blob: 32911f8f9bf75a64d56ac87dc42580cd2b543d35 [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
Rob Landleydcc28662004-11-25 07:21:47 +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 Vlasenkofe7a9f12007-01-29 14:31:47 +0000441 sed_cmd->sw_file = xfopen(sed_cmd->string, "w");
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{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000592 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
Rob Landleye3f5a3f2006-05-09 03:53:55 +0000627static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
Mark Whitley4f7fe772000-07-13 20:01:58 +0000628{
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000629 char *oldline = *line;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000630 int altered = 0;
Denis Vlasenko80778502006-10-25 12:46:46 +0000631 int match_count = 0;
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000632 regex_t *current_regex;
633
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000634 /* Handle empty regex. */
Glenn L McGrathc1d95072003-04-08 06:42:45 +0000635 if (sed_cmd->sub_match == NULL) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000636 current_regex = G.previous_regex_ptr;
Denis Vlasenko80778502006-10-25 12:46:46 +0000637 if (!current_regex)
Denis Vlasenkod3d004d2006-10-27 09:02:31 +0000638 bb_error_msg_and_die("no previous regexp");
Denis Vlasenko80778502006-10-25 12:46:46 +0000639 } else
Denis Vlasenko17a15262007-03-26 20:48:46 +0000640 G.previous_regex_ptr = current_regex = sed_cmd->sub_match;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000641
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000642 /* Find the first match */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000643 if (REG_NOMATCH == regexec(current_regex, oldline, 10, G.regmatch, 0))
Mark Whitley2dc192f2000-11-03 19:47:00 +0000644 return 0;
645
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000646 /* Initialize temporary output buffer. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000647 G.pipeline.buf = xmalloc(PIPE_GROW);
648 G.pipeline.len = PIPE_GROW;
649 G.pipeline.idx = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000650
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000651 /* Now loop through, substituting for matches */
652 do {
Mark Whitley4f7fe772000-07-13 20:01:58 +0000653 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000654
Eric Andersenc06f5682004-02-04 10:57:46 +0000655 /* Work around bug in glibc regexec, demonstrated by:
656 echo " a.b" | busybox sed 's [^ .]* x g'
657 The match_count check is so not to break
658 echo "hi" | busybox sed 's/^/!/g' */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000659 if (!G.regmatch[0].rm_so && !G.regmatch[0].rm_eo && match_count) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000660 pipe_putc(*oldline++);
Eric Andersenc06f5682004-02-04 10:57:46 +0000661 continue;
662 }
663
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000664 match_count++;
665
666 /* If we aren't interested in this match, output old line to
667 end of match and continue */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000668 if (sed_cmd->which_match && sed_cmd->which_match != match_count) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000669 for (i = 0; i < G.regmatch[0].rm_eo; i++)
Denis Vlasenko80778502006-10-25 12:46:46 +0000670 pipe_putc(*oldline++);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000671 continue;
672 }
673
Mark Whitley2dc192f2000-11-03 19:47:00 +0000674 /* print everything before the match */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000675 for (i = 0; i < G.regmatch[0].rm_so; i++)
Denis Vlasenko80778502006-10-25 12:46:46 +0000676 pipe_putc(oldline[i]);
Mark Whitley97562bd2000-07-17 20:06:42 +0000677
Mark Whitley2dc192f2000-11-03 19:47:00 +0000678 /* then print the substitution string */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000679 do_subst_w_backrefs(oldline, sed_cmd->string);
Mark Whitley97562bd2000-07-17 20:06:42 +0000680
Mark Whitley2dc192f2000-11-03 19:47:00 +0000681 /* advance past the match */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000682 oldline += G.regmatch[0].rm_eo;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000683 /* flag that something has changed */
684 altered++;
Mark Whitley97562bd2000-07-17 20:06:42 +0000685
Mark Whitley2dc192f2000-11-03 19:47:00 +0000686 /* if we're not doing this globally, get out now */
Denis Vlasenko284d0fa2008-02-16 13:18:17 +0000687 if (sed_cmd->which_match)
688 break;
Denis Vlasenko17a15262007-03-26 20:48:46 +0000689 } while (*oldline && (regexec(current_regex, oldline, 10, G.regmatch, 0) != REG_NOMATCH));
Mark Whitley2dc192f2000-11-03 19:47:00 +0000690
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000691 /* Copy rest of string into output pipeline */
692
Denis Vlasenko80778502006-10-25 12:46:46 +0000693 while (*oldline)
694 pipe_putc(*oldline++);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000695 pipe_putc(0);
Mark Whitley2dc192f2000-11-03 19:47:00 +0000696
Eric Andersenb76cb682001-08-22 05:58:16 +0000697 free(*line);
Denis Vlasenko17a15262007-03-26 20:48:46 +0000698 *line = G.pipeline.buf;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000699 return altered;
700}
Mark Whitley6315ce62000-07-10 22:55:51 +0000701
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000702/* Set command pointer to point to this label. (Does not handle null label.) */
Rob Landley4795e4e2006-07-26 17:25:08 +0000703static sed_cmd_t *branch_to(char *label)
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000704{
705 sed_cmd_t *sed_cmd;
Glenn L McGrathd4185b02003-04-11 17:10:23 +0000706
Denis Vlasenko17a15262007-03-26 20:48:46 +0000707 for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko80778502006-10-25 12:46:46 +0000708 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
709 return sed_cmd;
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000710 }
711 }
Denis Vlasenko80778502006-10-25 12:46:46 +0000712 bb_error_msg_and_die("can't find label for jump to '%s'", label);
Glenn L McGrath961c6c12003-03-28 04:43:39 +0000713}
Mark Whitley6315ce62000-07-10 22:55:51 +0000714
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000715static void append(char *s)
716{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000717 llist_add_to_end(&G.append_head, xstrdup(s));
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000718}
719
720static void flush_append(void)
721{
Rob Landley0b656282006-05-08 22:17:23 +0000722 char *data;
723
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000724 /* Output appended lines. */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000725 while ((data = (char *)llist_pop(&G.append_head))) {
726 fprintf(G.nonstdout, "%s\n", data);
Rob Landley0b656282006-05-08 22:17:23 +0000727 free(data);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000728 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000729}
730
Eric Andersen14f5c8d2005-04-16 19:39:00 +0000731static void add_input_file(FILE *file)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000732{
Denis Vlasenko17a15262007-03-26 20:48:46 +0000733 G.input_file_list = xrealloc(G.input_file_list,
734 (G.input_file_count + 1) * sizeof(FILE *));
735 G.input_file_list[G.input_file_count++] = file;
Rob Landleydcc28662004-11-25 07:21:47 +0000736}
737
Denis Vlasenko17a15262007-03-26 20:48:46 +0000738/* Get next line of input from G.input_file_list, flushing append buffer and
Rob Landleydcc28662004-11-25 07:21:47 +0000739 * noting if we ran out of files without a newline on the last line we read.
740 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000741enum {
742 NO_EOL_CHAR = 1,
Denis Vlasenko86811802007-01-29 17:10:19 +0000743 LAST_IS_NUL = 2,
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000744};
745static char *get_next_line(char *gets_char)
Rob Landleydcc28662004-11-25 07:21:47 +0000746{
Denis Vlasenko80778502006-10-25 12:46:46 +0000747 char *temp = NULL;
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000748 int len;
749 char gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000750
751 flush_append();
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000752
753 /* will be returned if last line in the file
754 * doesn't end with either '\n' or '\0' */
755 gc = NO_EOL_CHAR;
Denis Vlasenko17a15262007-03-26 20:48:46 +0000756 while (G.current_input_file < G.input_file_count) {
757 FILE *fp = G.input_file_list[G.current_input_file];
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000758 /* Read line up to a newline or NUL byte, inclusive,
759 * return malloc'ed char[]. length of the chunk read
760 * is stored in len. NULL if EOF/error */
Denis Vlasenko86811802007-01-29 17:10:19 +0000761 temp = bb_get_chunk_from_file(fp, &len);
Rob Landley2b26fd52006-02-24 02:30:39 +0000762 if (temp) {
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000763 /* len > 0 here, it's ok to do temp[len-1] */
764 char c = temp[len-1];
765 if (c == '\n' || c == '\0') {
766 temp[len-1] = '\0';
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000767 gc = c;
Denis Vlasenko86811802007-01-29 17:10:19 +0000768 if (c == '\0') {
769 int ch = fgetc(fp);
770 if (ch != EOF)
771 ungetc(ch, fp);
772 else
773 gc = LAST_IS_NUL;
774 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000775 }
Denis Vlasenkoc562bb72007-01-29 17:08:51 +0000776 /* else we put NO_EOL_CHAR into *gets_char */
777 break;
778
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000779 /* NB: I had the idea of peeking next file(s) and returning
780 * NO_EOL_CHAR only if it is the *last* non-empty
781 * input file. But there is a case where this won't work:
782 * file1: "a woo\nb woo"
783 * file2: "c no\nd no"
784 * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
785 * (note: *no* newline after "b bang"!) */
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000786 }
787 /* Close this file and advance to next one */
Denis Vlasenko86811802007-01-29 17:10:19 +0000788 fclose(fp);
Denis Vlasenko17a15262007-03-26 20:48:46 +0000789 G.current_input_file++;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000790 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000791 *gets_char = gc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000792 return temp;
793}
794
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000795/* Output line of text. */
796/* Note:
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000797 * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000798 * Without them, we had this:
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000799 * echo -n thingy >z1
800 * echo -n again >z2
801 * >znull
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000802 * sed "s/i/z/" z1 z2 znull | hexdump -vC
803 * output:
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000804 * gnu sed 4.1.5:
805 * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
806 * bbox:
807 * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000808 */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000809static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000810{
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000811 char lpc = *last_puts_char;
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000812
Denis Vlasenko86811802007-01-29 17:10:19 +0000813 /* Need to insert a '\n' between two files because first file's
814 * last line wasn't terminated? */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000815 if (lpc != '\n' && lpc != '\0') {
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000816 fputc('\n', file);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000817 lpc = '\n';
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000818 }
Denis Vlasenko8b22b072006-12-02 17:58:10 +0000819 fputs(s, file);
Denis Vlasenko86811802007-01-29 17:10:19 +0000820
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000821 /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
Denis Vlasenkoef44d9d2007-01-17 23:16:16 +0000822 if (s[0])
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000823 lpc = 'x';
Denis Vlasenko86811802007-01-29 17:10:19 +0000824
825 /* had trailing '\0' and it was last char of file? */
826 if (last_gets_char == LAST_IS_NUL) {
827 fputc('\0', file);
828 lpc = 'x'; /* */
829 } else
830 /* had trailing '\n' or '\0'? */
831 if (last_gets_char != NO_EOL_CHAR) {
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000832 fputc(last_gets_char, file);
833 lpc = last_gets_char;
Denis Vlasenko1375bc72006-12-02 20:12:12 +0000834 }
Denis Vlasenko86811802007-01-29 17:10:19 +0000835
Denis Vlasenko80778502006-10-25 12:46:46 +0000836 if (ferror(file)) {
Denis Vlasenko40920822006-10-03 20:28:06 +0000837 xfunc_error_retval = 4; /* It's what gnu sed exits with... */
Bernhard Reutner-Fischera3d4bf32006-06-03 21:40:11 +0000838 bb_error_msg_and_die(bb_msg_write_error);
Rob Landley53302f82004-02-18 09:54:15 +0000839 }
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000840 *last_puts_char = lpc;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000841}
842
Denis Vlasenko17a15262007-03-26 20:48:46 +0000843#define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000844
Denis Vlasenko8274e062007-08-06 03:41:08 +0000845static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
846{
847 int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
848 if (retval)
849 G.previous_regex_ptr = sed_cmd->beg_match;
850 return retval;
851}
852
Rob Landley2b26fd52006-02-24 02:30:39 +0000853/* Process all the lines in all the files */
854
Rob Landleydcc28662004-11-25 07:21:47 +0000855static void process_files(void)
Mark Whitley6315ce62000-07-10 22:55:51 +0000856{
Rob Landleyce4f0e92004-10-30 06:54:19 +0000857 char *pattern_space, *next_line;
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000858 int linenum = 0;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000859 char last_puts_char = '\n';
860 char last_gets_char, next_gets_char;
Denis Vlasenko2ea630f2006-12-10 02:52:19 +0000861 sed_cmd_t *sed_cmd;
862 int substituted;
Mark Whitley6315ce62000-07-10 22:55:51 +0000863
Rob Landley2b26fd52006-02-24 02:30:39 +0000864 /* Prime the pump */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000865 next_line = get_next_line(&next_gets_char);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000866
Rob Landleydcc28662004-11-25 07:21:47 +0000867 /* go through every line in each file */
Denis Vlasenkof39c7c02008-02-28 17:59:01 +0000868 again:
Denis Vlasenko2ea630f2006-12-10 02:52:19 +0000869 substituted = 0;
Glenn L McGrath505bd0f2003-03-08 05:21:02 +0000870
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000871 /* Advance to next line. Stop if out of lines. */
872 pattern_space = next_line;
873 if (!pattern_space) return;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000874 last_gets_char = next_gets_char;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000875
876 /* Read one line in advance so we can act on the last line,
877 * the '$' address */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000878 next_line = get_next_line(&next_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000879 linenum++;
Denis Vlasenkof39c7c02008-02-28 17:59:01 +0000880 restart:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000881 /* for every line, go through all the commands */
Denis Vlasenko17a15262007-03-26 20:48:46 +0000882 for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000883 int old_matched, matched;
Eric Andersen52a3c272003-12-23 08:53:51 +0000884
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000885 old_matched = sed_cmd->in_match;
Mark Whitley0915c4b2001-06-11 23:50:06 +0000886
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000887 /* Determine if this command matches this line: */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000888
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000889 /* Are we continuing a previous multi-line match? */
890 sed_cmd->in_match = sed_cmd->in_match
891 /* Or is no range necessary? */
892 || (!sed_cmd->beg_line && !sed_cmd->end_line
893 && !sed_cmd->beg_match && !sed_cmd->end_match)
894 /* Or did we match the start of a numerical range? */
895 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
896 /* Or does this line match our begin address regex? */
Denis Vlasenko8274e062007-08-06 03:41:08 +0000897 || (beg_match(sed_cmd, pattern_space))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000898 /* Or did we match last line of input? */
899 || (sed_cmd->beg_line == -1 && next_line == NULL);
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000900
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000901 /* Snapshot the value */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000902
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000903 matched = sed_cmd->in_match;
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000904
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000905 /* Is this line the end of the current match? */
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000906
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000907 if (matched) {
908 sed_cmd->in_match = !(
909 /* has the ending line come, or is this a single address command? */
910 (sed_cmd->end_line ?
911 sed_cmd->end_line == -1 ?
912 !next_line
913 : (sed_cmd->end_line <= linenum)
914 : !sed_cmd->end_match
915 )
916 /* or does this line matches our last address regex */
917 || (sed_cmd->end_match && old_matched
918 && (regexec(sed_cmd->end_match,
Denis Vlasenko8e858e22007-03-07 09:35:43 +0000919 pattern_space, 0, NULL, 0) == 0))
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000920 );
Erik Andersene49d5ec2000-02-08 19:58:47 +0000921 }
Erik Andersen1266a131999-12-29 22:19:46 +0000922
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000923 /* Skip blocks of commands we didn't match. */
924 if (sed_cmd->cmd == '{') {
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000925 if (sed_cmd->invert ? matched : !matched) {
926 while (sed_cmd->cmd != '}') {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000927 sed_cmd = sed_cmd->next;
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000928 if (!sed_cmd)
929 bb_error_msg_and_die("unterminated {");
930 }
931 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000932 continue;
933 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +0000934
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000935 /* Okay, so did this line match? */
936 if (sed_cmd->invert ? !matched : matched) {
937 /* Update last used regex in case a blank substitute BRE is found */
938 if (sed_cmd->beg_match) {
Denis Vlasenko17a15262007-03-26 20:48:46 +0000939 G.previous_regex_ptr = sed_cmd->beg_match;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000940 }
941
942 /* actual sedding */
943 switch (sed_cmd->cmd) {
944
945 /* Print line number */
946 case '=':
Denis Vlasenko17a15262007-03-26 20:48:46 +0000947 fprintf(G.nonstdout, "%d\n", linenum);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000948 break;
949
950 /* Write the current pattern space up to the first newline */
951 case 'P':
952 {
953 char *tmp = strchr(pattern_space, '\n');
954
955 if (tmp) {
956 *tmp = '\0';
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000957 /* TODO: explain why '\n' below */
958 sed_puts(pattern_space, '\n');
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000959 *tmp = '\n';
960 break;
961 }
962 /* Fall Through */
963 }
964
965 /* Write the current pattern space to output */
966 case 'p':
Denis Vlasenko826c85f2007-01-28 23:26:15 +0000967 /* NB: we print this _before_ the last line
968 * (of current file) is printed. Even if
969 * that line is nonterminated, we print
970 * '\n' here (gnu sed does the same) */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000971 sed_puts(pattern_space, '\n');
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000972 break;
973 /* Delete up through first newline */
974 case 'D':
975 {
976 char *tmp = strchr(pattern_space, '\n');
977
978 if (tmp) {
979 tmp = xstrdup(tmp+1);
980 free(pattern_space);
981 pattern_space = tmp;
982 goto restart;
983 }
984 }
985 /* discard this line. */
986 case 'd':
987 goto discard_line;
988
989 /* Substitute with regex */
990 case 's':
991 if (!do_subst_command(sed_cmd, &pattern_space))
992 break;
993 substituted |= 1;
994
995 /* handle p option */
996 if (sed_cmd->sub_p)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000997 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +0000998 /* handle w option */
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +0000999 if (sed_cmd->sw_file)
1000 puts_maybe_newline(
1001 pattern_space, sed_cmd->sw_file,
1002 &sed_cmd->sw_last_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001003 break;
1004
1005 /* Append line to linked list to be printed later */
1006 case 'a':
1007 append(sed_cmd->string);
1008 break;
1009
1010 /* Insert text before this line */
1011 case 'i':
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001012 sed_puts(sed_cmd->string, '\n');
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001013 break;
1014
1015 /* Cut and paste text (replace) */
1016 case 'c':
1017 /* Only triggers on last line of a matching range. */
1018 if (!sed_cmd->in_match)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001019 sed_puts(sed_cmd->string, NO_EOL_CHAR);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001020 goto discard_line;
1021
1022 /* Read file, append contents to output */
1023 case 'r':
1024 {
1025 FILE *rfile;
1026
1027 rfile = fopen(sed_cmd->string, "r");
1028 if (rfile) {
1029 char *line;
1030
1031 while ((line = xmalloc_getline(rfile))
1032 != NULL)
1033 append(line);
1034 xprint_and_close_file(rfile);
1035 }
1036
1037 break;
1038 }
1039
1040 /* Write pattern space to file. */
1041 case 'w':
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001042 puts_maybe_newline(
1043 pattern_space, sed_cmd->sw_file,
1044 &sed_cmd->sw_last_char, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001045 break;
1046
1047 /* Read next line from input */
1048 case 'n':
Denis Vlasenko17a15262007-03-26 20:48:46 +00001049 if (!G.be_quiet)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001050 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001051 if (next_line) {
1052 free(pattern_space);
1053 pattern_space = next_line;
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001054 last_gets_char = next_gets_char;
1055 next_line = get_next_line(&next_gets_char);
Denis Vlasenkof39c7c02008-02-28 17:59:01 +00001056 substituted = 0;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001057 linenum++;
1058 break;
1059 }
1060 /* fall through */
1061
1062 /* Quit. End of script, end of input. */
1063 case 'q':
1064 /* Exit the outer while loop */
1065 free(next_line);
1066 next_line = NULL;
1067 goto discard_commands;
1068
1069 /* Append the next line to the current line */
1070 case 'N':
1071 {
1072 int len;
1073 /* If no next line, jump to end of script and exit. */
1074 if (next_line == NULL) {
1075 /* Jump to end of script and exit */
1076 free(next_line);
1077 next_line = NULL;
1078 goto discard_line;
1079 /* append next_line, read new next_line. */
1080 }
1081 len = strlen(pattern_space);
1082 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1083 pattern_space[len] = '\n';
1084 strcpy(pattern_space + len+1, next_line);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001085 last_gets_char = next_gets_char;
1086 next_line = get_next_line(&next_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001087 linenum++;
1088 break;
1089 }
1090
1091 /* Test/branch if substitution occurred */
1092 case 't':
1093 if (!substituted) break;
1094 substituted = 0;
1095 /* Fall through */
1096 /* Test/branch if substitution didn't occur */
1097 case 'T':
1098 if (substituted) break;
1099 /* Fall through */
1100 /* Branch to label */
1101 case 'b':
1102 if (!sed_cmd->string) goto discard_commands;
1103 else sed_cmd = branch_to(sed_cmd->string);
1104 break;
1105 /* Transliterate characters */
1106 case 'y':
1107 {
1108 int i, j;
1109
1110 for (i = 0; pattern_space[i]; i++) {
1111 for (j = 0; sed_cmd->string[j]; j += 2) {
1112 if (pattern_space[i] == sed_cmd->string[j]) {
1113 pattern_space[i] = sed_cmd->string[j + 1];
1114 break;
1115 }
1116 }
1117 }
1118
1119 break;
1120 }
1121 case 'g': /* Replace pattern space with hold space */
1122 free(pattern_space);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001123 pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001124 break;
1125 case 'G': /* Append newline and hold space to pattern space */
1126 {
1127 int pattern_space_size = 2;
1128 int hold_space_size = 0;
1129
1130 if (pattern_space)
1131 pattern_space_size += strlen(pattern_space);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001132 if (G.hold_space)
1133 hold_space_size = strlen(G.hold_space);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001134 pattern_space = xrealloc(pattern_space,
1135 pattern_space_size + hold_space_size);
1136 if (pattern_space_size == 2)
1137 pattern_space[0] = 0;
1138 strcat(pattern_space, "\n");
Denis Vlasenko17a15262007-03-26 20:48:46 +00001139 if (G.hold_space)
1140 strcat(pattern_space, G.hold_space);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001141 last_gets_char = '\n';
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001142
1143 break;
1144 }
1145 case 'h': /* Replace hold space with pattern space */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001146 free(G.hold_space);
1147 G.hold_space = xstrdup(pattern_space);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001148 break;
1149 case 'H': /* Append newline and pattern space to hold space */
1150 {
1151 int hold_space_size = 2;
1152 int pattern_space_size = 0;
1153
Denis Vlasenko17a15262007-03-26 20:48:46 +00001154 if (G.hold_space)
1155 hold_space_size += strlen(G.hold_space);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001156 if (pattern_space)
1157 pattern_space_size = strlen(pattern_space);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001158 G.hold_space = xrealloc(G.hold_space,
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001159 hold_space_size + pattern_space_size);
1160
1161 if (hold_space_size == 2)
Denis Vlasenko17a15262007-03-26 20:48:46 +00001162 *G.hold_space = 0;
1163 strcat(G.hold_space, "\n");
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001164 if (pattern_space)
Denis Vlasenko17a15262007-03-26 20:48:46 +00001165 strcat(G.hold_space, pattern_space);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001166
1167 break;
1168 }
1169 case 'x': /* Exchange hold and pattern space */
1170 {
1171 char *tmp = pattern_space;
Denis Vlasenko17a15262007-03-26 20:48:46 +00001172 pattern_space = G.hold_space ? : xzalloc(1);
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001173 last_gets_char = '\n';
Denis Vlasenko17a15262007-03-26 20:48:46 +00001174 G.hold_space = tmp;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001175 break;
1176 }
1177 }
1178 }
Glenn L McGrathaa5a6022003-10-01 03:06:16 +00001179 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001180
1181 /*
1182 * exit point from sedding...
1183 */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001184 discard_commands:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001185 /* we will print the line unless we were told to be quiet ('-n')
1186 or if the line was suppressed (ala 'd'elete) */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001187 if (!G.be_quiet)
Denis Vlasenkofe7a9f12007-01-29 14:31:47 +00001188 sed_puts(pattern_space, last_gets_char);
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001189
1190 /* Delete and such jump here. */
Denis Vlasenko826c85f2007-01-28 23:26:15 +00001191 discard_line:
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001192 flush_append();
1193 free(pattern_space);
1194
1195 goto again;
Erik Andersen1266a131999-12-29 22:19:46 +00001196}
1197
Glenn L McGrath42c25732003-10-04 05:27:56 +00001198/* It is possible to have a command line argument with embedded
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001199 * newlines. This counts as multiple command lines.
1200 * However, newline can be escaped: 's/e/z\<newline>z/'
1201 * We check for this.
1202 */
Glenn L McGrath42c25732003-10-04 05:27:56 +00001203
1204static void add_cmd_block(char *cmdstr)
1205{
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001206 char *sv, *eol;
Glenn L McGrath42c25732003-10-04 05:27:56 +00001207
Denis Vlasenko4b0bb9e2007-03-16 23:36:58 +00001208 cmdstr = sv = xstrdup(cmdstr);
1209 do {
1210 eol = strchr(cmdstr, '\n');
1211 next:
1212 if (eol) {
1213 /* Count preceding slashes */
1214 int slashes = 0;
1215 char *sl = eol;
1216
1217 while (sl != cmdstr && *--sl == '\\')
1218 slashes++;
1219 /* Odd number of preceding slashes - newline is escaped */
1220 if (slashes & 1) {
1221 strcpy(eol-1, eol);
1222 eol = strchr(eol, '\n');
1223 goto next;
1224 }
1225 *eol = '\0';
1226 }
1227 add_cmd(cmdstr);
1228 cmdstr = eol + 1;
1229 } while (eol);
1230 free(sv);
Glenn L McGrath42c25732003-10-04 05:27:56 +00001231}
1232
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +00001233int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001234int sed_main(int argc ATTRIBUTE_UNUSED, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001235{
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001236 enum {
1237 OPT_in_place = 1 << 0,
1238 };
Denis Vlasenko67b23e62006-10-03 21:00:06 +00001239 unsigned opt;
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001240 llist_t *opt_e, *opt_f;
1241 int status = EXIT_SUCCESS;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001242
Denis Vlasenko74324c82007-06-04 10:16:52 +00001243 INIT_G();
Rob Landleye3f5a3f2006-05-09 03:53:55 +00001244
Mark Whitley858c1ad2000-07-11 21:38:47 +00001245 /* destroy command strings on exit */
Rob Landley0b656282006-05-08 22:17:23 +00001246 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
Mark Whitley858c1ad2000-07-11 21:38:47 +00001247
Rob Landleyfae1dc82005-11-20 07:44:35 +00001248 /* Lie to autoconf when it starts asking stupid questions. */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001249 if (argv[1] && !strcmp(argv[1], "--version")) {
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001250 puts("This is not GNU sed version 4.0");
1251 return 0;
Eric Andersenc06f5682004-02-04 10:57:46 +00001252 }
Eric Andersenc06f5682004-02-04 10:57:46 +00001253
Mark Whitley6315ce62000-07-10 22:55:51 +00001254 /* do normal option parsing */
Denis Vlasenkob97c9842006-10-01 21:05:12 +00001255 opt_e = opt_f = NULL;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001256 opt_complementary = "e::f::" /* can occur multiple times */
1257 "nn"; /* count -n */
Denis Vlasenkofe7cd642007-08-18 15:32:12 +00001258 opt = getopt32(argv, "irne:f:", &opt_e, &opt_f,
Denis Vlasenko17a15262007-03-26 20:48:46 +00001259 &G.be_quiet); /* counter for -n */
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001260 //argc -= optind;
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001261 argv += optind;
1262 if (opt & OPT_in_place) { // -i
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001263 atexit(cleanup_outname);
1264 }
Denis Vlasenko17a15262007-03-26 20:48:46 +00001265 if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r
1266 //if (opt & 0x4) G.be_quiet++; // -n
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001267 while (opt_e) { // -e
1268 add_cmd_block(opt_e->data);
1269 opt_e = opt_e->link;
1270 /* we leak opt_e here... */
Denis Vlasenko750fc6d2006-09-22 08:56:03 +00001271 }
Denis Vlasenko945bd3d2007-04-12 21:20:25 +00001272 while (opt_f) { // -f
1273 char *line;
1274 FILE *cmdfile;
1275 cmdfile = xfopen(opt_f->data, "r");
1276 while ((line = xmalloc_getline(cmdfile)) != NULL) {
1277 add_cmd(line);
1278 free(line);
1279 }
1280 fclose(cmdfile);
1281 opt_f = opt_f->link;
1282 /* we leak opt_f here... */
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001283 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001284 /* if we didn't get a pattern from -e or -f, use argv[0] */
Denis Vlasenko80778502006-10-25 12:46:46 +00001285 if (!(opt & 0x18)) {
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001286 if (!*argv)
Manuel Novoa III cad53642003-03-19 09:13:01 +00001287 bb_show_usage();
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001288 add_cmd_block(*argv++);
Mark Whitley6315ce62000-07-10 22:55:51 +00001289 }
Glenn L McGrath42c25732003-10-04 05:27:56 +00001290 /* Flush any unfinished commands. */
1291 add_cmd("");
Mark Whitley6315ce62000-07-10 22:55:51 +00001292
Rob Landley53302f82004-02-18 09:54:15 +00001293 /* By default, we write to stdout */
Denis Vlasenko17a15262007-03-26 20:48:46 +00001294 G.nonstdout = stdout;
Rob Landley53302f82004-02-18 09:54:15 +00001295
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001296 /* argv[0..(argc-1)] should be names of file to process. If no
Mark Whitley6315ce62000-07-10 22:55:51 +00001297 * files were specified or '-' was specified, take input from stdin.
1298 * Otherwise, we process all the files specified. */
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001299 if (argv[0] == NULL) {
1300 if (opt & OPT_in_place)
Denis Vlasenko80778502006-10-25 12:46:46 +00001301 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
Rob Landleydcc28662004-11-25 07:21:47 +00001302 add_input_file(stdin);
1303 process_files();
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001304 } else {
Mark Whitley6315ce62000-07-10 22:55:51 +00001305 int i;
1306 FILE *file;
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001307
Denis Vlasenko62a90cd2008-03-17 09:07:36 +00001308 for (i = 0; argv[i]; i++) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001309 struct stat statbuf;
1310 int nonstdoutfd;
1311
Denis Vlasenko9f739442006-12-16 23:49:13 +00001312 if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
Rob Landleydcc28662004-11-25 07:21:47 +00001313 add_input_file(stdin);
1314 process_files();
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001315 continue;
Glenn L McGrath8aac05b2003-09-14 04:06:12 +00001316 }
Denis Vlasenkoddec5af2006-10-26 23:25:17 +00001317 file = fopen_or_warn(argv[i], "r");
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001318 if (!file) {
1319 status = EXIT_FAILURE;
1320 continue;
1321 }
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001322 if (!(opt & OPT_in_place)) {
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001323 add_input_file(file);
1324 continue;
1325 }
1326
Denis Vlasenko17a15262007-03-26 20:48:46 +00001327 G.outname = xasprintf("%sXXXXXX", argv[i]);
1328 nonstdoutfd = mkstemp(G.outname);
Denis Vlasenko80778502006-10-25 12:46:46 +00001329 if (-1 == nonstdoutfd)
Denis Vlasenko17a15262007-03-26 20:48:46 +00001330 bb_perror_msg_and_die("cannot create temp file %s", G.outname);
1331 G.nonstdout = fdopen(nonstdoutfd, "w");
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001332
1333 /* Set permissions of output file */
1334
Denis Vlasenko2f8f71b2006-12-10 02:09:12 +00001335 fstat(fileno(file), &statbuf);
1336 fchmod(nonstdoutfd, statbuf.st_mode);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001337 add_input_file(file);
1338 process_files();
Denis Vlasenko17a15262007-03-26 20:48:46 +00001339 fclose(G.nonstdout);
Denis Vlasenkod18a3a22006-10-25 12:46:03 +00001340
Denis Vlasenko17a15262007-03-26 20:48:46 +00001341 G.nonstdout = stdout;
Denis Vlasenko80778502006-10-25 12:46:46 +00001342 /* unlink(argv[i]); */
Denis Vlasenkocb448fe2008-02-17 14:28:53 +00001343 xrename(G.outname, argv[i]);
Denis Vlasenko17a15262007-03-26 20:48:46 +00001344 free(G.outname);
Denis Vlasenko40276642007-11-13 16:48:10 +00001345 G.outname = NULL;
Mark Whitley6315ce62000-07-10 22:55:51 +00001346 }
Denis Vlasenko17a15262007-03-26 20:48:46 +00001347 if (G.input_file_count > G.current_input_file)
Denis Vlasenko80778502006-10-25 12:46:46 +00001348 process_files();
Mark Whitley6315ce62000-07-10 22:55:51 +00001349 }
Glenn L McGrath8d6395d2003-04-08 11:56:11 +00001350
Matt Kraaia5f09c62001-11-12 16:44:55 +00001351 return status;
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001352}