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";
96 int be_quiet, regex_type;
98 char *outname, *hold_space;
100 /* List of input files */
101 int input_file_count, current_input_file;
102 FILE **input_file_list;
104 regmatch_t regmatch[10];
105 regex_t *previous_regex_ptr;
107 /* linked list of sed commands */
108 sed_cmd_t sed_cmd_head, *sed_cmd_tail;
110 /* Linked list of append lines */
111 llist_t *append_head;
116 char *buf; /* Space to hold string */
117 int idx; /* Space used */
118 int len; /* Space allocated */
123 void sed_free_and_close_stuff(void);
124 #if ENABLE_FEATURE_CLEAN_UP
125 static void sed_free_and_close_stuff(void)
127 sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
129 llist_free(bbg.append_head, free);
132 sed_cmd_t *sed_cmd_next = sed_cmd->next;
135 xprint_and_close_file(sed_cmd->file);
137 if (sed_cmd->beg_match) {
138 regfree(sed_cmd->beg_match);
139 free(sed_cmd->beg_match);
141 if (sed_cmd->end_match) {
142 regfree(sed_cmd->end_match);
143 free(sed_cmd->end_match);
145 if (sed_cmd->sub_match) {
146 regfree(sed_cmd->sub_match);
147 free(sed_cmd->sub_match);
149 free(sed_cmd->string);
151 sed_cmd = sed_cmd_next;
154 if (bbg.hold_space) free(bbg.hold_space);
156 while (bbg.current_input_file < bbg.input_file_count)
157 fclose(bbg.input_file_list[bbg.current_input_file++]);
161 /* If something bad happens during -i operation, delete temp file */
163 static void cleanup_outname(void)
165 if (bbg.outname) unlink(bbg.outname);
168 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
170 static void parse_escapes(char *dest, char *string, int len, char from, char to)
175 if (string[i] == '\\') {
176 if (!to || string[i+1] == from) {
177 *(dest++) = to ? to : string[i+1];
180 } else *(dest++) = string[i++];
182 *(dest++) = string[i++];
187 static char *copy_parsing_escapes(char *string, int len)
189 char *dest = xmalloc(len + 1);
191 parse_escapes(dest, string, len, 'n', '\n');
197 * index_of_next_unescaped_regexp_delim - walks left to right through a string
198 * beginning at a specified index and returns the index of the next regular
199 * expression delimiter (typically a forward * slash ('/')) not preceded by
200 * a backslash ('\'). A negative delimiter disables square bracket checking.
202 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
211 delimiter = -delimiter;
214 for (; (ch = str[idx]); idx++) {
216 if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
217 && str[idx - 1] == '^')))
223 else if (bracket == -1 && ch == '[')
225 else if (ch == delimiter)
229 /* if we make it to here, we've hit the end of the string */
230 bb_error_msg_and_die("unmatched '%c'", delimiter);
234 * Returns the index of the third delimiter
236 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
238 char *cmdstr_ptr = cmdstr;
242 /* verify that the 's' or 'y' is followed by something. That something
243 * (typically a 'slash') is now our regexp delimiter... */
245 bb_error_msg_and_die("bad format in substitution expression");
246 delimiter = *cmdstr_ptr++;
248 /* save the match string */
249 idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
250 *match = copy_parsing_escapes(cmdstr_ptr, idx);
252 /* save the replacement string */
253 cmdstr_ptr += idx + 1;
254 idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
255 *replace = copy_parsing_escapes(cmdstr_ptr, idx);
257 return ((cmdstr_ptr - cmdstr) + idx);
261 * returns the index in the string just past where the address ends.
263 static int get_address(char *my_str, int *linenum, regex_t ** regex)
267 if (isdigit(*my_str)) {
268 *linenum = strtol(my_str, &pos, 10);
269 /* endstr shouldnt ever equal NULL */
270 } else if (*my_str == '$') {
273 } else if (*my_str == '/' || *my_str == '\\') {
279 if (*my_str == '\\') delimiter = *++pos;
280 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
281 temp = copy_parsing_escapes(pos, next);
282 *regex = xmalloc(sizeof(regex_t));
283 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
285 /* Move position to next character after last delimiter */
291 /* Grab a filename. Whitespace at start is skipped, then goes to EOL. */
292 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
294 int start = 0, idx, hack = 0;
296 /* Skip whitespace, then grab filename to end of line */
297 while (isspace(filecmdstr[start])) start++;
299 while (filecmdstr[idx] && filecmdstr[idx] != '\n') idx++;
301 /* If lines glued together, put backslash back. */
302 if (filecmdstr[idx] == '\n') hack = 1;
304 bb_error_msg_and_die("empty filename");
305 *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
306 if (hack) (*retval)[idx] = '\\';
311 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
313 int cflags = bbg.regex_type;
318 * A substitution command should look something like this:
319 * s/match/replace/ #gIpw
323 idx = parse_regex_delim(substr, &match, &sed_cmd->string);
325 /* determine the number of back references in the match string */
326 /* Note: we compute this here rather than in the do_subst_command()
327 * function to save processor time, at the expense of a little more memory
328 * (4 bits) per sed_cmd */
330 /* process the flags */
332 sed_cmd->which_match = 1;
333 while (substr[++idx]) {
334 /* Parse match number */
335 if (isdigit(substr[idx])) {
336 if (match[0] != '^') {
337 /* Match 0 treated as all, multiple matches we take the last one. */
338 char *pos = substr + idx;
339 /* FIXME: error check? */
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 */
809 /* Advance to next line. Stop if out of lines. */
810 pattern_space = next_line;
811 if (!pattern_space) return;
812 last_char = next_last_char;
814 /* Read one line in advance so we can act on the last line,
816 next_line = get_next_line(&next_last_char);
819 /* for every line, go through all the commands */
820 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
821 int old_matched, matched;
823 old_matched = sed_cmd->in_match;
825 /* Determine if this command matches this line: */
827 /* Are we continuing a previous multi-line match? */
828 sed_cmd->in_match = sed_cmd->in_match
829 /* Or is no range necessary? */
830 || (!sed_cmd->beg_line && !sed_cmd->end_line
831 && !sed_cmd->beg_match && !sed_cmd->end_match)
832 /* Or did we match the start of a numerical range? */
833 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
834 /* Or does this line match our begin address regex? */
835 || (sed_cmd->beg_match &&
836 !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
837 /* Or did we match last line of input? */
838 || (sed_cmd->beg_line == -1 && next_line == NULL);
840 /* Snapshot the value */
842 matched = sed_cmd->in_match;
844 /* Is this line the end of the current match? */
847 sed_cmd->in_match = !(
848 /* has the ending line come, or is this a single address command? */
850 sed_cmd->end_line == -1 ?
852 : (sed_cmd->end_line <= linenum)
853 : !sed_cmd->end_match
855 /* or does this line matches our last address regex */
856 || (sed_cmd->end_match && old_matched
857 && (regexec(sed_cmd->end_match,
858 pattern_space, 0, NULL, 0) == 0))
862 /* Skip blocks of commands we didn't match. */
863 if (sed_cmd->cmd == '{') {
864 if (sed_cmd->invert ? matched : !matched)
865 while (sed_cmd && sed_cmd->cmd != '}')
866 sed_cmd = sed_cmd->next;
867 if (!sed_cmd) bb_error_msg_and_die("unterminated {");
871 /* Okay, so did this line match? */
872 if (sed_cmd->invert ? !matched : matched) {
873 /* Update last used regex in case a blank substitute BRE is found */
874 if (sed_cmd->beg_match) {
875 bbg.previous_regex_ptr = sed_cmd->beg_match;
879 switch (sed_cmd->cmd) {
881 /* Print line number */
883 fprintf(bbg.nonstdout, "%d\n", linenum);
886 /* Write the current pattern space up to the first newline */
889 char *tmp = strchr(pattern_space, '\n');
893 sed_puts(pattern_space, 1);
900 /* Write the current pattern space to output */
902 sed_puts(pattern_space, last_char);
904 /* Delete up through first newline */
907 char *tmp = strchr(pattern_space, '\n');
910 tmp = xstrdup(tmp+1);
916 /* discard this line. */
920 /* Substitute with regex */
922 if (!do_subst_command(sed_cmd, &pattern_space))
926 /* handle p option */
928 sed_puts(pattern_space, last_char);
929 /* handle w option */
931 sed_cmd->last_char = puts_maybe_newline(
932 pattern_space, sed_cmd->file,
933 sed_cmd->last_char, last_char);
936 /* Append line to linked list to be printed later */
938 append(sed_cmd->string);
941 /* Insert text before this line */
943 sed_puts(sed_cmd->string, 1);
946 /* Cut and paste text (replace) */
948 /* Only triggers on last line of a matching range. */
949 if (!sed_cmd->in_match)
950 sed_puts(sed_cmd->string, 0);
953 /* Read file, append contents to output */
958 rfile = fopen(sed_cmd->string, "r");
962 while ((line = xmalloc_getline(rfile))
965 xprint_and_close_file(rfile);
971 /* Write pattern space to file. */
973 sed_cmd->last_char = puts_maybe_newline(
974 pattern_space, sed_cmd->file,
975 sed_cmd->last_char, last_char);
978 /* Read next line from input */
981 sed_puts(pattern_space, last_char);
984 pattern_space = next_line;
985 last_char = next_last_char;
986 next_line = get_next_line(&next_last_char);
992 /* Quit. End of script, end of input. */
994 /* Exit the outer while loop */
997 goto discard_commands;
999 /* Append the next line to the current line */
1003 /* If no next line, jump to end of script and exit. */
1004 if (next_line == NULL) {
1005 /* Jump to end of script and exit */
1009 /* append next_line, read new next_line. */
1011 len = strlen(pattern_space);
1012 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1013 pattern_space[len] = '\n';
1014 strcpy(pattern_space + len+1, next_line);
1015 last_char = next_last_char;
1016 next_line = get_next_line(&next_last_char);
1021 /* Test/branch if substitution occurred */
1023 if (!substituted) break;
1026 /* Test/branch if substitution didn't occur */
1028 if (substituted) break;
1030 /* Branch to label */
1032 if (!sed_cmd->string) goto discard_commands;
1033 else sed_cmd = branch_to(sed_cmd->string);
1035 /* Transliterate characters */
1040 for (i = 0; pattern_space[i]; i++) {
1041 for (j = 0; sed_cmd->string[j]; j += 2) {
1042 if (pattern_space[i] == sed_cmd->string[j]) {
1043 pattern_space[i] = sed_cmd->string[j + 1];
1051 case 'g': /* Replace pattern space with hold space */
1052 free(pattern_space);
1053 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1055 case 'G': /* Append newline and hold space to pattern space */
1057 int pattern_space_size = 2;
1058 int hold_space_size = 0;
1061 pattern_space_size += strlen(pattern_space);
1063 hold_space_size = strlen(bbg.hold_space);
1064 pattern_space = xrealloc(pattern_space,
1065 pattern_space_size + hold_space_size);
1066 if (pattern_space_size == 2)
1067 pattern_space[0] = 0;
1068 strcat(pattern_space, "\n");
1070 strcat(pattern_space, bbg.hold_space);
1075 case 'h': /* Replace hold space with pattern space */
1076 free(bbg.hold_space);
1077 bbg.hold_space = xstrdup(pattern_space);
1079 case 'H': /* Append newline and pattern space to hold space */
1081 int hold_space_size = 2;
1082 int pattern_space_size = 0;
1085 hold_space_size += strlen(bbg.hold_space);
1087 pattern_space_size = strlen(pattern_space);
1088 bbg.hold_space = xrealloc(bbg.hold_space,
1089 hold_space_size + pattern_space_size);
1091 if (hold_space_size == 2)
1092 *bbg.hold_space = 0;
1093 strcat(bbg.hold_space, "\n");
1095 strcat(bbg.hold_space, pattern_space);
1099 case 'x': /* Exchange hold and pattern space */
1101 char *tmp = pattern_space;
1102 pattern_space = bbg.hold_space ? : xzalloc(1);
1104 bbg.hold_space = tmp;
1112 * exit point from sedding...
1115 /* we will print the line unless we were told to be quiet ('-n')
1116 or if the line was suppressed (ala 'd'elete) */
1117 if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1119 /* Delete and such jump here. */
1122 free(pattern_space);
1127 /* It is possible to have a command line argument with embedded
1128 newlines. This counts as multiple command lines. */
1130 static void add_cmd_block(char *cmdstr)
1133 char *temp = xstrdup(cmdstr), *temp2 = temp;
1136 int len = strcspn(temp2, "\n");
1137 if (!temp2[len]) go = 0;
1138 else temp2[len] = 0;
1145 static void add_cmds_link(llist_t *opt_e)
1148 add_cmds_link(opt_e->link);
1149 add_cmd_block(opt_e->data);
1153 static void add_files_link(llist_t *opt_f)
1158 add_files_link(opt_f->link);
1159 cmdfile = xfopen(opt_f->data, "r");
1160 while ((line = xmalloc_getline(cmdfile)) != NULL) {
1164 xprint_and_close_file(cmdfile);
1168 int sed_main(int argc, char **argv)
1171 OPT_in_place = 1 << 0,
1174 llist_t *opt_e, *opt_f;
1175 int status = EXIT_SUCCESS;
1177 bbg.sed_cmd_tail = &bbg.sed_cmd_head;
1179 /* destroy command strings on exit */
1180 if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1182 /* Lie to autoconf when it starts asking stupid questions. */
1183 if (argc == 2 && !strcmp(argv[1], "--version")) {
1184 puts("This is not GNU sed version 4.0");
1188 /* do normal option parsing */
1189 opt_e = opt_f = NULL;
1190 opt_complementary = "e::f::" /* can occur multiple times */
1191 "nn"; /* count -n */
1192 opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f,
1193 &bbg.be_quiet); /* counter for -n */
1196 if (opt & OPT_in_place) { // -i
1197 atexit(cleanup_outname);
1199 if (opt & 0x2) bbg.regex_type |= REG_EXTENDED; // -r
1200 //if (opt & 0x4) bbg.be_quiet++; // -n
1201 if (opt & 0x8) { // -e
1202 /* getopt32 reverses order of arguments, handle it */
1203 add_cmds_link(opt_e);
1205 if (opt & 0x10) { // -f
1206 /* getopt32 reverses order of arguments, handle it */
1207 add_files_link(opt_f);
1209 /* if we didn't get a pattern from -e or -f, use argv[0] */
1210 if (!(opt & 0x18)) {
1213 add_cmd_block(*argv++);
1216 /* Flush any unfinished commands. */
1219 /* By default, we write to stdout */
1220 bbg.nonstdout = stdout;
1222 /* argv[0..(argc-1)] should be names of file to process. If no
1223 * files were specified or '-' was specified, take input from stdin.
1224 * Otherwise, we process all the files specified. */
1225 if (argv[0] == NULL) {
1226 if (opt & OPT_in_place)
1227 bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1228 add_input_file(stdin);
1234 for (i = 0; i < argc; i++) {
1235 struct stat statbuf;
1238 if (argv[i][0] == '-' && !argv[i][1]
1239 && !(opt & OPT_in_place)
1241 add_input_file(stdin);
1245 file = fopen_or_warn(argv[i], "r");
1247 status = EXIT_FAILURE;
1250 if (!(opt & OPT_in_place)) {
1251 add_input_file(file);
1255 bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1256 nonstdoutfd = mkstemp(bbg.outname);
1257 if (-1 == nonstdoutfd)
1258 bb_error_msg_and_die("no temp file");
1259 bbg.nonstdout = fdopen(nonstdoutfd, "w");
1261 /* Set permissions of output file */
1263 fstat(fileno(file), &statbuf);
1264 fchmod(nonstdoutfd, statbuf.st_mode);
1265 add_input_file(file);
1267 fclose(bbg.nonstdout);
1269 bbg.nonstdout = stdout;
1270 /* unlink(argv[i]); */
1271 // FIXME: error check / message?
1272 rename(bbg.outname, argv[i]);
1276 if (bbg.input_file_count > bbg.current_input_file)