c8bb503ea8cda2f6ea02ff4db5f3731b6081d064
[platform/upstream/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
9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
10  *
11  * MAINTAINER: Rob Landley <rob@landley.net>
12  *
13  * Licensed under GPLv2, see file LICENSE in this source tree.
14  */
15
16 /* Code overview.
17  *
18  * Files are laid out to avoid unnecessary function declarations.  So for
19  * example, every function add_cmd calls occurs before add_cmd in this file.
20  *
21  * add_cmd() is called on each line of sed command text (from a file or from
22  * the command line).  It calls get_address() and parse_cmd_args().  The
23  * resulting sed_cmd_t structures are appended to a linked list
24  * (G.sed_cmd_head/G.sed_cmd_tail).
25  *
26  * add_input_file() adds a FILE* to the list of input files.  We need to
27  * know all input sources ahead of time to find the last line for the $ match.
28  *
29  * process_files() does actual sedding, reading data lines from each input FILE*
30  * (which could be stdin) and applying the sed command list (sed_cmd_head) to
31  * each of the resulting lines.
32  *
33  * sed_main() is where external code calls into this, with a command line.
34  */
35
36 /* Supported features and commands in this version of sed:
37  *
38  * - comments ('#')
39  * - address matching: num|/matchstr/[,num|/matchstr/|$]command
40  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
41  * - edit commands: (a)ppend, (i)nsert, (c)hange
42  * - file commands: (r)ead
43  * - backreferences in substitution expressions (\0, \1, \2...\9)
44  * - grouped commands: {cmd1;cmd2}
45  * - transliteration (y/source-chars/dest-chars/)
46  * - pattern space hold space storing / swapping (g, h, x)
47  * - labels / branching (: label, b, t, T)
48  *
49  * (Note: Specifying an address (range) to match is *optional*; commands
50  * default to the whole pattern space if no specific address match was
51  * requested.)
52  *
53  * Todo:
54  * - Create a wrapper around regex to make libc's regex conform with sed
55  *
56  * Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
57  */
58
59 //usage:#define sed_trivial_usage
60 //usage:       "[-inr] [-f FILE]... [-e CMD]... [FILE]...\n"
61 //usage:       "or: sed [-inr] CMD [FILE]..."
62 //usage:#define sed_full_usage "\n\n"
63 //usage:       "        -e CMD  Add CMD to sed commands to be executed"
64 //usage:     "\n        -f FILE Add FILE contents to sed commands to be executed"
65 //usage:     "\n        -i      Edit files in-place (else sends result to stdout)"
66 //usage:     "\n        -n      Suppress automatic printing of pattern space"
67 //usage:     "\n        -r      Use extended regex syntax"
68 //usage:     "\n"
69 //usage:     "\nIf no -e or -f, the first non-option argument is the sed command string."
70 //usage:     "\nRemaining arguments are input files (stdin if none)."
71 //usage:
72 //usage:#define sed_example_usage
73 //usage:       "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
74 //usage:       "bar\n"
75
76 #include "libbb.h"
77 #include "xregex.h"
78
79 #if 0
80 # define dbg(...) bb_error_msg(__VA_ARGS__)
81 #else
82 # define dbg(...) ((void)0)
83 #endif
84
85
86 enum {
87         OPT_in_place = 1 << 0,
88 };
89
90 /* Each sed command turns into one of these structures. */
91 typedef struct sed_cmd_s {
92         /* Ordered by alignment requirements: currently 36 bytes on x86 */
93         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
94
95         /* address storage */
96         regex_t *beg_match;     /* sed -e '/match/cmd' */
97         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
98         regex_t *sub_match;     /* For 's/sub_match/string/' */
99         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
100         int beg_line_orig;      /* copy of the above, needed for -i */
101         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
102
103         FILE *sw_file;          /* File (sw) command writes to, -1 for none. */
104         char *string;           /* Data string for (saicytb) commands. */
105
106         unsigned which_match;   /* (s) Which match to replace (0 for all) */
107
108         /* Bitfields (gcc won't group them if we don't) */
109         unsigned invert:1;      /* the '!' after the address */
110         unsigned in_match:1;    /* Next line also included in match? */
111         unsigned sub_p:1;       /* (s) print option */
112
113         char sw_last_char;      /* Last line written by (sw) had no '\n' */
114
115         /* GENERAL FIELDS */
116         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
117 } sed_cmd_t;
118
119 static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
120
121 struct globals {
122         /* options */
123         int be_quiet, regex_type;
124         FILE *nonstdout;
125         char *outname, *hold_space;
126
127         /* List of input files */
128         int input_file_count, current_input_file;
129         FILE **input_file_list;
130
131         regmatch_t regmatch[10];
132         regex_t *previous_regex_ptr;
133
134         /* linked list of sed commands */
135         sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
136
137         /* Linked list of append lines */
138         llist_t *append_head;
139
140         char *add_cmd_line;
141
142         struct pipeline {
143                 char *buf;  /* Space to hold string */
144                 int idx;    /* Space used */
145                 int len;    /* Space allocated */
146         } pipeline;
147 } FIX_ALIASING;
148 #define G (*(struct globals*)&bb_common_bufsiz1)
149 struct BUG_G_too_big {
150         char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
151 };
152 #define INIT_G() do { \
153         G.sed_cmd_tail = &G.sed_cmd_head; \
154 } while (0)
155
156
157 #if ENABLE_FEATURE_CLEAN_UP
158 static void sed_free_and_close_stuff(void)
159 {
160         sed_cmd_t *sed_cmd = G.sed_cmd_head;
161
162         llist_free(G.append_head, free);
163
164         while (sed_cmd) {
165                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
166
167                 if (sed_cmd->sw_file)
168                         xprint_and_close_file(sed_cmd->sw_file);
169
170                 if (sed_cmd->beg_match) {
171                         regfree(sed_cmd->beg_match);
172                         free(sed_cmd->beg_match);
173                 }
174                 if (sed_cmd->end_match) {
175                         regfree(sed_cmd->end_match);
176                         free(sed_cmd->end_match);
177                 }
178                 if (sed_cmd->sub_match) {
179                         regfree(sed_cmd->sub_match);
180                         free(sed_cmd->sub_match);
181                 }
182                 free(sed_cmd->string);
183                 free(sed_cmd);
184                 sed_cmd = sed_cmd_next;
185         }
186
187         free(G.hold_space);
188
189         while (G.current_input_file < G.input_file_count)
190                 fclose(G.input_file_list[G.current_input_file++]);
191 }
192 #else
193 void sed_free_and_close_stuff(void);
194 #endif
195
196 /* If something bad happens during -i operation, delete temp file */
197
198 static void cleanup_outname(void)
199 {
200         if (G.outname) unlink(G.outname);
201 }
202
203 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
204
205 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
206 {
207         int i = 0;
208
209         while (i < len) {
210                 if (string[i] == '\\') {
211                         if (!to || string[i+1] == from) {
212                                 *dest++ = to ? to : string[i+1];
213                                 i += 2;
214                                 continue;
215                         }
216                         *dest++ = string[i++];
217                 }
218                 /* TODO: is it safe wrt a string with trailing '\\' ? */
219                 *dest++ = string[i++];
220         }
221         *dest = '\0';
222 }
223
224 static char *copy_parsing_escapes(const char *string, int len)
225 {
226         const char *s;
227         char *dest = xmalloc(len + 1);
228
229         /* sed recognizes \n */
230         /* GNU sed also recognizes \t and \r */
231         for (s = "\nn\tt\rr"; *s; s += 2) {
232                 parse_escapes(dest, string, len, s[1], s[0]);
233                 string = dest;
234                 len = strlen(dest);
235         }
236         return dest;
237 }
238
239
240 /*
241  * index_of_next_unescaped_regexp_delim - walks left to right through a string
242  * beginning at a specified index and returns the index of the next regular
243  * expression delimiter (typically a forward slash ('/')) not preceded by
244  * a backslash ('\').  A negative delimiter disables square bracket checking.
245  */
246 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
247 {
248         int bracket = -1;
249         int escaped = 0;
250         int idx = 0;
251         char ch;
252
253         if (delimiter < 0) {
254                 bracket--;
255                 delimiter = -delimiter;
256         }
257
258         for (; (ch = str[idx]) != '\0'; idx++) {
259                 if (bracket >= 0) {
260                         if (ch == ']'
261                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
262                         ) {
263                                 bracket = -1;
264                         }
265                 } else if (escaped)
266                         escaped = 0;
267                 else if (ch == '\\')
268                         escaped = 1;
269                 else if (bracket == -1 && ch == '[')
270                         bracket = idx;
271                 else if (ch == delimiter)
272                         return idx;
273         }
274
275         /* if we make it to here, we've hit the end of the string */
276         bb_error_msg_and_die("unmatched '%c'", delimiter);
277 }
278
279 /*
280  *  Returns the index of the third delimiter
281  */
282 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
283 {
284         const char *cmdstr_ptr = cmdstr;
285         char delimiter;
286         int idx = 0;
287
288         /* verify that the 's' or 'y' is followed by something.  That something
289          * (typically a 'slash') is now our regexp delimiter... */
290         if (*cmdstr == '\0')
291                 bb_error_msg_and_die("bad format in substitution expression");
292         delimiter = *cmdstr_ptr++;
293
294         /* save the match string */
295         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
296         *match = copy_parsing_escapes(cmdstr_ptr, idx);
297
298         /* save the replacement string */
299         cmdstr_ptr += idx + 1;
300         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
301         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
302
303         return ((cmdstr_ptr - cmdstr) + idx);
304 }
305
306 /*
307  * returns the index in the string just past where the address ends.
308  */
309 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
310 {
311         const char *pos = my_str;
312
313         if (isdigit(*my_str)) {
314                 *linenum = strtol(my_str, (char**)&pos, 10);
315                 /* endstr shouldnt ever equal NULL */
316         } else if (*my_str == '$') {
317                 *linenum = -1;
318                 pos++;
319         } else if (*my_str == '/' || *my_str == '\\') {
320                 int next;
321                 char delimiter;
322                 char *temp;
323
324                 delimiter = '/';
325                 if (*my_str == '\\') delimiter = *++pos;
326                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
327                 temp = copy_parsing_escapes(pos, next);
328                 *regex = xmalloc(sizeof(regex_t));
329                 xregcomp(*regex, temp, G.regex_type|REG_NEWLINE);
330                 free(temp);
331                 /* Move position to next character after last delimiter */
332                 pos += (next+1);
333         }
334         return pos - my_str;
335 }
336
337 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
338 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
339 {
340         int start = 0, idx, hack = 0;
341
342         /* Skip whitespace, then grab filename to end of line */
343         while (isspace(filecmdstr[start]))
344                 start++;
345         idx = start;
346         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
347                 idx++;
348
349         /* If lines glued together, put backslash back. */
350         if (filecmdstr[idx] == '\n')
351                 hack = 1;
352         if (idx == start)
353                 bb_error_msg_and_die("empty filename");
354         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
355         if (hack)
356                 (*retval)[idx] = '\\';
357
358         return idx;
359 }
360
361 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
362 {
363         int cflags = G.regex_type;
364         char *match;
365         int idx;
366
367         /*
368          * A substitution command should look something like this:
369          *    s/match/replace/ #gIpw
370          *    ||     |        |||
371          *    mandatory       optional
372          */
373         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
374
375         /* determine the number of back references in the match string */
376         /* Note: we compute this here rather than in the do_subst_command()
377          * function to save processor time, at the expense of a little more memory
378          * (4 bits) per sed_cmd */
379
380         /* process the flags */
381
382         sed_cmd->which_match = 1;
383         while (substr[++idx]) {
384                 /* Parse match number */
385                 if (isdigit(substr[idx])) {
386                         if (match[0] != '^') {
387                                 /* Match 0 treated as all, multiple matches we take the last one. */
388                                 const char *pos = substr + idx;
389 /* FIXME: error check? */
390                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
391                                 idx = pos - substr;
392                         }
393                         continue;
394                 }
395                 /* Skip spaces */
396                 if (isspace(substr[idx]))
397                         continue;
398
399                 switch (substr[idx]) {
400                 /* Replace all occurrences */
401                 case 'g':
402                         if (match[0] != '^')
403                                 sed_cmd->which_match = 0;
404                         break;
405                 /* Print pattern space */
406                 case 'p':
407                         sed_cmd->sub_p = 1;
408                         break;
409                 /* Write to file */
410                 case 'w':
411                 {
412                         char *temp;
413                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
414                         break;
415                 }
416                 /* Ignore case (gnu exension) */
417                 case 'I':
418                         cflags |= REG_ICASE;
419                         break;
420                 /* Comment */
421                 case '#':
422                         // while (substr[++idx]) continue;
423                         idx += strlen(substr + idx); // same
424                         /* Fall through */
425                 /* End of command */
426                 case ';':
427                 case '}':
428                         goto out;
429                 default:
430                         bb_error_msg_and_die("bad option in substitution expression");
431                 }
432         }
433  out:
434         /* compile the match string into a regex */
435         if (*match != '\0') {
436                 /* If match is empty, we use last regex used at runtime */
437                 sed_cmd->sub_match = xmalloc(sizeof(regex_t));
438                 xregcomp(sed_cmd->sub_match, match, cflags);
439         }
440         free(match);
441
442         return idx;
443 }
444
445 /*
446  *  Process the commands arguments
447  */
448 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
449 {
450         static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
451         enum {
452                 IDX_s = 0,
453                 IDX_a,
454                 IDX_i,
455                 IDX_c,
456                 IDX_r,
457                 IDX_w,
458                 IDX_colon,
459                 IDX_b,
460                 IDX_t,
461                 IDX_T,
462                 IDX_y,
463                 IDX_d,
464                 IDX_D,
465                 IDX_g,
466                 IDX_G,
467                 IDX_h,
468                 IDX_H,
469                 IDX_l,
470                 IDX_n,
471                 IDX_N,
472                 IDX_p,
473                 IDX_P,
474                 IDX_q,
475                 IDX_x,
476                 IDX_equal,
477                 IDX_lbrace,
478                 IDX_rbrace,
479                 IDX_nul
480         };
481         struct chk { char chk[sizeof(cmd_letters)-1 == IDX_nul ? 1 : -1]; };
482
483         unsigned idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
484
485         /* handle (s)ubstitution command */
486         if (idx == IDX_s) {
487                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
488         }
489         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
490         else if (idx <= IDX_c) { /* a,i,c */
491                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
492                         bb_error_msg_and_die("only a beginning address can be specified for edit commands");
493                 for (;;) {
494                         if (*cmdstr == '\n' || *cmdstr == '\\') {
495                                 cmdstr++;
496                                 break;
497                         }
498                         if (!isspace(*cmdstr))
499                                 break;
500                         cmdstr++;
501                 }
502                 sed_cmd->string = xstrdup(cmdstr);
503                 /* "\anychar" -> "anychar" */
504                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
505                 cmdstr += strlen(cmdstr);
506         }
507         /* handle file cmds: (r)ead */
508         else if (idx <= IDX_w) { /* r,w */
509                 if (sed_cmd->end_line || sed_cmd->end_match)
510                         bb_error_msg_and_die("command only uses one address");
511                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
512                 if (sed_cmd->cmd == 'w') {
513                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
514                         sed_cmd->sw_last_char = '\n';
515                 }
516         }
517         /* handle branch commands */
518         else if (idx <= IDX_T) { /* :,b,t,T */
519                 int length;
520
521                 cmdstr = skip_whitespace(cmdstr);
522                 length = strcspn(cmdstr, semicolon_whitespace);
523                 if (length) {
524                         sed_cmd->string = xstrndup(cmdstr, length);
525                         cmdstr += length;
526                 }
527         }
528         /* translation command */
529         else if (idx == IDX_y) {
530                 char *match, *replace;
531                 int i = cmdstr[0];
532
533                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
534                 /* \n already parsed, but \delimiter needs unescaping. */
535                 parse_escapes(match, match, strlen(match), i, i);
536                 parse_escapes(replace, replace, strlen(replace), i, i);
537
538                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
539                 for (i = 0; match[i] && replace[i]; i++) {
540                         sed_cmd->string[i*2] = match[i];
541                         sed_cmd->string[i*2+1] = replace[i];
542                 }
543                 free(match);
544                 free(replace);
545         }
546         /* if it wasnt a single-letter command that takes no arguments
547          * then it must be an invalid command.
548          */
549         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
550                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
551         }
552
553         /* give back whatever's left over */
554         return cmdstr;
555 }
556
557
558 /* Parse address+command sets, skipping comment lines. */
559
560 static void add_cmd(const char *cmdstr)
561 {
562         sed_cmd_t *sed_cmd;
563         unsigned len, n;
564
565         /* Append this line to any unfinished line from last time. */
566         if (G.add_cmd_line) {
567                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
568                 free(G.add_cmd_line);
569                 cmdstr = G.add_cmd_line = tp;
570         }
571
572         /* If this line ends with unescaped backslash, request next line. */
573         n = len = strlen(cmdstr);
574         while (n && cmdstr[n-1] == '\\')
575                 n--;
576         if ((len - n) & 1) { /* if odd number of trailing backslashes */
577                 if (!G.add_cmd_line)
578                         G.add_cmd_line = xstrdup(cmdstr);
579                 G.add_cmd_line[len-1] = '\0';
580                 return;
581         }
582
583         /* Loop parsing all commands in this line. */
584         while (*cmdstr) {
585                 /* Skip leading whitespace and semicolons */
586                 cmdstr += strspn(cmdstr, semicolon_whitespace);
587
588                 /* If no more commands, exit. */
589                 if (!*cmdstr) break;
590
591                 /* if this is a comment, jump past it and keep going */
592                 if (*cmdstr == '#') {
593                         /* "#n" is the same as using -n on the command line */
594                         if (cmdstr[1] == 'n')
595                                 G.be_quiet++;
596                         cmdstr = strpbrk(cmdstr, "\n\r");
597                         if (!cmdstr) break;
598                         continue;
599                 }
600
601                 /* parse the command
602                  * format is: [addr][,addr][!]cmd
603                  *            |----||-----||-|
604                  *            part1 part2  part3
605                  */
606
607                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
608
609                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
610                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
611                 sed_cmd->beg_line_orig = sed_cmd->beg_line;
612
613                 /* second part (if present) will begin with a comma */
614                 if (*cmdstr == ',') {
615                         int idx;
616
617                         cmdstr++;
618                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
619                         if (!idx)
620                                 bb_error_msg_and_die("no address after comma");
621                         cmdstr += idx;
622                 }
623
624                 /* skip whitespace before the command */
625                 cmdstr = skip_whitespace(cmdstr);
626
627                 /* Check for inversion flag */
628                 if (*cmdstr == '!') {
629                         sed_cmd->invert = 1;
630                         cmdstr++;
631
632                         /* skip whitespace before the command */
633                         cmdstr = skip_whitespace(cmdstr);
634                 }
635
636                 /* last part (mandatory) will be a command */
637                 if (!*cmdstr)
638                         bb_error_msg_and_die("missing command");
639                 sed_cmd->cmd = *cmdstr++;
640                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
641
642                 /* Add the command to the command array */
643                 *G.sed_cmd_tail = sed_cmd;
644                 G.sed_cmd_tail = &sed_cmd->next;
645         }
646
647         /* If we glued multiple lines together, free the memory. */
648         free(G.add_cmd_line);
649         G.add_cmd_line = NULL;
650 }
651
652 /* Append to a string, reallocating memory as necessary. */
653
654 #define PIPE_GROW 64
655
656 static void pipe_putc(char c)
657 {
658         if (G.pipeline.idx == G.pipeline.len) {
659                 G.pipeline.buf = xrealloc(G.pipeline.buf,
660                                 G.pipeline.len + PIPE_GROW);
661                 G.pipeline.len += PIPE_GROW;
662         }
663         G.pipeline.buf[G.pipeline.idx++] = c;
664 }
665
666 static void do_subst_w_backrefs(char *line, char *replace)
667 {
668         int i, j;
669
670         /* go through the replacement string */
671         for (i = 0; replace[i]; i++) {
672                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
673                 if (replace[i] == '\\') {
674                         unsigned backref = replace[++i] - '0';
675                         if (backref <= 9) {
676                                 /* print out the text held in G.regmatch[backref] */
677                                 if (G.regmatch[backref].rm_so != -1) {
678                                         j = G.regmatch[backref].rm_so;
679                                         while (j < G.regmatch[backref].rm_eo)
680                                                 pipe_putc(line[j++]);
681                                 }
682                                 continue;
683                         }
684                         /* I _think_ it is impossible to get '\' to be
685                          * the last char in replace string. Thus we dont check
686                          * for replace[i] == NUL. (counterexample anyone?) */
687                         /* if we find a backslash escaped character, print the character */
688                         pipe_putc(replace[i]);
689                         continue;
690                 }
691                 /* if we find an unescaped '&' print out the whole matched text. */
692                 if (replace[i] == '&') {
693                         j = G.regmatch[0].rm_so;
694                         while (j < G.regmatch[0].rm_eo)
695                                 pipe_putc(line[j++]);
696                         continue;
697                 }
698                 /* Otherwise just output the character. */
699                 pipe_putc(replace[i]);
700         }
701 }
702
703 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
704 {
705         char *line = *line_p;
706         int altered = 0;
707         unsigned match_count = 0;
708         regex_t *current_regex;
709
710         current_regex = sed_cmd->sub_match;
711         /* Handle empty regex. */
712         if (!current_regex) {
713                 current_regex = G.previous_regex_ptr;
714                 if (!current_regex)
715                         bb_error_msg_and_die("no previous regexp");
716         }
717         G.previous_regex_ptr = current_regex;
718
719         /* Find the first match */
720         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0))
721                 return 0;
722
723         /* Initialize temporary output buffer. */
724         G.pipeline.buf = xmalloc(PIPE_GROW);
725         G.pipeline.len = PIPE_GROW;
726         G.pipeline.idx = 0;
727
728         /* Now loop through, substituting for matches */
729         do {
730                 int i;
731
732                 /* Work around bug in glibc regexec, demonstrated by:
733                    echo " a.b" | busybox sed 's [^ .]* x g'
734                    The match_count check is so not to break
735                    echo "hi" | busybox sed 's/^/!/g' */
736                 if (!G.regmatch[0].rm_so && !G.regmatch[0].rm_eo && match_count) {
737                         pipe_putc(*line++);
738                         continue;
739                 }
740
741                 match_count++;
742
743                 /* If we aren't interested in this match, output old line to
744                    end of match and continue */
745                 if (sed_cmd->which_match
746                  && (sed_cmd->which_match != match_count)
747                 ) {
748                         for (i = 0; i < G.regmatch[0].rm_eo; i++)
749                                 pipe_putc(*line++);
750                         continue;
751                 }
752
753                 /* print everything before the match */
754                 for (i = 0; i < G.regmatch[0].rm_so; i++)
755                         pipe_putc(line[i]);
756
757                 /* then print the substitution string */
758                 do_subst_w_backrefs(line, sed_cmd->string);
759
760                 /* advance past the match */
761                 line += G.regmatch[0].rm_eo;
762                 /* flag that something has changed */
763                 altered++;
764
765                 /* if we're not doing this globally, get out now */
766                 if (sed_cmd->which_match)
767                         break;
768
769 //maybe (G.regmatch[0].rm_eo ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
770         } while (*line && regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
771
772         /* Copy rest of string into output pipeline */
773         while (1) {
774                 char c = *line++;
775                 pipe_putc(c);
776                 if (c == '\0')
777                         break;
778         }
779
780         free(*line_p);
781         *line_p = G.pipeline.buf;
782         return altered;
783 }
784
785 /* Set command pointer to point to this label.  (Does not handle null label.) */
786 static sed_cmd_t *branch_to(char *label)
787 {
788         sed_cmd_t *sed_cmd;
789
790         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
791                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
792                         return sed_cmd;
793                 }
794         }
795         bb_error_msg_and_die("can't find label for jump to '%s'", label);
796 }
797
798 static void append(char *s)
799 {
800         llist_add_to_end(&G.append_head, xstrdup(s));
801 }
802
803 static void flush_append(void)
804 {
805         char *data;
806
807         /* Output appended lines. */
808         while ((data = (char *)llist_pop(&G.append_head))) {
809                 fprintf(G.nonstdout, "%s\n", data);
810                 free(data);
811         }
812 }
813
814 static void add_input_file(FILE *file)
815 {
816         G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count);
817         G.input_file_list[G.input_file_count++] = file;
818 }
819
820 /* Get next line of input from G.input_file_list, flushing append buffer and
821  * noting if we ran out of files without a newline on the last line we read.
822  */
823 enum {
824         NO_EOL_CHAR = 1,
825         LAST_IS_NUL = 2,
826 };
827 static char *get_next_line(char *gets_char)
828 {
829         char *temp = NULL;
830         int len;
831         char gc;
832
833         flush_append();
834
835         /* will be returned if last line in the file
836          * doesn't end with either '\n' or '\0' */
837         gc = NO_EOL_CHAR;
838         while (G.current_input_file < G.input_file_count) {
839                 FILE *fp = G.input_file_list[G.current_input_file];
840                 /* Read line up to a newline or NUL byte, inclusive,
841                  * return malloc'ed char[]. length of the chunk read
842                  * is stored in len. NULL if EOF/error */
843                 temp = bb_get_chunk_from_file(fp, &len);
844                 if (temp) {
845                         /* len > 0 here, it's ok to do temp[len-1] */
846                         char c = temp[len-1];
847                         if (c == '\n' || c == '\0') {
848                                 temp[len-1] = '\0';
849                                 gc = c;
850                                 if (c == '\0') {
851                                         int ch = fgetc(fp);
852                                         if (ch != EOF)
853                                                 ungetc(ch, fp);
854                                         else
855                                                 gc = LAST_IS_NUL;
856                                 }
857                         }
858                         /* else we put NO_EOL_CHAR into *gets_char */
859                         break;
860
861                 /* NB: I had the idea of peeking next file(s) and returning
862                  * NO_EOL_CHAR only if it is the *last* non-empty
863                  * input file. But there is a case where this won't work:
864                  * file1: "a woo\nb woo"
865                  * file2: "c no\nd no"
866                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
867                  * (note: *no* newline after "b bang"!) */
868                 }
869                 /* Close this file and advance to next one */
870                 fclose(fp);
871                 G.current_input_file++;
872         }
873         *gets_char = gc;
874         return temp;
875 }
876
877 /* Output line of text. */
878 /* Note:
879  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
880  * Without them, we had this:
881  * echo -n thingy >z1
882  * echo -n again >z2
883  * >znull
884  * sed "s/i/z/" z1 z2 znull | hexdump -vC
885  * output:
886  * gnu sed 4.1.5:
887  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
888  * bbox:
889  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
890  */
891 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
892 {
893         char lpc = *last_puts_char;
894
895         /* Need to insert a '\n' between two files because first file's
896          * last line wasn't terminated? */
897         if (lpc != '\n' && lpc != '\0') {
898                 fputc('\n', file);
899                 lpc = '\n';
900         }
901         fputs(s, file);
902
903         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
904         if (s[0])
905                 lpc = 'x';
906
907         /* had trailing '\0' and it was last char of file? */
908         if (last_gets_char == LAST_IS_NUL) {
909                 fputc('\0', file);
910                 lpc = 'x'; /* */
911         } else
912         /* had trailing '\n' or '\0'? */
913         if (last_gets_char != NO_EOL_CHAR) {
914                 fputc(last_gets_char, file);
915                 lpc = last_gets_char;
916         }
917
918         if (ferror(file)) {
919                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
920                 bb_error_msg_and_die(bb_msg_write_error);
921         }
922         *last_puts_char = lpc;
923 }
924
925 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
926
927 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
928 {
929         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
930         if (retval)
931                 G.previous_regex_ptr = sed_cmd->beg_match;
932         return retval;
933 }
934
935 /* Process all the lines in all the files */
936
937 static void process_files(void)
938 {
939         char *pattern_space, *next_line;
940         int linenum = 0;
941         char last_puts_char = '\n';
942         char last_gets_char, next_gets_char;
943         sed_cmd_t *sed_cmd;
944         int substituted;
945
946         /* Prime the pump */
947         next_line = get_next_line(&next_gets_char);
948
949         /* Go through every line in each file */
950  again:
951         substituted = 0;
952
953         /* Advance to next line.  Stop if out of lines. */
954         pattern_space = next_line;
955         if (!pattern_space)
956                 return;
957         last_gets_char = next_gets_char;
958
959         /* Read one line in advance so we can act on the last line,
960          * the '$' address */
961         next_line = get_next_line(&next_gets_char);
962         linenum++;
963
964         /* For every line, go through all the commands */
965  restart:
966         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
967                 int old_matched, matched;
968
969                 old_matched = sed_cmd->in_match;
970
971                 /* Determine if this command matches this line: */
972
973                 dbg("match1:%d", sed_cmd->in_match);
974                 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
975                                 && !sed_cmd->beg_match && !sed_cmd->end_match));
976                 dbg("match3:%d", (sed_cmd->beg_line > 0
977                         && (sed_cmd->end_line || sed_cmd->end_match
978                             ? (sed_cmd->beg_line <= linenum)
979                             : (sed_cmd->beg_line == linenum)
980                             )
981                         ));
982                 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
983                 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
984
985                 /* Are we continuing a previous multi-line match? */
986                 sed_cmd->in_match = sed_cmd->in_match
987                         /* Or is no range necessary? */
988                         || (!sed_cmd->beg_line && !sed_cmd->end_line
989                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
990                         /* Or did we match the start of a numerical range? */
991                         || (sed_cmd->beg_line > 0
992                             && (sed_cmd->end_line || sed_cmd->end_match
993                                   /* note: even if end is numeric and is < linenum too,
994                                    * GNU sed matches! We match too, therefore we don't
995                                    * check here that linenum <= end.
996                                    * Example:
997                                    * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
998                                    * first three input lines are deleted;
999                                    * 4th line is matched and printed
1000                                    * by "2,3" (!) and by "4" ranges
1001                                    */
1002                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
1003                                 : (sed_cmd->beg_line == linenum)    /* N */
1004                                 )
1005                             )
1006                         /* Or does this line match our begin address regex? */
1007                         || (beg_match(sed_cmd, pattern_space))
1008                         /* Or did we match last line of input? */
1009                         || (sed_cmd->beg_line == -1 && next_line == NULL);
1010
1011                 /* Snapshot the value */
1012                 matched = sed_cmd->in_match;
1013
1014                 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1015                         sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
1016
1017                 /* Is this line the end of the current match? */
1018
1019                 if (matched) {
1020                         /* once matched, "n,xxx" range is dead, disabling it */
1021                         if (sed_cmd->beg_line > 0) {
1022                                 sed_cmd->beg_line = -2;
1023                         }
1024                         sed_cmd->in_match = !(
1025                                 /* has the ending line come, or is this a single address command? */
1026                                 (sed_cmd->end_line
1027                                         ? sed_cmd->end_line == -1
1028                                                 ? !next_line
1029                                                 : (sed_cmd->end_line <= linenum)
1030                                         : !sed_cmd->end_match
1031                                 )
1032                                 /* or does this line matches our last address regex */
1033                                 || (sed_cmd->end_match && old_matched
1034                                      && (regexec(sed_cmd->end_match,
1035                                                  pattern_space, 0, NULL, 0) == 0)
1036                                 )
1037                         );
1038                 }
1039
1040                 /* Skip blocks of commands we didn't match */
1041                 if (sed_cmd->cmd == '{') {
1042                         if (sed_cmd->invert ? matched : !matched) {
1043                                 unsigned nest_cnt = 0;
1044                                 while (1) {
1045                                         if (sed_cmd->cmd == '{')
1046                                                 nest_cnt++;
1047                                         if (sed_cmd->cmd == '}') {
1048                                                 nest_cnt--;
1049                                                 if (nest_cnt == 0)
1050                                                         break;
1051                                         }
1052                                         sed_cmd = sed_cmd->next;
1053                                         if (!sed_cmd)
1054                                                 bb_error_msg_and_die("unterminated {");
1055                                 }
1056                         }
1057                         continue;
1058                 }
1059
1060                 /* Okay, so did this line match? */
1061                 if (sed_cmd->invert ? matched : !matched)
1062                         continue; /* no */
1063
1064                 /* Update last used regex in case a blank substitute BRE is found */
1065                 if (sed_cmd->beg_match) {
1066                         G.previous_regex_ptr = sed_cmd->beg_match;
1067                 }
1068
1069                 /* actual sedding */
1070                 //bb_error_msg("pattern_space:'%s' next_line:'%s' cmd:%c",
1071                 //pattern_space, next_line, sed_cmd->cmd);
1072                 switch (sed_cmd->cmd) {
1073
1074                 /* Print line number */
1075                 case '=':
1076                         fprintf(G.nonstdout, "%d\n", linenum);
1077                         break;
1078
1079                 /* Write the current pattern space up to the first newline */
1080                 case 'P':
1081                 {
1082                         char *tmp = strchr(pattern_space, '\n');
1083                         if (tmp) {
1084                                 *tmp = '\0';
1085                                 /* TODO: explain why '\n' below */
1086                                 sed_puts(pattern_space, '\n');
1087                                 *tmp = '\n';
1088                                 break;
1089                         }
1090                         /* Fall Through */
1091                 }
1092
1093                 /* Write the current pattern space to output */
1094                 case 'p':
1095                         /* NB: we print this _before_ the last line
1096                          * (of current file) is printed. Even if
1097                          * that line is nonterminated, we print
1098                          * '\n' here (gnu sed does the same) */
1099                         sed_puts(pattern_space, '\n');
1100                         break;
1101                 /* Delete up through first newline */
1102                 case 'D':
1103                 {
1104                         char *tmp = strchr(pattern_space, '\n');
1105                         if (tmp) {
1106                                 overlapping_strcpy(pattern_space, tmp + 1);
1107                                 goto restart;
1108                         }
1109                 }
1110                 /* discard this line. */
1111                 case 'd':
1112                         goto discard_line;
1113
1114                 /* Substitute with regex */
1115                 case 's':
1116                         if (!do_subst_command(sed_cmd, &pattern_space))
1117                                 break;
1118                         substituted |= 1;
1119
1120                         /* handle p option */
1121                         if (sed_cmd->sub_p)
1122                                 sed_puts(pattern_space, last_gets_char);
1123                         /* handle w option */
1124                         if (sed_cmd->sw_file)
1125                                 puts_maybe_newline(
1126                                         pattern_space, sed_cmd->sw_file,
1127                                         &sed_cmd->sw_last_char, last_gets_char);
1128                         break;
1129
1130                 /* Append line to linked list to be printed later */
1131                 case 'a':
1132                         append(sed_cmd->string);
1133                         break;
1134
1135                 /* Insert text before this line */
1136                 case 'i':
1137                         sed_puts(sed_cmd->string, '\n');
1138                         break;
1139
1140                 /* Cut and paste text (replace) */
1141                 case 'c':
1142                         /* Only triggers on last line of a matching range. */
1143                         if (!sed_cmd->in_match)
1144                                 sed_puts(sed_cmd->string, '\n');
1145                         goto discard_line;
1146
1147                 /* Read file, append contents to output */
1148                 case 'r':
1149                 {
1150                         FILE *rfile;
1151                         rfile = fopen_for_read(sed_cmd->string);
1152                         if (rfile) {
1153                                 char *line;
1154
1155                                 while ((line = xmalloc_fgetline(rfile))
1156                                                 != NULL)
1157                                         append(line);
1158                                 xprint_and_close_file(rfile);
1159                         }
1160
1161                         break;
1162                 }
1163
1164                 /* Write pattern space to file. */
1165                 case 'w':
1166                         puts_maybe_newline(
1167                                 pattern_space, sed_cmd->sw_file,
1168                                 &sed_cmd->sw_last_char, last_gets_char);
1169                         break;
1170
1171                 /* Read next line from input */
1172                 case 'n':
1173                         if (!G.be_quiet)
1174                                 sed_puts(pattern_space, last_gets_char);
1175                         if (next_line) {
1176                                 free(pattern_space);
1177                                 pattern_space = next_line;
1178                                 last_gets_char = next_gets_char;
1179                                 next_line = get_next_line(&next_gets_char);
1180                                 substituted = 0;
1181                                 linenum++;
1182                                 break;
1183                         }
1184                         /* fall through */
1185
1186                 /* Quit.  End of script, end of input. */
1187                 case 'q':
1188                         /* Exit the outer while loop */
1189                         free(next_line);
1190                         next_line = NULL;
1191                         goto discard_commands;
1192
1193                 /* Append the next line to the current line */
1194                 case 'N':
1195                 {
1196                         int len;
1197                         /* If no next line, jump to end of script and exit. */
1198                         /* http://www.gnu.org/software/sed/manual/sed.html:
1199                          * "Most versions of sed exit without printing anything
1200                          * when the N command is issued on the last line of
1201                          * a file. GNU sed prints pattern space before exiting
1202                          * unless of course the -n command switch has been
1203                          * specified. This choice is by design."
1204                          */
1205                         if (next_line == NULL) {
1206                                 //goto discard_line;
1207                                 goto discard_commands; /* GNU behavior */
1208                         }
1209                         /* Append next_line, read new next_line. */
1210                         len = strlen(pattern_space);
1211                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1212                         pattern_space[len] = '\n';
1213                         strcpy(pattern_space + len+1, next_line);
1214                         last_gets_char = next_gets_char;
1215                         next_line = get_next_line(&next_gets_char);
1216                         linenum++;
1217                         break;
1218                 }
1219
1220                 /* Test/branch if substitution occurred */
1221                 case 't':
1222                         if (!substituted) break;
1223                         substituted = 0;
1224                         /* Fall through */
1225                 /* Test/branch if substitution didn't occur */
1226                 case 'T':
1227                         if (substituted) break;
1228                         /* Fall through */
1229                 /* Branch to label */
1230                 case 'b':
1231                         if (!sed_cmd->string) goto discard_commands;
1232                         else sed_cmd = branch_to(sed_cmd->string);
1233                         break;
1234                 /* Transliterate characters */
1235                 case 'y':
1236                 {
1237                         int i, j;
1238                         for (i = 0; pattern_space[i]; i++) {
1239                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1240                                         if (pattern_space[i] == sed_cmd->string[j]) {
1241                                                 pattern_space[i] = sed_cmd->string[j + 1];
1242                                                 break;
1243                                         }
1244                                 }
1245                         }
1246
1247                         break;
1248                 }
1249                 case 'g':       /* Replace pattern space with hold space */
1250                         free(pattern_space);
1251                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1252                         break;
1253                 case 'G':       /* Append newline and hold space to pattern space */
1254                 {
1255                         int pattern_space_size = 2;
1256                         int hold_space_size = 0;
1257
1258                         if (pattern_space)
1259                                 pattern_space_size += strlen(pattern_space);
1260                         if (G.hold_space)
1261                                 hold_space_size = strlen(G.hold_space);
1262                         pattern_space = xrealloc(pattern_space,
1263                                         pattern_space_size + hold_space_size);
1264                         if (pattern_space_size == 2)
1265                                 pattern_space[0] = 0;
1266                         strcat(pattern_space, "\n");
1267                         if (G.hold_space)
1268                                 strcat(pattern_space, G.hold_space);
1269                         last_gets_char = '\n';
1270
1271                         break;
1272                 }
1273                 case 'h':       /* Replace hold space with pattern space */
1274                         free(G.hold_space);
1275                         G.hold_space = xstrdup(pattern_space);
1276                         break;
1277                 case 'H':       /* Append newline and pattern space to hold space */
1278                 {
1279                         int hold_space_size = 2;
1280                         int pattern_space_size = 0;
1281
1282                         if (G.hold_space)
1283                                 hold_space_size += strlen(G.hold_space);
1284                         if (pattern_space)
1285                                 pattern_space_size = strlen(pattern_space);
1286                         G.hold_space = xrealloc(G.hold_space,
1287                                         hold_space_size + pattern_space_size);
1288
1289                         if (hold_space_size == 2)
1290                                 *G.hold_space = 0;
1291                         strcat(G.hold_space, "\n");
1292                         if (pattern_space)
1293                                 strcat(G.hold_space, pattern_space);
1294
1295                         break;
1296                 }
1297                 case 'x': /* Exchange hold and pattern space */
1298                 {
1299                         char *tmp = pattern_space;
1300                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1301                         last_gets_char = '\n';
1302                         G.hold_space = tmp;
1303                         break;
1304                 }
1305                 } /* switch */
1306         } /* for each cmd */
1307
1308         /*
1309          * Exit point from sedding...
1310          */
1311  discard_commands:
1312         /* we will print the line unless we were told to be quiet ('-n')
1313            or if the line was suppressed (ala 'd'elete) */
1314         if (!G.be_quiet)
1315                 sed_puts(pattern_space, last_gets_char);
1316
1317         /* Delete and such jump here. */
1318  discard_line:
1319         flush_append();
1320         free(pattern_space);
1321
1322         goto again;
1323 }
1324
1325 /* It is possible to have a command line argument with embedded
1326  * newlines.  This counts as multiple command lines.
1327  * However, newline can be escaped: 's/e/z\<newline>z/'
1328  * We check for this.
1329  */
1330
1331 static void add_cmd_block(char *cmdstr)
1332 {
1333         char *sv, *eol;
1334
1335         cmdstr = sv = xstrdup(cmdstr);
1336         do {
1337                 eol = strchr(cmdstr, '\n');
1338  next:
1339                 if (eol) {
1340                         /* Count preceding slashes */
1341                         int slashes = 0;
1342                         char *sl = eol;
1343
1344                         while (sl != cmdstr && *--sl == '\\')
1345                                 slashes++;
1346                         /* Odd number of preceding slashes - newline is escaped */
1347                         if (slashes & 1) {
1348                                 overlapping_strcpy(eol - 1, eol);
1349                                 eol = strchr(eol, '\n');
1350                                 goto next;
1351                         }
1352                         *eol = '\0';
1353                 }
1354                 add_cmd(cmdstr);
1355                 cmdstr = eol + 1;
1356         } while (eol);
1357         free(sv);
1358 }
1359
1360 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1361 int sed_main(int argc UNUSED_PARAM, char **argv)
1362 {
1363         unsigned opt;
1364         llist_t *opt_e, *opt_f;
1365         int status = EXIT_SUCCESS;
1366
1367         INIT_G();
1368
1369         /* destroy command strings on exit */
1370         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1371
1372         /* Lie to autoconf when it starts asking stupid questions. */
1373         if (argv[1] && !strcmp(argv[1], "--version")) {
1374                 puts("This is not GNU sed version 4.0");
1375                 return 0;
1376         }
1377
1378         /* do normal option parsing */
1379         opt_e = opt_f = NULL;
1380         opt_complementary = "e::f::" /* can occur multiple times */
1381                             "nn"; /* count -n */
1382         /* -i must be first, to match OPT_in_place definition */
1383         opt = getopt32(argv, "irne:f:", &opt_e, &opt_f,
1384                             &G.be_quiet); /* counter for -n */
1385         //argc -= optind;
1386         argv += optind;
1387         if (opt & OPT_in_place) { // -i
1388                 atexit(cleanup_outname);
1389         }
1390         if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r
1391         //if (opt & 0x4) G.be_quiet++; // -n
1392         while (opt_e) { // -e
1393                 add_cmd_block(llist_pop(&opt_e));
1394         }
1395         while (opt_f) { // -f
1396                 char *line;
1397                 FILE *cmdfile;
1398                 cmdfile = xfopen_for_read(llist_pop(&opt_f));
1399                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1400                         add_cmd(line);
1401                         free(line);
1402                 }
1403                 fclose(cmdfile);
1404         }
1405         /* if we didn't get a pattern from -e or -f, use argv[0] */
1406         if (!(opt & 0x18)) {
1407                 if (!*argv)
1408                         bb_show_usage();
1409                 add_cmd_block(*argv++);
1410         }
1411         /* Flush any unfinished commands. */
1412         add_cmd("");
1413
1414         /* By default, we write to stdout */
1415         G.nonstdout = stdout;
1416
1417         /* argv[0..(argc-1)] should be names of file to process. If no
1418          * files were specified or '-' was specified, take input from stdin.
1419          * Otherwise, we process all the files specified. */
1420         if (argv[0] == NULL) {
1421                 if (opt & OPT_in_place)
1422                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1423                 add_input_file(stdin);
1424         } else {
1425                 int i;
1426
1427                 for (i = 0; argv[i]; i++) {
1428                         struct stat statbuf;
1429                         int nonstdoutfd;
1430                         FILE *file;
1431                         sed_cmd_t *sed_cmd;
1432
1433                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1434                                 add_input_file(stdin);
1435                                 process_files();
1436                                 continue;
1437                         }
1438                         file = fopen_or_warn(argv[i], "r");
1439                         if (!file) {
1440                                 status = EXIT_FAILURE;
1441                                 continue;
1442                         }
1443                         add_input_file(file);
1444                         if (!(opt & OPT_in_place)) {
1445                                 continue;
1446                         }
1447
1448                         /* -i: process each FILE separately: */
1449
1450                         G.outname = xasprintf("%sXXXXXX", argv[i]);
1451                         nonstdoutfd = xmkstemp(G.outname);
1452                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1453
1454                         /* Set permissions/owner of output file */
1455                         fstat(fileno(file), &statbuf);
1456                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1457                          * but GNU sed 4.2.1 does not preserve them either */
1458                         fchmod(nonstdoutfd, statbuf.st_mode);
1459                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1460
1461                         process_files();
1462                         fclose(G.nonstdout);
1463                         G.nonstdout = stdout;
1464
1465                         /* unlink(argv[i]); */
1466                         xrename(G.outname, argv[i]);
1467                         free(G.outname);
1468                         G.outname = NULL;
1469
1470                         /* Re-enable disabled range matches */
1471                         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1472                                 sed_cmd->beg_line = sed_cmd->beg_line_orig;
1473                         }
1474                 }
1475                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1476                  * if (G.current_input_file >= G.input_file_count)
1477                  *      return status;
1478                  * but it's not needed since process_files() works correctly
1479                  * in this case too. */
1480         }
1481         process_files();
1482
1483         return status;
1484 }