continue;
663 }
664
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000665 match_count++;
666
667 /* If we aren't interested in this match, output old line to
668 end of match and continue */
Denis Vlasenko77ad97f2008-05-13 02:27:31 +0000669 if (sed_cmd->which_match
670 && (sed_cmd->which_match != match_count)
671 ) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000672 for (i = 0; i < G.regmatch[0].rm_eo; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200673 pipe_putc(*line++);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000674 continue;
675 }
676
Mark Whitley2dc192f2000-11-03 19:47:00 +0000677 /* print everything before the match */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000678 for (i = 0; i < G.regmatch[0].rm_so; i++)
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200679 pipe_putc(line[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000680
Mark Whitley2dc192f2000-11-03 19:47:00 +0000681 /* then print the substitution string */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200682 do_subst_w_backrefs(line, sed_cmd->string);
Mark Whitley97562bd2000-07-17 20:06:42 +0000683
Mark Whitley2dc192f2000-11-03 19:47:00 +0000684 /* advance past the match */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200685 line += G.regmatch[0].rm_eo;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000686 /* flag that something has changed */
687 altered++;
Mark Whitley97562bd2000-07-17 20:06:42 +0000688
Mark Whitley2dc192f2000-11-03 19:47:00 +0000689 /* if we're not doing this globally, get out now */
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000690 if (sed_cmd->which_match)
691 break;
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200692
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200693//maybe (G.regmatch[0].rm_eo ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
Denys Vlasenko11c82382009-09-22 03:02:21 +0200694 } while (*line && regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
Mark Whitley2dc192f2000-11-03 19:47:00 +0000695
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000696 /* Copy rest of string into output pipeline */
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200697 while (1) {
698 char c = *line++;
699 pipe_putc(c);
700 if (c == '\0')
701 break;
702 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000703
Denys Vlasenkof210cff2009-08-17 01:35:04 +0200704 free(*line_p);
705 *line_p = G.pipeline.buf;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000706 return altered;
707}
Mark Whitley6315ce62000-07-10 22:55:51 +0000708
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000709/* Set command pointer to point to this label. (Does not handle null label.) */
Rob Landley4795e4e2006-07-26 17:25:08 +0000710static sed_cmd_t *branch_to(char *label)
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000711{
712 sed_cmd_t *sed_cmd;
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000713
Denis Vlasenko17a15262007-03-26 20:48:46 +0000714 for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000715 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
716 return sed_cmd;
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000717 }
718 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000719 bb_error_msg_and_die("can't find label for jump to '%s'", label);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000720}
Mark Whitley6315ce62000-07-10 22:55:51 +0000721
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000722static void append(char *s)
723{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000724 llist_add_to_end(&G.append_head, xstrdup(s));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000725}
726
727static void flush_append(void)
728{
Rob Landley0b656282006-05-08 22:17:23 +0000729 char *data;
730
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000731 /* Output appended lines. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000732 while ((data = (char *)llist_pop(&G.append_head))) {
733 fprintf(G.nonstdout, "%s\n", data);
Rob Landley0b656282006-05-08 22:17:23 +0000734 free(data);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000735 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000736}
737
Eric Andersen14f5c8d2005-04-16 19:39:00 +0000738static void add_input_file(FILE *file)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000739{
Denis Vlasenkodeeed592008-07-08 05:14:36 +0000740 G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count);
Denis Vlasenko17a15262007-03-26 20:48:46 +0000741 G.input_file_list[G.input_file_count++] = file;
Rob Landleydcc28662004-11-25 07:21:47 +0000742}
743
Denis Vlasenko17a15262007-03-26 20:48:46 +0000744/* Get next line of input from G.input_file_list, flushing append buffer and
Rob Landleydcc28662004-11-25 07:21:47 +0000745 * noting if we ran out of files without a newline on the last line we read.
746 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000747enum {
748 NO_EOL_CHAR = 1,
Denis Vlasenko86811802007-01-29 17:10:19 +0000749 LAST_IS_NUL = 2,
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000750};
751static char *get_next_line(char *gets_char)
Rob Landleydcc28662004-11-25 07:21:47 +0000752{
Denis Vlasenko80778502006-10-25 12:46:46 +0000753 char *temp = NULL;
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000754 int len;
755 char gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000756
757 flush_append();
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000758
759 /* will be returned if last line in the file
760 * doesn't end with either '\n' or '\0' */
761 gc = NO_EOL_CHAR;
Denis Vlasenko17a15262007-03-26 20:48:46 +0000762 while (G.current_input_file < G.input_file_count) {
763 FILE *fp = G.input_file_list[G.current_input_file];
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000764 /* Read line up to a newline or NUL byte, inclusive,
765 * return malloc'ed char[]. length of the chunk read
766 * is stored in len. NULL if EOF/error */
Denis Vlasenko86811802007-01-29 17:10:19 +0000767 temp = bb_get_chunk_from_file(fp, &len);
Rob Landley2b26fd52006-02-24 02:30:39 +0000768 if (temp) {
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000769 /* len > 0 here, it's ok to do temp[len-1] */
770 char c = temp[len-1];
771 if (c == '\n' || c == '\0') {
772 temp[len-1] = '\0';
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000773 gc = c;
Denis Vlasenko86811802007-01-29 17:10:19 +0000774 if (c == '\0') {
775 int ch = fgetc(fp);
776 if (ch != EOF)
777 ungetc(ch, fp);
778 else
779 gc = LAST_IS_NUL;
780 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000781 }
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000782 /* else we put NO_EOL_CHAR into *gets_char */
783 break;
784
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000785 /* NB: I had the idea of peeking next file(s) and returning
786 * NO_EOL_CHAR only if it is the *last* non-empty
787 * input file. But there is a case where this won't work:
788 * file1: "a woo\nb woo"
789 * file2: "c no\nd no"
790 * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
791 * (note: *no* newline after "b bang"!) */
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000792 }
793 /* Close this file and advance to next one */
Denis Vlasenko86811802007-01-29 17:10:19 +0000794 fclose(fp);
Denis Vlasenko17a15262007-03-26 20:48:46 +0000795 G.current_input_file++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000796 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000797 *gets_char = gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000798 return temp;
799}
800
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000801/* Output line of text. */
802/* Note:
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000803 * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000804 * Without them, we had this:
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000805 * echo -n thingy >z1
806 * echo -n again >z2
807 * >znull
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000808 * sed "s/i/z/" z1 z2 znull | hexdump -vC
809 * output:
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000810 * gnu sed 4.1.5:
811 * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
812 * bbox:
813 * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000814 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000815static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000816{
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000817 char lpc = *last_puts_char;
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000818
Denis Vlasenko86811802007-01-29 17:10:19 +0000819 /* Need to insert a '\n' between two files because first file's
820 * last line wasn't terminated? */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000821 if (lpc != '\n' && lpc != '\0') {
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000822 fputc('\n', file);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000823 lpc = '\n';
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000824 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000825 fputs(s, file);
Denis Vlasenko86811802007-01-29 17:10:19 +0000826
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000827 /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000828 if (s[0])
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000829 lpc = 'x';
Denis Vlasenko86811802007-01-29 17:10:19 +0000830
831 /* had trailing '\0' and it was last char of file? */
832 if (last_gets_char == LAST_IS_NUL) {
833 fputc('\0', file);
834 lpc = 'x'; /* */
835 } else
836 /* had trailing '\n' or '\0'? */
837 if (last_gets_char != NO_EOL_CHAR) {
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000838 fputc(last_gets_char, file);
839 lpc = last_gets_char;
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000840 }
Denis Vlasenko86811802007-01-29 17:10:19 +0000841
Denis Vlasenko80778502006-10-25 12:46:46 +0000842 if (ferror(file)) {
Denis Vlasenko40920822006-10-03 20:28:06 +0000843 xfunc_error_retval = 4; /* It's what gnu sed exits with... */
Bernhard Reutner-Fischera3d4bf32006-06-03 21:40:11 +0000844 bb_error_msg_and_die(bb_msg_write_error);
Rob Landley53302f82004-02-18 09:54:15 +0000845 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000846 *last_puts_char = lpc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000847}
848
Denis Vlasenko17a15262007-03-26 20:48:46 +0000849#define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000850
Denis Vlasenko8274e062007-08-06 03:41:08 +0000851static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
852{
853 int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
854 if (retval)
855 G.previous_regex_ptr = sed_cmd->beg_match;
856 return retval;
857}
858
Rob Landley2b26fd52006-02-24 02:30:39 +0000859/* Process all the lines in all the files */
860
Rob Landleydcc28662004-11-25 07:21:47 +0000861static void process_files(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000862{
Rob Landleyce4f0e92004-10-30 06:54:19 +0000863 char *pattern_space, *next_line;
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000864 int linenum = 0;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000865 char last_puts_char = '\n';
866 char last_gets_char, next_gets_char;
Denis Vlasenko2ea630f2006-12-10 02:52:19 +0000867 sed_cmd_t *sed_cmd;
868 int substituted;
Mark Whitley6315ce62000-07-10 22:55:51 +0000869
Rob Landley2b26fd52006-02-24 02:30:39 +0000870 /* Prime the pump */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000871 next_line = get_next_line(&next_gets_char);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000872
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200873 /* Go through every line in each file */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +0000874 again:
Denis Vlasenko2ea630f2006-12-10 02:52:19 +0000875 substituted = 0;
Glenn L McGrath505bd0f2003-03-08 05:21:02 +0000876
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000877 /* Advance to next line. Stop if out of lines. */
878 pattern_space = next_line;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200879 if (!pattern_space)
880 return;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000881 last_gets_char = next_gets_char;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000882
883 /* Read one line in advance so we can act on the last line,
884 * the '$' address */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000885 next_line = get_next_line(&next_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000886 linenum++;
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200887
888 /* For every line, go through all the commands */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +0000889 restart:
Denis Vlasenko17a15262007-03-26 20:48:46 +0000890 for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000891 int old_matched, matched;
Eric Andersen52a3c272003-12-23 08:53:51 +0000892
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000893 old_matched = sed_cmd->in_match;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000894
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000895 /* Determine if this command matches this line: */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200896
897 /* Are we continuing a previous multi-line match? */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000898 sed_cmd->in_match = sed_cmd->in_match
899 /* Or is no range necessary? */
900 || (!sed_cmd->beg_line && !sed_cmd->end_line
901 && !sed_cmd->beg_match && !sed_cmd->end_match)
902 /* Or did we match the start of a numerical range? */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200903 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line <= linenum))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000904 /* Or does this line match our begin address regex? */
Denis Vlasenko8274e062007-08-06 03:41:08 +0000905 || (beg_match(sed_cmd, pattern_space))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000906 /* Or did we match last line of input? */
907 || (sed_cmd->beg_line == -1 && next_line == NULL);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000908
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200909 /* Snapshot the value */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000910 matched = sed_cmd->in_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000911
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200912 //bb_error_msg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
913 //sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000914
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200915 /* Is this line the end of the current match? */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200916
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000917 if (matched) {
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200918 /* once matched, "n,xxx" range is dead, disabling it */
919 if (sed_cmd->beg_line > 0)
920 sed_cmd->beg_line = -2;
921 sed_cmd->in_match = !(
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000922 /* has the ending line come, or is this a single address command? */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200923 (sed_cmd->end_line ?
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000924 sed_cmd->end_line == -1 ?
925 !next_line
926 : (sed_cmd->end_line <= linenum)
927 : !sed_cmd->end_match
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200928 )
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000929 /* or does this line matches our last address regex */
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200930 || (sed_cmd->end_match && old_matched
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000931 && (regexec(sed_cmd->end_match,
Denys Vlasenko8e96b5b2009-07-14 01:02:57 +0200932 pattern_space, 0, NULL, 0) == 0))
933 );
Erik Andersene49d5ec2000-02-08 19:58:47 +0000934 }
Erik Andersen1266a131999-12-29 22:19:46 +0000935
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200936 /* Skip blocks of commands we didn't match */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000937 if (sed_cmd->cmd == '{') {
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000938 if (sed_cmd->invert ? matched : !matched) {
939 while (sed_cmd->cmd != '}') {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000940 sed_cmd = sed_cmd->next;
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000941 if (!sed_cmd)
942 bb_error_msg_and_die("unterminated {");
943 }
944 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000945 continue;
946 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000947
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000948 /* Okay, so did this line match? */
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200949 if (sed_cmd->invert ? matched : !matched)
950 continue; /* no */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000951
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200952 /* Update last used regex in case a blank substitute BRE is found */
953 if (sed_cmd->beg_match) {
954 G.previous_regex_ptr = sed_cmd->beg_match;
955 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000956
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200957 /* actual sedding */
958 switch (sed_cmd->cmd) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000959
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200960 /* Print line number */
961 case '=':
962 fprintf(G.nonstdout, "%d\n", linenum);
963 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000964
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200965 /* Write the current pattern space up to the first newline */
966 case 'P':
967 {
968 char *tmp = strchr(pattern_space, '\n');
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000969
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200970 if (tmp) {
971 *tmp = '\0';
972 /* TODO: explain why '\n' below */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000973 sed_puts(pattern_space, '\n');
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200974 *tmp = '\n';
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000975 break;
976 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200977 /* Fall Through */
978 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000979
Denys Vlasenko8bca3e22009-06-30 19:19:37 +0200980 /* Write the current pattern space to output */
981 case 'p':
982 /* NB: we print this _before_ the last line
983 * (of current file) is printed. Even if
984 * that line is nonterminated, we print
985 * '\n' here (gnu sed does the same) */
986 sed_puts(pattern_space, '\n');
987 break;
988 /* Delete up through first newline */
989 case 'D':
990 {
991 char *tmp = strchr(pattern_space, '\n');
992
993 if (tmp) {
994 tmp = xstrdup(tmp+1);
995 free(pattern_space);
996 pattern_space = tmp;
997 goto restart;
998 }
999 }
1000 /* discard this line. */
1001 case 'd':
1002 goto discard_line;
1003
1004 /* Substitute with regex */
1005 case 's':
1006 if (!do_subst_command(sed_cmd, &pattern_space))
1007 break;
1008 substituted |= 1;
1009
1010 /* handle p option */
1011 if (sed_cmd->sub_p)
1012 sed_puts(pattern_space, last_gets_char);
1013 /* handle w option */
1014 if (sed_cmd->sw_file)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001015 puts_maybe_newline(
1016 pattern_space, sed_cmd->sw_file,
1017 &sed_cmd->sw_last_char, last_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001018 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001019
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001020 /* Append line to linked list to be printed later */
1021 case 'a':
1022 append(sed_cmd->string);
1023 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001024
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001025 /* Insert text before this line */
1026 case 'i':
1027 sed_puts(sed_cmd->string, '\n');
1028 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001029
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001030 /* Cut and paste text (replace) */
1031 case 'c':
1032 /* Only triggers on last line of a matching range. */
1033 if (!sed_cmd->in_match)
1034 sed_puts(sed_cmd->string, NO_EOL_CHAR);
1035 goto discard_line;
1036
1037 /* Read file, append contents to output */
1038 case 'r':
1039 {
1040 FILE *rfile;
1041
1042 rfile = fopen_for_read(sed_cmd->string);
1043 if (rfile) {
1044 char *line;
1045
1046 while ((line = xmalloc_fgetline(rfile))
1047 != NULL)
1048 append(line);
1049 xprint_and_close_file(rfile);
1050 }
1051
1052 break;
1053 }
1054
1055 /* Write pattern space to file. */
1056 case 'w':
1057 puts_maybe_newline(
1058 pattern_space, sed_cmd->sw_file,
1059 &sed_cmd->sw_last_char, last_gets_char);
1060 break;
1061
1062 /* Read next line from input */
1063 case 'n':
1064 if (!G.be_quiet)
1065 sed_puts(pattern_space, last_gets_char);
1066 if (next_line) {
1067 free(pattern_space);
1068 pattern_space = next_line;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001069 last_gets_char = next_gets_char;
1070 next_line = get_next_line(&next_gets_char);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001071 substituted = 0;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001072 linenum++;
1073 break;
1074 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001075 /* fall through */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001076
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001077 /* Quit. End of script, end of input. */
1078 case 'q':
1079 /* Exit the outer while loop */
1080 free(next_line);
1081 next_line = NULL;
1082 goto discard_commands;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001083
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001084 /* Append the next line to the current line */
1085 case 'N':
1086 {
1087 int len;
1088 /* If no next line, jump to end of script and exit. */
1089 if (next_line == NULL) {
1090 /* Jump to end of script and exit */
1091 free(next_line);
1092 next_line = NULL;
1093 goto discard_line;
1094 /* append next_line, read new next_line. */
1095 }
1096 len = strlen(pattern_space);
1097 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1098 pattern_space[len] = '\n';
1099 strcpy(pattern_space + len+1, next_line);
1100 last_gets_char = next_gets_char;
1101 next_line = get_next_line(&next_gets_char);
1102 linenum++;
1103 break;
1104 }
1105
1106 /* Test/branch if substitution occurred */
1107 case 't':
1108 if (!substituted) break;
1109 substituted = 0;
1110 /* Fall through */
1111 /* Test/branch if substitution didn't occur */
1112 case 'T':
1113 if (substituted) break;
1114 /* Fall through */
1115 /* Branch to label */
1116 case 'b':
1117 if (!sed_cmd->string) goto discard_commands;
1118 else sed_cmd = branch_to(sed_cmd->string);
1119 break;
1120 /* Transliterate characters */
1121 case 'y':
1122 {
1123 int i, j;
1124
1125 for (i = 0; pattern_space[i]; i++) {
1126 for (j = 0; sed_cmd->string[j]; j += 2) {
1127 if (pattern_space[i] == sed_cmd->string[j]) {
1128 pattern_space[i] = sed_cmd->string[j + 1];
1129 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001130 }
1131 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001132 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001133
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001134 break;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001135 }
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001136 case 'g': /* Replace pattern space with hold space */
1137 free(pattern_space);
1138 pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1139 break;
1140 case 'G': /* Append newline and hold space to pattern space */
1141 {
1142 int pattern_space_size = 2;
1143 int hold_space_size = 0;
1144
1145 if (pattern_space)
1146 pattern_space_size += strlen(pattern_space);
1147 if (G.hold_space)
1148 hold_space_size = strlen(G.hold_space);
1149 pattern_space = xrealloc(pattern_space,
1150 pattern_space_size + hold_space_size);
1151 if (pattern_space_size == 2)
1152 pattern_space[0] = 0;
1153 strcat(pattern_space, "\n");
1154 if (G.hold_space)
1155 strcat(pattern_space, G.hold_space);
1156 last_gets_char = '\n';
1157
1158 break;
1159 }
1160 case 'h': /* Replace hold space with pattern space */
1161 free(G.hold_space);
1162 G.hold_space = xstrdup(pattern_space);
1163 break;
1164 case 'H': /* Append newline and pattern space to hold space */
1165 {
1166 int hold_space_size = 2;
1167 int pattern_space_size = 0;
1168
1169 if (G.hold_space)
1170 hold_space_size += strlen(G.hold_space);
1171 if (pattern_space)
1172 pattern_space_size = strlen(pattern_space);
1173 G.hold_space = xrealloc(G.hold_space,
1174 hold_space_size + pattern_space_size);
1175
1176 if (hold_space_size == 2)
1177 *G.hold_space = 0;
1178 strcat(G.hold_space, "\n");
1179 if (pattern_space)
1180 strcat(G.hold_space, pattern_space);
1181
1182 break;
1183 }
1184 case 'x': /* Exchange hold and pattern space */
1185 {
1186 char *tmp = pattern_space;
Denys Vlasenko90a99042009-09-06 02:36:23 +02001187 pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001188 last_gets_char = '\n';
1189 G.hold_space = tmp;
1190 break;
1191 }
1192 } /* switch */
1193 } /* for each cmd */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001194
1195 /*
Denys Vlasenko8bca3e22009-06-30 19:19:37 +02001196 * Exit point from sedding...
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001197 */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001198 discard_commands:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001199 /* we will print the line unless we were told to be quiet ('-n')
1200 or if the line was suppressed (ala 'd'elete) */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001201 if (!G.be_quiet)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001202 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001203
1204 /* Delete and such jump here. */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001205 discard_line:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001206 flush_append();
1207 free(pattern_space);
1208
1209 goto again;
Erik Andersen1266a131999-12-29 22:19:46 +00001210}
1211
Glenn L McGrath42c25732003-10-04 05:27:56 +00001212/* It is possible to have a command line argument with embedded
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001213 * newlines. This counts as multiple command lines.
1214 * However, newline can be escaped: 's/e/z\<newline>z/'
1215 * We check for this.
1216 */
Glenn L McGrath42c25732003-10-04 05:27:56 +00001217
1218static void add_cmd_block(char *cmdstr)
1219{
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001220 char *sv, *eol;
Glenn L McGrath42c25732003-10-04 05:27:56 +00001221
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001222 cmdstr = sv = xstrdup(cmdstr);
1223 do {
1224 eol = strchr(cmdstr, '\n');
1225 next:
1226 if (eol) {
1227 /* Count preceding slashes */
1228 int slashes = 0;
1229 char *sl = eol;
1230
1231 while (sl != cmdstr && *--sl == '\\')
1232 slashes++;
1233 /* Odd number of preceding slashes - newline is escaped */
1234 if (slashes & 1) {
Denis Vlasenko0f293b92008-07-22 20:16:55 +00001235 overlapping_strcpy(eol - 1, eol);
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001236 eol = strchr(eol, '\n');
1237 goto next;
1238 }
1239 *eol = '\0';
1240 }
1241 add_cmd(cmdstr);
1242 cmdstr = eol + 1;
1243 } while (eol);
1244 free(sv);
Glenn L McGrath42c25732003-10-04 05:27:56 +00001245}
1246
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +00001247int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +00001248int sed_main(int argc UNUSED_PARAM, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001249{
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001250 enum {
1251 OPT_in_place = 1 << 0,
1252 };
Denis Vlasenko67b23e62006-10-03 21:00:06 +00001253 unsigned opt;
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001254 llist_t *opt_e, *opt_f;
1255 int status = EXIT_SUCCESS;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001256
Denis Vlasenko74324c82007-06-04 10:16:52 +00001257 INIT_G();
Rob Landleye3f5a3f2006-05-09 03:53:55 +00001258
Mark Whitley858c1ad2000-07-11 21:38:47 +00001259 /* destroy command strings on exit */
Rob Landley0b656282006-05-08 22:17:23 +00001260 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
Mark Whitley858c1ad2000-07-11 21:38:47 +00001261
Rob Landleyfae1dc82005-11-20 07:44:35 +00001262 /* Lie to autoconf when it starts asking stupid questions. */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001263 if (argv[1] && !strcmp(argv[1], "--version")) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001264 puts("This is not GNU sed version 4.0");
1265 return 0;
Eric Andersenc06f5682004-02-04 10:57:46 +00001266 }
Eric Andersenc06f5682004-02-04 10:57:46 +00001267
Mark Whitley6315ce62000-07-10 22:55:51 +00001268 /* do normal option parsing */
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001269 opt_e = opt_f = NULL;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001270 opt_complementary = "e::f::" /* can occur multiple times */
1271 "nn"; /* count -n */
Denis Vlasenkofe7cd642007-08-18 15:32:12 +00001272 opt = getopt32(argv, "irne:f:", &opt_e, &opt_f,
Denis Vlasenko17a15262007-03-26 20:48:46 +00001273 &G.be_quiet); /* counter for -n */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001274 //argc -= optind;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001275 argv += optind;
1276 if (opt & OPT_in_place) { // -i
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001277 atexit(cleanup_outname);
1278 }
Denis Vlasenko17a15262007-03-26 20:48:46 +00001279 if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r
1280 //if (opt & 0x4) G.be_quiet++; // -n
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001281 while (opt_e) { // -e
Denis Vlasenkod50dda82008-06-15 05:40:56 +00001282 add_cmd_block(llist_pop(&opt_e));
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001283 }
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001284 while (opt_f) { // -f
1285 char *line;
1286 FILE *cmdfile;
Denis Vlasenko5415c852008-07-21 23:05:26 +00001287 cmdfile = xfopen_for_read(llist_pop(&opt_f));
Denis Vlasenko8ee649a2008-03-26 20:04:27 +00001288 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001289 add_cmd(line);
1290 free(line);
1291 }
1292 fclose(cmdfile);
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001293 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001294 /* if we didn't get a pattern from -e or -f, use argv[0] */
Denis Vlasenko80778502006-10-25 12:46:46 +00001295 if (!(opt & 0x18)) {
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001296 if (!*argv)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001297 bb_show_usage();
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001298 add_cmd_block(*argv++);
Mark Whitley6315ce62000-07-10 22:55:51 +00001299 }
Glenn L McGrath42c25732003-10-04 05:27:56 +00001300 /* Flush any unfinished commands. */
1301 add_cmd("");
Mark Whitley6315ce62000-07-10 22:55:51 +00001302
Rob Landley53302f82004-02-18 09:54:15 +00001303 /* By default, we write to stdout */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001304 G.nonstdout = stdout;
Rob Landley53302f82004-02-18 09:54:15 +00001305
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001306 /* argv[0..(argc-1)] should be names of file to process. If no
Mark Whitley6315ce62000-07-10 22:55:51 +00001307 * files were specified or '-' was specified, take input from stdin.
1308 * Otherwise, we process all the files specified. */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001309 if (argv[0] == NULL) {
1310 if (opt & OPT_in_place)
Denis Vlasenko80778502006-10-25 12:46:46 +00001311 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
Rob Landleydcc28662004-11-25 07:21:47 +00001312 add_input_file(stdin);
1313 process_files();
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001314 } else {
Mark Whitley6315ce62000-07-10 22:55:51 +00001315 int i;
1316 FILE *file;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001317
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001318 for (i = 0; argv[i]; i++) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001319 struct stat statbuf;
1320 int nonstdoutfd;
1321
Denis Vlasenko9f739442006-12-16 23:49:13 +00001322 if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
Rob Landleydcc28662004-11-25 07:21:47 +00001323 add_input_file(stdin);
1324 process_files();
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001325 continue;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001326 }
Denis Vlasenkoddec5af2006-10-26 23:25:17 +00001327 file = fopen_or_warn(argv[i], "r");
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001328 if (!file) {
1329 status = EXIT_FAILURE;
1330 continue;
1331 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001332 if (!(opt & OPT_in_place)) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001333 add_input_file(file);
1334 continue;
1335 }
1336
Denis Vlasenko17a15262007-03-26 20:48:46 +00001337 G.outname = xasprintf("%sXXXXXX", argv[i]);
1338 nonstdoutfd = mkstemp(G.outname);
Denis Vlasenko80778502006-10-25 12:46:46 +00001339 if (-1 == nonstdoutfd)
Denis Vlasenko17a15262007-03-26 20:48:46 +00001340 bb_perror_msg_and_die("cannot create temp file %s", G.outname);
1341 G.nonstdout = fdopen(nonstdoutfd, "w");
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001342
1343 /* Set permissions of output file */
1344
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001345 fstat(fileno(file), &statbuf);
1346 fchmod(nonstdoutfd, statbuf.st_mode);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001347 add_input_file(file);
1348 process_files();
Denis Vlasenko17a15262007-03-26 20:48:46 +00001349 fclose(G.nonstdout);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001350
Denis Vlasenko17a15262007-03-26 20:48:46 +00001351 G.nonstdout = stdout;
Denis Vlasenko80778502006-10-25 12:46:46 +00001352 /* unlink(argv[i]); */
Denis Vlasenkocb448fe2008-02-17 14:28:53 +00001353 xrename(G.outname, argv[i]);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001354 free(G.outname);
Denis Vlasenko40276642007-11-13 16:48:10 +00001355 G.outname = NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +00001356 }
Denis Vlasenko17a15262007-03-26 20:48:46 +00001357 if (G.input_file_count > G.current_input_file)
Denis Vlasenko80778502006-10-25 12:46:46 +00001358 process_files();
Mark Whitley6315ce62000-07-10 22:55:51 +00001359 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001360
Matt Kraaia5f09c62001-11-12 16:44:55 +00001361 return status;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001362}