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 more.
  47
  48         Bugs:
  49
  50          - lots
  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         /* Runtime flag no not if the current command match's */
  79         int still_in_range;
  80
  81         /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
  82
  83         /* sed -e 's/sub_match/replace/' */
  84         regex_t *sub_match;
  85         char *replace;
  86
  87         /* EDIT COMMAND (a,i,c) SPECIFIC FIELDS */
  88         char *editline;
  89
  90         /* FILE COMMAND (r) SPECIFIC FIELDS */
  91         char *filename;
  92
  93         /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
  94
  95         unsigned int num_backrefs:4;    /* how many back references (\1..\9) */
  96         /* Note:  GNU/POSIX sed does not save more than nine backrefs, so
  97          * we only use 4 bits to hold the number */
  98         unsigned int sub_g:1;   /* sed -e 's/foo/bar/g' (global) */
  99         unsigned int sub_p:1;   /* sed -e 's/foo/bar/p' (print substitution) */
 100
 101         /* TRANSLATE COMMAND */
 102         char *translate;
 103
 104         /* GENERAL FIELDS */
 105         /* the command */
 106         char cmd;                       /* p,d,s (add more at your leisure :-) */
 107
 108         /* Branch commands */
 109         char *label;
 110
 111         /* next command in list (sequential list of specified commands) */
 112         struct sed_cmd_s *next;
 113
 114 } sed_cmd_t;
 115
 116
 117 /* externs */
 118 extern void xregcomp(regex_t * preg, const char *regex, int cflags);
 119 extern int optind;              /* in unistd.h */
 120 extern char *optarg;    /* ditto */
 121
 122 /* globals */
 123 /* options */
 124 static int be_quiet = 0;
 125 static const char bad_format_in_subst[] =
 126         "bad format in substitution expression";
 127
 128 /* linked list of sed commands */
 129 static sed_cmd_t sed_cmd_head;
 130 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
 131
 132 const char *const semicolon_whitespace = "; \n\r\t\v\0";
 133 static regex_t *previous_regex_ptr = NULL;
 134
 135
 136 #ifdef CONFIG_FEATURE_CLEAN_UP
 137 static void destroy_cmd_strs(void)
 138 {
 139         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 140
 141         while (sed_cmd) {
 142                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 143
 144                 if (sed_cmd->beg_match) {
 145                         regfree(sed_cmd->beg_match);
 146                         free(sed_cmd->beg_match);
 147                 }
 148                 if (sed_cmd->end_match) {
 149                         regfree(sed_cmd->end_match);
 150                         free(sed_cmd->end_match);
 151                 }
 152                 if (sed_cmd->sub_match) {
 153                         regfree(sed_cmd->sub_match);
 154                         free(sed_cmd->sub_match);
 155                 }
 156                 free(sed_cmd->replace);
 157                 free(sed_cmd);
 158                 sed_cmd = sed_cmd_next;
 159         }
 160 }
 161 #endif
 162
 163 /*
 164  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 165  * beginning at a specified index and returns the index of the next regular
 166  * expression delimiter (typically a forward * slash ('/')) not preceeded by
 167  * a backslash ('\').
 168  */
 169 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 170         const char *str)
 171 {
 172         int bracket = -1;
 173         int escaped = 0;
 174         int idx = 0;
 175         char ch;
 176
 177         for (; (ch = str[idx]); idx++) {
 178                 if (bracket != -1) {
 179                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 180                                                 && str[idx - 1] == '^')))
 181                                 bracket = -1;
 182                 } else if (escaped)
 183                         escaped = 0;
 184                 else if (ch == '\\')
 185                         escaped = 1;
 186                 else if (ch == '[')
 187                         bracket = idx;
 188                 else if (ch == delimiter)
 189                         return idx;
 190         }
 191
 192         /* if we make it to here, we've hit the end of the string */
 193         return -1;
 194 }
 195
 196 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 197 {
 198         const char *cmdstr_ptr = cmdstr;
 199         char delimiter;
 200         int idx = 0;
 201
 202         /* verify that the 's' is followed by something.  That something
 203          * (typically a 'slash') is now our regexp delimiter... */
 204         if (*cmdstr == '\0')
 205                 bb_error_msg_and_die(bad_format_in_subst);
 206         else
 207                 delimiter = *cmdstr_ptr;
 208
 209         cmdstr_ptr++;
 210
 211         /* save the match string */
 212         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 213         if (idx == -1) {
 214                 bb_error_msg_and_die(bad_format_in_subst);
 215         }
 216         *match = bb_xstrndup(cmdstr_ptr, idx);
 217
 218         /* save the replacement string */
 219         cmdstr_ptr += idx + 1;
 220         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 221         if (idx == -1) {
 222                 bb_error_msg_and_die(bad_format_in_subst);
 223         }
 224         *replace = bb_xstrndup(cmdstr_ptr, idx);
 225
 226         return ((cmdstr_ptr - cmdstr) + idx);
 227 }
 228
 229 /*
 230  * returns the index in the string just past where the address ends.
 231  */
 232 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 233 {
 234         int idx = 0;
 235
 236         if (isdigit(my_str[idx])) {
 237                 char *endstr;
 238
 239                 *linenum = strtol(my_str, &endstr, 10);
 240                 /* endstr shouldnt ever equal NULL */
 241                 idx = endstr - my_str;
 242         } else if (my_str[idx] == '$') {
 243                 *linenum = -1;
 244                 idx++;
 245         } else if (my_str[idx] == '/' || my_str[idx] == '\\') {
 246                 int idx_start = 1;
 247                 char delimiter;
 248
 249                 delimiter = '/';
 250                 if (my_str[idx] == '\\') {
 251                         idx_start++;
 252                         delimiter = my_str[++idx];
 253                 }
 254                 idx++;
 255                 idx += index_of_next_unescaped_regexp_delim(delimiter, my_str + idx);
 256                 if (idx == -1) {
 257                         bb_error_msg_and_die("unterminated match expression");
 258                 }
 259                 my_str[idx] = '\0';
 260
 261                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 262                 xregcomp(*regex, my_str + idx_start, REG_NEWLINE);
 263                 idx++;                  /* so it points to the next character after the last '/' */
 264         }
 265         return idx;
 266 }
 267
 268 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, const char *substr)
 269 {
 270         int cflags = 0;
 271         char *match;
 272         int idx = 0;
 273         int j;
 274
 275         /*
 276          * the string that gets passed to this function should look like this:
 277          *    s/match/replace/gIp
 278          *    ||     |        |||
 279          *    mandatory       optional
 280          *
 281          *    (all three of the '/' slashes are mandatory)
 282          */
 283         idx = parse_regex_delim(substr, &match, &sed_cmd->replace);
 284
 285         /* determine the number of back references in the match string */
 286         /* Note: we compute this here rather than in the do_subst_command()
 287          * function to save processor time, at the expense of a little more memory
 288          * (4 bits) per sed_cmd */
 289
 290         /* sed_cmd->num_backrefs = 0; *//* XXX: not needed? --apparently not */
 291         for (j = 0; match[j]; j++) {
 292                 /* GNU/POSIX sed does not save more than nine backrefs */
 293                 if (match[j] == '\\' && match[j + 1] == '('
 294                         && sed_cmd->num_backrefs <= 9)
 295                         sed_cmd->num_backrefs++;
 296         }
 297
 298         /* process the flags */
 299         while (substr[++idx]) {
 300                 switch (substr[idx]) {
 301                 case 'g':
 302                         if (match[0] != '^') {
 303                                 sed_cmd->sub_g = 1;
 304                         }
 305                         break;
 306                         /* Hmm, i dont see the I option mentioned in the standard */
 307                 case 'I':
 308                         cflags |= REG_ICASE;
 309                         break;
 310                 case 'p':
 311                         sed_cmd->sub_p = 1;
 312                         break;
 313                 default:
 314                         /* any whitespace or semicolon trailing after a s/// is ok */
 315                         if (strchr(semicolon_whitespace, substr[idx]))
 316                                 goto out;
 317                         /* else */
 318                         bb_error_msg_and_die("bad option in substitution expression");
 319                 }
 320         }
 321
 322   out:
 323         /* compile the match string into a regex */
 324         if (*match != '\0') {
 325                 /* If match is empty, we use last regex used at runtime */
 326                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 327                 xregcomp(sed_cmd->sub_match, match, cflags);
 328         }
 329         free(match);
 330
 331         return idx;
 332 }
 333
 334 static void replace_slash_n(char *string)
 335 {
 336         int i;
 337         int remaining = strlen(string);
 338
 339         for (i = 0; string[i]; i++) {
 340                 if ((string[i] == '\\') && (string[i + 1] == 'n')) {
 341                         string[i] = '\n';
 342                         memmove(string + i + 1, string + i + 1, remaining - 1);
 343                 } else {
 344                         remaining--;
 345                 }
 346         }
 347 }
 348
 349 static int parse_translate_cmd(sed_cmd_t * const sed_cmd, const char *cmdstr)
 350 {
 351         char *match;
 352         char *replace;
 353         int idx;
 354         int i;
 355
 356         idx = parse_regex_delim(cmdstr, &match, &replace);
 357         replace_slash_n(match);
 358         replace_slash_n(replace);
 359         sed_cmd->translate = xcalloc(1, (strlen(match) + 1) * 2);
 360         for (i = 0; (match[i] != 0) && (replace[i] != 0); i++) {
 361                 sed_cmd->translate[i * 2] = match[i];
 362                 sed_cmd->translate[(i * 2) + 1] = replace[i];
 363         }
 364         return (idx + 1);
 365 }
 366
 367 static int parse_edit_cmd(sed_cmd_t * sed_cmd, const char *editstr)
 368 {
 369         int i, j;
 370
 371         /*
 372          * the string that gets passed to this function should look like this:
 373          *
 374          *    need one of these
 375          *    |
 376          *    |    this backslash (immediately following the edit command) is mandatory
 377          *    |    |
 378          *    [aic]\
 379          *    TEXT1\
 380          *    TEXT2\
 381          *    TEXTN
 382          *
 383          * as soon as we hit a TEXT line that has no trailing '\', we're done.
 384          * this means a command like:
 385          *
 386          * i\
 387          * INSERTME
 388          *
 389          * is a-ok.
 390          *
 391          */
 392         if ((*editstr != '\\') || ((editstr[1] != '\n') && (editstr[1] != '\r'))) {
 393                 bb_error_msg_and_die("bad format in edit expression");
 394         }
 395
 396         /* store the edit line text */
 397         sed_cmd->editline = xmalloc(strlen(&editstr[2]) + 2);
 398         for (i = 2, j = 0;
 399                 editstr[i] != '\0' && strchr("\r\n", editstr[i]) == NULL; i++, j++) {
 400                 if ((editstr[i] == '\\') && strchr("\n\r", editstr[i + 1]) != NULL) {
 401                         sed_cmd->editline[j] = '\n';
 402                         i++;
 403                 } else
 404                         sed_cmd->editline[j] = editstr[i];
 405         }
 406
 407         /* figure out if we need to add a newline */
 408         if (sed_cmd->editline[j - 1] != '\n')
 409                 sed_cmd->editline[j++] = '\n';
 410
 411         /* terminate string */
 412         sed_cmd->editline[j] = '\0';
 413
 414         return i;
 415 }
 416
 417
 418 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr)
 419 {
 420         int idx = 0;
 421         int filenamelen = 0;
 422
 423         /*
 424          * the string that gets passed to this function should look like this:
 425          *    '[ ]filename'
 426          *      |  |
 427          *      |  a filename
 428          *      |
 429          *     optional whitespace
 430
 431          *   re: the file to be read, the GNU manual says the following: "Note that
 432          *   if filename cannot be read, it is treated as if it were an empty file,
 433          *   without any error indication." Thus, all of the following commands are
 434          *   perfectly leagal:
 435          *
 436          *   sed -e '1r noexist'
 437          *   sed -e '1r ;'
 438          *   sed -e '1r'
 439          */
 440
 441         /* the file command may be followed by whitespace; move past it. */
 442         while (isspace(filecmdstr[++idx])) {;
 443         }
 444
 445         /* the first non-whitespace we get is a filename. the filename ends when we
 446          * hit a normal sed command terminator or end of string */
 447         filenamelen = strcspn(&filecmdstr[idx], semicolon_whitespace);
 448         sed_cmd->filename = xmalloc(filenamelen + 1);
 449         safe_strncpy(sed_cmd->filename, &filecmdstr[idx], filenamelen + 1);
 450         return idx + filenamelen;
 451 }
 452
 453 /*
 454  *  Process the commands arguments
 455  */
 456 static char *parse_cmd_str(sed_cmd_t * sed_cmd, char *cmdstr)
 457 {
 458         /* handle (s)ubstitution command */
 459         if (sed_cmd->cmd == 's') {
 460                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 461         }
 462         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 463         else if (strchr("aic", sed_cmd->cmd)) {
 464                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 465                         bb_error_msg_and_die
 466                                 ("only a beginning address can be specified for edit commands");
 467                 cmdstr += parse_edit_cmd(sed_cmd, cmdstr);
 468         }
 469         /* handle file cmds: (r)ead */
 470         else if (sed_cmd->cmd == 'r') {
 471                 if (sed_cmd->end_line || sed_cmd->end_match)
 472                         bb_error_msg_and_die("Command only uses one address");
 473                 cmdstr += parse_file_cmd(sed_cmd, cmdstr);
 474         }
 475         /* handle branch commands */
 476         else if (strchr(":bt", sed_cmd->cmd)) {
 477                 int length;
 478
 479                 cmdstr += strspn(cmdstr, " ");
 480                 length = strcspn(cmdstr, "; \n");
 481                 sed_cmd->label = strndup(cmdstr, length);
 482                 cmdstr += length;
 483         }
 484         /* translation command */
 485         else if (sed_cmd->cmd == 'y') {
 486                 cmdstr += parse_translate_cmd(sed_cmd, cmdstr);
 487         }
 488         /* if it wasnt a single-letter command that takes no arguments
 489          * then it must be an invalid command.
 490          */
 491         else if (strchr("dgGhHnNpPqx={}", sed_cmd->cmd) == 0) {
 492                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 493         }
 494
 495         /* give back whatever's left over */
 496         return (cmdstr);
 497 }
 498
 499 static char *add_cmd(sed_cmd_t *sed_cmd, char *cmdstr)
 500 {
 501         /* Skip over leading whitespace and semicolons */
 502         cmdstr += strspn(cmdstr, semicolon_whitespace);
 503
 504         /* if we ate the whole thing, that means there was just trailing
 505          * whitespace or a final / no-op semicolon. either way, get out */
 506         if (*cmdstr == '\0') {
 507                 return (NULL);
 508         }
 509
 510         /* if this is a comment, jump past it and keep going */
 511         if (*cmdstr == '#') {
 512                 /* "#n" is the same as using -n on the command line */
 513                 if (cmdstr[1] == 'n') {
 514                         be_quiet++;
 515                 }
 516                 return (strpbrk(cmdstr, "\n\r"));
 517         }
 518
 519         /* parse the command
 520          * format is: [addr][,addr]cmd
 521          *            |----||-----||-|
 522          *            part1 part2  part3
 523          */
 524
 525         /* first part (if present) is an address: either a '$', a number or a /regex/ */
 526         cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 527
 528         /* second part (if present) will begin with a comma */
 529         if (*cmdstr == ',') {
 530                 int idx;
 531
 532                 cmdstr++;
 533                 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 534                 if (idx == 0) {
 535                         bb_error_msg_and_die("get_address: no address found in string\n"
 536                                 "\t(you probably didn't check the string you passed me)");
 537                 }
 538                 cmdstr += idx;
 539         }
 540
 541         /* skip whitespace before the command */
 542         while (isspace(*cmdstr)) {
 543                 cmdstr++;
 544         }
 545
 546         /* there my be the inversion flag between part2 and part3 */
 547         if (*cmdstr == '!') {
 548                 sed_cmd->invert = 1;
 549                 cmdstr++;
 550
 551 #ifdef SED_FEATURE_STRICT_CHECKING
 552                 /* According to the spec
 553                  * It is unspecified whether <blank>s can follow a '!' character,
 554                  * and conforming applications shall not follow a '!' character
 555                  * with <blank>s.
 556                  */
 557                 if (isblank(cmdstr[idx]) {
 558                         bb_error_msg_and_die("blank follows '!'");}
 559 #else
 560                 /* skip whitespace before the command */
 561                 while (isspace(*cmdstr)) {
 562                         cmdstr++;
 563                 }
 564 #endif
 565         }
 566
 567         /* last part (mandatory) will be a command */
 568         if (*cmdstr == '\0')
 569                 bb_error_msg_and_die("missing command");
 570
 571         sed_cmd->cmd = *cmdstr;
 572         cmdstr++;
 573
 574         cmdstr = parse_cmd_str(sed_cmd, cmdstr);
 575
 576         /* Add the command to the command array */
 577         sed_cmd_tail->next = sed_cmd;
 578         sed_cmd_tail = sed_cmd_tail->next;
 579
 580         return (cmdstr);
 581 }
 582
 583 static void add_cmd_str(char *cmdstr)
 584 {
 585 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 586         char *cmdstr_ptr = cmdstr;
 587
 588         /* HACK: convert "\n" to match tranlated '\n' string */
 589         while ((cmdstr_ptr = strstr(cmdstr_ptr, "\\n")) != NULL) {
 590                 cmdstr = xrealloc(cmdstr, strlen(cmdstr) + 2);
 591                 cmdstr_ptr = strstr(cmdstr, "\\n");
 592                 memmove(cmdstr_ptr + 1, cmdstr_ptr, strlen(cmdstr_ptr) + 1);
 593                 cmdstr_ptr[0] = '\\';
 594                 cmdstr_ptr += 3;
 595         }
 596 #endif
 597         do {
 598                 sed_cmd_t *sed_cmd;
 599
 600                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 601                 cmdstr = add_cmd(sed_cmd, cmdstr);
 602         } while (cmdstr && strlen(cmdstr));
 603 }
 604
 605
 606 static void load_cmd_file(char *filename)
 607 {
 608         FILE *cmdfile;
 609         char *line;
 610         char *nextline;
 611         char *e;
 612
 613         cmdfile = bb_xfopen(filename, "r");
 614
 615         while ((line = bb_get_line_from_file(cmdfile)) != NULL) {
 616                 /* if a line ends with '\' it needs the next line appended to it */
 617                 while (((e = last_char_is(line, '\n')) != NULL)
 618                         && (e > line) && (e[-1] == '\\')
 619                         && ((nextline = bb_get_line_from_file(cmdfile)) != NULL)) {
 620                         line = xrealloc(line, (e - line) + 1 + strlen(nextline) + 1);
 621                         strcat(line, nextline);
 622                         free(nextline);
 623                 }
 624                 /* eat trailing newline (if any) --if I don't do this, edit commands
 625                  * (aic) will print an extra newline */
 626                 chomp(line);
 627                 add_cmd_str(line);
 628                 free(line);
 629         }
 630 }
 631
 632 struct pipeline {
 633         char *buf;
 634         int idx;
 635         int len;
 636 };
 637
 638 #define PIPE_MAGIC 0x7f
 639 #define PIPE_GROW 64
 640
 641 void pipe_putc(struct pipeline *const pipeline, char c)
 642 {
 643         if (pipeline->buf[pipeline->idx] == PIPE_MAGIC) {
 644                 pipeline->buf = xrealloc(pipeline->buf, pipeline->len + PIPE_GROW);
 645                 memset(pipeline->buf + pipeline->len, 0, PIPE_GROW);
 646                 pipeline->len += PIPE_GROW;
 647                 pipeline->buf[pipeline->len - 1] = PIPE_MAGIC;
 648         }
 649         pipeline->buf[pipeline->idx++] = (c);
 650 }
 651
 652 #define pipeputc(c)     pipe_putc(pipeline, c)
 653
 654 #if 0
 655 {
 656         if (pipeline[pipeline_idx] == PIPE_MAGIC) {
 657                 pipeline = xrealloc(pipeline, pipeline_len + PIPE_GROW);
 658                 memset(pipeline + pipeline_len, 0, PIPE_GROW);
 659                 pipeline_len += PIPE_GROW;
 660                 pipeline[pipeline_len - 1] = PIPE_MAGIC;
 661         }
 662         pipeline[pipeline_idx++] = (c);
 663 }
 664 #endif
 665
 666 static void print_subst_w_backrefs(const char *line, const char *replace,
 667         regmatch_t * regmatch, struct pipeline *const pipeline, int matches)
 668 {
 669         int i;
 670
 671         /* go through the replacement string */
 672         for (i = 0; replace[i]; i++) {
 673                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 674                 if (replace[i] == '\\' && isdigit(replace[i + 1])) {
 675                         int j;
 676                         char tmpstr[2];
 677                         int backref;
 678
 679                         ++i;            /* i now indexes the backref number, instead of the leading slash */
 680                         tmpstr[0] = replace[i];
 681                         tmpstr[1] = 0;
 682                         backref = atoi(tmpstr);
 683                         /* print out the text held in regmatch[backref] */
 684                         if (backref <= matches && regmatch[backref].rm_so != -1)
 685                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo;
 686                                         j++)
 687                                         pipeputc(line[j]);
 688                 }
 689
 690                 /* if we find a backslash escaped character, print the character */
 691                 else if (replace[i] == '\\') {
 692                         ++i;
 693                         pipeputc(replace[i]);
 694                 }
 695
 696                 /* if we find an unescaped '&' print out the whole matched text.
 697                  * fortunately, regmatch[0] contains the indicies to the whole matched
 698                  * expression (kinda seems like it was designed for just such a
 699                  * purpose...) */
 700                 else if (replace[i] == '&') {
 701                         int j;
 702
 703                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 704                                 pipeputc(line[j]);
 705                 }
 706                 /* nothing special, just print this char of the replacement string to stdout */
 707                 else
 708                         pipeputc(replace[i]);
 709         }
 710 }
 711
 712 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 713 {
 714         char *hackline = *line;
 715         struct pipeline thepipe = { NULL, 0, 0 };
 716         struct pipeline *const pipeline = &thepipe;
 717         int altered = 0;
 718         int result;
 719         regmatch_t *regmatch = NULL;
 720         regex_t *current_regex;
 721
 722         if (sed_cmd->sub_match == NULL) {
 723                 current_regex = previous_regex_ptr;
 724         } else {
 725                 previous_regex_ptr = current_regex = sed_cmd->sub_match;
 726         }
 727         result = regexec(current_regex, hackline, 0, NULL, 0);
 728
 729         /* we only proceed if the substitution 'search' expression matches */
 730         if (result == REG_NOMATCH) {
 731                 return 0;
 732         }
 733
 734         /* whaddaya know, it matched. get the number of back references */
 735         regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs + 1));
 736
 737         /* allocate more PIPE_GROW bytes
 738            if replaced string is larger than original */
 739         thepipe.len = strlen(hackline) + PIPE_GROW;
 740         thepipe.buf = xcalloc(1, thepipe.len);
 741         /* buffer magic */
 742         thepipe.buf[thepipe.len - 1] = PIPE_MAGIC;
 743
 744         /* and now, as long as we've got a line to try matching and if we can match
 745          * the search string, we make substitutions */
 746         while ((*hackline || !altered)
 747                 && (regexec(current_regex, hackline, sed_cmd->num_backrefs + 1,
 748                                 regmatch, 0) != REG_NOMATCH)) {
 749                 int i;
 750
 751                 /* print everything before the match */
 752                 for (i = 0; i < regmatch[0].rm_so; i++)
 753                         pipeputc(hackline[i]);
 754
 755                 /* then print the substitution string */
 756                 print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch, pipeline,
 757                         sed_cmd->num_backrefs);
 758
 759                 /* advance past the match */
 760                 hackline += regmatch[0].rm_eo;
 761                 /* flag that something has changed */
 762                 altered++;
 763
 764                 /* if we're not doing this globally, get out now */
 765                 if (!sed_cmd->sub_g) {
 766                         break;
 767                 }
 768         }
 769         for (; *hackline; hackline++)
 770                 pipeputc(*hackline);
 771         if (thepipe.buf[thepipe.idx] == PIPE_MAGIC)
 772                 thepipe.buf[thepipe.idx] = 0;
 773
 774         /* cleanup */
 775         free(regmatch);
 776
 777         free(*line);
 778         *line = thepipe.buf;
 779         return altered;
 780 }
 781
 782 static sed_cmd_t *branch_to(const char *label)
 783 {
 784         sed_cmd_t *sed_cmd;
 785
 786         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 787                 if ((sed_cmd->label) && (strcmp(sed_cmd->label, label) == 0)) {
 788                         break;
 789                 }
 790         }
 791
 792         /* If no match returns last command */
 793         return (sed_cmd);
 794 }
 795
 796 static void process_file(FILE * file)
 797 {
 798         char *pattern_space;    /* Posix requires it be able to hold at least 8192 bytes */
 799         char *hold_space = NULL;        /* Posix requires it be able to hold at least 8192 bytes */
 800         static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
 801         int altered;
 802         int force_print;
 803
 804         pattern_space = bb_get_chomped_line_from_file(file);
 805         if (pattern_space == NULL) {
 806                 return;
 807         }
 808
 809         /* go through every line in the file */
 810         do {
 811                 char *next_line;
 812                 sed_cmd_t *sed_cmd;
 813                 int substituted = 0;
 814                 /* This enables whole blocks of commands to be mask'ed out if the lead address doesnt match */
 815                 int block_mask = 1;
 816
 817                 /* Read one line in advance so we can act on the last line, the '$' address */
 818                 next_line = bb_get_chomped_line_from_file(file);
 819
 820                 linenum++;
 821                 altered = 0;
 822                 force_print = 0;
 823
 824                 /* for every line, go through all the commands */
 825                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 826                         int deleted = 0;
 827
 828                         /*
 829                          * entry point into sedding...
 830                          */
 831                         int matched = (
 832                                 /* no range necessary */
 833                                 (sed_cmd->beg_line == 0 && sed_cmd->end_line == 0
 834                                         && sed_cmd->beg_match == NULL
 835                                         && sed_cmd->end_match == NULL) ||
 836                                 /* this line number is the first address we're looking for */
 837                                 (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum)) ||
 838                                 /* this line matches our first address regex */
 839                                 (sed_cmd->beg_match
 840                                         && (regexec(sed_cmd->beg_match, pattern_space, 0, NULL,
 841                                                         0) == 0)) ||
 842                                 /* we are currently within the beginning & ending address range */
 843                                 sed_cmd->still_in_range || ((sed_cmd->beg_line == -1)
 844                                         && (next_line == NULL))
 845                                 );
 846                         if (sed_cmd->cmd == '{') {
 847                                 block_mask = block_mask & matched;
 848                         }
 849 //                      matched &= block_mask;
 850
 851                         if (sed_cmd->invert ^ (matched & block_mask)) {
 852                                 /* Update last used regex incase a blank substitute BRE is found */
 853                                 if (sed_cmd->beg_match) {
 854                                         previous_regex_ptr = sed_cmd->beg_match;
 855                                 }
 856
 857                                 /*
 858                                  * actual sedding
 859                                  */
 860                                 switch (sed_cmd->cmd) {
 861                                 case '=':
 862                                         printf("%d\n", linenum);
 863                                         break;
 864                                 case 'P':{
 865                                         /* Write the current pattern space upto the first newline */
 866                                         char *tmp = strchr(pattern_space, '\n');
 867
 868                                         if (tmp) {
 869                                                 *tmp = '\0';
 870                                         }
 871                                 }
 872                                 case 'p':       /* Write the current pattern space to output */
 873                                         puts(pattern_space);
 874                                         break;
 875                                 case 'd':
 876                                         altered++;
 877                                         deleted = 1;
 878                                         break;
 879
 880                                 case 's':
 881
 882                                         /*
 883                                          * Some special cases for 's' printing to make it compliant with
 884                                          * GNU sed printing behavior (aka "The -n | s///p Matrix"):
 885                                          *
 886                                          *    -n ONLY = never print anything regardless of any successful
 887                                          *    substitution
 888                                          *
 889                                          *    s///p ONLY = always print successful substitutions, even if
 890                                          *    the pattern_space is going to be printed anyway (pattern_space
 891                                          *    will be printed twice).
 892                                          *
 893                                          *    -n AND s///p = print ONLY a successful substitution ONE TIME;
 894                                          *    no other lines are printed - this is the reason why the 'p'
 895                                          *    flag exists in the first place.
 896                                          */
 897
 898 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 899                                         /* HACK: escape newlines twice so regex can match them */
 900                                 {
 901                                         int offset = 0;
 902
 903                                         while (strchr(pattern_space + offset, '\n') != NULL) {
 904                                                 char *tmp;
 905
 906                                                 pattern_space =
 907                                                         xrealloc(pattern_space,
 908                                                         strlen(pattern_space) + 2);
 909                                                 tmp = strchr(pattern_space + offset, '\n');
 910                                                 memmove(tmp + 1, tmp, strlen(tmp) + 1);
 911                                                 tmp[0] = '\\';
 912                                                 tmp[1] = 'n';
 913                                                 offset = tmp - pattern_space + 2;
 914                                         }
 915                                 }
 916 #endif
 917                                         /* we print the pattern_space once, unless we were told to be quiet */
 918                                         substituted = do_subst_command(sed_cmd, &pattern_space);
 919
 920 #ifdef CONFIG_FEATURE_SED_EMBEDED_NEWLINE
 921                                         /* undo HACK: escape newlines twice so regex can match them */
 922                                         {
 923                                                 char *tmp = pattern_space;
 924
 925                                                 while ((tmp = strstr(tmp, "\\n")) != NULL) {
 926                                                         memmove(tmp, tmp + 1, strlen(tmp + 1) + 1);
 927                                                         tmp[0] = '\n';
 928                                                 }
 929                                         }
 930 #endif
 931                                         altered = substituted;
 932                                         if (!be_quiet && altered && ((sed_cmd->next == NULL)
 933                                                         || (sed_cmd->next->cmd != 's'))) {
 934                                                 force_print = 1;
 935                                         }
 936
 937                                         /* we also print the line if we were given the 'p' flag
 938                                          * (this is quite possibly the second printing) */
 939                                         if ((sed_cmd->sub_p) && altered) {
 940                                                 puts(pattern_space);
 941                                         }
 942                                         break;
 943                                 case 'a':
 944                                         puts(pattern_space);
 945                                         fputs(sed_cmd->editline, stdout);
 946                                         altered++;
 947                                         break;
 948
 949                                 case 'i':
 950                                         fputs(sed_cmd->editline, stdout);
 951                                         break;
 952
 953                                 case 'c':
 954                                         /* single-address case */
 955                                         if ((sed_cmd->end_match == NULL && sed_cmd->end_line == 0)
 956                                                 /* multi-address case */
 957                                                 /* - matching text */
 958                                                 || (sed_cmd->end_match
 959                                                         && (regexec(sed_cmd->end_match, pattern_space, 0,
 960                                                                         NULL, 0) == 0))
 961                                                 /* - matching line numbers */
 962                                                 || (sed_cmd->end_line > 0
 963                                                         && sed_cmd->end_line == linenum)) {
 964                                                 fputs(sed_cmd->editline, stdout);
 965                                         }
 966                                         altered++;
 967
 968                                         break;
 969
 970                                 case 'r':{
 971                                         FILE *outfile;
 972
 973                                         outfile = fopen(sed_cmd->filename, "r");
 974                                         if (outfile) {
 975                                                 char *line;
 976
 977                                                 while ((line =
 978                                                                 bb_get_chomped_line_from_file(outfile)) !=
 979                                                         NULL) {
 980                                                         pattern_space =
 981                                                                 xrealloc(pattern_space,
 982                                                                 strlen(line) + strlen(pattern_space) + 2);
 983                                                         strcat(pattern_space, "\n");
 984                                                         strcat(pattern_space, line);
 985                                                 }
 986                                                 bb_xprint_and_close_file(outfile);
 987                                         }
 988
 989                                 }
 990                                         break;
 991                                 case 'q':       /* Branch to end of script and quit */
 992                                         deleted = 1;
 993                                         /* Exit the outer while loop */
 994                                         free(next_line);
 995                                         next_line = NULL;
 996                                         break;
 997                                 case 'n':       /* Read next line from input */
 998                                         free(pattern_space);
 999                                         pattern_space = next_line;
1000                                         next_line = bb_get_chomped_line_from_file(file);
1001                                         linenum++;
1002                                         break;
1003                                 case 'N':       /* Append the next line to the current line */
1004                                         if (next_line) {
1005                                                 pattern_space =
1006                                                         realloc(pattern_space,
1007                                                         strlen(pattern_space) + strlen(next_line) + 2);
1008                                                 strcat(pattern_space, "\n");
1009                                                 strcat(pattern_space, next_line);
1010                                                 next_line = bb_get_chomped_line_from_file(file);
1011                                                 linenum++;
1012                                         }
1013                                         break;
1014                                 case 'b':
1015                                         sed_cmd = branch_to(sed_cmd->label);
1016                                         break;
1017                                 case 't':
1018                                         if (substituted) {
1019                                                 sed_cmd = branch_to(sed_cmd->label);
1020                                         }
1021                                         break;
1022                                 case 'y':{
1023                                         int i;
1024
1025                                         for (i = 0; pattern_space[i] != 0; i++) {
1026                                                 int j;
1027
1028                                                 for (j = 0; sed_cmd->translate[j]; j += 2) {
1029                                                         if (pattern_space[i] == sed_cmd->translate[j]) {
1030                                                                 pattern_space[i] = sed_cmd->translate[j + 1];
1031                                                         }
1032                                                 }
1033                                         }
1034                                 }
1035                                         break;
1036                                 case 'g':       /* Replace pattern space with hold space */
1037                                         free(pattern_space);
1038                                         pattern_space = strdup(hold_space);
1039                                         break;
1040                                 case 'G': {     /* Append newline and hold space to pattern space */
1041                                         int pattern_space_size = 0;
1042                                         if (pattern_space) {
1043                                                 pattern_space_size = strlen(pattern_space);
1044                                         }
1045                                         pattern_space = xrealloc(pattern_space, pattern_space_size + strlen(hold_space) + 2);
1046                                         strcat(pattern_space, "\n");
1047                                         strcat(pattern_space, hold_space);
1048                                         break;
1049                                 }
1050                                 case 'h':       /* Replace hold space with pattern space */
1051                                         free(hold_space);
1052                                         hold_space = strdup(pattern_space);
1053                                         break;
1054                                 case 'H': {     /* Append newline and pattern space to hold space */
1055                                         int hold_space_size = 0;
1056                                         if (hold_space) {
1057                                                 hold_space_size = strlen(hold_space);
1058                                         }
1059                                         hold_space = xrealloc(hold_space, hold_space_size + strlen(pattern_space) + 2);
1060                                         if (hold_space_size) {
1061                                                 strcat(hold_space, "\n");
1062                                         } else {
1063                                                 hold_space[0] = '\n';
1064                                         }
1065                                         strcat(hold_space, pattern_space);
1066                                         break;
1067                                 }
1068                                 case 'x':{
1069                                         /* Swap hold and pattern space */
1070                                         char *tmp = pattern_space;
1071                                         pattern_space = hold_space;
1072                                         hold_space = tmp;
1073                                         break;
1074                                 }
1075                                 }
1076                         }
1077
1078                         /*
1079                          * exit point from sedding...
1080                          */
1081                         if (matched) {
1082                                 if (
1083                                         /* this is a single-address command or... */
1084                                         (sed_cmd->end_line == 0 && sed_cmd->end_match == NULL)
1085                                         /* If only one address */
1086                                         /* we were in the middle of our address range (this
1087                                          * isn't the first time through) and.. */
1088                                         || ((sed_cmd->still_in_range == 1)
1089                                                 /* this line number is the last address we're looking for or... */
1090                                                 && ((sed_cmd->end_line > 0
1091                                                                 && (sed_cmd->end_line == linenum))
1092                                                         /* this line matches our last address regex */
1093                                                         || (sed_cmd->end_match
1094                                                                 && (regexec(sed_cmd->end_match, pattern_space,
1095                                                                                 0, NULL, 0) == 0))))) {
1096                                         /* we're out of our address range */
1097                                         sed_cmd->still_in_range = 0;
1098                                 } else {
1099                                         /* didn't hit the exit? then we're still in the middle of an address range */
1100                                         sed_cmd->still_in_range = 1;
1101                                 }
1102                         }
1103
1104                         if (sed_cmd->cmd == '}') {
1105                                 block_mask = 1;
1106                         }
1107
1108                         if (deleted)
1109                                 break;
1110
1111                 }
1112
1113                 /* we will print the line unless we were told to be quiet or if the
1114                  * line was altered (via a 'd'elete or 's'ubstitution), in which case
1115                  * the altered line was already printed */
1116                 if ((!be_quiet && !altered) || force_print) {
1117                         puts(pattern_space);
1118                 }
1119                 free(pattern_space);
1120                 pattern_space = next_line;
1121         } while (pattern_space);
1122 }
1123
1124 extern int sed_main(int argc, char **argv)
1125 {
1126         int opt, status = EXIT_SUCCESS;
1127
1128 #ifdef CONFIG_FEATURE_CLEAN_UP
1129         /* destroy command strings on exit */
1130         if (atexit(destroy_cmd_strs) == -1)
1131                 bb_perror_msg_and_die("atexit");
1132 #endif
1133
1134         /* do normal option parsing */
1135         while ((opt = getopt(argc, argv, "ne:f:")) > 0) {
1136                 switch (opt) {
1137                 case 'n':
1138                         be_quiet++;
1139                         break;
1140                 case 'e':{
1141                         char *str_cmd = strdup(optarg);
1142
1143                         add_cmd_str(str_cmd);
1144                         free(str_cmd);
1145                         break;
1146                 }
1147                 case 'f':
1148                         load_cmd_file(optarg);
1149                         break;
1150                 default:
1151                         bb_show_usage();
1152                 }
1153         }
1154
1155         /* if we didn't get a pattern from a -e and no command file was specified,
1156          * argv[optind] should be the pattern. no pattern, no worky */
1157         if (sed_cmd_head.next == NULL) {
1158                 if (argv[optind] == NULL)
1159                         bb_show_usage();
1160                 else {
1161                         char *str_cmd = strdup(argv[optind]);
1162
1163                         add_cmd_str(strd_cmd);
1164                         free(str_cmd);
1165                         optind++;
1166                 }
1167         }
1168
1169         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1170          * files were specified or '-' was specified, take input from stdin.
1171          * Otherwise, we process all the files specified. */
1172         if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
1173                 process_file(stdin);
1174         } else {
1175                 int i;
1176                 FILE *file;
1177
1178                 for (i = optind; i < argc; i++) {
1179                         file = bb_wfopen(argv[i], "r");
1180                         if (file) {
1181                                 process_file(file);
1182                                 fclose(file);
1183                         } else
1184                                 status = EXIT_FAILURE;
1185                 }
1186         }
1187
1188         return status;
1189 }