fix bug #474:
[people/mcb30/busybox.git] / editors / sed.c
1 /* vi: set sw=4 ts=4: */
2 /*
3  * sed.c - very minimalist version of sed
4  *
5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
7  * Copyright (C) 2002  Matt Kraai
8  * Copyright (C) 2003 by Glenn McGrath <bug1@iinet.net.au>
9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
10  *
11  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
12  */
13
14 /* Code overview.
15
16   Files are laid out to avoid unnecessary function declarations.  So for
17   example, every function add_cmd calls occurs before add_cmd in this file.
18
19   add_cmd() is called on each line of sed command text (from a file or from
20   the command line).  It calls get_address() and parse_cmd_args().  The
21   resulting sed_cmd_t structures are appended to a linked list
22   (sed_cmd_head/sed_cmd_tail).
23
24   add_input_file() adds a FILE * to the list of input files.  We need to
25   know all input sources ahead of time to find the last line for the $ match.
26
27   process_files() does actual sedding, reading data lines from each input FILE *
28   (which could be stdin) and applying the sed command list (sed_cmd_head) to
29   each of the resulting lines.
30
31   sed_main() is where external code calls into this, with a command line.
32 */
33
34
35 /*
36         Supported features and commands in this version of sed:
37
38          - comments ('#')
39          - address matching: num|/matchstr/[,num|/matchstr/|$]command
40          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
41          - edit commands: (a)ppend, (i)nsert, (c)hange
42          - file commands: (r)ead
43          - backreferences in substitution expressions (\0, \1, \2...\9)
44          - grouped commands: {cmd1;cmd2}
45          - transliteration (y/source-chars/dest-chars/)
46          - pattern space hold space storing / swapping (g, h, x)
47          - labels / branching (: label, b, t, T)
48
49          (Note: Specifying an address (range) to match is *optional*; commands
50          default to the whole pattern space if no specific address match was
51          requested.)
52
53         Todo:
54          - Create a wrapper around regex to make libc's regex conform with sed
55
56         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
57 */
58
59 #include <stdio.h>
60 #include <unistd.h>             /* for getopt() */
61 #include <string.h>             /* for strdup() */
62 #include <errno.h>
63 #include <ctype.h>              /* for isspace() */
64 #include <stdlib.h>
65 #include "busybox.h"
66 #include "xregex.h"
67
68 typedef struct sed_cmd_s {
69     /* Ordered by alignment requirements: currently 36 bytes on x86 */
70
71     /* address storage */
72     regex_t *beg_match; /* sed -e '/match/cmd' */
73     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
74     regex_t *sub_match; /* For 's/sub_match/string/' */
75     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
76     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
77
78     FILE *file;                 /* File (sr) command writes to, -1 for none. */
79     char *string;               /* Data string for (saicytb) commands. */
80
81     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
82
83     /* Bitfields (gcc won't group them if we don't) */
84     unsigned int invert:1;                      /* the '!' after the address */
85     unsigned int in_match:1;            /* Next line also included in match? */
86     unsigned int no_newline:1;          /* Last line written by (sr) had no '\n' */
87     unsigned int sub_p:1;                       /* (s) print option */
88
89
90     /* GENERAL FIELDS */
91     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
92     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
93 } sed_cmd_t;
94
95 /* globals */
96 /* options */
97 static int be_quiet, in_place, regex_type;
98 static FILE *nonstdout;
99 static char *outname,*hold_space;
100
101 /* List of input files */
102 static int input_file_count,current_input_file;
103 static FILE **input_file_list;
104
105 static const char bad_format_in_subst[] =
106         "bad format in substitution expression";
107 static const char *const semicolon_whitespace = "; \n\r\t\v";
108
109 static regmatch_t regmatch[10];
110 static regex_t *previous_regex_ptr;
111
112 /* linked list of sed commands */
113 static sed_cmd_t sed_cmd_head;
114 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
115
116 /* Linked list of append lines */
117 struct append_list {
118         char *string;
119         struct append_list *next;
120 };
121 static struct append_list *append_head=NULL, *append_tail=NULL;
122
123 #ifdef CONFIG_FEATURE_CLEAN_UP
124 static void free_and_close_stuff(void)
125 {
126         sed_cmd_t *sed_cmd = sed_cmd_head.next;
127
128         while(append_head) {
129                 append_tail=append_head->next;
130                 free(append_head->string);
131                 free(append_head);
132                 append_head=append_tail;
133         }
134
135         while (sed_cmd) {
136                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
137
138                 if(sed_cmd->file)
139                         bb_xprint_and_close_file(sed_cmd->file);
140
141                 if (sed_cmd->beg_match) {
142                         regfree(sed_cmd->beg_match);
143                         free(sed_cmd->beg_match);
144                 }
145                 if (sed_cmd->end_match) {
146                         regfree(sed_cmd->end_match);
147                         free(sed_cmd->end_match);
148                 }
149                 if (sed_cmd->sub_match) {
150                         regfree(sed_cmd->sub_match);
151                         free(sed_cmd->sub_match);
152                 }
153                 free(sed_cmd->string);
154                 free(sed_cmd);
155                 sed_cmd = sed_cmd_next;
156         }
157
158         if(hold_space) free(hold_space);
159
160     while(current_input_file<input_file_count)
161                 fclose(input_file_list[current_input_file++]);
162 }
163 #endif
164
165 /* If something bad happens during -i operation, delete temp file */
166
167 static void cleanup_outname(void)
168 {
169   if(outname) unlink(outname);
170 }
171
172 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
173
174 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
175 {
176         int i=0;
177
178         while(i<len) {
179                 if(string[i] == '\\') {
180                         if(!to || string[i+1] == from) {
181                                 *(dest++) = to ? to : string[i+1];
182                                 i+=2;
183                                 continue;
184                         } else *(dest++)=string[i++];
185                 }
186                 *(dest++) = string[i++];
187         }
188         *dest=0;
189 }
190
191 static char *copy_parsing_escapes(const char *string, int len)
192 {
193         char *dest=xmalloc(len+1);
194
195         parse_escapes(dest,string,len,'n','\n');
196         return dest;
197 }
198
199
200 /*
201  * index_of_next_unescaped_regexp_delim - walks left to right through a string
202  * beginning at a specified index and returns the index of the next regular
203  * expression delimiter (typically a forward * slash ('/')) not preceded by
204  * a backslash ('\').
205  */
206 static int index_of_next_unescaped_regexp_delim(const char delimiter,
207         const char *str)
208 {
209         int bracket = -1;
210         int escaped = 0;
211         int idx = 0;
212         char ch;
213
214         for (; (ch = str[idx]); idx++) {
215                 if (bracket != -1) {
216                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
217                                         && str[idx - 1] == '^')))
218                                 bracket = -1;
219                 } else if (escaped)
220                         escaped = 0;
221                 else if (ch == '\\')
222                         escaped = 1;
223                 else if (ch == '[')
224                         bracket = idx;
225                 else if (ch == delimiter)
226                         return idx;
227         }
228
229         /* if we make it to here, we've hit the end of the string */
230         return -1;
231 }
232
233 /*
234  *  Returns the index of the third delimiter
235  */
236 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
237 {
238         const char *cmdstr_ptr = cmdstr;
239         char delimiter;
240         int idx = 0;
241
242         /* verify that the 's' or 'y' is followed by something.  That something
243          * (typically a 'slash') is now our regexp delimiter... */
244         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
245         delimiter = *(cmdstr_ptr++);
246
247         /* save the match string */
248         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
249         if (idx == -1) {
250                 bb_error_msg_and_die(bad_format_in_subst);
251         }
252         *match = copy_parsing_escapes(cmdstr_ptr, idx);
253
254         /* save the replacement string */
255         cmdstr_ptr += idx + 1;
256         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
257         if (idx == -1) {
258                 bb_error_msg_and_die(bad_format_in_subst);
259         }
260         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
261
262         return ((cmdstr_ptr - cmdstr) + idx);
263 }
264
265 /*
266  * returns the index in the string just past where the address ends.
267  */
268 static int get_address(char *my_str, int *linenum, regex_t ** regex)
269 {
270         char *pos = my_str;
271
272         if (isdigit(*my_str)) {
273                 *linenum = strtol(my_str, &pos, 10);
274                 /* endstr shouldnt ever equal NULL */
275         } else if (*my_str == '$') {
276                 *linenum = -1;
277                 pos++;
278         } else if (*my_str == '/' || *my_str == '\\') {
279                 int next;
280                 char delimiter;
281                 char *temp;
282
283                 if (*my_str == '\\') delimiter = *(++pos);
284                 else delimiter = '/';
285                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
286                 if (next == -1)
287                         bb_error_msg_and_die("unterminated match expression");
288
289                 temp=copy_parsing_escapes(pos,next);
290                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
291                 xregcomp(*regex, temp, regex_type|REG_NEWLINE);
292                 free(temp);
293                 /* Move position to next character after last delimiter */
294                 pos+=(next+1);
295         }
296         return pos - my_str;
297 }
298
299 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
300 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
301 {
302         int start = 0, idx, hack=0;
303
304         /* Skip whitespace, then grab filename to end of line */
305         while (isspace(filecmdstr[start])) start++;
306         idx=start;
307         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
308         /* If lines glued together, put backslash back. */
309         if(filecmdstr[idx]=='\n') hack=1;
310         if(idx==start) bb_error_msg_and_die("Empty filename");
311         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
312         if(hack) *(idx+*retval)='\\';
313
314         return idx;
315 }
316
317 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
318 {
319         int cflags = regex_type;
320         char *match;
321         int idx = 0;
322
323         /*
324          * A substitution command should look something like this:
325          *    s/match/replace/ #gIpw
326          *    ||     |        |||
327          *    mandatory       optional
328          */
329         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
330
331         /* determine the number of back references in the match string */
332         /* Note: we compute this here rather than in the do_subst_command()
333          * function to save processor time, at the expense of a little more memory
334          * (4 bits) per sed_cmd */
335
336         /* process the flags */
337
338         sed_cmd->which_match=1;
339         while (substr[++idx]) {
340                 /* Parse match number */
341                 if(isdigit(substr[idx])) {
342                         if(match[0]!='^') {
343                                 /* Match 0 treated as all, multiple matches we take the last one. */
344                                 char *pos=substr+idx;
345                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
346                                 idx=pos-substr;
347                         }
348                         continue;
349                 }
350                 /* Skip spaces */
351                 if(isspace(substr[idx])) continue;
352
353                 switch (substr[idx]) {
354                         /* Replace all occurrences */
355                         case 'g':
356                                 if (match[0] != '^') sed_cmd->which_match = 0;
357                                 break;
358                         /* Print pattern space */
359                         case 'p':
360                                 sed_cmd->sub_p = 1;
361                                 break;
362                         /* Write to file */
363                         case 'w':
364                         {
365                                 char *temp;
366                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
367
368                                 break;
369                         }
370                         /* Ignore case (gnu exension) */
371                         case 'I':
372                                 cflags |= REG_ICASE;
373                                 break;
374                         /* Comment */
375                         case '#':
376                                 while(substr[++idx]);
377                                 /* Fall through */
378                         /* End of command */
379                         case ';':
380                         case '}':
381                                 goto out;
382                         default:
383                                 bb_error_msg_and_die("bad option in substitution expression");
384                 }
385         }
386 out:
387         /* compile the match string into a regex */
388         if (*match != '\0') {
389                 /* If match is empty, we use last regex used at runtime */
390                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
391                 xregcomp(sed_cmd->sub_match, match, cflags);
392         }
393         free(match);
394
395         return idx;
396 }
397
398 /*
399  *  Process the commands arguments
400  */
401 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
402 {
403         /* handle (s)ubstitution command */
404         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
405         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
406         else if (strchr("aic", sed_cmd->cmd)) {
407                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
408                         bb_error_msg_and_die
409                                 ("only a beginning address can be specified for edit commands");
410                 for(;;) {
411                         if(*cmdstr=='\n' || *cmdstr=='\\') {
412                                 cmdstr++;
413                                 break;
414                         } else if(isspace(*cmdstr)) cmdstr++;
415                         else break;
416                 }
417                 sed_cmd->string = bb_xstrdup(cmdstr);
418                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
419                 cmdstr += strlen(cmdstr);
420         /* handle file cmds: (r)ead */
421         } else if(strchr("rw", sed_cmd->cmd)) {
422                 if (sed_cmd->end_line || sed_cmd->end_match)
423                         bb_error_msg_and_die("Command only uses one address");
424                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
425                 if(sed_cmd->cmd=='w')
426                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
427         /* handle branch commands */
428         } else if (strchr(":btT", sed_cmd->cmd)) {
429                 int length;
430
431                 while(isspace(*cmdstr)) cmdstr++;
432                 length = strcspn(cmdstr, semicolon_whitespace);
433                 if (length) {
434                         sed_cmd->string = strndup(cmdstr, length);
435                         cmdstr += length;
436                 }
437         }
438         /* translation command */
439         else if (sed_cmd->cmd == 'y') {
440                 char *match, *replace;
441                 int i=cmdstr[0];
442
443                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
444                 /* \n already parsed, but \delimiter needs unescaping. */
445                 parse_escapes(match,match,strlen(match),i,i);
446                 parse_escapes(replace,replace,strlen(replace),i,i);
447
448                 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
449                 for (i = 0; match[i] && replace[i]; i++) {
450                         sed_cmd->string[i * 2] = match[i];
451                         sed_cmd->string[(i * 2) + 1] = replace[i];
452                 }
453                 free(match);
454                 free(replace);
455         }
456         /* if it wasnt a single-letter command that takes no arguments
457          * then it must be an invalid command.
458          */
459         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
460                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
461         }
462
463         /* give back whatever's left over */
464         return (cmdstr);
465 }
466
467
468 /* Parse address+command sets, skipping comment lines. */
469
470 static void add_cmd(char *cmdstr)
471 {
472         static char *add_cmd_line=NULL;
473         sed_cmd_t *sed_cmd;
474         int temp;
475
476         /* Append this line to any unfinished line from last time. */
477         if(add_cmd_line) {
478                 int lastlen=strlen(add_cmd_line);
479                 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
480
481                 memcpy(tmp,add_cmd_line,lastlen);
482                 tmp[lastlen]='\n';
483                 strcpy(tmp+lastlen+1,cmdstr);
484                 free(add_cmd_line);
485                 cmdstr=add_cmd_line=tmp;
486         } else add_cmd_line=NULL;
487
488         /* If this line ends with backslash, request next line. */
489         temp=strlen(cmdstr);
490         if(temp && cmdstr[temp-1]=='\\') {
491                 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
492                 add_cmd_line[temp-1]=0;
493                 return;
494         }
495
496         /* Loop parsing all commands in this line. */
497         while(*cmdstr) {
498                 /* Skip leading whitespace and semicolons */
499                 cmdstr += strspn(cmdstr, semicolon_whitespace);
500
501                 /* If no more commands, exit. */
502                 if(!*cmdstr) break;
503
504                 /* if this is a comment, jump past it and keep going */
505                 if (*cmdstr == '#') {
506                         /* "#n" is the same as using -n on the command line */
507                         if (cmdstr[1] == 'n') be_quiet++;
508                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
509                         continue;
510                 }
511
512                 /* parse the command
513                  * format is: [addr][,addr][!]cmd
514                  *            |----||-----||-|
515                  *            part1 part2  part3
516                  */
517
518                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
519
520                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
521                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
522
523                 /* second part (if present) will begin with a comma */
524                 if (*cmdstr == ',') {
525                         int idx;
526
527                         cmdstr++;
528                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
529                         if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
530                         cmdstr += idx;
531                 }
532
533                 /* skip whitespace before the command */
534                 while (isspace(*cmdstr)) cmdstr++;
535
536                 /* Check for inversion flag */
537                 if (*cmdstr == '!') {
538                         sed_cmd->invert = 1;
539                         cmdstr++;
540
541                         /* skip whitespace before the command */
542                         while (isspace(*cmdstr)) cmdstr++;
543                 }
544
545                 /* last part (mandatory) will be a command */
546                 if (!*cmdstr) bb_error_msg_and_die("missing command");
547                 sed_cmd->cmd = *(cmdstr++);
548                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
549
550                 /* Add the command to the command array */
551                 sed_cmd_tail->next = sed_cmd;
552                 sed_cmd_tail = sed_cmd_tail->next;
553         }
554
555         /* If we glued multiple lines together, free the memory. */
556         if(add_cmd_line) {
557                 free(add_cmd_line);
558                 add_cmd_line=NULL;
559         }
560 }
561
562 /* Append to a string, reallocating memory as necessary. */
563
564 static struct pipeline {
565         char *buf;      /* Space to hold string */
566         int idx;        /* Space used */
567         int len;        /* Space allocated */
568 } pipeline;
569
570 #define PIPE_GROW 64
571
572 static void pipe_putc(char c)
573 {
574         if(pipeline.idx==pipeline.len) {
575                 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
576                 pipeline.len+=PIPE_GROW;
577         }
578         pipeline.buf[pipeline.idx++] = (c);
579 }
580
581 static void do_subst_w_backrefs(const char *line, const char *replace)
582 {
583         int i,j;
584
585         /* go through the replacement string */
586         for (i = 0; replace[i]; i++) {
587                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
588                 if (replace[i] == '\\' && replace[i+1]>='0' && replace[i+1]<='9') {
589                         int backref=replace[++i]-'0';
590
591                         /* print out the text held in regmatch[backref] */
592                         if(regmatch[backref].rm_so != -1)
593                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
594                                         pipe_putc(line[j]);
595                 }
596
597                 /* if we find a backslash escaped character, print the character */
598                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
599
600                 /* if we find an unescaped '&' print out the whole matched text. */
601                 else if (replace[i] == '&')
602                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
603                                 pipe_putc(line[j]);
604                 /* Otherwise just output the character. */
605                 else pipe_putc(replace[i]);
606         }
607 }
608
609 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
610 {
611         char *oldline = *line;
612         int altered = 0;
613         int match_count=0;
614         regex_t *current_regex;
615
616         /* Handle empty regex. */
617         if (sed_cmd->sub_match == NULL) {
618                 current_regex = previous_regex_ptr;
619                 if(!current_regex)
620                         bb_error_msg_and_die("No previous regexp.");
621         } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
622
623         /* Find the first match */
624         if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
625                 return 0;
626
627         /* Initialize temporary output buffer. */
628         pipeline.buf=xmalloc(PIPE_GROW);
629         pipeline.len=PIPE_GROW;
630         pipeline.idx=0;
631
632         /* Now loop through, substituting for matches */
633         do {
634                 int i;
635
636                 /* Work around bug in glibc regexec, demonstrated by:
637                    echo " a.b" | busybox sed 's [^ .]* x g'
638                    The match_count check is so not to break
639                    echo "hi" | busybox sed 's/^/!/g' */
640                 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
641                         pipe_putc(*(oldline++));
642                         continue;
643                 }
644
645                 match_count++;
646
647                 /* If we aren't interested in this match, output old line to
648                    end of match and continue */
649                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
650                         for(i=0;i<regmatch[0].rm_eo;i++)
651                                 pipe_putc(oldline[i]);
652                         continue;
653                 }
654
655                 /* print everything before the match */
656                 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
657
658                 /* then print the substitution string */
659                 do_subst_w_backrefs(oldline, sed_cmd->string);
660
661                 /* advance past the match */
662                 oldline += regmatch[0].rm_eo;
663                 /* flag that something has changed */
664                 altered++;
665
666                 /* if we're not doing this globally, get out now */
667                 if (sed_cmd->which_match) break;
668         } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
669
670         /* Copy rest of string into output pipeline */
671
672         while(*oldline) pipe_putc(*(oldline++));
673         pipe_putc(0);
674
675         free(*line);
676         *line = pipeline.buf;
677         return altered;
678 }
679
680 /* Set command pointer to point to this label.  (Does not handle null label.) */
681 static sed_cmd_t *branch_to(const char *label)
682 {
683         sed_cmd_t *sed_cmd;
684
685         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
686                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
687                         return (sed_cmd);
688                 }
689         }
690         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
691 }
692
693 /* Append copy of string to append buffer */
694 static void append(char *s)
695 {
696         struct append_list *temp=calloc(1,sizeof(struct append_list));
697
698         if(append_head)
699                 append_tail=(append_tail->next=temp);
700         else append_head=append_tail=temp;
701         temp->string=strdup(s);
702 }
703
704 static void flush_append(void)
705 {
706         /* Output appended lines. */
707         while(append_head) {
708                 fprintf(nonstdout,"%s\n",append_head->string);
709                 append_tail=append_head->next;
710                 free(append_head->string);
711                 free(append_head);
712                 append_head=append_tail;
713         }
714         append_head=append_tail=NULL;
715 }
716
717 static void add_input_file(FILE *file)
718 {
719         input_file_list=xrealloc(input_file_list,(input_file_count+1)*sizeof(FILE *));
720         input_file_list[input_file_count++]=file;
721 }
722
723 /* Get next line of input from input_file_list, flushing append buffer and
724  * noting if we ran out of files without a newline on the last line we read.
725  */
726 static char *get_next_line(int *no_newline)
727 {
728         char *temp=NULL;
729         int len;
730
731         flush_append();
732         while(current_input_file<input_file_count) {
733                 temp=bb_get_line_from_file(input_file_list[current_input_file]);
734                 if(temp) {
735                         len=strlen(temp);
736                         *no_newline=!(len && temp[len-1]=='\n');
737                         if(!*no_newline) temp[len-1]=0;
738                         break;
739                 } else fclose(input_file_list[current_input_file++]);
740         }
741
742         return temp;
743 }
744
745 /* Output line of text.  missing_newline means the last line output did not
746    end with a newline.  no_newline means this line does not end with a
747    newline. */
748
749 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
750 {
751         if(missing_newline) fputc('\n',file);
752         fputs(s,file);
753         if(!no_newline) fputc('\n',file);
754
755     if(ferror(file)) {
756                 fprintf(stderr,"Write failed.\n");
757                 exit(4);  /* It's what gnu sed exits with... */
758         }
759
760         return no_newline;
761 }
762
763 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,nonstdout,missing_newline,n)
764
765 static void process_files(void)
766 {
767         char *pattern_space, *next_line;
768         int linenum = 0, missing_newline=0;
769         int no_newline,next_no_newline=0;
770
771         next_line = get_next_line(&next_no_newline);
772
773         /* go through every line in each file */
774         for(;;) {
775                 sed_cmd_t *sed_cmd;
776                 int substituted=0;
777
778                 /* Advance to next line.  Stop if out of lines. */
779                 if(!(pattern_space=next_line)) break;
780                 no_newline=next_no_newline;
781
782                 /* Read one line in advance so we can act on the last line, the '$' address */
783                 next_line = get_next_line(&next_no_newline);
784                 linenum++;
785 restart:
786                 /* for every line, go through all the commands */
787                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
788                         int old_matched, matched;
789
790                         old_matched = sed_cmd->in_match;
791
792                         /* Determine if this command matches this line: */
793
794                         /* Are we continuing a previous multi-line match? */
795
796                         sed_cmd->in_match = sed_cmd->in_match
797
798                         /* Or is no range necessary? */
799                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
800                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
801
802                         /* Or did we match the start of a numerical range? */
803                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
804
805                         /* Or does this line match our begin address regex? */
806                                 || (sed_cmd->beg_match &&
807                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
808
809                         /* Or did we match last line of input? */
810                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
811
812                         /* Snapshot the value */
813
814                         matched = sed_cmd->in_match;
815
816                         /* Is this line the end of the current match? */
817
818                         if(matched) {
819                                 sed_cmd->in_match = !(
820                                         /* has the ending line come, or is this a single address command? */
821                                         (sed_cmd->end_line ?
822                                                 sed_cmd->end_line==-1 ?
823                                                         !next_line
824                                                         : sed_cmd->end_line<=linenum
825                                                 : !sed_cmd->end_match)
826                                         /* or does this line matches our last address regex */
827                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
828                                 );
829                         }
830
831                         /* Skip blocks of commands we didn't match. */
832                         if (sed_cmd->cmd == '{') {
833                                 if(sed_cmd->invert ? matched : !matched)
834                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
835                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
836                                 continue;
837                         }
838
839                         /* Okay, so did this line match? */
840                         if (sed_cmd->invert ? !matched : matched) {
841                                 /* Update last used regex in case a blank substitute BRE is found */
842                                 if (sed_cmd->beg_match) {
843                                         previous_regex_ptr = sed_cmd->beg_match;
844                                 }
845
846                                 /* actual sedding */
847                                 switch (sed_cmd->cmd) {
848
849                                         /* Print line number */
850                                         case '=':
851                                                 fprintf(nonstdout,"%d\n", linenum);
852                                                 break;
853
854                                         /* Write the current pattern space up to the first newline */
855                                         case 'P':
856                                         {
857                                                 char *tmp = strchr(pattern_space, '\n');
858
859                                                 if (tmp) {
860                                                         *tmp = '\0';
861                                                         sed_puts(pattern_space,1);
862                                                         *tmp = '\n';
863                                                         break;
864                                                 }
865                                                 /* Fall Through */
866                                         }
867
868                                         /* Write the current pattern space to output */
869                                         case 'p':
870                                                 sed_puts(pattern_space,no_newline);
871                                                 break;
872                                         /* Delete up through first newline */
873                                         case 'D':
874                                         {
875                                                 char *tmp = strchr(pattern_space,'\n');
876
877                                                 if(tmp) {
878                                                         tmp=bb_xstrdup(tmp+1);
879                                                         free(pattern_space);
880                                                         pattern_space=tmp;
881                                                         goto restart;
882                                                 }
883                                         }
884                                         /* discard this line. */
885                                         case 'd':
886                                                 goto discard_line;
887
888                                         /* Substitute with regex */
889                                         case 's':
890                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
891                                                         substituted|=1;
892
893                                                         /* handle p option */
894                                                         if(sed_cmd->sub_p)
895                                                                 sed_puts(pattern_space,no_newline);
896                                                         /* handle w option */
897                                                         if(sed_cmd->file)
898                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
899
900                                                 }
901                                                 break;
902
903                                         /* Append line to linked list to be printed later */
904                                         case 'a':
905                                         {
906                                                 append(sed_cmd->string);
907                                                 break;
908                                         }
909
910                                         /* Insert text before this line */
911                                         case 'i':
912                                                 sed_puts(sed_cmd->string,1);
913                                                 break;
914
915                                         /* Cut and paste text (replace) */
916                                         case 'c':
917                                                 /* Only triggers on last line of a matching range. */
918                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,0);
919                                                 goto discard_line;
920
921                                         /* Read file, append contents to output */
922                                         case 'r':
923                                         {
924                                                 FILE *outfile;
925
926                                                 outfile = fopen(sed_cmd->string, "r");
927                                                 if (outfile) {
928                                                         char *line;
929
930                                                         while ((line = bb_get_chomped_line_from_file(outfile))
931                                                                         != NULL)
932                                                                 append(line);
933                                                         bb_xprint_and_close_file(outfile);
934                                                 }
935
936                                                 break;
937                                         }
938
939                                         /* Write pattern space to file. */
940                                         case 'w':
941                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
942                                                 break;
943
944                                         /* Read next line from input */
945                                         case 'n':
946                                                 if (!be_quiet)
947                                                         sed_puts(pattern_space,no_newline);
948                                                 if (next_line) {
949                                                         free(pattern_space);
950                                                         pattern_space = next_line;
951                                                         no_newline=next_no_newline;
952                                                         next_line = get_next_line(&next_no_newline);
953                                                         linenum++;
954                                                         break;
955                                                 }
956                                                 /* fall through */
957
958                                         /* Quit.  End of script, end of input. */
959                                         case 'q':
960                                                 /* Exit the outer while loop */
961                                                 free(next_line);
962                                                 next_line = NULL;
963                                                 goto discard_commands;
964
965                                         /* Append the next line to the current line */
966                                         case 'N':
967                                         {
968                                                 /* If no next line, jump to end of script and exit. */
969                                                 if (next_line == NULL) {
970                                                         /* Jump to end of script and exit */
971                                                         free(next_line);
972                                                         next_line = NULL;
973                                                         goto discard_line;
974                                                 /* append next_line, read new next_line. */
975                                                 } else {
976                                                         int len=strlen(pattern_space);
977
978                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
979                                                         pattern_space[len]='\n';
980                                                         strcpy(pattern_space+len+1, next_line);
981                                                         no_newline=next_no_newline;
982                                                         next_line = get_next_line(&next_no_newline);
983                                                         linenum++;
984                                                 }
985                                                 break;
986                                         }
987
988                                         /* Test/branch if substitution occurred */
989                                         case 't':
990                                                 if(!substituted) break;
991                                                 substituted=0;
992                                                 /* Fall through */
993                                         /* Test/branch if substitution didn't occur */
994                                         case 'T':
995                                                 if (substituted) break;
996                                                 /* Fall through */
997                                         /* Branch to label */
998                                         case 'b':
999                                                 if (!sed_cmd->string) goto discard_commands;
1000                                                 else sed_cmd = branch_to(sed_cmd->string);
1001                                                 break;
1002                                         /* Transliterate characters */
1003                                         case 'y':
1004                                         {
1005                                                 int i;
1006
1007                                                 for (i = 0; pattern_space[i]; i++) {
1008                                                         int j;
1009
1010                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1011                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1012                                                                         pattern_space[i] = sed_cmd->string[j + 1];
1013                                                                         break;
1014                                                                 }
1015                                                         }
1016                                                 }
1017
1018                                                 break;
1019                                         }
1020                                         case 'g':       /* Replace pattern space with hold space */
1021                                                 free(pattern_space);
1022                                                 pattern_space = strdup(hold_space ? hold_space : "");
1023                                                 break;
1024                                         case 'G':       /* Append newline and hold space to pattern space */
1025                                         {
1026                                                 int pattern_space_size = 2;
1027                                                 int hold_space_size = 0;
1028
1029                                                 if (pattern_space)
1030                                                         pattern_space_size += strlen(pattern_space);
1031                                                 if (hold_space) hold_space_size = strlen(hold_space);
1032                                                 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1033                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1034                                                 strcat(pattern_space, "\n");
1035                                                 if (hold_space) strcat(pattern_space, hold_space);
1036                                                 no_newline=0;
1037
1038                                                 break;
1039                                         }
1040                                         case 'h':       /* Replace hold space with pattern space */
1041                                                 free(hold_space);
1042                                                 hold_space = strdup(pattern_space);
1043                                                 break;
1044                                         case 'H':       /* Append newline and pattern space to hold space */
1045                                         {
1046                                                 int hold_space_size = 2;
1047                                                 int pattern_space_size = 0;
1048
1049                                                 if (hold_space) hold_space_size += strlen(hold_space);
1050                                                 if (pattern_space)
1051                                                         pattern_space_size = strlen(pattern_space);
1052                                                 hold_space = xrealloc(hold_space,
1053                                                                                         hold_space_size + pattern_space_size);
1054
1055                                                 if (hold_space_size == 2) hold_space[0]=0;
1056                                                 strcat(hold_space, "\n");
1057                                                 if (pattern_space) strcat(hold_space, pattern_space);
1058
1059                                                 break;
1060                                         }
1061                                         case 'x': /* Exchange hold and pattern space */
1062                                         {
1063                                                 char *tmp = pattern_space;
1064                                                 pattern_space = hold_space;
1065                                                 no_newline=0;
1066                                                 hold_space = tmp;
1067                                                 break;
1068                                         }
1069                                 }
1070                         }
1071                 }
1072
1073                 /*
1074                  * exit point from sedding...
1075                  */
1076 discard_commands:
1077                 /* we will print the line unless we were told to be quiet ('-n')
1078                    or if the line was suppressed (ala 'd'elete) */
1079                 if (!be_quiet) sed_puts(pattern_space,no_newline);
1080
1081                 /* Delete and such jump here. */
1082 discard_line:
1083                 flush_append();
1084                 free(pattern_space);
1085         }
1086 }
1087
1088 /* It is possible to have a command line argument with embedded
1089    newlines.  This counts as multiple command lines. */
1090
1091 static void add_cmd_block(char *cmdstr)
1092 {
1093         int go=1;
1094         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1095
1096         while(go) {
1097                 int len=strcspn(temp2,"\n");
1098                 if(!temp2[len]) go=0;
1099                 else temp2[len]=0;
1100                 add_cmd(temp2);
1101                 temp2+=len+1;
1102         }
1103         free(temp);
1104 }
1105
1106 extern int sed_main(int argc, char **argv)
1107 {
1108         int status = EXIT_SUCCESS, opt, getpat = 1;
1109
1110 #ifdef CONFIG_FEATURE_CLEAN_UP
1111         /* destroy command strings on exit */
1112         if (atexit(free_and_close_stuff) == -1)
1113                 bb_perror_msg_and_die("atexit");
1114 #endif
1115
1116         /* Lie to autoconf when it starts asking stupid questions. */
1117         if(argc==2 && !strcmp(argv[1],"--version")) {
1118                 printf("This is not GNU sed version 4.0\n");
1119                 exit(0);
1120         }
1121
1122         /* do normal option parsing */
1123         while ((opt = getopt(argc, argv, "irne:f:")) > 0) {
1124                 switch (opt) {
1125                 case 'i':
1126                         in_place++;
1127                         atexit(cleanup_outname);
1128                         break;
1129                 case 'r':
1130                         regex_type|=REG_EXTENDED;
1131                         break;
1132                 case 'n':
1133                         be_quiet++;
1134                         break;
1135                 case 'e':
1136                         add_cmd_block(optarg);
1137                         getpat=0;
1138                         break;
1139                 case 'f':
1140                 {
1141                         FILE *cmdfile;
1142                         char *line;
1143
1144                         cmdfile = bb_xfopen(optarg, "r");
1145
1146                         while ((line = bb_get_chomped_line_from_file(cmdfile)) != NULL) {
1147                                 add_cmd(line);
1148                                 getpat=0;
1149                                 free(line);
1150                         }
1151                         bb_xprint_and_close_file(cmdfile);
1152
1153                         break;
1154                 }
1155                 default:
1156                         bb_show_usage();
1157                 }
1158         }
1159
1160         /* if we didn't get a pattern from -e or -f, use argv[optind] */
1161         if(getpat) {
1162                 if (argv[optind] == NULL)
1163                         bb_show_usage();
1164                 else
1165                         add_cmd_block(argv[optind++]);
1166         }
1167         /* Flush any unfinished commands. */
1168         add_cmd("");
1169
1170         /* By default, we write to stdout */
1171         nonstdout=stdout;
1172
1173         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1174          * files were specified or '-' was specified, take input from stdin.
1175          * Otherwise, we process all the files specified. */
1176         if (argv[optind] == NULL) {
1177                 if(in_place) bb_error_msg_and_die("Filename required for -i");
1178                 add_input_file(stdin);
1179                 process_files();
1180         } else {
1181                 int i;
1182                 FILE *file;
1183
1184                 for (i = optind; i < argc; i++) {
1185                         if(!strcmp(argv[i], "-") && !in_place) {
1186                                 add_input_file(stdin);
1187                                 process_files();
1188                         } else {
1189                                 file = bb_wfopen(argv[i], "r");
1190                                 if (file) {
1191                                         if(in_place) {
1192                                                 struct stat statbuf;
1193                                                 int nonstdoutfd;
1194                                                 
1195                                                 outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1196                                                 strcat(outname,"XXXXXX");
1197                                                 if(-1==(nonstdoutfd=mkstemp(outname)))
1198                                                         bb_error_msg_and_die("no temp file");
1199                                                 nonstdout=fdopen(nonstdoutfd,"w");
1200                                                 /* Set permissions of output file */
1201                                                 fstat(fileno(file),&statbuf);
1202                                                 fchmod(nonstdoutfd,statbuf.st_mode);
1203                                                 add_input_file(file);
1204                                                 process_files();
1205                                                 fclose(nonstdout);
1206                                                 nonstdout=stdout;
1207                                                 unlink(argv[i]);
1208                                                 rename(outname,argv[i]);
1209                                                 free(outname);
1210                                                 outname=0;
1211                                         } else add_input_file(file);
1212                                 } else {
1213                                         status = EXIT_FAILURE;
1214                                 }
1215                         }
1216                 }
1217                 if(input_file_count>current_input_file) process_files();
1218         }
1219
1220         return status;
1221 }