blob: 3901813754bd3d832f14ce6a5bf1d2be7e469286 [file] [log] [blame]
Eric Andersen6b6b3f61999-10-28 16:06:25 +00001/*
Mark Whitley6315ce62000-07-10 22:55:51 +00002 * sed.c - very minimalist version of sed
Eric Andersen6b6b3f61999-10-28 16:06:25 +00003 *
Eric Andersen8ec10a92001-01-27 09:33:39 +00004 * Copyright (C) 1999,2000,2001 by Lineo, inc.
Mark Whitley6c6ea6c2001-01-04 22:21:13 +00005 * Written by Mark Whitley <markw@lineo.com>, <markw@codepoet.org>
Erik Andersen1266a131999-12-29 22:19:46 +00006 *
Eric Andersen6b6b3f61999-10-28 16:06:25 +00007 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 *
21 */
22
Mark Whitley6315ce62000-07-10 22:55:51 +000023/*
24 Supported features and commands in this version of sed:
25
26 - comments ('#')
Mark Whitley94074a92000-07-14 00:00:15 +000027 - address matching: num|/matchstr/[,num|/matchstr/|$]command
28 - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
29 - edit commands: (a)ppend, (i)nsert, (c)hange
Mark Whitley97562bd2000-07-17 20:06:42 +000030 - backreferences in substitution expressions (\1, \2...\9)
Mark Whitley6315ce62000-07-10 22:55:51 +000031
32 (Note: Specifying an address (range) to match is *optional*; commands
33 default to the whole pattern space if no specific address match was
34 requested.)
35
36 Unsupported features:
37
38 - transliteration (y/source-chars/dest-chars/) (use 'tr')
Mark Whitley6315ce62000-07-10 22:55:51 +000039 - no pattern space hold space storing / swapping (x, etc.)
40 - no labels / branching (: label, b, t, and friends)
41 - and lots, lots more.
Mark Whitley6315ce62000-07-10 22:55:51 +000042*/
43
Eric Andersen6b6b3f61999-10-28 16:06:25 +000044#include <stdio.h>
Mark Whitley6315ce62000-07-10 22:55:51 +000045#include <unistd.h> /* for getopt() */
46#include <regex.h>
47#include <string.h> /* for strdup() */
Eric Andersen6b6b3f61999-10-28 16:06:25 +000048#include <errno.h>
Mark Whitley6315ce62000-07-10 22:55:51 +000049#include <ctype.h> /* for isspace() */
Eric Andersened3ef502001-01-27 08:24:39 +000050#include <stdlib.h>
Eric Andersen3570a342000-09-25 21:45:58 +000051#include "busybox.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000052
Mark Whitley6315ce62000-07-10 22:55:51 +000053/* externs */
Eric Andersened3ef502001-01-27 08:24:39 +000054extern void xregcomp(regex_t *preg, const char *regex, int cflags);
Mark Whitley6315ce62000-07-10 22:55:51 +000055extern int optind; /* in unistd.h */
56extern char *optarg; /* ditto */
57
58/* options */
59static int be_quiet = 0;
60
Mark Whitley0e4cec02000-08-21 21:29:20 +000061
Mark Whitley6315ce62000-07-10 22:55:51 +000062struct sed_cmd {
63
Mark Whitley2dc192f2000-11-03 19:47:00 +000064
65 /* GENERAL FIELDS */
Eric Andersen28b3c532001-01-02 11:01:31 +000066 char delimiter; /* The delimiter used to separate regexps */
Mark Whitley2dc192f2000-11-03 19:47:00 +000067
Mark Whitley6315ce62000-07-10 22:55:51 +000068 /* address storage */
69 int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
70 int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
71 regex_t *beg_match; /* sed -e '/match/cmd' */
72 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
73
74 /* the command */
75 char cmd; /* p,d,s (add more at your leisure :-) */
76
Mark Whitley2dc192f2000-11-03 19:47:00 +000077
78 /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
79
80 /* sed -e 's/sub_match/replace/' */
81 regex_t *sub_match;
82 char *replace;
Mark Whitley97562bd2000-07-17 20:06:42 +000083 unsigned int num_backrefs:4; /* how many back references (\1..\9) */
84 /* Note: GNU/POSIX sed does not save more than nine backrefs, so
85 * we only use 4 bits to hold the number */
Mark Whitley2dc192f2000-11-03 19:47:00 +000086 unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
87 unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
Mark Whitley94074a92000-07-14 00:00:15 +000088
Mark Whitley2dc192f2000-11-03 19:47:00 +000089
90 /* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
91
Mark Whitley94074a92000-07-14 00:00:15 +000092 char *editline;
Mark Whitley6315ce62000-07-10 22:55:51 +000093};
94
95/* globals */
96static struct sed_cmd *sed_cmds = NULL; /* growable arrary holding a sequence of sed cmds */
97static int ncmds = 0; /* number of sed commands */
98
99/*static char *cur_file = NULL;*/ /* file currently being processed XXX: do I need this? */
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000100
Eric Andersenb040d4f2000-07-25 18:01:20 +0000101#ifdef BB_FEATURE_CLEAN_UP
Mark Whitley6315ce62000-07-10 22:55:51 +0000102static void destroy_cmd_strs()
103{
104 if (sed_cmds == NULL)
105 return;
106
107 /* destroy all the elements in the array */
108 while (--ncmds >= 0) {
109
110 if (sed_cmds[ncmds].beg_match) {
111 regfree(sed_cmds[ncmds].beg_match);
112 free(sed_cmds[ncmds].beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000113 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000114 if (sed_cmds[ncmds].end_match) {
115 regfree(sed_cmds[ncmds].end_match);
116 free(sed_cmds[ncmds].end_match);
117 }
118 if (sed_cmds[ncmds].sub_match) {
119 regfree(sed_cmds[ncmds].sub_match);
120 free(sed_cmds[ncmds].sub_match);
121 }
122 if (sed_cmds[ncmds].replace)
123 free(sed_cmds[ncmds].replace);
124 }
Erik Andersene49d5ec2000-02-08 19:58:47 +0000125
Mark Whitley6315ce62000-07-10 22:55:51 +0000126 /* destroy the array */
127 free(sed_cmds);
128 sed_cmds = NULL;
129}
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000130#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000131
Mark Whitley6315ce62000-07-10 22:55:51 +0000132
133/*
Eric Andersen28b3c532001-01-02 11:01:31 +0000134 * index_of_next_unescaped_regexp_delim - walks left to right through a string
135 * beginning at a specified index and returns the index of the next regular
136 * expression delimiter (typically a forward * slash ('/')) not preceeded by
137 * a backslash ('\').
Mark Whitley6315ce62000-07-10 22:55:51 +0000138 */
Eric Andersen28b3c532001-01-02 11:01:31 +0000139static int index_of_next_unescaped_regexp_delim(struct sed_cmd *sed_cmd, const char *str, int idx)
Mark Whitley6315ce62000-07-10 22:55:51 +0000140{
Mark Whitley97562bd2000-07-17 20:06:42 +0000141 for ( ; str[idx]; idx++) {
Eric Andersen28b3c532001-01-02 11:01:31 +0000142 if (str[idx] == sed_cmd->delimiter && str[idx-1] != '\\')
Mark Whitley97562bd2000-07-17 20:06:42 +0000143 return idx;
144 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000145
Mark Whitley97562bd2000-07-17 20:06:42 +0000146 /* if we make it to here, we've hit the end of the string */
147 return -1;
Mark Whitley6315ce62000-07-10 22:55:51 +0000148}
149
150/*
151 * returns the index in the string just past where the address ends.
152 */
Eric Andersen28b3c532001-01-02 11:01:31 +0000153static int get_address(struct sed_cmd *sed_cmd, const char *str, int *line, regex_t **regex)
Mark Whitley6315ce62000-07-10 22:55:51 +0000154{
155 char *my_str = strdup(str);
156 int idx = 0;
157
158 if (isdigit(my_str[idx])) {
159 do {
160 idx++;
161 } while (isdigit(my_str[idx]));
162 my_str[idx] = 0;
163 *line = atoi(my_str);
Mark Whitley6315ce62000-07-10 22:55:51 +0000164 }
165 else if (my_str[idx] == '$') {
166 *line = -1;
Mark Whitley6315ce62000-07-10 22:55:51 +0000167 idx++;
168 }
169 else if (my_str[idx] == '/') {
Eric Andersen28b3c532001-01-02 11:01:31 +0000170 idx = index_of_next_unescaped_regexp_delim(sed_cmd, my_str, ++idx);
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000171 if (idx == -1)
Matt Kraaidd19c692001-01-31 19:00:21 +0000172 error_msg_and_die("unterminated match expression");
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000173 my_str[idx] = '\0';
174 *regex = (regex_t *)xmalloc(sizeof(regex_t));
Mark Whitleyeb69ead2000-11-03 20:23:49 +0000175 xregcomp(*regex, my_str+1, 0);
Mark Whitley496e33f2000-07-13 22:52:02 +0000176 idx++; /* so it points to the next character after the last '/' */
Mark Whitley6315ce62000-07-10 22:55:51 +0000177 }
178 else {
Mark Whitleyf57c9442000-12-07 19:56:48 +0000179 error_msg("get_address: no address found in string\n"
Matt Kraaidd19c692001-01-31 19:00:21 +0000180 "\t(you probably didn't check the string you passed me)");
Mark Whitley6315ce62000-07-10 22:55:51 +0000181 idx = -1;
182 }
Erik Andersene49d5ec2000-02-08 19:58:47 +0000183
Mark Whitley6315ce62000-07-10 22:55:51 +0000184 free(my_str);
185 return idx;
186}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000187
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000188static char *strdup_substr(const char *str, int start, int end)
189{
190 int size = end - start + 1;
191 char *newstr = xmalloc(size);
192 memcpy(newstr, str+start, size-1);
193 newstr[size-1] = '\0';
194 return newstr;
195}
196
Mark Whitley70705d72000-07-14 19:06:30 +0000197static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000198{
199 int oldidx, cflags = REG_NEWLINE;
200 char *match;
201 int idx = 0;
Mark Whitley97562bd2000-07-17 20:06:42 +0000202 int j;
Mark Whitley06f35292000-07-13 19:58:04 +0000203
204 /*
205 * the string that gets passed to this function should look like this:
Mark Whitley0e4cec02000-08-21 21:29:20 +0000206 * s/match/replace/gIp
207 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000208 * mandatory optional
209 *
210 * (all three of the '/' slashes are mandatory)
211 */
212
Eric Andersen28b3c532001-01-02 11:01:31 +0000213 /* verify that the 's' is followed by something. That something
214 * (typically a 'slash') is now our regexp delimiter... */
215 if (!substr[++idx])
Matt Kraaidd19c692001-01-31 19:00:21 +0000216 error_msg_and_die("bad format in substitution expression");
Eric Andersen28b3c532001-01-02 11:01:31 +0000217 else
218 sed_cmd->delimiter=substr[idx];
Mark Whitley06f35292000-07-13 19:58:04 +0000219
220 /* save the match string */
221 oldidx = idx+1;
Eric Andersen28b3c532001-01-02 11:01:31 +0000222 idx = index_of_next_unescaped_regexp_delim(sed_cmd, substr, ++idx);
Mark Whitley06f35292000-07-13 19:58:04 +0000223 if (idx == -1)
Matt Kraaidd19c692001-01-31 19:00:21 +0000224 error_msg_and_die("bad format in substitution expression");
Mark Whitley06f35292000-07-13 19:58:04 +0000225 match = strdup_substr(substr, oldidx, idx);
226
Mark Whitley97562bd2000-07-17 20:06:42 +0000227 /* determine the number of back references in the match string */
228 /* Note: we compute this here rather than in the do_subst_command()
229 * function to save processor time, at the expense of a little more memory
230 * (4 bits) per sed_cmd */
231
232 /* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
233 for (j = 0; match[j]; j++) {
234 /* GNU/POSIX sed does not save more than nine backrefs */
Mark Whitley2dc192f2000-11-03 19:47:00 +0000235 if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
Mark Whitley97562bd2000-07-17 20:06:42 +0000236 sed_cmd->num_backrefs++;
237 }
238
Mark Whitley06f35292000-07-13 19:58:04 +0000239 /* save the replacement string */
240 oldidx = idx+1;
Eric Andersen28b3c532001-01-02 11:01:31 +0000241 idx = index_of_next_unescaped_regexp_delim(sed_cmd, substr, ++idx);
Mark Whitley06f35292000-07-13 19:58:04 +0000242 if (idx == -1)
Matt Kraaidd19c692001-01-31 19:00:21 +0000243 error_msg_and_die("bad format in substitution expression");
Mark Whitley06f35292000-07-13 19:58:04 +0000244 sed_cmd->replace = strdup_substr(substr, oldidx, idx);
245
246 /* process the flags */
247 while (substr[++idx]) {
248 switch (substr[idx]) {
Mark Whitley70705d72000-07-14 19:06:30 +0000249 case 'g':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000250 sed_cmd->sub_g = 1;
Mark Whitley70705d72000-07-14 19:06:30 +0000251 break;
252 case 'I':
253 cflags |= REG_ICASE;
254 break;
Mark Whitley0e4cec02000-08-21 21:29:20 +0000255 case 'p':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000256 sed_cmd->sub_p = 1;
Mark Whitley0e4cec02000-08-21 21:29:20 +0000257 break;
Mark Whitley70705d72000-07-14 19:06:30 +0000258 default:
259 /* any whitespace or semicolon trailing after a s/// is ok */
260 if (strchr("; \t\v\n\r", substr[idx]))
261 goto out;
262 /* else */
Matt Kraaidd19c692001-01-31 19:00:21 +0000263 error_msg_and_die("bad option in substitution expression");
Mark Whitley06f35292000-07-13 19:58:04 +0000264 }
265 }
Mark Whitley70705d72000-07-14 19:06:30 +0000266
267out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000268 /* compile the match string into a regex */
Mark Whitley06f35292000-07-13 19:58:04 +0000269 sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
270 xregcomp(sed_cmd->sub_match, match, cflags);
271 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000272
273 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000274}
275
Mark Whitley70705d72000-07-14 19:06:30 +0000276static int parse_edit_cmd(struct sed_cmd *sed_cmd, const char *editstr)
Mark Whitley94074a92000-07-14 00:00:15 +0000277{
278 int idx = 0;
Mark Whitley464c5de2000-07-14 23:24:00 +0000279 int slashes_eaten = 0;
Mark Whitley94074a92000-07-14 00:00:15 +0000280 char *ptr; /* shorthand */
281
282 /*
283 * the string that gets passed to this function should look like this:
284 *
285 * need one of these
286 * |
287 * | this backslash (immediately following the edit command) is mandatory
288 * | |
289 * [aic]\
290 * TEXT1\
291 * TEXT2\
292 * TEXTN
293 *
294 * as soon as we hit a TEXT line that has no trailing '\', we're done.
295 * this means a command like:
296 *
297 * i\
298 * INSERTME
299 *
300 * is a-ok.
301 *
302 */
303
304 if (editstr[1] != '\\' && (editstr[2] != '\n' || editstr[2] != '\r'))
Matt Kraaidd19c692001-01-31 19:00:21 +0000305 error_msg_and_die("bad format in edit expression");
Mark Whitley94074a92000-07-14 00:00:15 +0000306
307 /* store the edit line text */
Mark Whitley34623db2000-07-14 00:49:59 +0000308 /* make editline big enough to accomodate the extra '\n' we will tack on
309 * to the end */
310 sed_cmd->editline = xmalloc(strlen(&editstr[3]) + 2);
311 strcpy(sed_cmd->editline, &editstr[3]);
Mark Whitley94074a92000-07-14 00:00:15 +0000312 ptr = sed_cmd->editline;
313
314 /* now we need to go through * and: s/\\[\r\n]$/\n/g on the edit line */
315 while (ptr[idx]) {
316 while (ptr[idx] != '\\' && (ptr[idx+1] != '\n' || ptr[idx+1] != '\r')) {
317 idx++;
318 if (!ptr[idx]) {
Mark Whitley464c5de2000-07-14 23:24:00 +0000319 goto out;
Mark Whitley94074a92000-07-14 00:00:15 +0000320 }
321 }
322 /* move the newline over the '\' before it (effectively eats the '\') */
323 memmove(&ptr[idx], &ptr[idx+1], strlen(&ptr[idx+1]));
324 ptr[strlen(ptr)-1] = 0;
Mark Whitley464c5de2000-07-14 23:24:00 +0000325 slashes_eaten++;
Mark Whitley94074a92000-07-14 00:00:15 +0000326 /* substitue \r for \n if needed */
327 if (ptr[idx] == '\r')
328 ptr[idx] = '\n';
329 }
Mark Whitley70705d72000-07-14 19:06:30 +0000330
Mark Whitley464c5de2000-07-14 23:24:00 +0000331out:
332 ptr[idx] = '\n';
333 ptr[idx+1] = 0;
334
335 /* this accounts for discrepancies between the modified string and the
336 * original string passed in to this function */
337 idx += slashes_eaten;
338
339 /* this accounts for the fact that A) we started at index 3, not at index
340 * 0 and B) that we added an extra '\n' at the end (if you think the next
341 * line should read 'idx += 4' remember, arrays are zero-based) */
342
343 idx += 3;
344
Mark Whitley70705d72000-07-14 19:06:30 +0000345 return idx;
Mark Whitley94074a92000-07-14 00:00:15 +0000346}
347
Mark Whitley70705d72000-07-14 19:06:30 +0000348static char *parse_cmd_str(struct sed_cmd *sed_cmd, const char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000349{
350 int idx = 0;
351
352 /* parse the command
353 * format is: [addr][,addr]cmd
354 * |----||-----||-|
355 * part1 part2 part3
356 */
357
Mark Whitley70705d72000-07-14 19:06:30 +0000358
Mark Whitley6315ce62000-07-10 22:55:51 +0000359 /* first part (if present) is an address: either a number or a /regex/ */
360 if (isdigit(cmdstr[idx]) || cmdstr[idx] == '/')
Eric Andersen28b3c532001-01-02 11:01:31 +0000361 idx = get_address(sed_cmd, cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000362
363 /* second part (if present) will begin with a comma */
364 if (cmdstr[idx] == ',')
Eric Andersen28b3c532001-01-02 11:01:31 +0000365 idx += get_address(sed_cmd, &cmdstr[++idx], &sed_cmd->end_line, &sed_cmd->end_match);
Mark Whitley6315ce62000-07-10 22:55:51 +0000366
367 /* last part (mandatory) will be a command */
368 if (cmdstr[idx] == '\0')
Matt Kraaidd19c692001-01-31 19:00:21 +0000369 error_msg_and_die("missing command");
Mark Whitley94074a92000-07-14 00:00:15 +0000370 if (!strchr("pdsaic", cmdstr[idx])) /* <-- XXX add new commands here */
Matt Kraaidd19c692001-01-31 19:00:21 +0000371 error_msg_and_die("invalid command");
Mark Whitley6315ce62000-07-10 22:55:51 +0000372 sed_cmd->cmd = cmdstr[idx];
Mark Whitley6315ce62000-07-10 22:55:51 +0000373
Mark Whitley06f35292000-07-13 19:58:04 +0000374 /* special-case handling for (s)ubstitution */
Mark Whitley70705d72000-07-14 19:06:30 +0000375 if (sed_cmd->cmd == 's') {
376 idx += parse_subst_cmd(sed_cmd, &cmdstr[idx]);
377 }
Mark Whitley94074a92000-07-14 00:00:15 +0000378 /* special-case handling for (a)ppend, (i)nsert, and (c)hange */
Mark Whitley70705d72000-07-14 19:06:30 +0000379 else if (strchr("aic", cmdstr[idx])) {
Mark Whitley02008342000-07-14 00:13:52 +0000380 if (sed_cmd->end_line || sed_cmd->end_match)
Matt Kraaidd19c692001-01-31 19:00:21 +0000381 error_msg_and_die("only a beginning address can be specified for edit commands");
Mark Whitley70705d72000-07-14 19:06:30 +0000382 idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
Mark Whitley02008342000-07-14 00:13:52 +0000383 }
Mark Whitley464c5de2000-07-14 23:24:00 +0000384 /* if it was a single-letter command (such as 'p' or 'd') we need to
385 * increment the index past that command */
386 else
387 idx++;
Mark Whitley70705d72000-07-14 19:06:30 +0000388
389 /* give back whatever's left over */
Mark Whitley464c5de2000-07-14 23:24:00 +0000390 return (char *)&cmdstr[idx];
Eric Andersen50d63601999-11-09 01:47:36 +0000391}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000392
Mark Whitley6315ce62000-07-10 22:55:51 +0000393static void add_cmd_str(const char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000394{
Mark Whitley70705d72000-07-14 19:06:30 +0000395 char *mystr = (char *)cmdstr;
Erik Andersen1266a131999-12-29 22:19:46 +0000396
Mark Whitley70705d72000-07-14 19:06:30 +0000397 do {
Mark Whitley6315ce62000-07-10 22:55:51 +0000398
Mark Whitley70705d72000-07-14 19:06:30 +0000399 /* trim leading whitespace and semicolons */
400 memmove(mystr, &mystr[strspn(mystr, "; \n\r\t\v")], strlen(mystr));
401 /* if we ate the whole thing, that means there was just trailing
Mark Whitley464c5de2000-07-14 23:24:00 +0000402 * whitespace or a final / no-op semicolon. either way, get out */
Mark Whitley70705d72000-07-14 19:06:30 +0000403 if (strlen(mystr) == 0)
404 return;
405 /* if this is a comment, jump past it and keep going */
406 if (mystr[0] == '#') {
407 mystr = strpbrk(mystr, ";\n\r");
408 continue;
409 }
410 /* grow the array */
Matt Kraai322ae932000-09-13 02:46:14 +0000411 sed_cmds = xrealloc(sed_cmds, sizeof(struct sed_cmd) * (++ncmds));
Mark Whitley70705d72000-07-14 19:06:30 +0000412 /* zero new element */
413 memset(&sed_cmds[ncmds-1], 0, sizeof(struct sed_cmd));
414 /* load command string into new array element, get remainder */
415 mystr = parse_cmd_str(&sed_cmds[ncmds-1], mystr);
416
Mark Whitley464c5de2000-07-14 23:24:00 +0000417 } while (mystr && strlen(mystr));
Mark Whitley6315ce62000-07-10 22:55:51 +0000418}
419
420
421static void load_cmd_file(char *filename)
422{
423 FILE *cmdfile;
424 char *line;
Mark Whitley94074a92000-07-14 00:00:15 +0000425 char *nextline;
Mark Whitley6315ce62000-07-10 22:55:51 +0000426
Matt Kraaibbaef662000-09-27 02:43:35 +0000427 cmdfile = xfopen(filename, "r");
Mark Whitley6315ce62000-07-10 22:55:51 +0000428
429 while ((line = get_line_from_file(cmdfile)) != NULL) {
Mark Whitley94074a92000-07-14 00:00:15 +0000430 /* if a line ends with '\' it needs the next line appended to it */
431 while (line[strlen(line)-2] == '\\' &&
432 (nextline = get_line_from_file(cmdfile)) != NULL) {
Matt Kraai322ae932000-09-13 02:46:14 +0000433 line = xrealloc(line, strlen(line) + strlen(nextline) + 1);
Mark Whitley94074a92000-07-14 00:00:15 +0000434 strcat(line, nextline);
Mark Whitley464c5de2000-07-14 23:24:00 +0000435 free(nextline);
Mark Whitley94074a92000-07-14 00:00:15 +0000436 }
Mark Whitley464c5de2000-07-14 23:24:00 +0000437 /* eat trailing newline (if any) --if I don't do this, edit commands
438 * (aic) will print an extra newline */
Matt Kraai05e782d2001-02-01 16:49:30 +0000439 chomp(line);
Mark Whitley6315ce62000-07-10 22:55:51 +0000440 add_cmd_str(line);
441 free(line);
442 }
443}
444
Mark Whitley97562bd2000-07-17 20:06:42 +0000445static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch)
446{
447 int i;
448
449 /* go through the replacement string */
450 for (i = 0; replace[i]; i++) {
451 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
452 if (replace[i] == '\\' && isdigit(replace[i+1])) {
453 int j;
454 char tmpstr[2];
455 int backref;
456 ++i; /* i now indexes the backref number, instead of the leading slash */
457 tmpstr[0] = replace[i];
458 tmpstr[1] = 0;
459 backref = atoi(tmpstr);
460 /* print out the text held in regmatch[backref] */
461 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
462 fputc(line[j], stdout);
463 }
464
Mark Whitley83e85f62000-07-25 20:48:44 +0000465 /* if we find a backslash escaped character, print the character */
466 else if (replace[i] == '\\') {
467 ++i;
468 fputc(replace[i], stdout);
469 }
470
Mark Whitley97562bd2000-07-17 20:06:42 +0000471 /* if we find an unescaped '&' print out the whole matched text.
472 * fortunately, regmatch[0] contains the indicies to the whole matched
473 * expression (kinda seems like it was designed for just such a
474 * purpose...) */
475 else if (replace[i] == '&' && replace[i-1] != '\\') {
476 int j;
477 for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
478 fputc(line[j], stdout);
479 }
480 /* nothing special, just print this char of the replacement string to stdout */
481 else
482 fputc(replace[i], stdout);
483 }
484}
485
Mark Whitley4f7fe772000-07-13 20:01:58 +0000486static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
487{
Mark Whitley2dc192f2000-11-03 19:47:00 +0000488 char *hackline = (char *)line;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000489 int altered = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000490 regmatch_t *regmatch = NULL;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000491
Mark Whitley2dc192f2000-11-03 19:47:00 +0000492 /* we only proceed if the substitution 'search' expression matches */
493 if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
494 return 0;
495
496 /* whaddaya know, it matched. get the number of back references */
497 regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
498
499 /* and now, as long as we've got a line to try matching and if we can match
500 * the search string, we make substitutions */
501 while (*hackline && (regexec(sed_cmd->sub_match, hackline,
502 sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
Mark Whitley4f7fe772000-07-13 20:01:58 +0000503 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000504
Mark Whitley2dc192f2000-11-03 19:47:00 +0000505 /* print everything before the match */
506 for (i = 0; i < regmatch[0].rm_so; i++)
507 fputc(hackline[i], stdout);
Mark Whitley97562bd2000-07-17 20:06:42 +0000508
Mark Whitley2dc192f2000-11-03 19:47:00 +0000509 /* then print the substitution string */
510 print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
Mark Whitley97562bd2000-07-17 20:06:42 +0000511
Mark Whitley2dc192f2000-11-03 19:47:00 +0000512 /* advance past the match */
513 hackline += regmatch[0].rm_eo;
514 /* flag that something has changed */
515 altered++;
Mark Whitley97562bd2000-07-17 20:06:42 +0000516
Mark Whitley2dc192f2000-11-03 19:47:00 +0000517 /* if we're not doing this globally, get out now */
518 if (!sed_cmd->sub_g)
519 break;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000520 }
521
Mark Whitley2dc192f2000-11-03 19:47:00 +0000522 /* if there's anything left of the line, print it */
523 if (*hackline)
524 fputs(hackline, stdout);
525
526 /* cleanup */
527 free(regmatch);
528
Mark Whitley4f7fe772000-07-13 20:01:58 +0000529 return altered;
530}
Mark Whitley6315ce62000-07-10 22:55:51 +0000531
532static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
533{
534 int altered = 0;
535
536 switch (sed_cmd->cmd) {
537
538 case 'p':
539 fputs(line, stdout);
540 break;
541
542 case 'd':
543 altered++;
544 break;
545
Mark Whitley4f7fe772000-07-13 20:01:58 +0000546 case 's':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000547
548 /*
549 * Some special cases for 's' printing to make it compliant with
550 * GNU sed printing behavior (aka "The -n | s///p Matrix"):
551 *
552 * -n ONLY = never print anything regardless of any successful
553 * substitution
554 *
555 * s///p ONLY = always print successful substitutions, even if
556 * the line is going to be printed anyway (line will be printed
557 * twice).
558 *
559 * -n AND s///p = print ONLY a successful substitution ONE TIME;
560 * no other lines are printed - this is the reason why the 'p'
561 * flag exists in the first place.
562 */
563
Mark Whitleye7ff2842000-11-03 20:02:35 +0000564 /* if the user specified that they didn't want anything printed (i.e. a -n
565 * flag and no 'p' flag after the s///), then there's really no point doing
566 * anything here. */
567 if (be_quiet && !sed_cmd->sub_p)
568 break;
569
Mark Whitley2dc192f2000-11-03 19:47:00 +0000570 /* we print the line once, unless we were told to be quiet */
571 if (!be_quiet)
572 altered = do_subst_command(sed_cmd, line);
573
574 /* we also print the line if we were given the 'p' flag
575 * (this is quite possibly the second printing) */
576 if (sed_cmd->sub_p)
577 altered = do_subst_command(sed_cmd, line);
578
Mark Whitley6315ce62000-07-10 22:55:51 +0000579 break;
Mark Whitley94074a92000-07-14 00:00:15 +0000580
581 case 'a':
582 fputs(line, stdout);
583 fputs(sed_cmd->editline, stdout);
584 altered++;
585 break;
586
587 case 'i':
588 fputs(sed_cmd->editline, stdout);
589 break;
590
591 case 'c':
592 fputs(sed_cmd->editline, stdout);
593 altered++;
594 break;
Mark Whitley6315ce62000-07-10 22:55:51 +0000595 }
596
597 return altered;
598}
599
600static void process_file(FILE *file)
601{
602 char *line = NULL;
603 static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
604 unsigned int still_in_range = 0;
605 int line_altered;
606 int i;
607
608 /* go through every line in the file */
609 while ((line = get_line_from_file(file)) != NULL) {
610
611 linenum++;
612 line_altered = 0;
613
614 /* for every line, go through all the commands */
615 for (i = 0; i < ncmds; i++) {
616
617 /* are we acting on a range of matched lines? */
618 if (sed_cmds[i].beg_match && sed_cmds[i].end_match) {
619 if (still_in_range || regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0) {
620 line_altered += do_sed_command(&sed_cmds[i], line);
621 still_in_range = 1;
622 if (regexec(sed_cmds[i].end_match, line, 0, NULL, 0) == 0)
623 still_in_range = 0;
624 }
625 }
626
627 /* are we trying to match a single line? */
628 else if (sed_cmds[i].beg_match) {
629 if (regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0)
630 line_altered += do_sed_command(&sed_cmds[i], line);
631 }
632
633 /* are we acting on a range of line numbers? */
Mark Whitley40406e62000-08-10 00:09:47 +0000634 else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
Mark Whitley2dc192f2000-11-03 19:47:00 +0000635 if (linenum >= sed_cmds[i].beg_line &&
636 (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
Mark Whitley6315ce62000-07-10 22:55:51 +0000637 line_altered += do_sed_command(&sed_cmds[i], line);
638 }
639
640 /* are we acting on a specified line number */
641 else if (sed_cmds[i].beg_line > 0) {
642 if (linenum == sed_cmds[i].beg_line)
643 line_altered += do_sed_command(&sed_cmds[i], line);
644 }
645
646 /* not acting on matches or line numbers. act on every line */
647 else
648 line_altered += do_sed_command(&sed_cmds[i], line);
649
Erik Andersene49d5ec2000-02-08 19:58:47 +0000650 }
Erik Andersen1266a131999-12-29 22:19:46 +0000651
Mark Whitley2dc192f2000-11-03 19:47:00 +0000652 /* we will print the line unless we were told to be quiet or if the
653 * line was altered (via a 'd'elete or 's'ubstitution), in which case
654 * the altered line was already printed */
655 if (!be_quiet && !line_altered)
Mark Whitley6315ce62000-07-10 22:55:51 +0000656 fputs(line, stdout);
657
658 free(line);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000659 }
Erik Andersen1266a131999-12-29 22:19:46 +0000660}
661
662extern int sed_main(int argc, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000663{
Mark Whitley6315ce62000-07-10 22:55:51 +0000664 int opt;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000665
Eric Andersenb040d4f2000-07-25 18:01:20 +0000666#ifdef BB_FEATURE_CLEAN_UP
Mark Whitley858c1ad2000-07-11 21:38:47 +0000667 /* destroy command strings on exit */
Matt Kraaia9819b22000-12-22 01:48:07 +0000668 if (atexit(destroy_cmd_strs) == -1)
669 perror_msg_and_die("atexit");
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000670#endif
Mark Whitley858c1ad2000-07-11 21:38:47 +0000671
Mark Whitley6315ce62000-07-10 22:55:51 +0000672 /* do normal option parsing */
Mark Whitley1f45b262000-07-20 23:08:40 +0000673 while ((opt = getopt(argc, argv, "hne:f:")) > 0) {
Mark Whitley6315ce62000-07-10 22:55:51 +0000674 switch (opt) {
Mark Whitley6315ce62000-07-10 22:55:51 +0000675 case 'h':
Eric Andersen67991cf2001-02-14 21:23:06 +0000676 show_usage();
Mark Whitley6315ce62000-07-10 22:55:51 +0000677 break;
Erik Andersene916d242000-03-06 19:20:35 +0000678 case 'n':
Mark Whitley6315ce62000-07-10 22:55:51 +0000679 be_quiet++;
Erik Andersene916d242000-03-06 19:20:35 +0000680 break;
681 case 'e':
Mark Whitley6315ce62000-07-10 22:55:51 +0000682 add_cmd_str(optarg);
Erik Andersene916d242000-03-06 19:20:35 +0000683 break;
Mark Whitley6315ce62000-07-10 22:55:51 +0000684 case 'f':
685 load_cmd_file(optarg);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000686 break;
Erik Andersene49d5ec2000-02-08 19:58:47 +0000687 }
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000688 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000689
690 /* if we didn't get a pattern from a -e and no command file was specified,
691 * argv[optind] should be the pattern. no pattern, no worky */
692 if (ncmds == 0) {
693 if (argv[optind] == NULL)
Eric Andersen67991cf2001-02-14 21:23:06 +0000694 show_usage();
Mark Whitley6315ce62000-07-10 22:55:51 +0000695 else {
696 add_cmd_str(argv[optind]);
697 optind++;
698 }
699 }
700
701
702 /* argv[(optind)..(argc-1)] should be names of file to process. If no
703 * files were specified or '-' was specified, take input from stdin.
704 * Otherwise, we process all the files specified. */
705 if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
706 process_file(stdin);
707 }
708 else {
709 int i;
710 FILE *file;
711 for (i = optind; i < argc; i++) {
712 file = fopen(argv[i], "r");
713 if (file == NULL) {
Matt Kraai1fa1ade2000-12-18 03:57:16 +0000714 perror_msg("%s", argv[i]);
Mark Whitley6315ce62000-07-10 22:55:51 +0000715 } else {
716 process_file(file);
717 fclose(file);
718 }
719 }
720 }
721
Mark Whitley6315ce62000-07-10 22:55:51 +0000722 return 0;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000723}