blob: a7152e52a038b9b7e7f0cd4d0a655fc92154ef24 [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 *
Erik Andersen61677fe2000-04-13 01:18:56 +00004 * Copyright (C) 1999,2000 by Lineo, inc.
Mark Whitley6315ce62000-07-10 22:55:51 +00005 * Written by Mark Whitley <markw@lineo.com>, <markw@enol.com>
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')
39 - no support for characters other than the '/' character for regex matches
40 - no pattern space hold space storing / swapping (x, etc.)
41 - no labels / branching (: label, b, t, and friends)
42 - and lots, lots more.
43
44*/
45
Eric Andersen6b6b3f61999-10-28 16:06:25 +000046#include <stdio.h>
Mark Whitley6315ce62000-07-10 22:55:51 +000047#include <unistd.h> /* for getopt() */
48#include <regex.h>
49#include <string.h> /* for strdup() */
Eric Andersen6b6b3f61999-10-28 16:06:25 +000050#include <errno.h>
Mark Whitley6315ce62000-07-10 22:55:51 +000051#include <ctype.h> /* for isspace() */
Eric Andersen3570a342000-09-25 21:45:58 +000052#include "busybox.h"
Mark Whitley6315ce62000-07-10 22:55:51 +000053
Mark Whitley6315ce62000-07-10 22:55:51 +000054/* externs */
55extern 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 */
66
Mark Whitley6315ce62000-07-10 22:55:51 +000067 /* address storage */
68 int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
69 int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
70 regex_t *beg_match; /* sed -e '/match/cmd' */
71 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
72
73 /* the command */
74 char cmd; /* p,d,s (add more at your leisure :-) */
75
Mark Whitley2dc192f2000-11-03 19:47:00 +000076
77 /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
78
79 /* sed -e 's/sub_match/replace/' */
80 regex_t *sub_match;
81 char *replace;
Mark Whitley97562bd2000-07-17 20:06:42 +000082 unsigned int num_backrefs:4; /* how many back references (\1..\9) */
83 /* Note: GNU/POSIX sed does not save more than nine backrefs, so
84 * we only use 4 bits to hold the number */
Mark Whitley2dc192f2000-11-03 19:47:00 +000085 unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
86 unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
Mark Whitley94074a92000-07-14 00:00:15 +000087
Mark Whitley2dc192f2000-11-03 19:47:00 +000088
89 /* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
90
Mark Whitley94074a92000-07-14 00:00:15 +000091 char *editline;
Mark Whitley6315ce62000-07-10 22:55:51 +000092};
93
94/* globals */
95static struct sed_cmd *sed_cmds = NULL; /* growable arrary holding a sequence of sed cmds */
96static int ncmds = 0; /* number of sed commands */
97
98/*static char *cur_file = NULL;*/ /* file currently being processed XXX: do I need this? */
Eric Andersen6b6b3f61999-10-28 16:06:25 +000099
Eric Andersenb040d4f2000-07-25 18:01:20 +0000100#ifdef BB_FEATURE_CLEAN_UP
Mark Whitley6315ce62000-07-10 22:55:51 +0000101static void destroy_cmd_strs()
102{
103 if (sed_cmds == NULL)
104 return;
105
106 /* destroy all the elements in the array */
107 while (--ncmds >= 0) {
108
109 if (sed_cmds[ncmds].beg_match) {
110 regfree(sed_cmds[ncmds].beg_match);
111 free(sed_cmds[ncmds].beg_match);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000112 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000113 if (sed_cmds[ncmds].end_match) {
114 regfree(sed_cmds[ncmds].end_match);
115 free(sed_cmds[ncmds].end_match);
116 }
117 if (sed_cmds[ncmds].sub_match) {
118 regfree(sed_cmds[ncmds].sub_match);
119 free(sed_cmds[ncmds].sub_match);
120 }
121 if (sed_cmds[ncmds].replace)
122 free(sed_cmds[ncmds].replace);
123 }
Erik Andersene49d5ec2000-02-08 19:58:47 +0000124
Mark Whitley6315ce62000-07-10 22:55:51 +0000125 /* destroy the array */
126 free(sed_cmds);
127 sed_cmds = NULL;
128}
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000129#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000130
Mark Whitley70705d72000-07-14 19:06:30 +0000131#if 0
Mark Whitley6315ce62000-07-10 22:55:51 +0000132/*
133 * trim_str - trims leading and trailing space from a string
134 *
135 * Note: This returns a malloc'ed string so you must store and free it
136 * XXX: This should be in the utility.c file.
Mark Whitley70705d72000-07-14 19:06:30 +0000137 * XXX: This is now obsolete. Maybe it belongs nowhere.
Mark Whitley6315ce62000-07-10 22:55:51 +0000138 */
139static char *trim_str(const char *str)
140{
141 int i;
142 char *retstr = strdup(str);
143
144 /* trim leading whitespace */
145 memmove(retstr, &retstr[strspn(retstr, " \n\t\v")], strlen(retstr));
146
147 /* trim trailing whitespace */
148 i = strlen(retstr) - 1;
149 while (isspace(retstr[i]))
150 i--;
151 retstr[++i] = 0;
152
153 /* Aside:
154 *
155 * you know, a strrspn() would really be nice cuz then we could say:
156 *
Mark Whitley70705d72000-07-14 19:06:30 +0000157 * retstr[strrspn(retstr, " \n\t\v") + 1] = 0;
Mark Whitley6315ce62000-07-10 22:55:51 +0000158 */
159
160 return retstr;
161}
Mark Whitley70705d72000-07-14 19:06:30 +0000162#endif
163
164#if 0
165/*
166 * strrspn - works just like strspn() but goes from right to left instead of
167 * left to right
168 */
169static size_t strrspn(const char *s, const char *accept)
170{
171 size_t i = strlen(s);
172
173 while (strchr(accept, s[--i]))
174 ;
175
176 return i;
177}
178#endif
Mark Whitley6315ce62000-07-10 22:55:51 +0000179
180/*
Mark Whitley97562bd2000-07-17 20:06:42 +0000181 * index_of_next_unescaped_slash - walks left to right through a string
182 * beginning at a specified index and returns the index of the next forward
183 * slash ('/') not preceeded by a backslash ('\').
Mark Whitley6315ce62000-07-10 22:55:51 +0000184 */
Mark Whitley06f35292000-07-13 19:58:04 +0000185static int index_of_next_unescaped_slash(const char *str, int idx)
Mark Whitley6315ce62000-07-10 22:55:51 +0000186{
Mark Whitley97562bd2000-07-17 20:06:42 +0000187 for ( ; str[idx]; idx++) {
188 if (str[idx] == '/' && str[idx-1] != '\\')
189 return idx;
190 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000191
Mark Whitley97562bd2000-07-17 20:06:42 +0000192 /* if we make it to here, we've hit the end of the string */
193 return -1;
Mark Whitley6315ce62000-07-10 22:55:51 +0000194}
195
196/*
197 * returns the index in the string just past where the address ends.
198 */
199static int get_address(const char *str, int *line, regex_t **regex)
200{
201 char *my_str = strdup(str);
202 int idx = 0;
203
204 if (isdigit(my_str[idx])) {
205 do {
206 idx++;
207 } while (isdigit(my_str[idx]));
208 my_str[idx] = 0;
209 *line = atoi(my_str);
Mark Whitley6315ce62000-07-10 22:55:51 +0000210 }
211 else if (my_str[idx] == '$') {
212 *line = -1;
Mark Whitley6315ce62000-07-10 22:55:51 +0000213 idx++;
214 }
215 else if (my_str[idx] == '/') {
Mark Whitley97562bd2000-07-17 20:06:42 +0000216 idx = index_of_next_unescaped_slash(my_str, ++idx);
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000217 if (idx == -1)
Mark Whitleyf57c9442000-12-07 19:56:48 +0000218 error_msg_and_die("unterminated match expression\n");
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000219 my_str[idx] = '\0';
220 *regex = (regex_t *)xmalloc(sizeof(regex_t));
Mark Whitleyeb69ead2000-11-03 20:23:49 +0000221 xregcomp(*regex, my_str+1, 0);
Mark Whitley496e33f2000-07-13 22:52:02 +0000222 idx++; /* so it points to the next character after the last '/' */
Mark Whitley6315ce62000-07-10 22:55:51 +0000223 }
224 else {
Mark Whitleyf57c9442000-12-07 19:56:48 +0000225 error_msg("get_address: no address found in string\n"
Matt Kraaid537a952000-07-14 01:51:25 +0000226 "\t(you probably didn't check the string you passed me)\n");
Mark Whitley6315ce62000-07-10 22:55:51 +0000227 idx = -1;
228 }
Erik Andersene49d5ec2000-02-08 19:58:47 +0000229
Mark Whitley6315ce62000-07-10 22:55:51 +0000230 free(my_str);
231 return idx;
232}
Erik Andersene49d5ec2000-02-08 19:58:47 +0000233
Mark Whitleydf5f6ba2000-07-11 16:53:56 +0000234static char *strdup_substr(const char *str, int start, int end)
235{
236 int size = end - start + 1;
237 char *newstr = xmalloc(size);
238 memcpy(newstr, str+start, size-1);
239 newstr[size-1] = '\0';
240 return newstr;
241}
242
Mark Whitley70705d72000-07-14 19:06:30 +0000243static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
Mark Whitley06f35292000-07-13 19:58:04 +0000244{
245 int oldidx, cflags = REG_NEWLINE;
246 char *match;
247 int idx = 0;
Mark Whitley97562bd2000-07-17 20:06:42 +0000248 int j;
Mark Whitley06f35292000-07-13 19:58:04 +0000249
250 /*
251 * the string that gets passed to this function should look like this:
Mark Whitley0e4cec02000-08-21 21:29:20 +0000252 * s/match/replace/gIp
253 * || | |||
Mark Whitley06f35292000-07-13 19:58:04 +0000254 * mandatory optional
255 *
256 * (all three of the '/' slashes are mandatory)
257 */
258
259 /* verify that the 's' is followed by a 'slash' */
260 if (substr[++idx] != '/')
Mark Whitleyf57c9442000-12-07 19:56:48 +0000261 error_msg_and_die("bad format in substitution expression\n");
Mark Whitley06f35292000-07-13 19:58:04 +0000262
263 /* save the match string */
264 oldidx = idx+1;
Mark Whitley97562bd2000-07-17 20:06:42 +0000265 idx = index_of_next_unescaped_slash(substr, ++idx);
Mark Whitley06f35292000-07-13 19:58:04 +0000266 if (idx == -1)
Mark Whitleyf57c9442000-12-07 19:56:48 +0000267 error_msg_and_die("bad format in substitution expression\n");
Mark Whitley06f35292000-07-13 19:58:04 +0000268 match = strdup_substr(substr, oldidx, idx);
269
Mark Whitley97562bd2000-07-17 20:06:42 +0000270 /* determine the number of back references in the match string */
271 /* Note: we compute this here rather than in the do_subst_command()
272 * function to save processor time, at the expense of a little more memory
273 * (4 bits) per sed_cmd */
274
275 /* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
276 for (j = 0; match[j]; j++) {
277 /* GNU/POSIX sed does not save more than nine backrefs */
Mark Whitley2dc192f2000-11-03 19:47:00 +0000278 if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
Mark Whitley97562bd2000-07-17 20:06:42 +0000279 sed_cmd->num_backrefs++;
280 }
281
Mark Whitley06f35292000-07-13 19:58:04 +0000282 /* save the replacement string */
283 oldidx = idx+1;
Mark Whitley97562bd2000-07-17 20:06:42 +0000284 idx = index_of_next_unescaped_slash(substr, ++idx);
Mark Whitley06f35292000-07-13 19:58:04 +0000285 if (idx == -1)
Mark Whitleyf57c9442000-12-07 19:56:48 +0000286 error_msg_and_die("bad format in substitution expression\n");
Mark Whitley06f35292000-07-13 19:58:04 +0000287 sed_cmd->replace = strdup_substr(substr, oldidx, idx);
288
289 /* process the flags */
290 while (substr[++idx]) {
291 switch (substr[idx]) {
Mark Whitley70705d72000-07-14 19:06:30 +0000292 case 'g':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000293 sed_cmd->sub_g = 1;
Mark Whitley70705d72000-07-14 19:06:30 +0000294 break;
295 case 'I':
296 cflags |= REG_ICASE;
297 break;
Mark Whitley0e4cec02000-08-21 21:29:20 +0000298 case 'p':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000299 sed_cmd->sub_p = 1;
Mark Whitley0e4cec02000-08-21 21:29:20 +0000300 break;
Mark Whitley70705d72000-07-14 19:06:30 +0000301 default:
302 /* any whitespace or semicolon trailing after a s/// is ok */
303 if (strchr("; \t\v\n\r", substr[idx]))
304 goto out;
305 /* else */
Mark Whitleyf57c9442000-12-07 19:56:48 +0000306 error_msg_and_die("bad option in substitution expression\n");
Mark Whitley06f35292000-07-13 19:58:04 +0000307 }
308 }
Mark Whitley70705d72000-07-14 19:06:30 +0000309
310out:
Mark Whitley97562bd2000-07-17 20:06:42 +0000311 /* compile the match string into a regex */
Mark Whitley06f35292000-07-13 19:58:04 +0000312 sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
313 xregcomp(sed_cmd->sub_match, match, cflags);
314 free(match);
Mark Whitley70705d72000-07-14 19:06:30 +0000315
316 return idx;
Mark Whitley06f35292000-07-13 19:58:04 +0000317}
318
Mark Whitley70705d72000-07-14 19:06:30 +0000319static int parse_edit_cmd(struct sed_cmd *sed_cmd, const char *editstr)
Mark Whitley94074a92000-07-14 00:00:15 +0000320{
321 int idx = 0;
Mark Whitley464c5de2000-07-14 23:24:00 +0000322 int slashes_eaten = 0;
Mark Whitley94074a92000-07-14 00:00:15 +0000323 char *ptr; /* shorthand */
324
325 /*
326 * the string that gets passed to this function should look like this:
327 *
328 * need one of these
329 * |
330 * | this backslash (immediately following the edit command) is mandatory
331 * | |
332 * [aic]\
333 * TEXT1\
334 * TEXT2\
335 * TEXTN
336 *
337 * as soon as we hit a TEXT line that has no trailing '\', we're done.
338 * this means a command like:
339 *
340 * i\
341 * INSERTME
342 *
343 * is a-ok.
344 *
345 */
346
347 if (editstr[1] != '\\' && (editstr[2] != '\n' || editstr[2] != '\r'))
Mark Whitleyf57c9442000-12-07 19:56:48 +0000348 error_msg_and_die("bad format in edit expression\n");
Mark Whitley94074a92000-07-14 00:00:15 +0000349
350 /* store the edit line text */
Mark Whitley34623db2000-07-14 00:49:59 +0000351 /* make editline big enough to accomodate the extra '\n' we will tack on
352 * to the end */
353 sed_cmd->editline = xmalloc(strlen(&editstr[3]) + 2);
354 strcpy(sed_cmd->editline, &editstr[3]);
Mark Whitley94074a92000-07-14 00:00:15 +0000355 ptr = sed_cmd->editline;
356
357 /* now we need to go through * and: s/\\[\r\n]$/\n/g on the edit line */
358 while (ptr[idx]) {
359 while (ptr[idx] != '\\' && (ptr[idx+1] != '\n' || ptr[idx+1] != '\r')) {
360 idx++;
361 if (!ptr[idx]) {
Mark Whitley464c5de2000-07-14 23:24:00 +0000362 goto out;
Mark Whitley94074a92000-07-14 00:00:15 +0000363 }
364 }
365 /* move the newline over the '\' before it (effectively eats the '\') */
366 memmove(&ptr[idx], &ptr[idx+1], strlen(&ptr[idx+1]));
367 ptr[strlen(ptr)-1] = 0;
Mark Whitley464c5de2000-07-14 23:24:00 +0000368 slashes_eaten++;
Mark Whitley94074a92000-07-14 00:00:15 +0000369 /* substitue \r for \n if needed */
370 if (ptr[idx] == '\r')
371 ptr[idx] = '\n';
372 }
Mark Whitley70705d72000-07-14 19:06:30 +0000373
Mark Whitley464c5de2000-07-14 23:24:00 +0000374out:
375 ptr[idx] = '\n';
376 ptr[idx+1] = 0;
377
378 /* this accounts for discrepancies between the modified string and the
379 * original string passed in to this function */
380 idx += slashes_eaten;
381
382 /* this accounts for the fact that A) we started at index 3, not at index
383 * 0 and B) that we added an extra '\n' at the end (if you think the next
384 * line should read 'idx += 4' remember, arrays are zero-based) */
385
386 idx += 3;
387
Mark Whitley70705d72000-07-14 19:06:30 +0000388 return idx;
Mark Whitley94074a92000-07-14 00:00:15 +0000389}
390
Mark Whitley70705d72000-07-14 19:06:30 +0000391static char *parse_cmd_str(struct sed_cmd *sed_cmd, const char *cmdstr)
Mark Whitley6315ce62000-07-10 22:55:51 +0000392{
393 int idx = 0;
394
395 /* parse the command
396 * format is: [addr][,addr]cmd
397 * |----||-----||-|
398 * part1 part2 part3
399 */
400
Mark Whitley70705d72000-07-14 19:06:30 +0000401
Mark Whitley6315ce62000-07-10 22:55:51 +0000402 /* first part (if present) is an address: either a number or a /regex/ */
403 if (isdigit(cmdstr[idx]) || cmdstr[idx] == '/')
404 idx = get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
405
406 /* second part (if present) will begin with a comma */
407 if (cmdstr[idx] == ',')
408 idx += get_address(&cmdstr[++idx], &sed_cmd->end_line, &sed_cmd->end_match);
409
410 /* last part (mandatory) will be a command */
411 if (cmdstr[idx] == '\0')
Mark Whitleyf57c9442000-12-07 19:56:48 +0000412 error_msg_and_die("missing command\n");
Mark Whitley94074a92000-07-14 00:00:15 +0000413 if (!strchr("pdsaic", cmdstr[idx])) /* <-- XXX add new commands here */
Mark Whitleyf57c9442000-12-07 19:56:48 +0000414 error_msg_and_die("invalid command\n");
Mark Whitley6315ce62000-07-10 22:55:51 +0000415 sed_cmd->cmd = cmdstr[idx];
Mark Whitley6315ce62000-07-10 22:55:51 +0000416
Mark Whitley06f35292000-07-13 19:58:04 +0000417 /* special-case handling for (s)ubstitution */
Mark Whitley70705d72000-07-14 19:06:30 +0000418 if (sed_cmd->cmd == 's') {
419 idx += parse_subst_cmd(sed_cmd, &cmdstr[idx]);
420 }
Mark Whitley94074a92000-07-14 00:00:15 +0000421 /* special-case handling for (a)ppend, (i)nsert, and (c)hange */
Mark Whitley70705d72000-07-14 19:06:30 +0000422 else if (strchr("aic", cmdstr[idx])) {
Mark Whitley02008342000-07-14 00:13:52 +0000423 if (sed_cmd->end_line || sed_cmd->end_match)
Mark Whitleyf57c9442000-12-07 19:56:48 +0000424 error_msg_and_die("only a beginning address can be specified for edit commands\n");
Mark Whitley70705d72000-07-14 19:06:30 +0000425 idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
Mark Whitley02008342000-07-14 00:13:52 +0000426 }
Mark Whitley464c5de2000-07-14 23:24:00 +0000427 /* if it was a single-letter command (such as 'p' or 'd') we need to
428 * increment the index past that command */
429 else
430 idx++;
Mark Whitley70705d72000-07-14 19:06:30 +0000431
432 /* give back whatever's left over */
Mark Whitley464c5de2000-07-14 23:24:00 +0000433 return (char *)&cmdstr[idx];
Eric Andersen50d63601999-11-09 01:47:36 +0000434}
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000435
Mark Whitley6315ce62000-07-10 22:55:51 +0000436static void add_cmd_str(const char *cmdstr)
Erik Andersen1266a131999-12-29 22:19:46 +0000437{
Mark Whitley70705d72000-07-14 19:06:30 +0000438 char *mystr = (char *)cmdstr;
Erik Andersen1266a131999-12-29 22:19:46 +0000439
Mark Whitley70705d72000-07-14 19:06:30 +0000440 do {
Mark Whitley6315ce62000-07-10 22:55:51 +0000441
Mark Whitley70705d72000-07-14 19:06:30 +0000442 /* trim leading whitespace and semicolons */
443 memmove(mystr, &mystr[strspn(mystr, "; \n\r\t\v")], strlen(mystr));
444 /* if we ate the whole thing, that means there was just trailing
Mark Whitley464c5de2000-07-14 23:24:00 +0000445 * whitespace or a final / no-op semicolon. either way, get out */
Mark Whitley70705d72000-07-14 19:06:30 +0000446 if (strlen(mystr) == 0)
447 return;
448 /* if this is a comment, jump past it and keep going */
449 if (mystr[0] == '#') {
450 mystr = strpbrk(mystr, ";\n\r");
451 continue;
452 }
453 /* grow the array */
Matt Kraai322ae932000-09-13 02:46:14 +0000454 sed_cmds = xrealloc(sed_cmds, sizeof(struct sed_cmd) * (++ncmds));
Mark Whitley70705d72000-07-14 19:06:30 +0000455 /* zero new element */
456 memset(&sed_cmds[ncmds-1], 0, sizeof(struct sed_cmd));
457 /* load command string into new array element, get remainder */
458 mystr = parse_cmd_str(&sed_cmds[ncmds-1], mystr);
459
Mark Whitley464c5de2000-07-14 23:24:00 +0000460 } while (mystr && strlen(mystr));
Mark Whitley6315ce62000-07-10 22:55:51 +0000461}
462
463
464static void load_cmd_file(char *filename)
465{
466 FILE *cmdfile;
467 char *line;
Mark Whitley94074a92000-07-14 00:00:15 +0000468 char *nextline;
Mark Whitley6315ce62000-07-10 22:55:51 +0000469
Matt Kraaibbaef662000-09-27 02:43:35 +0000470 cmdfile = xfopen(filename, "r");
Mark Whitley6315ce62000-07-10 22:55:51 +0000471
472 while ((line = get_line_from_file(cmdfile)) != NULL) {
Mark Whitley94074a92000-07-14 00:00:15 +0000473 /* if a line ends with '\' it needs the next line appended to it */
474 while (line[strlen(line)-2] == '\\' &&
475 (nextline = get_line_from_file(cmdfile)) != NULL) {
Matt Kraai322ae932000-09-13 02:46:14 +0000476 line = xrealloc(line, strlen(line) + strlen(nextline) + 1);
Mark Whitley94074a92000-07-14 00:00:15 +0000477 strcat(line, nextline);
Mark Whitley464c5de2000-07-14 23:24:00 +0000478 free(nextline);
Mark Whitley94074a92000-07-14 00:00:15 +0000479 }
Mark Whitley464c5de2000-07-14 23:24:00 +0000480 /* eat trailing newline (if any) --if I don't do this, edit commands
481 * (aic) will print an extra newline */
482 if (line[strlen(line)-1] == '\n')
483 line[strlen(line)-1] = 0;
Mark Whitley6315ce62000-07-10 22:55:51 +0000484 add_cmd_str(line);
485 free(line);
486 }
487}
488
Mark Whitley97562bd2000-07-17 20:06:42 +0000489static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch)
490{
491 int i;
492
493 /* go through the replacement string */
494 for (i = 0; replace[i]; i++) {
495 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
496 if (replace[i] == '\\' && isdigit(replace[i+1])) {
497 int j;
498 char tmpstr[2];
499 int backref;
500 ++i; /* i now indexes the backref number, instead of the leading slash */
501 tmpstr[0] = replace[i];
502 tmpstr[1] = 0;
503 backref = atoi(tmpstr);
504 /* print out the text held in regmatch[backref] */
505 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
506 fputc(line[j], stdout);
507 }
508
Mark Whitley83e85f62000-07-25 20:48:44 +0000509 /* if we find a backslash escaped character, print the character */
510 else if (replace[i] == '\\') {
511 ++i;
512 fputc(replace[i], stdout);
513 }
514
Mark Whitley97562bd2000-07-17 20:06:42 +0000515 /* if we find an unescaped '&' print out the whole matched text.
516 * fortunately, regmatch[0] contains the indicies to the whole matched
517 * expression (kinda seems like it was designed for just such a
518 * purpose...) */
519 else if (replace[i] == '&' && replace[i-1] != '\\') {
520 int j;
521 for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
522 fputc(line[j], stdout);
523 }
524 /* nothing special, just print this char of the replacement string to stdout */
525 else
526 fputc(replace[i], stdout);
527 }
528}
529
Mark Whitley4f7fe772000-07-13 20:01:58 +0000530static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
531{
Mark Whitley2dc192f2000-11-03 19:47:00 +0000532 char *hackline = (char *)line;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000533 int altered = 0;
Mark Whitley2dc192f2000-11-03 19:47:00 +0000534 regmatch_t *regmatch = NULL;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000535
Mark Whitley2dc192f2000-11-03 19:47:00 +0000536 /* we only proceed if the substitution 'search' expression matches */
537 if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
538 return 0;
539
540 /* whaddaya know, it matched. get the number of back references */
541 regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
542
543 /* and now, as long as we've got a line to try matching and if we can match
544 * the search string, we make substitutions */
545 while (*hackline && (regexec(sed_cmd->sub_match, hackline,
546 sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
Mark Whitley4f7fe772000-07-13 20:01:58 +0000547 int i;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000548
Mark Whitley2dc192f2000-11-03 19:47:00 +0000549 /* print everything before the match */
550 for (i = 0; i < regmatch[0].rm_so; i++)
551 fputc(hackline[i], stdout);
Mark Whitley97562bd2000-07-17 20:06:42 +0000552
Mark Whitley2dc192f2000-11-03 19:47:00 +0000553 /* then print the substitution string */
554 print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
Mark Whitley97562bd2000-07-17 20:06:42 +0000555
Mark Whitley2dc192f2000-11-03 19:47:00 +0000556 /* advance past the match */
557 hackline += regmatch[0].rm_eo;
558 /* flag that something has changed */
559 altered++;
Mark Whitley97562bd2000-07-17 20:06:42 +0000560
Mark Whitley2dc192f2000-11-03 19:47:00 +0000561 /* if we're not doing this globally, get out now */
562 if (!sed_cmd->sub_g)
563 break;
Mark Whitley4f7fe772000-07-13 20:01:58 +0000564 }
565
Mark Whitley2dc192f2000-11-03 19:47:00 +0000566 /* if there's anything left of the line, print it */
567 if (*hackline)
568 fputs(hackline, stdout);
569
570 /* cleanup */
571 free(regmatch);
572
Mark Whitley4f7fe772000-07-13 20:01:58 +0000573 return altered;
574}
Mark Whitley6315ce62000-07-10 22:55:51 +0000575
576static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
577{
578 int altered = 0;
579
580 switch (sed_cmd->cmd) {
581
582 case 'p':
583 fputs(line, stdout);
584 break;
585
586 case 'd':
587 altered++;
588 break;
589
Mark Whitley4f7fe772000-07-13 20:01:58 +0000590 case 's':
Mark Whitley2dc192f2000-11-03 19:47:00 +0000591
592 /*
593 * Some special cases for 's' printing to make it compliant with
594 * GNU sed printing behavior (aka "The -n | s///p Matrix"):
595 *
596 * -n ONLY = never print anything regardless of any successful
597 * substitution
598 *
599 * s///p ONLY = always print successful substitutions, even if
600 * the line is going to be printed anyway (line will be printed
601 * twice).
602 *
603 * -n AND s///p = print ONLY a successful substitution ONE TIME;
604 * no other lines are printed - this is the reason why the 'p'
605 * flag exists in the first place.
606 */
607
Mark Whitleye7ff2842000-11-03 20:02:35 +0000608 /* if the user specified that they didn't want anything printed (i.e. a -n
609 * flag and no 'p' flag after the s///), then there's really no point doing
610 * anything here. */
611 if (be_quiet && !sed_cmd->sub_p)
612 break;
613
Mark Whitley2dc192f2000-11-03 19:47:00 +0000614 /* we print the line once, unless we were told to be quiet */
615 if (!be_quiet)
616 altered = do_subst_command(sed_cmd, line);
617
618 /* we also print the line if we were given the 'p' flag
619 * (this is quite possibly the second printing) */
620 if (sed_cmd->sub_p)
621 altered = do_subst_command(sed_cmd, line);
622
Mark Whitley6315ce62000-07-10 22:55:51 +0000623 break;
Mark Whitley94074a92000-07-14 00:00:15 +0000624
625 case 'a':
626 fputs(line, stdout);
627 fputs(sed_cmd->editline, stdout);
628 altered++;
629 break;
630
631 case 'i':
632 fputs(sed_cmd->editline, stdout);
633 break;
634
635 case 'c':
636 fputs(sed_cmd->editline, stdout);
637 altered++;
638 break;
Mark Whitley6315ce62000-07-10 22:55:51 +0000639 }
640
641 return altered;
642}
643
644static void process_file(FILE *file)
645{
646 char *line = NULL;
647 static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
648 unsigned int still_in_range = 0;
649 int line_altered;
650 int i;
651
652 /* go through every line in the file */
653 while ((line = get_line_from_file(file)) != NULL) {
654
655 linenum++;
656 line_altered = 0;
657
658 /* for every line, go through all the commands */
659 for (i = 0; i < ncmds; i++) {
660
661 /* are we acting on a range of matched lines? */
662 if (sed_cmds[i].beg_match && sed_cmds[i].end_match) {
663 if (still_in_range || regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0) {
664 line_altered += do_sed_command(&sed_cmds[i], line);
665 still_in_range = 1;
666 if (regexec(sed_cmds[i].end_match, line, 0, NULL, 0) == 0)
667 still_in_range = 0;
668 }
669 }
670
671 /* are we trying to match a single line? */
672 else if (sed_cmds[i].beg_match) {
673 if (regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0)
674 line_altered += do_sed_command(&sed_cmds[i], line);
675 }
676
677 /* are we acting on a range of line numbers? */
Mark Whitley40406e62000-08-10 00:09:47 +0000678 else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
Mark Whitley2dc192f2000-11-03 19:47:00 +0000679 if (linenum >= sed_cmds[i].beg_line &&
680 (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
Mark Whitley6315ce62000-07-10 22:55:51 +0000681 line_altered += do_sed_command(&sed_cmds[i], line);
682 }
683
684 /* are we acting on a specified line number */
685 else if (sed_cmds[i].beg_line > 0) {
686 if (linenum == sed_cmds[i].beg_line)
687 line_altered += do_sed_command(&sed_cmds[i], line);
688 }
689
690 /* not acting on matches or line numbers. act on every line */
691 else
692 line_altered += do_sed_command(&sed_cmds[i], line);
693
Erik Andersene49d5ec2000-02-08 19:58:47 +0000694 }
Erik Andersen1266a131999-12-29 22:19:46 +0000695
Mark Whitley2dc192f2000-11-03 19:47:00 +0000696 /* we will print the line unless we were told to be quiet or if the
697 * line was altered (via a 'd'elete or 's'ubstitution), in which case
698 * the altered line was already printed */
699 if (!be_quiet && !line_altered)
Mark Whitley6315ce62000-07-10 22:55:51 +0000700 fputs(line, stdout);
701
702 free(line);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000703 }
Erik Andersen1266a131999-12-29 22:19:46 +0000704}
705
706extern int sed_main(int argc, char **argv)
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000707{
Mark Whitley6315ce62000-07-10 22:55:51 +0000708 int opt;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000709
Eric Andersenb040d4f2000-07-25 18:01:20 +0000710#ifdef BB_FEATURE_CLEAN_UP
Mark Whitley858c1ad2000-07-11 21:38:47 +0000711 /* destroy command strings on exit */
712 if (atexit(destroy_cmd_strs) == -1) {
713 perror("sed");
714 exit(1);
715 }
Mark Whitleyc41e8c82000-07-12 23:35:21 +0000716#endif
Mark Whitley858c1ad2000-07-11 21:38:47 +0000717
Mark Whitley6315ce62000-07-10 22:55:51 +0000718 /* do normal option parsing */
Mark Whitley1f45b262000-07-20 23:08:40 +0000719 while ((opt = getopt(argc, argv, "hne:f:")) > 0) {
Mark Whitley6315ce62000-07-10 22:55:51 +0000720 switch (opt) {
Mark Whitley6315ce62000-07-10 22:55:51 +0000721 case 'h':
722 usage(sed_usage);
723 break;
Erik Andersene916d242000-03-06 19:20:35 +0000724 case 'n':
Mark Whitley6315ce62000-07-10 22:55:51 +0000725 be_quiet++;
Erik Andersene916d242000-03-06 19:20:35 +0000726 break;
727 case 'e':
Mark Whitley6315ce62000-07-10 22:55:51 +0000728 add_cmd_str(optarg);
Erik Andersene916d242000-03-06 19:20:35 +0000729 break;
Mark Whitley6315ce62000-07-10 22:55:51 +0000730 case 'f':
731 load_cmd_file(optarg);
Erik Andersene49d5ec2000-02-08 19:58:47 +0000732 break;
Erik Andersene49d5ec2000-02-08 19:58:47 +0000733 }
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000734 }
Mark Whitley6315ce62000-07-10 22:55:51 +0000735
736 /* if we didn't get a pattern from a -e and no command file was specified,
737 * argv[optind] should be the pattern. no pattern, no worky */
738 if (ncmds == 0) {
739 if (argv[optind] == NULL)
740 usage(sed_usage);
741 else {
742 add_cmd_str(argv[optind]);
743 optind++;
744 }
745 }
746
747
748 /* argv[(optind)..(argc-1)] should be names of file to process. If no
749 * files were specified or '-' was specified, take input from stdin.
750 * Otherwise, we process all the files specified. */
751 if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
752 process_file(stdin);
753 }
754 else {
755 int i;
756 FILE *file;
757 for (i = optind; i < argc; i++) {
758 file = fopen(argv[i], "r");
759 if (file == NULL) {
Matt Kraai1fa1ade2000-12-18 03:57:16 +0000760 perror_msg("%s", argv[i]);
Mark Whitley6315ce62000-07-10 22:55:51 +0000761 } else {
762 process_file(file);
763 fclose(file);
764 }
765 }
766 }
767
Mark Whitley6315ce62000-07-10 22:55:51 +0000768 return 0;
Eric Andersen6b6b3f61999-10-28 16:06:25 +0000769}