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