editors/sed.c

   1 /*
   2  * sed.c - very minimalist version of sed
   3  *
   4  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   5  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   6  * Copyright (C) 2002  Matt Kraai
   7  * Copyright (C) 2003 by Glenn McGrath <bug1@optushome.com.au>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17  * General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  22  *
  23  */
  24
  25 /*
  26         Supported features and commands in this version of sed:
  27
  28          - comments ('#')
  29          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  30          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  31          - edit commands: (a)ppend, (i)nsert, (c)hange
  32          - file commands: (r)ead
  33          - backreferences in substitution expressions (\1, \2...\9)
  34          - grouped commands: {cmd1;cmd2}
  35          - transliteration (y/source-chars/dest-chars/)
  36          - pattern space hold space storing / swapping (g, h, x)
  37          - labels / branching (: label, b, t)
  38
  39          (Note: Specifying an address (range) to match is *optional*; commands
  40          default to the whole pattern space if no specific address match was
  41          requested.)
  42
  43         Unsupported features:
  44
  45          - GNU extensions
  46          - and lots, lots more.
  47
  48         Bugs:
  49
  50          - Cant subst globally using ^ or $ in regex, eg. "aah" | sed 's/^a/b/g'
  51
  52         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  53 */
  54
  55 #include <stdio.h>
  56 #include <unistd.h>             /* for getopt() */
  57 #include <regex.h>
  58 #include <string.h>             /* for strdup() */
  59 #include <errno.h>
  60 #include <ctype.h>              /* for isspace() */
  61 #include <stdlib.h>
  62 #include "busybox.h"
  63
  64 typedef struct sed_cmd_s {
  65         /* Order by alignment requirements */
  66
  67         /* address storage */
  68         regex_t *beg_match;     /* sed -e '/match/cmd' */
  69         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
  70
  71         int beg_line;           /* 'sed 1p'   0 == no begining line, apply commands to all lines */
  72         int end_line;           /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
  73
  74         /* inversion flag */
  75         int invert;                     /* the '!' after the address */
  76 //      int block_cmd;  /* This command is part of a group that has a command address */
  77
  78         /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
  79
  80         /* sed -e 's/sub_match/replace/' */
  81         regex_t *sub_match;
  82         char *replace;
  83
  84         /* EDIT COMMAND (a,i,c) SPECIFIC FIELDS */
  85         char *editline;
  86
  87         /* FILE COMMAND (r) SPECIFIC FIELDS */
  88         char *filename;
  89
  90         /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
  91
  92         unsigned int num_backrefs:4;    /* how many back references (\1..\9) */
  93         /* Note:  GNU/POSIX sed does not save more than nine backrefs, so
  94          * we only use 4 bits to hold the number */
  95         unsigned int sub_g:1;   /* sed -e 's/foo/bar/g' (global) */
  96         unsigned int sub_p:1;   /* sed -e 's/foo/bar/p' (print substitution) */
  97
  98         /* TRANSLATE COMMAND */
  99         char *translate;
 100
 101         /* GENERAL FIELDS */
 102         /* the command */
 103         char cmd;                       /* p,d,s (add more at your leisure :-) */
 104
 105         /* Branch commands */
 106         char *label;
 107
 108         /* next command in list (sequential list of specified commands) */
 109         struct sed_cmd_s *next;
 110
 111 } sed_cmd_t;
 112
 113
 114 /* externs */
 115 extern void xregcomp(regex_t * preg, const char *regex, int cflags);
 116 extern int optind;              /* in unistd.h */
 117 extern char *optarg;    /* ditto */
 118
 119 /* globals */
 120 /* options */
 121 static int be_quiet = 0;
 122 static const char bad_format_in_subst[] =
 123         "bad format in substitution expression";
 124
 125 /* linked list of sed commands */
 126 static sed_cmd_t sed_cmd_head;
 127 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
 128
 129 const char *const semicolon_whitespace = "; \n\r\t\v\0";
 130 static regex_t *previous_regex_ptr = NULL;
 131
 132
 133 #ifdef CONFIG_FEATURE_CLEAN_UP
 134 static void destroy_cmd_strs(void)
 135 {
 136         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 137
 138         while (sed_cmd) {
 139                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 140
 141                 if (sed_cmd->beg_match) {
 142                         regfree(sed_cmd->beg_match);
 143                         free(sed_cmd->beg_match);
 144                 }
 145                 if (sed_cmd->end_match) {
 146                         regfree(sed_cmd->end_match);
 147                         free(sed_cmd->end_match);
 148                 }
 149                 if (sed_cmd->sub_match) {
 150                         regfree(sed_cmd->sub_match);
 151                         free(sed_cmd->sub_match);
 152                 }
 153                 free(sed_cmd->replace);
 154                 free(sed_cmd);
 155                 sed_cmd = sed_cmd_next;
 156         }
 157 }
 158 #endif
 159
 160 /*
 161  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 162  * beginning at a specified index and returns the index of the next regular
 163  * expression delimiter (typically a forward * slash ('/')) not preceeded by
 164  * a backslash ('\').
 165  */
 166 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 167         const char *str)
 168 {
 169         int bracket = -1;
 170         int escaped = 0;
 171         int idx = 0;
 172         char ch;
 173
 174         for (; (ch = str[idx]); idx++) {
 175                 if (bracket != -1) {
 176                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 177                                                 && str[idx - 1] == '^')))
 178                                 bracket = -1;
 179                 } else if (escaped)
 180                         escaped = 0;
 181                 else if (ch == '\\')
 182                         escaped = 1;
 183                 else if (ch == '[')
 184                         bracket = idx;
 185                 else if (ch == delimiter)
 186                         return idx;
 187         }
 188
 189         /* if we make it to here, we've hit the end of the string */
 190         return -1;
 191 }
 192
 193 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 194 {
 195         const char *cmdstr_ptr = cmdstr;
 196         char delimiter;
 197         int idx = 0;
 198
 199         /* verify that the 's' is followed by something.  That something
 200          * (typically a 'slash') is now our regexp delimiter... */
 201         if (*cmdstr == '\0')
 202                 bb_error_msg_and_die(bad_format_in_subst);
 203         else
 204                 delimiter = *cmdstr_ptr;
 205
 206         cmdstr_ptr++;
 207
 208         /* save the match string */
 209         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 210         if (idx == -1) {
 211                 bb_error_msg_and_die(bad_format_in_subst);
 212         }
 213         *match = bb_xstrndup(cmdstr_ptr, idx);
 214
 215         /* save the replacement string */
 216         cmdstr_ptr += idx + 1;
 217         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 218         if (idx == -1) {
 219                 bb_error_msg_and_die(bad_format_in_subst);
 220         }
 221         *replace = bb_xstrndup(cmdstr_ptr, idx);
 222
 223         return ((cmdstr_ptr - cmdstr) + idx);
 224 }
 225
 226 /*
 227  * returns the index in the string just past where the address ends.
 228  */
 229 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 230 {
 231         int idx = 0;
 232
 233         if (isdigit(my_str[idx])) {
 234                 char *endstr;
 235
 236                 *linenum = strtol(my_str, &endstr, 10);
 237                 /* endstr shouldnt ever equal NULL */
 238                 idx = endstr - my_str;
 239         } else if (my_str[idx] == '$') {
 240                 *linenum = -1;
 241                 idx++;
 242         } else if (my_str[idx] == '/' || my_str[idx] == '\\') {
 243                 int idx_start = 1;
 244                 char delimiter;
 245
 246                 delimiter = '/';
 247                 if (my_str[idx] == '\\') {
 248                         idx_start++;
 249                         delimiter = my_str[++idx];
 250                 }
 251                 idx++;
 252                 idx += index_of_next_unescaped_regexp_delim(delimiter, my_str + idx);
 253                 if (idx == -1) {
 254                         bb_error_msg_and_die("unterminated match expression");
 255                 }
 256                 my_str[idx] = '\0';
 257
 258                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 259                 xregcomp(*regex, my_str + idx_start, REG_NEWLINE);
 260                 idx++;                  /* so it points to the next character after the last '/' */
 261         }
 262         return idx;
 263 }
 264
 265 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, const char *substr)
 266 {
 267         int cflags = 0;
 268         char *match;
 269         int idx = 0;
 270         int j;
 271
 272         /*
 273          * the string that gets passed to this function should look like this:
 274          *    s/match/replace/gIp
 275          *    ||     |        |||
 276          *    mandatory       optional
 277          *
 278          *    (all three of the '/' slashes are mandatory)
 279          */
 280         idx = parse_regex_delim(substr, &match, &sed_cmd->replace);
 281
 282         /* determine the number of back references in the match string */
 283         /* Note: we compute this here rather than in the do_subst_command()
 284          * function to save processor time, at the expense of a little more memory
 285          * (4 bits) per sed_cmd */
 286
 287         /* sed_cmd->num_backrefs = 0; *//* XXX: not needed? --apparently not */
 288         for (j = 0; match[j]; j++) {
 289                 /* GNU/POSIX sed does not save more than nine backrefs */
 290                 if (match[j] == '\\' && match[j + 1] == '('
 291                         && sed_cmd->num_backrefs <= 9)
 292                         sed_cmd->num_backrefs++;
 293         }
 294
 295         /* process the flags */
 296         while (substr[++idx]) {
 297                 switch (substr[idx]) {
 298                 case 'g':
 299                         sed_cmd->sub_g = 1;
 300                         break;
 301                         /* Hmm, i dont see the I option mentioned in the standard */
 302                 case 'I':
 303                         cflags |= REG_ICASE;
 304                         break;
 305                 case 'p':
 306                         sed_cmd->sub_p = 1;
 307                         break;
 308                 default:
 309                         /* any whitespace or semicolon trailing after a s/// is ok */
 310                         if (strchr(semicolon_whitespace, substr[idx]))
 311                                 goto out;
 312                         /* else */
 313                         bb_error_msg_and_die("bad option in substitution expression");
 314                 }
 315         }
 316
 317   out:
 318         /* compile the match string into a regex */
 319         if (*match != '\0') {
 320                 /* If match is empty, we use last regex used at runtime */
 321                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 322                 xregcomp(sed_cmd->sub_match, match, cflags);
 323         }
 324         free(match);
 325
 326         return idx;
 327 }
 328
 329 static void replace_slash_n(char *string)
 330 {
 331         int i;
 332         int remaining = strlen(string);
 333
 334         for (i = 0; string[i]; i++) {
 335                 if ((string[i] == '\\') && (string[i + 1] == 'n')) {
 336                         string[i] = '\n';
 337                         memmove(string + i + 1, string + i + 1, remaining - 1);
 338                 } else {
 339                         remaining--;
 340                 }
 341         }
 342 }
 343
 344 static int parse_translate_cmd(sed_cmd_t * const sed_cmd, const char *cmdstr)
 345 {
 346         char *match;
 347         char *replace;
 348         int idx;
 349         int i;
 350
 351         idx = parse_regex_delim(cmdstr, &match, &replace);
 352         replace_slash_n(match);
 353         replace_slash_n(replace);
 354         sed_cmd->translate = xcalloc(1, (strlen(match) + 1) * 2);
 355         for (i = 0; (match[i] != 0) && (replace[i] != 0); i++) {
 356                 sed_cmd->translate[i * 2] = match[i];
 357                 sed_cmd->translate[(i * 2) + 1] = replace[i];
 358         }
 359         return (idx + 1);
 360 }
 361
 362 static int parse_edit_cmd(sed_cmd_t * sed_cmd, const char *editstr)
 363 {
 364         int i, j;
 365
 366         /*
 367          * the string that gets passed to this function should look like this:
 368          *
 369          *    need one of these
 370          *    |
 371          *    |    this backslash (immediately following the edit command) is mandatory
 372          *    |    |
 373          *    [aic]\
 374          *    TEXT1\
 375          *    TEXT2\
 376          *    TEXTN
 377          *
 378          * as soon as we hit a TEXT line that has no trailing '\', we're done.
 379          * this means a command like:
 380          *
 381          * i\
 382          * INSERTME
 383          *
 384          * is a-ok.
 385          *
 386          */
 387         if ((*editstr != '\\') || ((editstr[1] != '\n') && (editstr[1] != '\r'))) {
 388                 bb_error_msg_and_die("bad format in edit expression");
 389         }
 390
 391         /* store the edit line text */
 392         sed_cmd->editline = xmalloc(strlen(&editstr[2]) + 2);
 393         for (i = 2, j = 0;
 394                 editstr[i] != '\0' && strchr("\r\n", editstr[i]) == NULL; i++, j++) {
 395                 if ((editstr[i] == '\\') && strchr("\n\r", editstr[i + 1]) != NULL) {
 396                         sed_cmd->editline[j] = '\n';
 397                         i++;
 398                 } else
 399                         sed_cmd->editline[j] = editstr[i];
 400         }
 401
 402         /* figure out if we need to add a newline */
 403         if (sed_cmd->editline[j - 1] != '\n')
 404                 sed_cmd->editline[j++] = '\n';
 405
 406         /* terminate string */
 407         sed_cmd->editline[j] = '\0';
 408
 409         return i;
 410 }
 411
 412
 413 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr)
 414 {
 415         int idx = 0;
 416         int filenamelen = 0;
 417
 418         /*
 419          * the string that gets passed to this function should look like this:
 420          *    '[ ]filename'
 421          *      |  |
 422          *      |  a filename
 423          *      |
 424          *     optional whitespace
 425
 426          *   re: the file to be read, the GNU manual says the following: "Note that
 427          *   if filename cannot be read, it is treated as if it were an empty file,
 428          *   without any error indication." Thus, all of the following commands are
 429          *   perfectly leagal:
 430          *
 431          *   sed -e '1r noexist'
 432          *   sed -e '1r ;'
 433          *   sed -e '1r'
 434          */
 435
 436         /* the file command may be followed by whitespace; move past it. */
 437         while (isspace(filecmdstr[++idx])) {;
 438         }
 439
 440         /* the first non-whitespace we get is a filename. the filename ends when we
 441          * hit a normal sed command terminator or end of string */
 442         filenamelen = strcspn(&filecmdstr[idx], semicolon_whitespace);
 443         sed_cmd->filename = xmalloc(filenamelen + 1);
 444         safe_strncpy(sed_cmd->filename, &filecmdstr[idx], filenamelen + 1);
 445         return idx + filenamelen;
 446 }
 447
 448 /*
 449  *  Process the commands arguments
 450  */
 451 static char *parse_cmd_str(sed_cmd_t * sed_cmd, char *cmdstr)
 452 {
 453         /* handle (s)ubstitution command */
 454         if (sed_cmd->cmd == 's') {
 455                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 456         }
 457         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 458         else if (strchr("aic", sed_cmd->cmd)) {
 459                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 460                         bb_error_msg_and_die
 461                                 ("only a beginning address can be specified for edit commands");
 462                 cmdstr += parse_edit_cmd(sed_cmd, cmdstr);
 463         }
 464         /* handle file cmds: (r)ead */
 465         else if (sed_cmd->cmd == 'r') {
 466                 if (sed_cmd->end_line || sed_cmd->end_match)
 467                         bb_error_msg_and_die("Command only uses one address");
 468                 cmdstr += parse_file_cmd(sed_cmd, cmdstr);
 469         }
 470         /* handle branch commands */
 471         else if (strchr(":bt", sed_cmd->cmd)) {
 472                 int length;
 473
 474                 cmdstr += strspn(cmdstr, " ");
 475                 length = strcspn(cmdstr, "; \n");
 476                 sed_cmd->label = strndup(cmdstr, length);
 477                 cmdstr += length;
 478         }
 479         /* translation command */
 480         else if (sed_cmd->cmd == 'y') {
 481                 cmdstr += parse_translate_cmd(sed_cmd, cmdstr);
 482         }
 483         /* if it wasnt a single-letter command that takes no arguments
 484          * then it must be an invalid command.
 485          */
 486         else if (strchr("dgGhHnNpPqx={}", sed_cmd->cmd) == 0) {
 487                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 488         }
 489
 490         /* give back whatever's left over */
 491         return (cmdstr);
 492 }
 493
 494 static char *add_cmd(sed_cmd_t * sed_cmd, char *cmdstr)
 495 {
 496         /* Skip over leading whitespace and semicolons */
 497         cmdstr += strspn(cmdstr, semicolon_whitespace);
 498
 499         /* if we ate the whole thing, that means there was just trailing
 500          * whitespace or a final / no-op semicolon. either way, get out */
 501         if (*cmdstr == '\0') {
 502                 return (NULL);
 503         }
 504
 505         /* if this is a comment, jump past it and keep going */
 506         if (*cmdstr == '#') {
 507                 /* "#n" is the same as using -n on the command line */
 508                 if (cmdstr[1] == 'n') {
 509                         be_quiet++;
 510                 }
 511                 return (strpbrk(cmdstr, "\n\r"));
 512         }
 513
 514         /* parse the command
 515          * format is: [addr][,addr]cmd
 516          *            |----||-----||-|
 517          *            part1 part2  part3
 518          */
 519
 520         /* first part (if present) is an address: either a '$', a number or a /regex/ */
 521         cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 522
 523         /* second part (if present) will begin with a comma */
 524         if (*cmdstr == ',') {
 525                 int idx;
 526
 527                 cmdstr++;
 528                 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 529                 if (idx == 0) {
 530                         bb_error_msg_and_die("get_address: no address found in string\n"
 531                                 "\t(you probably didn't check the string you passed me)");
 532                 }
 533                 cmdstr += idx;
 534         }
 535
 536         /* skip whitespace before the command */
 537         while (isspace(*cmdstr)) {
 538                 cmdstr++;
 539         }
 540
 541         /* there my be the inversion flag between part2 and part3 */
 542         if (*cmdstr == '!') {
 543                 sed_cmd->invert = 1;
 544                 cmdstr++;
 545
 546 #ifdef SED_FEATURE_STRICT_CHECKING
 547                 /* According to the spec
 548                  * It is unspecified whether <blank>s can follow a '!' character,
 549                  * and conforming applications shall not follow a '!' character
 550                  * with <blank>s.
 551                  */
 552                 if (isblank(cmdstr[idx]) {
 553                         bb_error_msg_and_die("blank follows '!'");}
 554 #else
 555                 /* skip whitespace before the command */
 556                 while (isspace(*cmdstr)) {
 557                         cmdstr++;
 558                 }
 559 #endif
 560         }
 561
 562         /* last part (mandatory) will be a command */
 563         if (*cmdstr == '\0')
 564                 bb_error_msg_and_die("missing command");
 565
 566         sed_cmd->cmd = *cmdstr;
 567         cmdstr++;
 568
 569         cmdstr = parse_cmd_str(sed_cmd, cmdstr);
 570
 571         /* Add the command to the command array */
 572         sed_cmd_tail->next = sed_cmd;
 573         sed_cmd_tail = sed_cmd_tail->next;
 574
 575         return (cmdstr);
 576 }
 577
 578 static void add_cmd_str(char *cmdstr)
 579 {
 580 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 581         char *cmdstr_ptr = cmdstr;
 582
 583         /* HACK: convert "\n" to match tranlated '\n' string */
 584         while ((cmdstr_ptr = strstr(cmdstr_ptr, "\\n")) != NULL) {
 585                 cmdstr = xrealloc(cmdstr, strlen(cmdstr) + 2);
 586                 cmdstr_ptr = strstr(cmdstr, "\\n");
 587                 memmove(cmdstr_ptr + 1, cmdstr_ptr, strlen(cmdstr_ptr) + 1);
 588                 cmdstr_ptr[0] = '\\';
 589                 cmdstr_ptr += 3;
 590         }
 591 #endif
 592         do {
 593                 sed_cmd_t *sed_cmd;
 594
 595                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 596                 cmdstr = add_cmd(sed_cmd, cmdstr);
 597         } while (cmdstr && strlen(cmdstr));
 598 }
 599
 600
 601 static void load_cmd_file(char *filename)
 602 {
 603         FILE *cmdfile;
 604         char *line;
 605         char *nextline;
 606         char *e;
 607
 608         cmdfile = bb_xfopen(filename, "r");
 609
 610         while ((line = bb_get_line_from_file(cmdfile)) != NULL) {
 611                 /* if a line ends with '\' it needs the next line appended to it */
 612                 while (((e = last_char_is(line, '\n')) != NULL)
 613                         && (e > line) && (e[-1] == '\\')
 614                         && ((nextline = bb_get_line_from_file(cmdfile)) != NULL)) {
 615                         line = xrealloc(line, (e - line) + 1 + strlen(nextline) + 1);
 616                         strcat(line, nextline);
 617                         free(nextline);
 618                 }
 619                 /* eat trailing newline (if any) --if I don't do this, edit commands
 620                  * (aic) will print an extra newline */
 621                 chomp(line);
 622                 add_cmd_str(line);
 623                 free(line);
 624         }
 625 }
 626
 627 struct pipeline {
 628         char *buf;
 629         int idx;
 630         int len;
 631 };
 632
 633 #define PIPE_MAGIC 0x7f
 634 #define PIPE_GROW 64
 635
 636 void pipe_putc(struct pipeline *const pipeline, char c)
 637 {
 638         if (pipeline->buf[pipeline->idx] == PIPE_MAGIC) {
 639                 pipeline->buf = xrealloc(pipeline->buf, pipeline->len + PIPE_GROW);
 640                 memset(pipeline->buf + pipeline->len, 0, PIPE_GROW);
 641                 pipeline->len += PIPE_GROW;
 642                 pipeline->buf[pipeline->len - 1] = PIPE_MAGIC;
 643         }
 644         pipeline->buf[pipeline->idx++] = (c);
 645 }
 646
 647 #define pipeputc(c)     pipe_putc(pipeline, c)
 648
 649 #if 0
 650 {
 651         if (pipeline[pipeline_idx] == PIPE_MAGIC) {
 652                 pipeline = xrealloc(pipeline, pipeline_len + PIPE_GROW);
 653                 memset(pipeline + pipeline_len, 0, PIPE_GROW);
 654                 pipeline_len += PIPE_GROW;
 655                 pipeline[pipeline_len - 1] = PIPE_MAGIC;
 656         }
 657         pipeline[pipeline_idx++] = (c);
 658 }
 659 #endif
 660
 661 static void print_subst_w_backrefs(const char *line, const char *replace,
 662         regmatch_t * regmatch, struct pipeline *const pipeline, int matches)
 663 {
 664         int i;
 665
 666         /* go through the replacement string */
 667         for (i = 0; replace[i]; i++) {
 668                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 669                 if (replace[i] == '\\' && isdigit(replace[i + 1])) {
 670                         int j;
 671                         char tmpstr[2];
 672                         int backref;
 673
 674                         ++i;            /* i now indexes the backref number, instead of the leading slash */
 675                         tmpstr[0] = replace[i];
 676                         tmpstr[1] = 0;
 677                         backref = atoi(tmpstr);
 678                         /* print out the text held in regmatch[backref] */
 679                         if (backref <= matches && regmatch[backref].rm_so != -1)
 680                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo;
 681                                         j++)
 682                                         pipeputc(line[j]);
 683                 }
 684
 685                 /* if we find a backslash escaped character, print the character */
 686                 else if (replace[i] == '\\') {
 687                         ++i;
 688                         pipeputc(replace[i]);
 689                 }
 690
 691                 /* if we find an unescaped '&' print out the whole matched text.
 692                  * fortunately, regmatch[0] contains the indicies to the whole matched
 693                  * expression (kinda seems like it was designed for just such a
 694                  * purpose...) */
 695                 else if (replace[i] == '&') {
 696                         int j;
 697
 698                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 699                                 pipeputc(line[j]);
 700                 }
 701                 /* nothing special, just print this char of the replacement string to stdout */
 702                 else
 703                         pipeputc(replace[i]);
 704         }
 705 }
 706
 707 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 708 {
 709         char *hackline = *line;
 710         struct pipeline thepipe = { NULL, 0, 0 };
 711         struct pipeline *const pipeline = &thepipe;
 712         int altered = 0;
 713         int result;
 714         regmatch_t *regmatch = NULL;
 715         regex_t *current_regex;
 716
 717         if (sed_cmd->sub_match == NULL) {
 718                 current_regex = previous_regex_ptr;
 719         } else {
 720                 previous_regex_ptr = current_regex = sed_cmd->sub_match;
 721         }
 722         result = regexec(current_regex, hackline, 0, NULL, 0);
 723
 724         /* we only proceed if the substitution 'search' expression matches */
 725         if (result == REG_NOMATCH) {
 726                 return 0;
 727         }
 728
 729         /* whaddaya know, it matched. get the number of back references */
 730         regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs + 1));
 731
 732         /* allocate more PIPE_GROW bytes
 733            if replaced string is larger than original */
 734         thepipe.len = strlen(hackline) + PIPE_GROW;
 735         thepipe.buf = xcalloc(1, thepipe.len);
 736         /* buffer magic */
 737         thepipe.buf[thepipe.len - 1] = PIPE_MAGIC;
 738
 739         /* and now, as long as we've got a line to try matching and if we can match
 740          * the search string, we make substitutions */
 741         while ((*hackline || !altered)
 742                 && (regexec(current_regex, hackline, sed_cmd->num_backrefs + 1,
 743                                 regmatch, 0) != REG_NOMATCH)) {
 744                 int i;
 745
 746                 /* print everything before the match */
 747                 for (i = 0; i < regmatch[0].rm_so; i++)
 748                         pipeputc(hackline[i]);
 749
 750                 /* then print the substitution string */
 751                 print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch, pipeline,
 752                         sed_cmd->num_backrefs);
 753
 754                 /* advance past the match */
 755                 hackline += regmatch[0].rm_eo;
 756                 /* flag that something has changed */
 757                 altered++;
 758
 759                 /* if we're not doing this globally, get out now */
 760                 if (!sed_cmd->sub_g) {
 761                         break;
 762                 }
 763         }
 764         for (; *hackline; hackline++)
 765                 pipeputc(*hackline);
 766         if (thepipe.buf[thepipe.idx] == PIPE_MAGIC)
 767                 thepipe.buf[thepipe.idx] = 0;
 768
 769         /* cleanup */
 770         free(regmatch);
 771
 772         free(*line);
 773         *line = thepipe.buf;
 774         return altered;
 775 }
 776
 777 static sed_cmd_t *branch_to(const char *label)
 778 {
 779         sed_cmd_t *sed_cmd;
 780
 781         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 782                 if ((sed_cmd->label) && (strcmp(sed_cmd->label, label) == 0)) {
 783                         break;
 784                 }
 785         }
 786
 787         /* If no match returns last command */
 788         return (sed_cmd);
 789 }
 790
 791 static void process_file(FILE * file)
 792 {
 793         char *pattern_space;    /* Posix requires it be able to hold at least 8192 bytes */
 794         char *hold_space = NULL;        /* Posix requires it be able to hold at least 8192 bytes */
 795         static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
 796         unsigned int still_in_range = 0;
 797         int altered;
 798         int force_print;
 799
 800         pattern_space = bb_get_chomped_line_from_file(file);
 801         if (pattern_space == NULL) {
 802                 return;
 803         }
 804
 805         /* go through every line in the file */
 806         do {
 807                 char *next_line;
 808                 sed_cmd_t *sed_cmd;
 809                 int substituted = 0;
 810                 /* This enables whole blocks of commands to be mask'ed out if the lead address doesnt match */
 811                 int block_mask = 1;
 812
 813                 /* Read one line in advance so we can act on the last line, the '$' address */
 814                 next_line = bb_get_chomped_line_from_file(file);
 815
 816                 linenum++;
 817                 altered = 0;
 818                 force_print = 0;
 819
 820                 /* for every line, go through all the commands */
 821                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 822                         int deleted = 0;
 823
 824                         /*
 825                          * entry point into sedding...
 826                          */
 827                         int matched = (
 828                                 /* no range necessary */
 829                                 (sed_cmd->beg_line == 0 && sed_cmd->end_line == 0
 830                                         && sed_cmd->beg_match == NULL
 831                                         && sed_cmd->end_match == NULL) ||
 832                                 /* this line number is the first address we're looking for */
 833                                 (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum)) ||
 834                                 /* this line matches our first address regex */
 835                                 (sed_cmd->beg_match
 836                                         && (regexec(sed_cmd->beg_match, pattern_space, 0, NULL,
 837                                                         0) == 0)) ||
 838                                 /* we are currently within the beginning & ending address range */
 839                                 still_in_range || ((sed_cmd->beg_line == -1)
 840                                         && (next_line == NULL))
 841                                 );
 842
 843                         if (sed_cmd->cmd == '{') {
 844                                 block_mask = block_mask & matched;
 845                         }
 846 //                      matched &= block_mask;
 847
 848                         if (sed_cmd->invert ^ (matched & block_mask)) {
 849                                 /* Update last used regex incase a blank substitute BRE is found */
 850                                 if (sed_cmd->beg_match) {
 851                                         previous_regex_ptr = sed_cmd->beg_match;
 852                                 }
 853
 854                                 /*
 855                                  * actual sedding
 856                                  */
 857                                 switch (sed_cmd->cmd) {
 858                                 case '=':
 859                                         printf("%d\n", linenum);
 860                                         break;
 861                                 case 'P':{
 862                                         /* Write the current pattern space upto the first newline */
 863                                         char *tmp = strchr(pattern_space, '\n');
 864
 865                                         if (tmp) {
 866                                                 *tmp = '\0';
 867                                         }
 868                                 }
 869                                 case 'p':       /* Write the current pattern space to output */
 870                                         puts(pattern_space);
 871                                         break;
 872                                 case 'd':
 873                                         altered++;
 874                                         deleted = 1;
 875                                         break;
 876
 877                                 case 's':
 878
 879                                         /*
 880                                          * Some special cases for 's' printing to make it compliant with
 881                                          * GNU sed printing behavior (aka "The -n | s///p Matrix"):
 882                                          *
 883                                          *    -n ONLY = never print anything regardless of any successful
 884                                          *    substitution
 885                                          *
 886                                          *    s///p ONLY = always print successful substitutions, even if
 887                                          *    the pattern_space is going to be printed anyway (pattern_space
 888                                          *    will be printed twice).
 889                                          *
 890                                          *    -n AND s///p = print ONLY a successful substitution ONE TIME;
 891                                          *    no other lines are printed - this is the reason why the 'p'
 892                                          *    flag exists in the first place.
 893                                          */
 894
 895 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 896                                         /* HACK: escape newlines twice so regex can match them */
 897                                 {
 898                                         int offset = 0;
 899
 900                                         while (strchr(pattern_space + offset, '\n') != NULL) {
 901                                                 char *tmp;
 902
 903                                                 pattern_space =
 904                                                         xrealloc(pattern_space,
 905                                                         strlen(pattern_space) + 2);
 906                                                 tmp = strchr(pattern_space + offset, '\n');
 907                                                 memmove(tmp + 1, tmp, strlen(tmp) + 1);
 908                                                 tmp[0] = '\\';
 909                                                 tmp[1] = 'n';
 910                                                 offset = tmp - pattern_space + 2;
 911                                         }
 912                                 }
 913 #endif
 914                                         /* we print the pattern_space once, unless we were told to be quiet */
 915                                         substituted = do_subst_command(sed_cmd, &pattern_space);
 916
 917 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 918                                         /* undo HACK: escape newlines twice so regex can match them */
 919                                         {
 920                                                 char *tmp = pattern_space;
 921
 922                                                 while ((tmp = strstr(tmp, "\\n")) != NULL) {
 923                                                         memmove(tmp, tmp + 1, strlen(tmp + 1) + 1);
 924                                                         tmp[0] = '\n';
 925                                                 }
 926                                         }
 927 #endif
 928                                         altered |= substituted;
 929                                         if (!be_quiet && altered && ((sed_cmd->next == NULL)
 930                                                         || (sed_cmd->next->cmd != 's'))) {
 931                                                 force_print = 1;
 932                                         }
 933
 934                                         /* we also print the line if we were given the 'p' flag
 935                                          * (this is quite possibly the second printing) */
 936                                         if ((sed_cmd->sub_p) && altered) {
 937                                                 puts(pattern_space);
 938                                         }
 939                                         break;
 940                                 case 'a':
 941                                         puts(pattern_space);
 942                                         fputs(sed_cmd->editline, stdout);
 943                                         altered++;
 944                                         break;
 945
 946                                 case 'i':
 947                                         fputs(sed_cmd->editline, stdout);
 948                                         break;
 949
 950                                 case 'c':
 951                                         /* single-address case */
 952                                         if ((sed_cmd->end_match == NULL && sed_cmd->end_line == 0)
 953                                                 /* multi-address case */
 954                                                 /* - matching text */
 955                                                 || (sed_cmd->end_match
 956                                                         && (regexec(sed_cmd->end_match, pattern_space, 0,
 957                                                                         NULL, 0) == 0))
 958                                                 /* - matching line numbers */
 959                                                 || (sed_cmd->end_line > 0
 960                                                         && sed_cmd->end_line == linenum)) {
 961                                                 fputs(sed_cmd->editline, stdout);
 962                                         }
 963                                         altered++;
 964
 965                                         break;
 966
 967                                 case 'r':{
 968                                         FILE *outfile;
 969
 970                                         outfile = fopen(sed_cmd->filename, "r");
 971                                         if (outfile) {
 972                                                 char *line;
 973
 974                                                 while ((line =
 975                                                                 bb_get_chomped_line_from_file(outfile)) !=
 976                                                         NULL) {
 977                                                         pattern_space =
 978                                                                 xrealloc(pattern_space,
 979                                                                 strlen(line) + strlen(pattern_space) + 2);
 980                                                         strcat(pattern_space, "\n");
 981                                                         strcat(pattern_space, line);
 982                                                 }
 983                                                 bb_xprint_and_close_file(outfile);
 984                                         }
 985
 986                                 }
 987                                         break;
 988                                 case 'q':       /* Branch to end of script and quit */
 989                                         deleted = 1;
 990                                         /* Exit the outer while loop */
 991                                         free(next_line);
 992                                         next_line = NULL;
 993                                         break;
 994                                 case 'n':       /* Read next line from input */
 995                                         free(pattern_space);
 996                                         pattern_space = next_line;
 997                                         next_line = bb_get_chomped_line_from_file(file);
 998                                         linenum++;
 999                                         break;
1000                                 case 'N':       /* Append the next line to the current line */
1001                                         if (next_line) {
1002                                                 pattern_space =
1003                                                         realloc(pattern_space,
1004                                                         strlen(pattern_space) + strlen(next_line) + 2);
1005                                                 strcat(pattern_space, "\n");
1006                                                 strcat(pattern_space, next_line);
1007                                                 next_line = bb_get_chomped_line_from_file(file);
1008                                                 linenum++;
1009                                         }
1010                                         break;
1011                                 case 'b':
1012                                         sed_cmd = branch_to(sed_cmd->label);
1013                                         break;
1014                                 case 't':
1015                                         if (substituted) {
1016                                                 sed_cmd = branch_to(sed_cmd->label);
1017                                         }
1018                                         break;
1019                                 case 'y':{
1020                                         int i;
1021
1022                                         for (i = 0; pattern_space[i] != 0; i++) {
1023                                                 int j;
1024
1025                                                 for (j = 0; sed_cmd->translate[j]; j += 2) {
1026                                                         if (pattern_space[i] == sed_cmd->translate[j]) {
1027                                                                 pattern_space[i] = sed_cmd->translate[j + 1];
1028                                                         }
1029                                                 }
1030                                         }
1031                                 }
1032                                         break;
1033                                 case 'g':       /* Replace pattern space with hold space */
1034                                         free(pattern_space);
1035                                         pattern_space = strdup(hold_space);
1036                                         break;
1037                                 case 'G': {     /* Append newline and hold space to pattern space */
1038                                         int pattern_space_size = 0;
1039                                         if (pattern_space) {
1040                                                 pattern_space_size = strlen(pattern_space);
1041                                         }
1042                                         pattern_space = xrealloc(pattern_space, pattern_space_size + strlen(hold_space) + 2);
1043                                         strcat(pattern_space, "\n");
1044                                         strcat(pattern_space, hold_space);
1045                                         break;
1046                                 }
1047                                 case 'h':       /* Replace hold space with pattern space */
1048                                         free(hold_space);
1049                                         hold_space = strdup(pattern_space);
1050                                         break;
1051                                 case 'H': {     /* Append newline and pattern space to hold space */
1052                                         int hold_space_size = 0;
1053                                         if (hold_space) {
1054                                                 hold_space_size = strlen(hold_space);
1055                                         }
1056                                         hold_space = xrealloc(hold_space, hold_space_size + strlen(pattern_space) + 2);
1057                                         strcat(hold_space, "\n");
1058                                         strcat(hold_space, pattern_space);
1059                                         break;
1060                                 }
1061                                 case 'x':{
1062                                         /* Swap hold and pattern space */
1063                                         char *tmp = pattern_space;
1064                                         pattern_space = hold_space;
1065                                         hold_space = tmp;
1066                                         break;
1067                                 }
1068                                 }
1069                         }
1070
1071                         /*
1072                          * exit point from sedding...
1073                          */
1074                         if (matched) {
1075                                 if (
1076                                         /* this is a single-address command or... */
1077                                         (sed_cmd->end_line == 0 && sed_cmd->end_match == NULL)
1078                                         /* If only one address */
1079                                         /* we were in the middle of our address range (this
1080                                          * isn't the first time through) and.. */
1081                                         || ((still_in_range == 1)
1082                                                 /* this line number is the last address we're looking for or... */
1083                                                 && ((sed_cmd->end_line > 0
1084                                                                 && (sed_cmd->end_line == linenum))
1085                                                         /* this line matches our last address regex */
1086                                                         || (sed_cmd->end_match
1087                                                                 && (regexec(sed_cmd->end_match, pattern_space,
1088                                                                                 0, NULL, 0) == 0))))) {
1089                                         /* we're out of our address range */
1090                                         still_in_range = 0;
1091                                 } else {
1092                                         /* didn't hit the exit? then we're still in the middle of an address range */
1093                                         still_in_range = 1;
1094                                 }
1095                         }
1096
1097                         if (sed_cmd->cmd == '}') {
1098                                 block_mask = 1;
1099                         }
1100
1101                         if (deleted)
1102                                 break;
1103                 }
1104
1105                 /* we will print the line unless we were told to be quiet or if the
1106                  * line was altered (via a 'd'elete or 's'ubstitution), in which case
1107                  * the altered line was already printed */
1108                 if ((!be_quiet && !altered) || force_print) {
1109                         puts(pattern_space);
1110                 }
1111                 free(pattern_space);
1112                 pattern_space = next_line;
1113         } while (pattern_space);
1114 }
1115
1116 extern int sed_main(int argc, char **argv)
1117 {
1118         int opt, status = EXIT_SUCCESS;
1119
1120 #ifdef CONFIG_FEATURE_CLEAN_UP
1121         /* destroy command strings on exit */
1122         if (atexit(destroy_cmd_strs) == -1)
1123                 bb_perror_msg_and_die("atexit");
1124 #endif
1125
1126         /* do normal option parsing */
1127         while ((opt = getopt(argc, argv, "ne:f:")) > 0) {
1128                 switch (opt) {
1129                 case 'n':
1130                         be_quiet++;
1131                         break;
1132                 case 'e':{
1133                         char *str_cmd = strdup(optarg);
1134
1135                         add_cmd_str(str_cmd);
1136                         free(str_cmd);
1137                         break;
1138                 }
1139                 case 'f':
1140                         load_cmd_file(optarg);
1141                         break;
1142                 default:
1143                         bb_show_usage();
1144                 }
1145         }
1146
1147         /* if we didn't get a pattern from a -e and no command file was specified,
1148          * argv[optind] should be the pattern. no pattern, no worky */
1149         if (sed_cmd_head.next == NULL) {
1150                 if (argv[optind] == NULL)
1151                         bb_show_usage();
1152                 else {
1153                         char *str_cmd = strdup(argv[optind]);
1154
1155                         add_cmd_str(strdup(str_cmd));
1156                         free(str_cmd);
1157                         optind++;
1158                 }
1159         }
1160
1161         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1162          * files were specified or '-' was specified, take input from stdin.
1163          * Otherwise, we process all the files specified. */
1164         if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
1165                 process_file(stdin);
1166         } else {
1167                 int i;
1168                 FILE *file;
1169
1170                 for (i = optind; i < argc; i++) {
1171                         file = bb_wfopen(argv[i], "r");
1172                         if (file) {
1173                                 process_file(file);
1174                                 fclose(file);
1175                         } else
1176                                 status = EXIT_FAILURE;
1177                 }
1178         }
1179
1180         return status;
1181 }