1 /* vi: set sw=4 ts=4: */
3 * sed.c - very minimalist version of sed
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>
11 * MAINTAINER: Rob Landley <rob@landley.net>
13 * Licensed under GPL version 2, see file LICENSE in this tarball for details.
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.
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 (bbg.sed_cmd_head/bbg.sed_cmd_tail).
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.
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.
33 sed_main() is where external code calls into this, with a command line.
38 Supported features and commands in this version of sed:
41 - address matching: num|/matchstr/[,num|/matchstr/|$]command
42 - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
43 - edit commands: (a)ppend, (i)nsert, (c)hange
44 - file commands: (r)ead
45 - backreferences in substitution expressions (\0, \1, \2...\9)
46 - grouped commands: {cmd1;cmd2}
47 - transliteration (y/source-chars/dest-chars/)
48 - pattern space hold space storing / swapping (g, h, x)
49 - labels / branching (: label, b, t, T)
51 (Note: Specifying an address (range) to match is *optional*; commands
52 default to the whole pattern space if no specific address match was
56 - Create a wrapper around regex to make libc's regex conform with sed
58 Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
64 /* Each sed command turns into one of these structures. */
65 typedef struct sed_cmd_s {
66 /* Ordered by alignment requirements: currently 36 bytes on x86 */
69 regex_t *beg_match; /* sed -e '/match/cmd' */
70 regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
71 regex_t *sub_match; /* For 's/sub_match/string/' */
72 int beg_line; /* 'sed 1p' 0 == apply commands to all lines */
73 int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
75 FILE *file; /* File (sw) command writes to, -1 for none. */
76 char *string; /* Data string for (saicytb) commands. */
78 unsigned short which_match; /* (s) Which match to replace (0 for all) */
80 /* Bitfields (gcc won't group them if we don't) */
81 unsigned int invert:1; /* the '!' after the address */
82 unsigned int in_match:1; /* Next line also included in match? */
83 unsigned int sub_p:1; /* (s) print option */
85 int last_char; /* Last line written by (sw) had no '\n' */
88 char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
89 struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
92 static const char *const semicolon_whitespace = "; \n\r\t\v";
97 int be_quiet, in_place, regex_type;
99 char *outname, *hold_space;
101 /* List of input files */
102 int input_file_count,current_input_file;
103 FILE **input_file_list;
105 regmatch_t regmatch[10];
106 regex_t *previous_regex_ptr;
108 /* linked list of sed commands */
109 sed_cmd_t sed_cmd_head, *sed_cmd_tail;
111 /* Linked list of append lines */
112 llist_t *append_head;
117 char *buf; /* Space to hold string */
118 int idx; /* Space used */
119 int len; /* Space allocated */
124 void sed_free_and_close_stuff(void);
125 #if ENABLE_FEATURE_CLEAN_UP
126 void sed_free_and_close_stuff(void)
128 sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
130 llist_free(bbg.append_head, free);
133 sed_cmd_t *sed_cmd_next = sed_cmd->next;
136 xprint_and_close_file(sed_cmd->file);
138 if (sed_cmd->beg_match) {
139 regfree(sed_cmd->beg_match);
140 free(sed_cmd->beg_match);
142 if (sed_cmd->end_match) {
143 regfree(sed_cmd->end_match);
144 free(sed_cmd->end_match);
146 if (sed_cmd->sub_match) {
147 regfree(sed_cmd->sub_match);
148 free(sed_cmd->sub_match);
150 free(sed_cmd->string);
152 sed_cmd = sed_cmd_next;
155 if (bbg.hold_space) free(bbg.hold_space);
157 while (bbg.current_input_file < bbg.input_file_count)
158 fclose(bbg.input_file_list[bbg.current_input_file++]);
162 /* If something bad happens during -i operation, delete temp file */
164 static void cleanup_outname(void)
166 if (bbg.outname) unlink(bbg.outname);
169 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
171 static void parse_escapes(char *dest, char *string, int len, char from, char to)
176 if (string[i] == '\\') {
177 if (!to || string[i+1] == from) {
178 *(dest++) = to ? to : string[i+1];
181 } else *(dest++) = string[i++];
183 *(dest++) = string[i++];
188 static char *copy_parsing_escapes(char *string, int len)
190 char *dest = xmalloc(len+1);
192 parse_escapes(dest,string,len,'n','\n');
198 * index_of_next_unescaped_regexp_delim - walks left to right through a string
199 * beginning at a specified index and returns the index of the next regular
200 * expression delimiter (typically a forward * slash ('/')) not preceded by
201 * a backslash ('\'). A negative delimiter disables square bracket checking.
203 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
215 for (; (ch = str[idx]); idx++) {
217 if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
218 && str[idx - 1] == '^')))
224 else if (bracket == -1 && ch == '[')
226 else if (ch == delimiter)
230 /* if we make it to here, we've hit the end of the string */
231 bb_error_msg_and_die("unmatched '%c'",delimiter);
235 * Returns the index of the third delimiter
237 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
239 char *cmdstr_ptr = cmdstr;
243 /* verify that the 's' or 'y' is followed by something. That something
244 * (typically a 'slash') is now our regexp delimiter... */
246 bb_error_msg_and_die("bad format in substitution expression");
247 delimiter = *cmdstr_ptr++;
249 /* save the match string */
250 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
251 *match = copy_parsing_escapes(cmdstr_ptr, idx);
253 /* save the replacement string */
254 cmdstr_ptr += idx + 1;
255 idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
256 *replace = copy_parsing_escapes(cmdstr_ptr, idx);
258 return ((cmdstr_ptr - cmdstr) + idx);
262 * returns the index in the string just past where the address ends.
264 static int get_address(char *my_str, int *linenum, regex_t ** regex)
268 if (isdigit(*my_str)) {
269 *linenum = strtol(my_str, &pos, 10);
270 /* endstr shouldnt ever equal NULL */
271 } else if (*my_str == '$') {
274 } else if (*my_str == '/' || *my_str == '\\') {
280 if (*my_str == '\\') delimiter = *++pos;
281 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
282 temp = copy_parsing_escapes(pos,next);
283 *regex = xmalloc(sizeof(regex_t));
284 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
286 /* Move position to next character after last delimiter */
292 /* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
293 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
295 int start = 0, idx, hack = 0;
297 /* Skip whitespace, then grab filename to end of line */
298 while (isspace(filecmdstr[start])) start++;
300 while (filecmdstr[idx] && filecmdstr[idx] != '\n') idx++;
302 /* If lines glued together, put backslash back. */
303 if (filecmdstr[idx] == '\n') hack = 1;
305 bb_error_msg_and_die("empty filename");
306 *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
307 if (hack) (*retval)[idx] = '\\';
312 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
314 int cflags = bbg.regex_type;
319 * A substitution command should look something like this:
320 * s/match/replace/ #gIpw
324 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
326 /* determine the number of back references in the match string */
327 /* Note: we compute this here rather than in the do_subst_command()
328 * function to save processor time, at the expense of a little more memory
329 * (4 bits) per sed_cmd */
331 /* process the flags */
333 sed_cmd->which_match = 1;
334 while (substr[++idx]) {
335 /* Parse match number */
336 if (isdigit(substr[idx])) {
337 if (match[0] != '^') {
338 /* Match 0 treated as all, multiple matches we take the last one. */
339 char *pos = substr+idx;
340 sed_cmd->which_match = (unsigned short)strtol(substr+idx,&pos,10);
346 if (isspace(substr[idx])) continue;
348 switch (substr[idx]) {
349 /* Replace all occurrences */
351 if (match[0] != '^') sed_cmd->which_match = 0;
353 /* Print pattern space */
361 idx += parse_file_cmd(sed_cmd,substr+idx,&temp);
365 /* Ignore case (gnu exension) */
371 while (substr[++idx]) /*skip all*/;
378 bb_error_msg_and_die("bad option in substitution expression");
382 /* compile the match string into a regex */
383 if (*match != '\0') {
384 /* If match is empty, we use last regex used at runtime */
385 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
386 xregcomp(sed_cmd->sub_match, match, cflags);
394 * Process the commands arguments
396 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
398 /* handle (s)ubstitution command */
399 if (sed_cmd->cmd == 's')
400 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
401 /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
402 else if (strchr("aic", sed_cmd->cmd)) {
403 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
405 ("only a beginning address can be specified for edit commands");
407 if (*cmdstr == '\n' || *cmdstr == '\\') {
410 } else if (isspace(*cmdstr))
415 sed_cmd->string = xstrdup(cmdstr);
416 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
417 cmdstr += strlen(cmdstr);
418 /* handle file cmds: (r)ead */
419 } else if (strchr("rw", sed_cmd->cmd)) {
420 if (sed_cmd->end_line || sed_cmd->end_match)
421 bb_error_msg_and_die("command only uses one address");
422 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
423 if (sed_cmd->cmd == 'w')
424 sed_cmd->file = xfopen(sed_cmd->string,"w");
425 /* handle branch commands */
426 } else if (strchr(":btT", sed_cmd->cmd)) {
429 cmdstr = skip_whitespace(cmdstr);
430 length = strcspn(cmdstr, semicolon_whitespace);
432 sed_cmd->string = xstrndup(cmdstr, length);
436 /* translation command */
437 else if (sed_cmd->cmd == 'y') {
438 char *match, *replace;
441 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
442 /* \n already parsed, but \delimiter needs unescaping. */
443 parse_escapes(match,match,strlen(match),i,i);
444 parse_escapes(replace,replace,strlen(replace),i,i);
446 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
447 for (i = 0; match[i] && replace[i]; i++) {
448 sed_cmd->string[i*2] = match[i];
449 sed_cmd->string[i*2+1] = replace[i];
454 /* if it wasnt a single-letter command that takes no arguments
455 * then it must be an invalid command.
457 else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
458 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
461 /* give back whatever's left over */
466 /* Parse address+command sets, skipping comment lines. */
468 static void add_cmd(char *cmdstr)
473 /* Append this line to any unfinished line from last time. */
474 if (bbg.add_cmd_line) {
475 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
476 free(bbg.add_cmd_line);
477 bbg.add_cmd_line = cmdstr;
480 /* If this line ends with backslash, request next line. */
481 temp = strlen(cmdstr);
482 if (temp && cmdstr[temp-1] == '\\') {
483 if (!bbg.add_cmd_line)
484 bbg.add_cmd_line = xstrdup(cmdstr);
485 bbg.add_cmd_line[temp-1] = 0;
489 /* Loop parsing all commands in this line. */
491 /* Skip leading whitespace and semicolons */
492 cmdstr += strspn(cmdstr, semicolon_whitespace);
494 /* If no more commands, exit. */
497 /* if this is a comment, jump past it and keep going */
498 if (*cmdstr == '#') {
499 /* "#n" is the same as using -n on the command line */
500 if (cmdstr[1] == 'n')
502 cmdstr = strpbrk(cmdstr, "\n\r");
508 * format is: [addr][,addr][!]cmd
513 sed_cmd = xzalloc(sizeof(sed_cmd_t));
515 /* first part (if present) is an address: either a '$', a number or a /regex/ */
516 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
518 /* second part (if present) will begin with a comma */
519 if (*cmdstr == ',') {
523 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
525 bb_error_msg_and_die("no address after comma");
529 /* skip whitespace before the command */
530 cmdstr = skip_whitespace(cmdstr);
532 /* Check for inversion flag */
533 if (*cmdstr == '!') {
537 /* skip whitespace before the command */
538 cmdstr = skip_whitespace(cmdstr);
541 /* last part (mandatory) will be a command */
543 bb_error_msg_and_die("missing command");
544 sed_cmd->cmd = *(cmdstr++);
545 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
547 /* Add the command to the command array */
548 bbg.sed_cmd_tail->next = sed_cmd;
549 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
552 /* If we glued multiple lines together, free the memory. */
553 free(bbg.add_cmd_line);
554 bbg.add_cmd_line = NULL;
557 /* Append to a string, reallocating memory as necessary. */
561 static void pipe_putc(char c)
563 if (bbg.pipeline.idx == bbg.pipeline.len) {
564 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
565 bbg.pipeline.len + PIPE_GROW);
566 bbg.pipeline.len += PIPE_GROW;
568 bbg.pipeline.buf[bbg.pipeline.idx++] = c;
571 static void do_subst_w_backrefs(char *line, char *replace)
575 /* go through the replacement string */
576 for (i = 0; replace[i]; i++) {
577 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
578 if (replace[i] == '\\' && replace[i+1] >= '0' && replace[i+1] <= '9') {
579 int backref = replace[++i]-'0';
581 /* print out the text held in bbg.regmatch[backref] */
582 if (bbg.regmatch[backref].rm_so != -1) {
583 j = bbg.regmatch[backref].rm_so;
584 while (j < bbg.regmatch[backref].rm_eo)
585 pipe_putc(line[j++]);
589 /* if we find a backslash escaped character, print the character */
590 else if (replace[i] == '\\') pipe_putc(replace[++i]);
592 /* if we find an unescaped '&' print out the whole matched text. */
593 else if (replace[i] == '&') {
594 j = bbg.regmatch[0].rm_so;
595 while (j < bbg.regmatch[0].rm_eo)
596 pipe_putc(line[j++]);
598 /* Otherwise just output the character. */
599 else pipe_putc(replace[i]);
603 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
605 char *oldline = *line;
608 regex_t *current_regex;
610 /* Handle empty regex. */
611 if (sed_cmd->sub_match == NULL) {
612 current_regex = bbg.previous_regex_ptr;
614 bb_error_msg_and_die("no previous regexp");
616 bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
618 /* Find the first match */
619 if (REG_NOMATCH == regexec(current_regex, oldline, 10, bbg.regmatch, 0))
622 /* Initialize temporary output buffer. */
623 bbg.pipeline.buf = xmalloc(PIPE_GROW);
624 bbg.pipeline.len = PIPE_GROW;
625 bbg.pipeline.idx = 0;
627 /* Now loop through, substituting for matches */
631 /* Work around bug in glibc regexec, demonstrated by:
632 echo " a.b" | busybox sed 's [^ .]* x g'
633 The match_count check is so not to break
634 echo "hi" | busybox sed 's/^/!/g' */
635 if (!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
636 pipe_putc(*oldline++);
642 /* If we aren't interested in this match, output old line to
643 end of match and continue */
644 if (sed_cmd->which_match && sed_cmd->which_match!=match_count) {
645 for (i = 0; i < bbg.regmatch[0].rm_eo; i++)
646 pipe_putc(*oldline++);
650 /* print everything before the match */
651 for (i = 0; i < bbg.regmatch[0].rm_so; i++)
652 pipe_putc(oldline[i]);
654 /* then print the substitution string */
655 do_subst_w_backrefs(oldline, sed_cmd->string);
657 /* advance past the match */
658 oldline += bbg.regmatch[0].rm_eo;
659 /* flag that something has changed */
662 /* if we're not doing this globally, get out now */
663 if (sed_cmd->which_match) break;
664 } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
666 /* Copy rest of string into output pipeline */
669 pipe_putc(*oldline++);
673 *line = bbg.pipeline.buf;
677 /* Set command pointer to point to this label. (Does not handle null label.) */
678 static sed_cmd_t *branch_to(char *label)
682 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
683 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
687 bb_error_msg_and_die("can't find label for jump to '%s'", label);
690 static void append(char *s)
692 llist_add_to_end(&bbg.append_head, xstrdup(s));
695 static void flush_append(void)
699 /* Output appended lines. */
700 while ((data = (char *)llist_pop(&bbg.append_head))) {
701 fprintf(bbg.nonstdout,"%s\n",data);
706 static void add_input_file(FILE *file)
708 bbg.input_file_list = xrealloc(bbg.input_file_list,
709 (bbg.input_file_count + 1) * sizeof(FILE *));
710 bbg.input_file_list[bbg.input_file_count++] = file;
713 /* Get next line of input from bbg.input_file_list, flushing append buffer and
714 * noting if we ran out of files without a newline on the last line we read.
716 static char *get_next_line(int *last_char)
723 while (bbg.current_input_file < bbg.input_file_count) {
724 temp = bb_get_chunk_from_file(
725 bbg.input_file_list[bbg.current_input_file], &len);
727 /* len > 0 here, it's ok to do temp[len-1] */
728 char c = temp[len-1];
729 if (c == '\n' || c == '\0') {
731 lc |= (unsigned char)c;
734 /* will be returned if last line in the file
735 * doesn't end with either '\n' or '\0' */
739 /* Close this file and advance to next one */
740 fclose(bbg.input_file_list[bbg.current_input_file++]);
741 /* "this is the first line from new input file" */
748 /* Output line of text. */
750 * The tricks with 0x200 and last_puts_char are there to emulate gnu sed.
751 * Without them, we had this:
755 * sed "s/i/z/" z1 z2 znull | hexdump -vC output:
757 * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn|
759 * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn|
762 static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_char)
764 static char last_puts_char;
766 /* Is this a first line from new file
767 * and old file didn't end with '\n'? */
768 if ((last_char & 0x200) && last_puts_char != '\n') {
770 last_puts_char = '\n';
773 /* 'x': we don't care what is it, but we know it isn't '\n' */
774 if (s[0]) last_puts_char = 'x';
775 if (!(last_char & 0x100)) { /* had trailing '\n' or '\0'? */
777 fputc(last_char, file);
778 last_puts_char = last_char;
782 xfunc_error_retval = 4; /* It's what gnu sed exits with... */
783 bb_error_msg_and_die(bb_msg_write_error);
789 #define sed_puts(s, n) \
790 (prev_last_char = puts_maybe_newline(s, bbg.nonstdout, prev_last_char, n))
792 /* Process all the lines in all the files */
794 static void process_files(void)
796 char *pattern_space, *next_line;
797 int linenum = 0, prev_last_char = 0;
798 int last_char, next_last_char = 0;
801 next_line = get_next_line(&next_last_char);
803 /* go through every line in each file */
808 /* Advance to next line. Stop if out of lines. */
809 pattern_space = next_line;
810 if (!pattern_space) break;
811 last_char = next_last_char;
813 /* Read one line in advance so we can act on the last line,
815 next_line = get_next_line(&next_last_char);
818 /* for every line, go through all the commands */
819 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
820 int old_matched, matched;
822 old_matched = sed_cmd->in_match;
824 /* Determine if this command matches this line: */
826 /* Are we continuing a previous multi-line match? */
827 sed_cmd->in_match = sed_cmd->in_match
828 /* Or is no range necessary? */
829 || (!sed_cmd->beg_line && !sed_cmd->end_line
830 && !sed_cmd->beg_match && !sed_cmd->end_match)
831 /* Or did we match the start of a numerical range? */
832 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
833 /* Or does this line match our begin address regex? */
834 || (sed_cmd->beg_match &&
835 !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
836 /* Or did we match last line of input? */
837 || (sed_cmd->beg_line == -1 && next_line == NULL);
839 /* Snapshot the value */
841 matched = sed_cmd->in_match;
843 /* Is this line the end of the current match? */
846 sed_cmd->in_match = !(
847 /* has the ending line come, or is this a single address command? */
849 sed_cmd->end_line==-1 ?
851 : sed_cmd->end_line<=linenum
852 : !sed_cmd->end_match)
853 /* or does this line matches our last address regex */
854 || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
858 /* Skip blocks of commands we didn't match. */
859 if (sed_cmd->cmd == '{') {
860 if (sed_cmd->invert ? matched : !matched)
861 while (sed_cmd && sed_cmd->cmd != '}')
862 sed_cmd = sed_cmd->next;
863 if (!sed_cmd) bb_error_msg_and_die("unterminated {");
867 /* Okay, so did this line match? */
868 if (sed_cmd->invert ? !matched : matched) {
869 /* Update last used regex in case a blank substitute BRE is found */
870 if (sed_cmd->beg_match) {
871 bbg.previous_regex_ptr = sed_cmd->beg_match;
875 switch (sed_cmd->cmd) {
877 /* Print line number */
879 fprintf(bbg.nonstdout, "%d\n", linenum);
882 /* Write the current pattern space up to the first newline */
885 char *tmp = strchr(pattern_space, '\n');
889 sed_puts(pattern_space,1);
896 /* Write the current pattern space to output */
898 sed_puts(pattern_space, last_char);
900 /* Delete up through first newline */
903 char *tmp = strchr(pattern_space,'\n');
906 tmp = xstrdup(tmp+1);
912 /* discard this line. */
916 /* Substitute with regex */
918 if (!do_subst_command(sed_cmd, &pattern_space))
922 /* handle p option */
924 sed_puts(pattern_space, last_char);
925 /* handle w option */
927 sed_cmd->last_char = puts_maybe_newline(
928 pattern_space, sed_cmd->file,
929 sed_cmd->last_char, last_char);
932 /* Append line to linked list to be printed later */
934 append(sed_cmd->string);
937 /* Insert text before this line */
939 sed_puts(sed_cmd->string,1);
942 /* Cut and paste text (replace) */
944 /* Only triggers on last line of a matching range. */
945 if (!sed_cmd->in_match)
946 sed_puts(sed_cmd->string,0);
949 /* Read file, append contents to output */
954 rfile = fopen(sed_cmd->string, "r");
958 while ((line = xmalloc_getline(rfile))
961 xprint_and_close_file(rfile);
967 /* Write pattern space to file. */
969 sed_cmd->last_char = puts_maybe_newline(
970 pattern_space,sed_cmd->file,
971 sed_cmd->last_char, last_char);
974 /* Read next line from input */
977 sed_puts(pattern_space, last_char);
980 pattern_space = next_line;
981 last_char = next_last_char;
982 next_line = get_next_line(&next_last_char);
988 /* Quit. End of script, end of input. */
990 /* Exit the outer while loop */
993 goto discard_commands;
995 /* Append the next line to the current line */
998 /* If no next line, jump to end of script and exit. */
999 if (next_line == NULL) {
1000 /* Jump to end of script and exit */
1004 /* append next_line, read new next_line. */
1006 int len = strlen(pattern_space);
1008 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1009 pattern_space[len] = '\n';
1010 strcpy(pattern_space + len+1, next_line);
1011 last_char = next_last_char;
1012 next_line = get_next_line(&next_last_char);
1018 /* Test/branch if substitution occurred */
1020 if (!substituted) break;
1023 /* Test/branch if substitution didn't occur */
1025 if (substituted) break;
1027 /* Branch to label */
1029 if (!sed_cmd->string) goto discard_commands;
1030 else sed_cmd = branch_to(sed_cmd->string);
1032 /* Transliterate characters */
1037 for (i = 0; pattern_space[i]; i++) {
1040 for (j = 0; sed_cmd->string[j]; j += 2) {
1041 if (pattern_space[i] == sed_cmd->string[j]) {
1042 pattern_space[i] = sed_cmd->string[j + 1];
1050 case 'g': /* Replace pattern space with hold space */
1051 free(pattern_space);
1052 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1054 case 'G': /* Append newline and hold space to pattern space */
1056 int pattern_space_size = 2;
1057 int hold_space_size = 0;
1060 pattern_space_size += strlen(pattern_space);
1062 hold_space_size = strlen(bbg.hold_space);
1063 pattern_space = xrealloc(pattern_space,
1064 pattern_space_size + hold_space_size);
1065 if (pattern_space_size == 2)
1066 pattern_space[0] = 0;
1067 strcat(pattern_space, "\n");
1069 strcat(pattern_space, bbg.hold_space);
1074 case 'h': /* Replace hold space with pattern space */
1075 free(bbg.hold_space);
1076 bbg.hold_space = xstrdup(pattern_space);
1078 case 'H': /* Append newline and pattern space to hold space */
1080 int hold_space_size = 2;
1081 int pattern_space_size = 0;
1084 hold_space_size += strlen(bbg.hold_space);
1086 pattern_space_size = strlen(pattern_space);
1087 bbg.hold_space = xrealloc(bbg.hold_space,
1088 hold_space_size + pattern_space_size);
1090 if (hold_space_size == 2)
1091 *bbg.hold_space = 0;
1092 strcat(bbg.hold_space, "\n");
1094 strcat(bbg.hold_space, pattern_space);
1098 case 'x': /* Exchange hold and pattern space */
1100 char *tmp = pattern_space;
1101 pattern_space = bbg.hold_space ? : xzalloc(1);
1103 bbg.hold_space = tmp;
1111 * exit point from sedding...
1114 /* we will print the line unless we were told to be quiet ('-n')
1115 or if the line was suppressed (ala 'd'elete) */
1116 if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1118 /* Delete and such jump here. */
1121 free(pattern_space);
1125 /* It is possible to have a command line argument with embedded
1126 newlines. This counts as multiple command lines. */
1128 static void add_cmd_block(char *cmdstr)
1131 char *temp = xstrdup(cmdstr), *temp2 = temp;
1134 int len = strcspn(temp2,"\n");
1135 if (!temp2[len]) go = 0;
1136 else temp2[len] = 0;
1143 static void add_cmds_link(llist_t *opt_e)
1146 add_cmds_link(opt_e->link);
1147 add_cmd_block(opt_e->data);
1151 static void add_files_link(llist_t *opt_f)
1156 add_files_link(opt_f->link);
1157 cmdfile = xfopen(opt_f->data, "r");
1158 while ((line = xmalloc_getline(cmdfile)) != NULL) {
1162 xprint_and_close_file(cmdfile);
1166 int sed_main(int argc, char **argv)
1169 llist_t *opt_e, *opt_f;
1170 int status = EXIT_SUCCESS;
1172 bbg.sed_cmd_tail=&bbg.sed_cmd_head;
1174 /* destroy command strings on exit */
1175 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1177 /* Lie to autoconf when it starts asking stupid questions. */
1178 if (argc==2 && !strcmp(argv[1],"--version")) {
1179 printf("This is not GNU sed version 4.0\n");
1183 /* do normal option parsing */
1184 opt_e = opt_f = NULL;
1185 opt_complementary = "e::f::"; /* can occur multiple times */
1186 opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f);
1187 if (opt & 0x1) { // -i
1189 atexit(cleanup_outname);
1191 if (opt & 0x2) bbg.regex_type|=REG_EXTENDED; // -r
1192 if (opt & 0x4) bbg.be_quiet++; // -n
1193 if (opt & 0x8) { // -e
1194 /* getopt32 reverses order of arguments, handle it */
1195 add_cmds_link(opt_e);
1197 if (opt & 0x10) { // -f
1198 /* getopt32 reverses order of arguments, handle it */
1199 add_files_link(opt_f);
1201 /* if we didn't get a pattern from -e or -f, use argv[optind] */
1202 if (!(opt & 0x18)) {
1203 if (argv[optind] == NULL)
1206 add_cmd_block(argv[optind++]);
1208 /* Flush any unfinished commands. */
1211 /* By default, we write to stdout */
1212 bbg.nonstdout = stdout;
1214 /* argv[(optind)..(argc-1)] should be names of file to process. If no
1215 * files were specified or '-' was specified, take input from stdin.
1216 * Otherwise, we process all the files specified. */
1217 if (argv[optind] == NULL) {
1219 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1220 add_input_file(stdin);
1226 for (i = optind; i < argc; i++) {
1227 struct stat statbuf;
1230 if (!strcmp(argv[i], "-") && !bbg.in_place) {
1231 add_input_file(stdin);
1235 file = fopen_or_warn(argv[i], "r");
1237 status = EXIT_FAILURE;
1240 if (!bbg.in_place) {
1241 add_input_file(file);
1245 bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1246 nonstdoutfd = mkstemp(bbg.outname);
1247 if (-1 == nonstdoutfd)
1248 bb_error_msg_and_die("no temp file");
1249 bbg.nonstdout = fdopen(nonstdoutfd,"w");
1251 /* Set permissions of output file */
1253 fstat(fileno(file),&statbuf);
1254 fchmod(nonstdoutfd,statbuf.st_mode);
1255 add_input_file(file);
1257 fclose(bbg.nonstdout);
1259 bbg.nonstdout = stdout;
1260 /* unlink(argv[i]); */
1261 // FIXME: error check / message?
1262 rename(bbg.outname,argv[i]);
1266 if (bbg.input_file_count > bbg.current_input_file)