editors/sed.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * sed.c - very minimalist version of sed
   4  *
   5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   7  * Copyright (C) 2002  Matt Kraai
   8  * Copyright (C) 2003 by Glenn McGrath <bug1@iinet.net.au>
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * This program is free software; you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation; either version 2 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  19  * General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program; if not, write to the Free Software
  23  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  24  *
  25  */
  26
  27 /* Code overview.
  28
  29   Files are laid out to avoid unnecessary function declarations.  So for
  30   example, every function add_cmd calls occurs before add_cmd in this file.
  31
  32   add_cmd() is called on each line of sed command text (from a file or from
  33   the command line).  It calls get_address() and parse_cmd_args().  The
  34   resulting sed_cmd_t structures are appended to a linked list
  35   (sed_cmd_head/sed_cmd_tail).
  36
  37   process_file() does actual sedding, reading data lines from an input FILE *
  38   (which could be stdin) and applying the sed command list (sed_cmd_head) to
  39   each of the resulting lines.
  40
  41   sed_main() is where external code calls into this, with a command line.
  42 */
  43
  44
  45 /*
  46         Supported features and commands in this version of sed:
  47
  48          - comments ('#')
  49          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  50          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  51          - edit commands: (a)ppend, (i)nsert, (c)hange
  52          - file commands: (r)ead
  53          - backreferences in substitution expressions (\1, \2...\9)
  54          - grouped commands: {cmd1;cmd2}
  55          - transliteration (y/source-chars/dest-chars/)
  56          - pattern space hold space storing / swapping (g, h, x)
  57          - labels / branching (: label, b, t)
  58
  59          (Note: Specifying an address (range) to match is *optional*; commands
  60          default to the whole pattern space if no specific address match was
  61          requested.)
  62
  63         Unsupported features:
  64
  65          - GNU extensions
  66          - and more.
  67
  68         Todo:
  69
  70          - Create a wrapper around regex to make libc's regex conform with sed
  71          - Fix bugs
  72
  73
  74         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  75 */
  76
  77 #include <stdio.h>
  78 #include <unistd.h>             /* for getopt() */
  79 #include <regex.h>
  80 #include <string.h>             /* for strdup() */
  81 #include <errno.h>
  82 #include <ctype.h>              /* for isspace() */
  83 #include <stdlib.h>
  84 #include "busybox.h"
  85
  86 typedef struct sed_cmd_s {
  87     /* Ordered by alignment requirements: currently 36 bytes on x86 */
  88
  89     /* address storage */
  90     regex_t *beg_match; /* sed -e '/match/cmd' */
  91     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
  92     regex_t *sub_match; /* For 's/sub_match/string/' */
  93     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
  94     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  95
  96     FILE *file;                 /* File (sr) command writes to, -1 for none. */
  97     char *string;               /* Data string for (saicytb) commands. */
  98
  99     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
 100
 101     /* Bitfields (gcc won't group them if we don't) */
 102     unsigned int invert:1;                      /* the '!' after the address */
 103     unsigned int in_match:1;            /* Next line also included in match? */
 104     unsigned int no_newline:1;          /* Last line written by (sr) had no '\n' */
 105     unsigned int sub_p:1;                       /* (s) print option */
 106
 107
 108     /* GENERAL FIELDS */
 109     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 110     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
 111 } sed_cmd_t;
 112
 113 /* globals */
 114 /* options */
 115 static int be_quiet, in_place, regex_type;
 116 FILE *nonstdout;
 117 char *outname,*hold_space;
 118
 119
 120 static const char bad_format_in_subst[] =
 121         "bad format in substitution expression";
 122 const char *const semicolon_whitespace = "; \n\r\t\v";
 123
 124 regmatch_t regmatch[10];
 125 static regex_t *previous_regex_ptr;
 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 /* Linked list of append lines */
 132 struct append_list {
 133         char *string;
 134         struct append_list *next;
 135 };
 136 struct append_list *append_head=NULL, *append_tail=NULL;
 137
 138 #ifdef CONFIG_FEATURE_CLEAN_UP
 139 static void free_and_close_stuff(void)
 140 {
 141         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 142
 143         while(append_head) {
 144                 append_tail=append_head->next;
 145                 free(append_head->string);
 146                 free(append_head);
 147                 append_head=append_tail;
 148         }
 149
 150         while (sed_cmd) {
 151                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 152
 153                 if(sed_cmd->file)
 154                         bb_xprint_and_close_file(sed_cmd->file);
 155
 156                 if (sed_cmd->beg_match) {
 157                         regfree(sed_cmd->beg_match);
 158                         free(sed_cmd->beg_match);
 159                 }
 160                 if (sed_cmd->end_match) {
 161                         regfree(sed_cmd->end_match);
 162                         free(sed_cmd->end_match);
 163                 }
 164                 if (sed_cmd->sub_match) {
 165                         regfree(sed_cmd->sub_match);
 166                         free(sed_cmd->sub_match);
 167                 }
 168                 free(sed_cmd->string);
 169                 free(sed_cmd);
 170                 sed_cmd = sed_cmd_next;
 171         }
 172
 173         if(hold_space) free(hold_space);
 174 }
 175 #endif
 176
 177 /* If something bad happens during -i operation, delete temp file */
 178
 179 static void cleanup_outname(void)
 180 {
 181   if(outname) unlink(outname);
 182 }
 183
 184 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 185
 186 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 187 {
 188         int i=0;
 189
 190         while(i<len) {
 191                 if(string[i] == '\\') {
 192                         if(!to || string[i+1] == from) {
 193                                 *(dest++) = to ? to : string[i+1];
 194                                 i+=2;
 195                                 continue;
 196                         } else *(dest++)=string[i++];
 197                 }
 198                 *(dest++) = string[i++];
 199         }
 200         *dest=0;
 201 }
 202
 203 static char *copy_parsing_slashn(const char *string, int len)
 204 {
 205         char *dest=xmalloc(len+1);
 206
 207         parse_escapes(dest,string,len,'n','\n');
 208         return dest;
 209 }
 210
 211
 212 /*
 213  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 214  * beginning at a specified index and returns the index of the next regular
 215  * expression delimiter (typically a forward * slash ('/')) not preceded by
 216  * a backslash ('\').
 217  */
 218 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 219         const char *str)
 220 {
 221         int bracket = -1;
 222         int escaped = 0;
 223         int idx = 0;
 224         char ch;
 225
 226         for (; (ch = str[idx]); idx++) {
 227                 if (bracket != -1) {
 228                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 229                                         && str[idx - 1] == '^')))
 230                                 bracket = -1;
 231                 } else if (escaped)
 232                         escaped = 0;
 233                 else if (ch == '\\')
 234                         escaped = 1;
 235                 else if (ch == '[')
 236                         bracket = idx;
 237                 else if (ch == delimiter)
 238                         return idx;
 239         }
 240
 241         /* if we make it to here, we've hit the end of the string */
 242         return -1;
 243 }
 244
 245 /*
 246  *  Returns the index of the third delimiter
 247  */
 248 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 249 {
 250         const char *cmdstr_ptr = cmdstr;
 251         char delimiter;
 252         int idx = 0;
 253
 254         /* verify that the 's' or 'y' is followed by something.  That something
 255          * (typically a 'slash') is now our regexp delimiter... */
 256         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
 257         delimiter = *(cmdstr_ptr++);
 258
 259         /* save the match string */
 260         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 261         if (idx == -1) {
 262                 bb_error_msg_and_die(bad_format_in_subst);
 263         }
 264         *match = copy_parsing_slashn(cmdstr_ptr, idx);
 265
 266         /* save the replacement string */
 267         cmdstr_ptr += idx + 1;
 268         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 269         if (idx == -1) {
 270                 bb_error_msg_and_die(bad_format_in_subst);
 271         }
 272         *replace = copy_parsing_slashn(cmdstr_ptr, idx);
 273
 274         return ((cmdstr_ptr - cmdstr) + idx);
 275 }
 276
 277 /*
 278  * returns the index in the string just past where the address ends.
 279  */
 280 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 281 {
 282         char *pos = my_str;
 283
 284         if (isdigit(*my_str)) {
 285                 *linenum = strtol(my_str, &pos, 10);
 286                 /* endstr shouldnt ever equal NULL */
 287         } else if (*my_str == '$') {
 288                 *linenum = -1;
 289                 pos++;
 290         } else if (*my_str == '/' || *my_str == '\\') {
 291                 int next;
 292                 char delimiter;
 293                 char *temp;
 294
 295                 if (*my_str == '\\') delimiter = *(++pos);
 296                 else delimiter = '/';
 297                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 298                 if (next == -1)
 299                         bb_error_msg_and_die("unterminated match expression");
 300
 301                 temp=copy_parsing_slashn(pos,next);
 302                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 303                 xregcomp(*regex, temp, regex_type|REG_NEWLINE);
 304                 free(temp);
 305                 /* Move position to next character after last delimiter */
 306                 pos+=(next+1);
 307         }
 308         return pos - my_str;
 309 }
 310
 311 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 312 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
 313 {
 314         int start = 0, idx, hack=0;
 315
 316         /* Skip whitespace, then grab filename to end of line */
 317         while (isspace(filecmdstr[start])) start++;
 318         idx=start;
 319         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 320         /* If lines glued together, put backslash back. */
 321         if(filecmdstr[idx]=='\n') hack=1;
 322         if(idx==start) bb_error_msg_and_die("Empty filename");
 323         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
 324         if(hack) *(idx+*retval)='\\';
 325
 326         return idx;
 327 }
 328
 329 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
 330 {
 331         int cflags = regex_type;
 332         char *match;
 333         int idx = 0;
 334
 335         /*
 336          * A substitution command should look something like this:
 337          *    s/match/replace/ #gIpw
 338          *    ||     |        |||
 339          *    mandatory       optional
 340          */
 341         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 342
 343         /* determine the number of back references in the match string */
 344         /* Note: we compute this here rather than in the do_subst_command()
 345          * function to save processor time, at the expense of a little more memory
 346          * (4 bits) per sed_cmd */
 347
 348         /* process the flags */
 349
 350         sed_cmd->which_match=1;
 351         while (substr[++idx]) {
 352                 /* Parse match number */
 353                 if(isdigit(substr[idx])) {
 354                         if(match[0]!='^') {
 355                                 /* Match 0 treated as all, multiple matches we take the last one. */
 356                                 char *pos=substr+idx;
 357                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 358                                 idx=pos-substr;
 359                         }
 360                         continue;
 361                 }
 362                 /* Skip spaces */
 363                 if(isspace(substr[idx])) continue;
 364
 365                 switch (substr[idx]) {
 366                         /* Replace all occurrences */
 367                         case 'g':
 368                                 if (match[0] != '^') sed_cmd->which_match = 0;
 369                                 break;
 370                         /* Print pattern space */
 371                         case 'p':
 372                                 sed_cmd->sub_p = 1;
 373                                 break;
 374                         case 'w':
 375                         {
 376                                 char *temp;
 377                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 378
 379                                 break;
 380                         }
 381                         /* Ignore case (gnu exension) */
 382                         case 'I':
 383                                 cflags |= REG_ICASE;
 384                                 break;
 385                         case ';':
 386                         case '}':
 387                                 goto out;
 388                         default:
 389                                 bb_error_msg_and_die("bad option in substitution expression");
 390                 }
 391         }
 392 out:
 393         /* compile the match string into a regex */
 394         if (*match != '\0') {
 395                 /* If match is empty, we use last regex used at runtime */
 396                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 397                 xregcomp(sed_cmd->sub_match, match, cflags);
 398         }
 399         free(match);
 400
 401         return idx;
 402 }
 403
 404 /*
 405  *  Process the commands arguments
 406  */
 407 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 408 {
 409         /* handle (s)ubstitution command */
 410         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 411         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 412         else if (strchr("aic", sed_cmd->cmd)) {
 413                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 414                         bb_error_msg_and_die
 415                                 ("only a beginning address can be specified for edit commands");
 416                 for(;;) {
 417                         if(*cmdstr=='\n' || *cmdstr=='\\') {
 418                                 cmdstr++;
 419                                 break;
 420                         } else if(isspace(*cmdstr)) cmdstr++;
 421                         else break;
 422                 }
 423                 sed_cmd->string = bb_xstrdup(cmdstr);
 424                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 425                 cmdstr += strlen(cmdstr);
 426         /* handle file cmds: (r)ead */
 427         } else if(strchr("rw", sed_cmd->cmd)) {
 428                 if (sed_cmd->end_line || sed_cmd->end_match)
 429                         bb_error_msg_and_die("Command only uses one address");
 430                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 431                 if(sed_cmd->cmd=='w')
 432                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
 433         /* handle branch commands */
 434         } else if (strchr(":bt", sed_cmd->cmd)) {
 435                 int length;
 436
 437                 while(isspace(*cmdstr)) cmdstr++;
 438                 length = strcspn(cmdstr, semicolon_whitespace);
 439                 if (length) {
 440                         sed_cmd->string = strndup(cmdstr, length);
 441                         cmdstr += length;
 442                 }
 443         }
 444         /* translation command */
 445         else if (sed_cmd->cmd == 'y') {
 446                 char *match, *replace;
 447                 int i=cmdstr[0];
 448
 449                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 450                 /* \n already parsed, but \delimiter needs unescaping. */
 451                 parse_escapes(match,match,strlen(match),i,i);
 452                 parse_escapes(replace,replace,strlen(replace),i,i);
 453
 454                 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
 455                 for (i = 0; match[i] && replace[i]; i++) {
 456                         sed_cmd->string[i * 2] = match[i];
 457                         sed_cmd->string[(i * 2) + 1] = replace[i];
 458                 }
 459                 free(match);
 460                 free(replace);
 461         }
 462         /* if it wasnt a single-letter command that takes no arguments
 463          * then it must be an invalid command.
 464          */
 465         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 466                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 467         }
 468
 469         /* give back whatever's left over */
 470         return (cmdstr);
 471 }
 472
 473
 474 /* Parse address+command sets, skipping comment lines. */
 475
 476 void add_cmd(char *cmdstr)
 477 {
 478         static char *add_cmd_line=NULL;
 479         sed_cmd_t *sed_cmd;
 480         int temp;
 481
 482         /* Append this line to any unfinished line from last time. */
 483         if(add_cmd_line) {
 484                 int lastlen=strlen(add_cmd_line);
 485                 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
 486
 487                 memcpy(tmp,add_cmd_line,lastlen);
 488                 tmp[lastlen]='\n';
 489                 strcpy(tmp+lastlen+1,cmdstr);
 490                 free(add_cmd_line);
 491                 cmdstr=add_cmd_line=tmp;
 492         } else add_cmd_line=NULL;
 493
 494         /* If this line ends with backslash, request next line. */
 495         temp=strlen(cmdstr);
 496         if(temp && cmdstr[temp-1]=='\\') {
 497                 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
 498                 add_cmd_line[temp-1]=0;
 499                 return;
 500         }
 501
 502         /* Loop parsing all commands in this line. */
 503         while(*cmdstr) {
 504                 /* Skip leading whitespace and semicolons */
 505                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 506
 507                 /* If no more commands, exit. */
 508                 if(!*cmdstr) break;
 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') be_quiet++;
 514                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 515                         continue;
 516                 }
 517
 518                 /* parse the command
 519                  * format is: [addr][,addr][!]cmd
 520                  *            |----||-----||-|
 521                  *            part1 part2  part3
 522                  */
 523
 524                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 525
 526                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 527                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 528
 529                 /* second part (if present) will begin with a comma */
 530                 if (*cmdstr == ',') {
 531                         int idx;
 532
 533                         cmdstr++;
 534                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 535                         if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
 536                         cmdstr += idx;
 537                 }
 538
 539                 /* skip whitespace before the command */
 540                 while (isspace(*cmdstr)) cmdstr++;
 541
 542                 /* Check for inversion flag */
 543                 if (*cmdstr == '!') {
 544                         sed_cmd->invert = 1;
 545                         cmdstr++;
 546
 547                         /* skip whitespace before the command */
 548                         while (isspace(*cmdstr)) cmdstr++;
 549                 }
 550
 551                 /* last part (mandatory) will be a command */
 552                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 553                 sed_cmd->cmd = *(cmdstr++);
 554                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 555
 556                 /* Add the command to the command array */
 557                 sed_cmd_tail->next = sed_cmd;
 558                 sed_cmd_tail = sed_cmd_tail->next;
 559         }
 560
 561         /* If we glued multiple lines together, free the memory. */
 562         if(add_cmd_line) {
 563                 free(add_cmd_line);
 564                 add_cmd_line=NULL;
 565         }
 566 }
 567
 568 struct pipeline {
 569         char *buf;      /* Space to hold string */
 570         int idx;        /* Space used */
 571         int len;        /* Space allocated */
 572 } pipeline;
 573
 574 #define PIPE_GROW 64
 575
 576 void pipe_putc(char c)
 577 {
 578         if(pipeline.idx==pipeline.len) {
 579                 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
 580                 pipeline.len+=PIPE_GROW;
 581         }
 582         pipeline.buf[pipeline.idx++] = (c);
 583 }
 584
 585 static void do_subst_w_backrefs(const char *line, const char *replace)
 586 {
 587         int i,j;
 588
 589         /* go through the replacement string */
 590         for (i = 0; replace[i]; i++) {
 591                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 592                 if (replace[i] == '\\' && replace[i+1]>'0' && replace[i+1]<='9') {
 593                         int backref=replace[++i]-'0';
 594
 595                         /* print out the text held in regmatch[backref] */
 596                         if(regmatch[backref].rm_so != -1)
 597                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
 598                                         pipe_putc(line[j]);
 599                 }
 600
 601                 /* if we find a backslash escaped character, print the character */
 602                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 603
 604                 /* if we find an unescaped '&' print out the whole matched text. */
 605                 else if (replace[i] == '&')
 606                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 607                                 pipe_putc(line[j]);
 608                 /* Otherwise just output the character. */
 609                 else pipe_putc(replace[i]);
 610         }
 611 }
 612
 613 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 614 {
 615         char *oldline = *line;
 616         int altered = 0;
 617         int match_count=0;
 618         regex_t *current_regex;
 619
 620         /* Handle empty regex. */
 621         if (sed_cmd->sub_match == NULL) {
 622                 current_regex = previous_regex_ptr;
 623                 if(!current_regex)
 624                         bb_error_msg_and_die("No previous regexp.");
 625         } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
 626
 627         /* Find the first match */
 628         if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
 629                 return 0;
 630
 631         /* Initialize temporary output buffer. */
 632         pipeline.buf=xmalloc(PIPE_GROW);
 633         pipeline.len=PIPE_GROW;
 634         pipeline.idx=0;
 635
 636         /* Now loop through, substituting for matches */
 637         do {
 638                 int i;
 639
 640                 /* Work around bug in glibc regexec, demonstrated by:
 641                    echo " a.b" | busybox sed 's [^ .]* x g'
 642                    The match_count check is so not to break
 643                    echo "hi" | busybox sed 's/^/!/g' */
 644                 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
 645                         pipe_putc(*(oldline++));
 646                         continue;
 647                 }
 648
 649                 match_count++;
 650
 651                 /* If we aren't interested in this match, output old line to
 652                    end of match and continue */
 653                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 654                         for(i=0;i<regmatch[0].rm_eo;i++)
 655                                 pipe_putc(oldline[i]);
 656                         continue;
 657                 }
 658
 659                 /* print everything before the match */
 660                 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 661
 662                 /* then print the substitution string */
 663                 do_subst_w_backrefs(oldline, sed_cmd->string);
 664
 665                 /* advance past the match */
 666                 oldline += regmatch[0].rm_eo;
 667                 /* flag that something has changed */
 668                 altered++;
 669
 670                 /* if we're not doing this globally, get out now */
 671                 if (sed_cmd->which_match) break;
 672         } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
 673
 674         /* Copy rest of string into output pipeline */
 675
 676         while(*oldline) pipe_putc(*(oldline++));
 677         pipe_putc(0);
 678
 679         free(*line);
 680         *line = pipeline.buf;
 681         return altered;
 682 }
 683
 684 /* Set command pointer to point to this label.  (Does not handle null label.) */
 685 static sed_cmd_t *branch_to(const char *label)
 686 {
 687         sed_cmd_t *sed_cmd;
 688
 689         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 690                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 691                         return (sed_cmd);
 692                 }
 693         }
 694         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 695 }
 696
 697 /* Append copy of string to append buffer */
 698 static void append(char *s)
 699 {
 700         struct append_list *temp=calloc(1,sizeof(struct append_list));
 701
 702         if(append_head)
 703                 append_tail=(append_tail->next=temp);
 704         else append_head=append_tail=temp;
 705         temp->string=strdup(s);
 706 }
 707
 708 static void flush_append(void)
 709 {
 710         /* Output appended lines. */
 711         while(append_head) {
 712                 fprintf(nonstdout,"%s\n",append_head->string);
 713                 append_tail=append_head->next;
 714                 free(append_head->string);
 715                 free(append_head);
 716                 append_head=append_tail;
 717         }
 718         append_head=append_tail=NULL;
 719 }
 720
 721 /* Get next line of input, flushing append buffer and noting if we hit EOF
 722  * without a newline on the last line.
 723  */
 724 static char *get_next_line(FILE * file, int *no_newline)
 725 {
 726         char *temp;
 727         int len;
 728
 729         flush_append();
 730         temp=bb_get_line_from_file(file);
 731         if(temp) {
 732                 len=strlen(temp);
 733                 if(len && temp[len-1]=='\n') temp[len-1]=0;
 734                 else *no_newline=1;
 735         }
 736
 737         return temp;
 738 }
 739
 740 /* Output line of text.  missing_newline means the last line output did not
 741    end with a newline.  no_newline means this line does not end with a
 742    newline. */
 743
 744 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 745 {
 746         if(missing_newline) fputc('\n',file);
 747         fputs(s,file);
 748         if(!no_newline) fputc('\n',file);
 749
 750     if(ferror(file)) {
 751                 fprintf(stderr,"Write failed.\n");
 752                 exit(4);  /* It's what gnu sed exits with... */
 753         }
 754
 755         return no_newline;
 756 }
 757
 758 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,nonstdout,missing_newline,n)
 759
 760 static void process_file(FILE *file)
 761 {
 762         char *pattern_space, *next_line;
 763         static int linenum = 0, missing_newline=0;
 764         int no_newline,next_no_newline=0;
 765
 766         next_line = get_next_line(file,&next_no_newline);
 767
 768         /* go through every line in the file */
 769         for(;;) {
 770                 sed_cmd_t *sed_cmd;
 771                 int substituted=0;
 772
 773                 /* Advance to next line.  Stop if out of lines. */
 774                 if(!(pattern_space=next_line)) break;
 775                 no_newline=next_no_newline;
 776
 777                 /* Read one line in advance so we can act on the last line, the '$' address */
 778                 next_line = get_next_line(file,&next_no_newline);
 779                 linenum++;
 780 restart:
 781                 /* for every line, go through all the commands */
 782                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 783                         int old_matched, matched;
 784
 785                         old_matched = sed_cmd->in_match;
 786
 787                         /* Determine if this command matches this line: */
 788
 789                         /* Are we continuing a previous multi-line match? */
 790
 791                         sed_cmd->in_match = sed_cmd->in_match
 792
 793                         /* Or is no range necessary? */
 794                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 795                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 796
 797                         /* Or did we match the start of a numerical range? */
 798                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 799
 800                         /* Or does this line match our begin address regex? */
 801                                 || (sed_cmd->beg_match &&
 802                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 803
 804                         /* Or did we match last line of input? */
 805                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 806
 807                         /* Snapshot the value */
 808
 809                         matched = sed_cmd->in_match;
 810
 811                         /* Is this line the end of the current match? */
 812
 813                         if(matched) {
 814                                 sed_cmd->in_match = !(
 815                                         /* has the ending line come, or is this a single address command? */
 816                                         (sed_cmd->end_line ?
 817                                                 sed_cmd->end_line==-1 ?
 818                                                         !next_line
 819                                                         : sed_cmd->end_line<=linenum
 820                                                 : !sed_cmd->end_match)
 821                                         /* or does this line matches our last address regex */
 822                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 823                                 );
 824                         }
 825
 826                         /* Skip blocks of commands we didn't match. */
 827                         if (sed_cmd->cmd == '{') {
 828                                 if(sed_cmd->invert ? matched : !matched)
 829                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 830                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 831                                 continue;
 832                         }
 833
 834                         /* Okay, so did this line match? */
 835                         if (sed_cmd->invert ? !matched : matched) {
 836                                 /* Update last used regex in case a blank substitute BRE is found */
 837                                 if (sed_cmd->beg_match) {
 838                                         previous_regex_ptr = sed_cmd->beg_match;
 839                                 }
 840
 841                                 /* actual sedding */
 842                                 switch (sed_cmd->cmd) {
 843
 844                                         /* Print line number */
 845                                         case '=':
 846                                                 fprintf(nonstdout,"%d\n", linenum);
 847                                                 break;
 848
 849                                         /* Write the current pattern space up to the first newline */
 850                                         case 'P':
 851                                         {
 852                                                 char *tmp = strchr(pattern_space, '\n');
 853
 854                                                 if (tmp) {
 855                                                         *tmp = '\0';
 856                                                         sed_puts(pattern_space,1);
 857                                                         *tmp = '\n';
 858                                                         break;
 859                                                 }
 860                                                 /* Fall Through */
 861                                         }
 862
 863                                         /* Write the current pattern space to output */
 864                                         case 'p':
 865                                                 sed_puts(pattern_space,no_newline);
 866                                                 break;
 867                                         /* Delete up through first newline */
 868                                         case 'D':
 869                                         {
 870                                                 char *tmp = strchr(pattern_space,'\n');
 871
 872                                                 if(tmp) {
 873                                                         tmp=bb_xstrdup(tmp+1);
 874                                                         free(pattern_space);
 875                                                         pattern_space=tmp;
 876                                                         goto restart;
 877                                                 }
 878                                         }
 879                                         /* discard this line. */
 880                                         case 'd':
 881                                                 goto discard_line;
 882
 883                                         /* Substitute with regex */
 884                                         case 's':
 885                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 886                                                         substituted|=1;
 887
 888                                                         /* handle p option */
 889                                                         if(sed_cmd->sub_p)
 890                                                                 sed_puts(pattern_space,no_newline);
 891                                                         /* handle w option */
 892                                                         if(sed_cmd->file)
 893                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 894
 895                                                 }
 896                                                 break;
 897
 898                                         /* Append line to linked list to be printed later */
 899                                         case 'a':
 900                                         {
 901                                                 append(sed_cmd->string);
 902                                                 break;
 903                                         }
 904
 905                                         /* Insert text before this line */
 906                                         case 'i':
 907                                                 sed_puts(sed_cmd->string,1);
 908                                                 break;
 909
 910                                         /* Cut and paste text (replace) */
 911                                         case 'c':
 912                                                 /* Only triggers on last line of a matching range. */
 913                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,0);
 914                                                 goto discard_line;
 915
 916                                         /* Read file, append contents to output */
 917                                         case 'r':
 918                                         {
 919                                                 FILE *outfile;
 920
 921                                                 outfile = fopen(sed_cmd->string, "r");
 922                                                 if (outfile) {
 923                                                         char *line;
 924
 925                                                         while ((line = bb_get_chomped_line_from_file(outfile))
 926                                                                         != NULL)
 927                                                                 append(line);
 928                                                         bb_xprint_and_close_file(outfile);
 929                                                 }
 930
 931                                                 break;
 932                                         }
 933
 934                                         /* Write pattern space to file. */
 935                                         case 'w':
 936                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 937                                                 break;
 938
 939                                         /* Read next line from input */
 940                                         case 'n':
 941                                                 if (!be_quiet)
 942                                                         sed_puts(pattern_space,no_newline);
 943                                                 if (next_line) {
 944                                                         free(pattern_space);
 945                                                         pattern_space = next_line;
 946                                                         no_newline=next_no_newline;
 947                                                         next_line = get_next_line(file,&next_no_newline);
 948                                                         linenum++;
 949                                                         break;
 950                                                 }
 951                                                 /* fall through */
 952
 953                                         /* Quit.  End of script, end of input. */
 954                                         case 'q':
 955                                                 /* Exit the outer while loop */
 956                                                 free(next_line);
 957                                                 next_line = NULL;
 958                                                 goto discard_commands;
 959
 960                                         /* Append the next line to the current line */
 961                                         case 'N':
 962                                         {
 963                                                 /* If no next line, jump to end of script and exit. */
 964                                                 if (next_line == NULL) {
 965                                                         /* Jump to end of script and exit */
 966                                                         free(next_line);
 967                                                         next_line = NULL;
 968                                                         goto discard_line;
 969                                                 /* append next_line, read new next_line. */
 970                                                 } else {
 971                                                         int len=strlen(pattern_space);
 972
 973                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 974                                                         pattern_space[len]='\n';
 975                                                         strcpy(pattern_space+len+1, next_line);
 976                                                         no_newline=next_no_newline;
 977                                                         next_line = get_next_line(file,&next_no_newline);
 978                                                         linenum++;
 979                                                 }
 980                                                 break;
 981                                         }
 982
 983                                         /* Test if substition worked, branch if so. */
 984                                         case 't':
 985                                                 if (!substituted) break;
 986                                                 substituted=0;
 987                                                         /* Fall through */
 988                                         /* Branch to label */
 989                                         case 'b':
 990                                                 if (!sed_cmd->string) goto discard_commands;
 991                                                 else sed_cmd = branch_to(sed_cmd->string);
 992                                                 break;
 993                                         /* Transliterate characters */
 994                                         case 'y':
 995                                         {
 996                                                 int i;
 997
 998                                                 for (i = 0; pattern_space[i]; i++) {
 999                                                         int j;
1000
1001                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1002                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1003                                                                         pattern_space[i] = sed_cmd->string[j + 1];
1004                                                                 }
1005                                                         }
1006                                                 }
1007
1008                                                 break;
1009                                         }
1010                                         case 'g':       /* Replace pattern space with hold space */
1011                                                 free(pattern_space);
1012                                                 pattern_space = strdup(hold_space ? hold_space : "");
1013                                                 break;
1014                                         case 'G':       /* Append newline and hold space to pattern space */
1015                                         {
1016                                                 int pattern_space_size = 2;
1017                                                 int hold_space_size = 0;
1018
1019                                                 if (pattern_space)
1020                                                         pattern_space_size += strlen(pattern_space);
1021                                                 if (hold_space) hold_space_size = strlen(hold_space);
1022                                                 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1023                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1024                                                 strcat(pattern_space, "\n");
1025                                                 if (hold_space) strcat(pattern_space, hold_space);
1026                                                 no_newline=0;
1027
1028                                                 break;
1029                                         }
1030                                         case 'h':       /* Replace hold space with pattern space */
1031                                                 free(hold_space);
1032                                                 hold_space = strdup(pattern_space);
1033                                                 break;
1034                                         case 'H':       /* Append newline and pattern space to hold space */
1035                                         {
1036                                                 int hold_space_size = 2;
1037                                                 int pattern_space_size = 0;
1038
1039                                                 if (hold_space) hold_space_size += strlen(hold_space);
1040                                                 if (pattern_space)
1041                                                         pattern_space_size = strlen(pattern_space);
1042                                                 hold_space = xrealloc(hold_space,
1043                                                                                         hold_space_size + pattern_space_size);
1044
1045                                                 if (hold_space_size == 2) hold_space[0]=0;
1046                                                 strcat(hold_space, "\n");
1047                                                 if (pattern_space) strcat(hold_space, pattern_space);
1048
1049                                                 break;
1050                                         }
1051                                         case 'x': /* Exchange hold and pattern space */
1052                                         {
1053                                                 char *tmp = pattern_space;
1054                                                 pattern_space = hold_space;
1055                                                 no_newline=0;
1056                                                 hold_space = tmp;
1057                                                 break;
1058                                         }
1059                                 }
1060                         }
1061                 }
1062
1063                 /*
1064                  * exit point from sedding...
1065                  */
1066 discard_commands:
1067                 /* we will print the line unless we were told to be quiet ('-n')
1068                    or if the line was suppressed (ala 'd'elete) */
1069                 if (!be_quiet) sed_puts(pattern_space,no_newline);
1070
1071                 /* Delete and such jump here. */
1072 discard_line:
1073                 flush_append();
1074                 free(pattern_space);
1075         }
1076 }
1077
1078 /* It is possible to have a command line argument with embedded
1079    newlines.  This counts as multiple command lines. */
1080
1081 static void add_cmd_block(char *cmdstr)
1082 {
1083         int go=1;
1084         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1085
1086         while(go) {
1087                 int len=strcspn(temp2,"\n");
1088                 if(!temp2[len]) go=0;
1089                 else temp2[len]=0;
1090                 add_cmd(temp2);
1091                 temp2+=len+1;
1092         }
1093         free(temp);
1094 }
1095
1096 extern int sed_main(int argc, char **argv)
1097 {
1098         int status = EXIT_SUCCESS, opt, getpat = 1;
1099
1100 #ifdef CONFIG_FEATURE_CLEAN_UP
1101         /* destroy command strings on exit */
1102         if (atexit(free_and_close_stuff) == -1)
1103                 bb_perror_msg_and_die("atexit");
1104 #endif
1105
1106 #define LIE_TO_AUTOCONF
1107 #ifdef LIE_TO_AUTOCONF
1108         if(argc==2 && !strcmp(argv[1],"--version")) {
1109                 printf("This is not GNU sed version 4.0\n");
1110                 exit(0);
1111         }
1112 #endif
1113
1114         /* do normal option parsing */
1115         while ((opt = getopt(argc, argv, "irne:f:")) > 0) {
1116                 switch (opt) {
1117                 case 'i':
1118                         in_place++;
1119                         atexit(cleanup_outname);
1120                         break;
1121                 case 'r':
1122                         regex_type|=REG_EXTENDED;
1123                         break;
1124                 case 'n':
1125                         be_quiet++;
1126                         break;
1127                 case 'e':
1128                         add_cmd_block(optarg);
1129                         getpat=0;
1130                         break;
1131                 case 'f':
1132                 {
1133                         FILE *cmdfile;
1134                         char *line;
1135
1136                         cmdfile = bb_xfopen(optarg, "r");
1137
1138                         while ((line = bb_get_chomped_line_from_file(cmdfile))
1139                                  != NULL) {
1140                                 add_cmd(line);
1141                                 getpat=0;
1142                                 free(line);
1143                         }
1144                         bb_xprint_and_close_file(cmdfile);
1145
1146                         break;
1147                 }
1148                 default:
1149                         bb_show_usage();
1150                 }
1151         }
1152
1153         /* if we didn't get a pattern from a -e and no command file was specified,
1154          * argv[optind] should be the pattern. no pattern, no worky */
1155         if(getpat) {
1156                 if (argv[optind] == NULL)
1157                         bb_show_usage();
1158                 else
1159                         add_cmd_block(argv[optind++]);
1160         }
1161         /* Flush any unfinished commands. */
1162         add_cmd("");
1163
1164         /* By default, we write to stdout */
1165         nonstdout=stdout;
1166
1167         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1168          * files were specified or '-' was specified, take input from stdin.
1169          * Otherwise, we process all the files specified. */
1170         if (argv[optind] == NULL) {
1171                 if(in_place) {
1172                         fprintf(stderr,"sed: Filename required for -i\n");
1173                         exit(1);
1174                 }
1175                 process_file(stdin);
1176         } else {
1177                 int i;
1178                 FILE *file;
1179
1180                 for (i = optind; i < argc; i++) {
1181                         if(!strcmp(argv[i], "-") && !in_place) {
1182                                 process_file(stdin);
1183                         } else {
1184                                 file = bb_wfopen(argv[i], "r");
1185                                 if (file) {
1186                                         if(in_place) {
1187                                                 struct stat statbuf;
1188                                                 outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1189                                                 strcat(outname,"XXXXXX");
1190                                                 /* Set permissions of output file */
1191                                                 fstat(fileno(file),&statbuf);
1192                                                 mkstemp(outname);
1193                                                 nonstdout=bb_wfopen(outname,"w");
1194                                                 /* Set permissions of output file */
1195                                                 fstat(fileno(file),&statbuf);
1196                                                 fchmod(fileno(nonstdout),statbuf.st_mode);
1197                                                 atexit(cleanup_outname);
1198                                         }
1199                                         process_file(file);
1200                                         fclose(file);
1201                                         if(in_place) {
1202                                                 fclose(nonstdout);
1203                                                 nonstdout=stdout;
1204                                                 unlink(argv[i]);
1205                                                 rename(outname,argv[i]);
1206                                                 free(outname);
1207                                                 outname=0;
1208                                         }
1209                                 } else {
1210                                         status = EXIT_FAILURE;
1211                                 }
1212                         }
1213                 }
1214         }
1215
1216         return status;
1217 }