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@optushome.com.au>
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18  * General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with this program; if not, write to the Free Software
  22  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  23  *
  24  */
  25
  26 /* Code overview.
  27
  28   Files are laid out to avoid unnecessary function declarations.  So for
  29   example, every function add_cmd calls occurs before add_cmd in this file.
  30
  31   add_cmd() is called on each line of sed command text (from a file or from
  32   the command line).  It calls get_address() and parse_cmd_args().  The
  33   resulting sed_cmd_t structures are appended to a linked list
  34   (sed_cmd_head/sed_cmd_tail).
  35
  36   process_file() does actual sedding, reading data lines from an input FILE *
  37   (which could be stdin) and applying the sed command list (sed_cmd_head) to
  38   each of the resulting lines.
  39
  40   sed_main() is where external code calls into this, with a command line.
  41 */
  42
  43
  44 /*
  45         Supported features and commands in this version of sed:
  46
  47          - comments ('#')
  48          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  49          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  50          - edit commands: (a)ppend, (i)nsert, (c)hange
  51          - file commands: (r)ead
  52          - backreferences in substitution expressions (\1, \2...\9)
  53          - grouped commands: {cmd1;cmd2}
  54          - transliteration (y/source-chars/dest-chars/)
  55          - pattern space hold space storing / swapping (g, h, x)
  56          - labels / branching (: label, b, t)
  57
  58          (Note: Specifying an address (range) to match is *optional*; commands
  59          default to the whole pattern space if no specific address match was
  60          requested.)
  61
  62         Unsupported features:
  63
  64          - GNU extensions
  65          - and more.
  66
  67         Todo:
  68
  69          - Create a wrapper around regex to make libc's regex conform with sed
  70          - Fix bugs
  71
  72
  73         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  74 */
  75
  76 #include <stdio.h>
  77 #include <unistd.h>             /* for getopt() */
  78 #include <regex.h>
  79 #include <string.h>             /* for strdup() */
  80 #include <errno.h>
  81 #include <ctype.h>              /* for isspace() */
  82 #include <stdlib.h>
  83 #include "busybox.h"
  84
  85 typedef struct sed_cmd_s {
  86     /* Ordered by alignment requirements: currently 36 bytes on x86 */
  87
  88     /* address storage */
  89     regex_t *beg_match; /* sed -e '/match/cmd' */
  90     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
  91     regex_t *sub_match; /* For 's/sub_match/string/' */
  92     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
  93     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  94
  95     FILE *file;                 /* File (sr) command writes to, -1 for none. */
  96     char *string;               /* Data string for (saicytb) commands. */
  97
  98     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
  99
 100     /* Bitfields (gcc won't group them if we don't) */
 101     unsigned int invert:1;                      /* the '!' after the address */
 102     unsigned int in_match:1;            /* Next line also included in match? */
 103     unsigned int no_newline:1;          /* Last line written by (sr) had no '\n' */
 104     unsigned int sub_p:1;                       /* (s) print option */
 105
 106
 107     /* GENERAL FIELDS */
 108     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 109     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
 110 } sed_cmd_t;
 111
 112 /* globals */
 113 /* options */
 114 static int be_quiet = 0, in_place=0;
 115 FILE *nonstdout;
 116 char *outname;
 117
 118
 119 static const char bad_format_in_subst[] =
 120         "bad format in substitution expression";
 121 const char *const semicolon_whitespace = "; \n\r\t\v";
 122
 123 regmatch_t regmatch[10];
 124 static regex_t *previous_regex_ptr = NULL;
 125
 126 /* linked list of sed commands */
 127 static sed_cmd_t sed_cmd_head;
 128 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
 129
 130 /* Linked list of append lines */
 131 struct append_list {
 132         char *string;
 133         struct append_list *next;
 134 };
 135 struct append_list *append_head=NULL, *append_tail=NULL;
 136
 137 #ifdef CONFIG_FEATURE_CLEAN_UP
 138 static void free_and_close_stuff(void)
 139 {
 140         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 141
 142         while(append_head) {
 143                 append_tail=append_head->next;
 144                 free(append_head->string);
 145                 free(append_head);
 146                 append_head=append_tail;
 147         }
 148
 149         while (sed_cmd) {
 150                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 151
 152                 if(sed_cmd->file)
 153                         bb_xprint_and_close_file(sed_cmd->file);
 154
 155                 if (sed_cmd->beg_match) {
 156                         regfree(sed_cmd->beg_match);
 157                         free(sed_cmd->beg_match);
 158                 }
 159                 if (sed_cmd->end_match) {
 160                         regfree(sed_cmd->end_match);
 161                         free(sed_cmd->end_match);
 162                 }
 163                 if (sed_cmd->sub_match) {
 164                         regfree(sed_cmd->sub_match);
 165                         free(sed_cmd->sub_match);
 166                 }
 167                 free(sed_cmd->string);
 168                 free(sed_cmd);
 169                 sed_cmd = sed_cmd_next;
 170         }
 171 }
 172 #endif
 173
 174 /* If something bad happens during -i operation, delete temp file */
 175
 176 static void cleanup_outname(void)
 177 {
 178   if(outname) unlink(outname);
 179 }
 180
 181 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 182
 183 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 184 {
 185         int i=0;
 186
 187         while(i<len) {
 188                 if(string[i] == '\\') {
 189                         if(!to || string[i+1] == from) {
 190                                 *(dest++) = to ? to : string[i+1];
 191                                 i+=2;
 192                                 continue;
 193                         } else *(dest++)=string[i++];
 194                 }
 195                 *(dest++) = string[i++];
 196         }
 197         *dest=0;
 198 }
 199
 200 static char *copy_parsing_slashn(const char *string, int len)
 201 {
 202         char *dest=xmalloc(len+1);
 203
 204         parse_escapes(dest,string,len,'n','\n');
 205         return dest;
 206 }
 207
 208
 209 /*
 210  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 211  * beginning at a specified index and returns the index of the next regular
 212  * expression delimiter (typically a forward * slash ('/')) not preceeded by
 213  * a backslash ('\').
 214  */
 215 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 216         const char *str)
 217 {
 218         int bracket = -1;
 219         int escaped = 0;
 220         int idx = 0;
 221         char ch;
 222
 223         for (; (ch = str[idx]); idx++) {
 224                 if (bracket != -1) {
 225                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 226                                         && str[idx - 1] == '^')))
 227                                 bracket = -1;
 228                 } else if (escaped)
 229                         escaped = 0;
 230                 else if (ch == '\\')
 231                         escaped = 1;
 232                 else if (ch == '[')
 233                         bracket = idx;
 234                 else if (ch == delimiter)
 235                         return idx;
 236         }
 237
 238         /* if we make it to here, we've hit the end of the string */
 239         return -1;
 240 }
 241
 242 /*
 243  *  Returns the index of the third delimiter
 244  */
 245 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 246 {
 247         const char *cmdstr_ptr = cmdstr;
 248         char delimiter;
 249         int idx = 0;
 250
 251         /* verify that the 's' or 'y' is followed by something.  That something
 252          * (typically a 'slash') is now our regexp delimiter... */
 253         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
 254         delimiter = *(cmdstr_ptr++);
 255
 256         /* save the match string */
 257         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 258         if (idx == -1) {
 259                 bb_error_msg_and_die(bad_format_in_subst);
 260         }
 261         *match = copy_parsing_slashn(cmdstr_ptr, idx);
 262
 263         /* save the replacement string */
 264         cmdstr_ptr += idx + 1;
 265         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 266         if (idx == -1) {
 267                 bb_error_msg_and_die(bad_format_in_subst);
 268         }
 269         *replace = copy_parsing_slashn(cmdstr_ptr, idx);
 270
 271         return ((cmdstr_ptr - cmdstr) + idx);
 272 }
 273
 274 /*
 275  * returns the index in the string just past where the address ends.
 276  */
 277 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 278 {
 279         char *pos = my_str;
 280
 281         if (isdigit(*my_str)) {
 282                 *linenum = strtol(my_str, &pos, 10);
 283                 /* endstr shouldnt ever equal NULL */
 284         } else if (*my_str == '$') {
 285                 *linenum = -1;
 286                 pos++;
 287         } else if (*my_str == '/' || *my_str == '\\') {
 288                 int next;
 289                 char delimiter;
 290                 char *temp;
 291
 292                 if (*my_str == '\\') delimiter = *(++pos);
 293                 else delimiter = '/';
 294                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 295                 if (next == -1)
 296                         bb_error_msg_and_die("unterminated match expression");
 297
 298                 temp=copy_parsing_slashn(pos,next);
 299                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 300                 xregcomp(*regex, temp, REG_NEWLINE);
 301                 free(temp);
 302                 /* Move position to next character after last delimiter */
 303                 pos+=(next+1);
 304         }
 305         return pos - my_str;
 306 }
 307
 308 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 309 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
 310 {
 311         int start = 0, idx, hack=0;
 312
 313         /* Skip whitespace, then grab filename to end of line */
 314         while (isspace(filecmdstr[start])) start++;
 315         idx=start;
 316         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 317         /* If lines glued together, put backslash back. */
 318         if(filecmdstr[idx]=='\n') hack=1;
 319         if(idx==start) bb_error_msg_and_die("Empty filename");
 320         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
 321         if(hack) *(idx+*retval)='\\';
 322
 323         return idx;
 324 }
 325
 326 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
 327 {
 328         int cflags = 0;
 329         char *match;
 330         int idx = 0;
 331
 332         /*
 333          * A substitution command should look something like this:
 334          *    s/match/replace/ #gIpw
 335          *    ||     |        |||
 336          *    mandatory       optional
 337          */
 338         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 339
 340         /* determine the number of back references in the match string */
 341         /* Note: we compute this here rather than in the do_subst_command()
 342          * function to save processor time, at the expense of a little more memory
 343          * (4 bits) per sed_cmd */
 344
 345         /* process the flags */
 346
 347         sed_cmd->which_match=1;
 348         while (substr[++idx]) {
 349                 /* Parse match number */
 350                 if(isdigit(substr[idx])) {
 351                         if(match[0]!='^') {
 352                                 /* Match 0 treated as all, multiple matches we take the last one. */
 353                                 char *pos=substr+idx;
 354                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 355                                 idx=pos-substr;
 356                         }
 357                         continue;
 358                 }
 359                 /* Skip spaces */
 360                 if(isspace(substr[idx])) continue;
 361
 362                 switch (substr[idx]) {
 363                         /* Replace all occurrences */
 364                         case 'g':
 365                                 if (match[0] != '^') sed_cmd->which_match = 0;
 366                                 break;
 367                         /* Print pattern space */
 368                         case 'p':
 369                                 sed_cmd->sub_p = 1;
 370                                 break;
 371                         case 'w':
 372                         {
 373                                 char *temp;
 374                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 375
 376                                 break;
 377                         }
 378                         /* Ignore case (gnu exension) */
 379                         case 'I':
 380                                 cflags |= REG_ICASE;
 381                                 break;
 382                         case ';':
 383                         case '}':
 384                                 goto out;
 385                         default:
 386                                 bb_error_msg_and_die("bad option in substitution expression");
 387                 }
 388         }
 389 out:
 390         /* compile the match string into a regex */
 391         if (*match != '\0') {
 392                 /* If match is empty, we use last regex used at runtime */
 393                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 394                 xregcomp(sed_cmd->sub_match, match, cflags);
 395         }
 396         free(match);
 397
 398         return idx;
 399 }
 400
 401 /*
 402  *  Process the commands arguments
 403  */
 404 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 405 {
 406         /* handle (s)ubstitution command */
 407         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 408         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 409         else if (strchr("aic", sed_cmd->cmd)) {
 410                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 411                         bb_error_msg_and_die
 412                                 ("only a beginning address can be specified for edit commands");
 413                 while(isspace(*cmdstr)) cmdstr++;
 414                 sed_cmd->string = bb_xstrdup(cmdstr);
 415                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 416                 cmdstr += strlen(cmdstr);
 417         /* handle file cmds: (r)ead */
 418         } else if(strchr("rw", sed_cmd->cmd)) {
 419                 if (sed_cmd->end_line || sed_cmd->end_match)
 420                         bb_error_msg_and_die("Command only uses one address");
 421                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 422                 if(sed_cmd->cmd=='w')
 423                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
 424         /* handle branch commands */
 425         } else if (strchr(":bt", sed_cmd->cmd)) {
 426                 int length;
 427
 428                 while(isspace(*cmdstr)) cmdstr++;
 429                 length = strcspn(cmdstr, semicolon_whitespace);
 430                 if (length) {
 431                         sed_cmd->string = strndup(cmdstr, length);
 432                         cmdstr += length;
 433                 }
 434         }
 435         /* translation command */
 436         else if (sed_cmd->cmd == 'y') {
 437                 char *match, *replace;
 438                 int i=cmdstr[0];
 439
 440                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 441                 /* \n already parsed, but \delimiter needs unescaping. */
 442                 parse_escapes(match,match,strlen(match),i,i);
 443                 parse_escapes(replace,replace,strlen(replace),i,i);
 444
 445                 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
 446                 for (i = 0; match[i] && replace[i]; i++) {
 447                         sed_cmd->string[i * 2] = match[i];
 448                         sed_cmd->string[(i * 2) + 1] = replace[i];
 449                 }
 450                 free(match);
 451                 free(replace);
 452         }
 453         /* if it wasnt a single-letter command that takes no arguments
 454          * then it must be an invalid command.
 455          */
 456         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 457                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 458         }
 459
 460         /* give back whatever's left over */
 461         return (cmdstr);
 462 }
 463
 464
 465 /* Parse address+command sets, skipping comment lines. */
 466
 467 void add_cmd(char *cmdstr)
 468 {
 469         static char *add_cmd_line=NULL;
 470         sed_cmd_t *sed_cmd;
 471         int temp;
 472
 473         /* Append this line to any unfinished line from last time. */
 474         if(add_cmd_line) {
 475                 int lastlen=strlen(add_cmd_line);
 476                 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
 477
 478                 memcpy(tmp,add_cmd_line,lastlen);
 479                 tmp[lastlen]='\n';
 480                 strcpy(tmp+lastlen+1,cmdstr);
 481                 free(add_cmd_line);
 482                 cmdstr=add_cmd_line=tmp;
 483         } else add_cmd_line=NULL;
 484
 485         /* If this line ends with backslash, request next line. */
 486         temp=strlen(cmdstr);
 487         if(temp && cmdstr[temp-1]=='\\') {
 488                 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
 489                 add_cmd_line[temp-1]=0;
 490                 return;
 491         }
 492
 493         /* Loop parsing all commands in this line. */
 494         while(*cmdstr) {
 495                 /* Skip leading whitespace and semicolons */
 496                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 497
 498                 /* If no more commands, exit. */
 499                 if(!*cmdstr) break;
 500
 501                 /* if this is a comment, jump past it and keep going */
 502                 if (*cmdstr == '#') {
 503                         /* "#n" is the same as using -n on the command line */
 504                         if (cmdstr[1] == 'n') be_quiet++;
 505                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 506                         continue;
 507                 }
 508
 509                 /* parse the command
 510                  * format is: [addr][,addr][!]cmd
 511                  *            |----||-----||-|
 512                  *            part1 part2  part3
 513                  */
 514
 515                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 516
 517                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 518                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 519
 520                 /* second part (if present) will begin with a comma */
 521                 if (*cmdstr == ',') {
 522                         int idx;
 523
 524                         cmdstr++;
 525                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 526                         if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
 527                         cmdstr += idx;
 528                 }
 529
 530                 /* skip whitespace before the command */
 531                 while (isspace(*cmdstr)) cmdstr++;
 532
 533                 /* Check for inversion flag */
 534                 if (*cmdstr == '!') {
 535                         sed_cmd->invert = 1;
 536                         cmdstr++;
 537
 538                         /* skip whitespace before the command */
 539                         while (isspace(*cmdstr)) cmdstr++;
 540                 }
 541
 542                 /* last part (mandatory) will be a command */
 543                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 544                 sed_cmd->cmd = *(cmdstr++);
 545                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 546
 547                 /* Add the command to the command array */
 548                 sed_cmd_tail->next = sed_cmd;
 549                 sed_cmd_tail = sed_cmd_tail->next;
 550         }
 551
 552         /* If we glued multiple lines together, free the memory. */
 553         if(add_cmd_line) {
 554                 free(add_cmd_line);
 555                 add_cmd_line=NULL;
 556         }
 557 }
 558
 559 struct pipeline {
 560         char *buf;      /* Space to hold string */
 561         int idx;        /* Space used */
 562         int len;        /* Space allocated */
 563 } pipeline;
 564
 565 #define PIPE_GROW 64
 566
 567 void pipe_putc(char c)
 568 {
 569         if(pipeline.idx==pipeline.len) {
 570                 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
 571                 pipeline.len+=PIPE_GROW;
 572         }
 573         pipeline.buf[pipeline.idx++] = (c);
 574 }
 575
 576 static void do_subst_w_backrefs(const char *line, const char *replace)
 577 {
 578         int i,j;
 579
 580         /* go through the replacement string */
 581         for (i = 0; replace[i]; i++) {
 582                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 583                 if (replace[i] == '\\' && replace[i+1]>'0' && replace[i+1]<='9') {
 584                         int backref=replace[++i]-'0';
 585
 586                         /* print out the text held in regmatch[backref] */
 587                         if(regmatch[backref].rm_so != -1)
 588                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
 589                                         pipe_putc(line[j]);
 590                 }
 591
 592                 /* if we find a backslash escaped character, print the character */
 593                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 594
 595                 /* if we find an unescaped '&' print out the whole matched text. */
 596                 else if (replace[i] == '&')
 597                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 598                                 pipe_putc(line[j]);
 599                 /* Otherwise just output the character. */
 600                 else pipe_putc(replace[i]);
 601         }
 602 }
 603
 604 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 605 {
 606         char *oldline = *line;
 607         int altered = 0;
 608         int match_count=0;
 609         regex_t *current_regex;
 610
 611         /* Handle empty regex. */
 612         if (sed_cmd->sub_match == NULL) {
 613                 current_regex = previous_regex_ptr;
 614                 if(!current_regex)
 615                         bb_error_msg_and_die("No previous regexp.");
 616         } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
 617
 618         /* Find the first match */
 619         if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
 620                 return 0;
 621
 622         /* Initialize temporary output buffer. */
 623         pipeline.buf=xmalloc(PIPE_GROW);
 624         pipeline.len=PIPE_GROW;
 625         pipeline.idx=0;
 626
 627         /* Now loop through, substituting for matches */
 628         do {
 629                 int i;
 630
 631                 /* Work around bug in glibc regexec, demonstrated by:
 632                    echo " a.b" | busybox sed 's [^ .]* x g'
 633                    The match_count check is so not to break
 634                    echo "hi" | busybox sed 's/^/!/g' */
 635                 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
 636                         pipe_putc(*(oldline++));
 637                         continue;
 638                 }
 639
 640                 match_count++;
 641
 642                 /* If we aren't interested in this match, output old line to
 643                    end of match and continue */
 644                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 645                         for(i=0;i<regmatch[0].rm_eo;i++)
 646                                 pipe_putc(oldline[i]);
 647                         continue;
 648                 }
 649
 650                 /* print everything before the match */
 651                 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 652
 653                 /* then print the substitution string */
 654                 do_subst_w_backrefs(oldline, sed_cmd->string);
 655
 656                 /* advance past the match */
 657                 oldline += regmatch[0].rm_eo;
 658                 /* flag that something has changed */
 659                 altered++;
 660
 661                 /* if we're not doing this globally, get out now */
 662                 if (sed_cmd->which_match) break;
 663         } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
 664
 665         /* Copy rest of string into output pipeline */
 666
 667         while(*oldline) pipe_putc(*(oldline++));
 668         pipe_putc(0);
 669
 670         free(*line);
 671         *line = pipeline.buf;
 672         return altered;
 673 }
 674
 675 /* Set command pointer to point to this label.  (Does not handle null label.) */
 676 static sed_cmd_t *branch_to(const char *label)
 677 {
 678         sed_cmd_t *sed_cmd;
 679
 680         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 681                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 682                         return (sed_cmd);
 683                 }
 684         }
 685         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 686 }
 687
 688 /* Append copy of string to append buffer */
 689 static void append(char *s)
 690 {
 691         struct append_list *temp=calloc(1,sizeof(struct append_list));
 692
 693         if(append_head)
 694                 append_tail=(append_tail->next=temp);
 695         else append_head=append_tail=temp;
 696         temp->string=strdup(s);
 697 }
 698
 699 static void flush_append(void)
 700 {
 701         /* Output appended lines. */
 702         while(append_head) {
 703                 fprintf(nonstdout,"%s\n",append_head->string);
 704                 append_tail=append_head->next;
 705                 free(append_head->string);
 706                 free(append_head);
 707                 append_head=append_tail;
 708         }
 709         append_head=append_tail=NULL;
 710 }
 711
 712 /* Get next line of input, flushing append buffer and noting if we hit EOF
 713  * without a newline on the last line.
 714  */
 715 static char *get_next_line(FILE * file, int *no_newline)
 716 {
 717         char *temp;
 718         int len;
 719
 720         flush_append();
 721         temp=bb_get_line_from_file(file);
 722         if(temp) {
 723                 len=strlen(temp);
 724                 if(len && temp[len-1]=='\n') temp[len-1]=0;
 725                 else *no_newline=1;
 726         }
 727
 728         return temp;
 729 }
 730
 731 /* Output line of text.  missing_newline means the last line output did not
 732    end with a newline.  no_newline means this line does not end with a
 733    newline. */
 734
 735 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 736 {
 737         if(missing_newline) fputc('\n',file);
 738         fputs(s,file);
 739         if(!no_newline) fputc('\n',file);
 740
 741     if(ferror(file)) {
 742                 fprintf(stderr,"Write failed.\n");
 743                 exit(4);  /* It's what gnu sed exits with... */
 744         }
 745
 746         return no_newline;
 747 }
 748
 749 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,nonstdout,missing_newline,n)
 750
 751 static void process_file(FILE *file)
 752 {
 753         char *pattern_space, *next_line, *hold_space=NULL;
 754         static int linenum = 0, missing_newline=0;
 755         int no_newline,next_no_newline=0;
 756
 757         next_line = get_next_line(file,&next_no_newline);
 758
 759         /* go through every line in the file */
 760         for(;;) {
 761                 sed_cmd_t *sed_cmd;
 762                 int substituted=0;
 763
 764                 /* Advance to next line.  Stop if out of lines. */
 765                 if(!(pattern_space=next_line)) break;
 766                 no_newline=next_no_newline;
 767
 768                 /* Read one line in advance so we can act on the last line, the '$' address */
 769                 next_line = get_next_line(file,&next_no_newline);
 770                 linenum++;
 771 restart:
 772                 /* for every line, go through all the commands */
 773                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 774                         int old_matched, matched;
 775
 776                         old_matched = sed_cmd->in_match;
 777
 778                         /* Determine if this command matches this line: */
 779
 780                         /* Are we continuing a previous multi-line match? */
 781
 782                         sed_cmd->in_match = sed_cmd->in_match
 783
 784                         /* Or is no range necessary? */
 785                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 786                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 787
 788                         /* Or did we match the start of a numerical range? */
 789                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 790
 791                         /* Or does this line match our begin address regex? */
 792                                 || (sed_cmd->beg_match &&
 793                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 794
 795                         /* Or did we match last line of input? */
 796                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 797
 798                         /* Snapshot the value */
 799
 800                         matched = sed_cmd->in_match;
 801
 802                         /* Is this line the end of the current match? */
 803
 804                         if(matched) {
 805                                 sed_cmd->in_match = !(
 806                                         /* has the ending line come, or is this a single address command? */
 807                                         (sed_cmd->end_line ?
 808                                                 sed_cmd->end_line==-1 ?
 809                                                         !next_line
 810                                                         : sed_cmd->end_line<=linenum
 811                                                 : !sed_cmd->end_match)
 812                                         /* or does this line matches our last address regex */
 813                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 814                                 );
 815                         }
 816
 817                         /* Skip blocks of commands we didn't match. */
 818                         if (sed_cmd->cmd == '{') {
 819                                 if(sed_cmd->invert ? matched : !matched)
 820                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 821                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 822                                 continue;
 823                         }
 824
 825                         /* Okay, so did this line match? */
 826                         if (sed_cmd->invert ? !matched : matched) {
 827                                 /* Update last used regex in case a blank substitute BRE is found */
 828                                 if (sed_cmd->beg_match) {
 829                                         previous_regex_ptr = sed_cmd->beg_match;
 830                                 }
 831
 832                                 /* actual sedding */
 833                                 switch (sed_cmd->cmd) {
 834
 835                                         /* Print line number */
 836                                         case '=':
 837                                                 fprintf(nonstdout,"%d\n", linenum);
 838                                                 break;
 839
 840                                         /* Write the current pattern space up to the first newline */
 841                                         case 'P':
 842                                         {
 843                                                 char *tmp = strchr(pattern_space, '\n');
 844
 845                                                 if (tmp) {
 846                                                         *tmp = '\0';
 847                                                         sed_puts(pattern_space,1);
 848                                                         *tmp = '\n';
 849                                                         break;
 850                                                 }
 851                                                 /* Fall Through */
 852                                         }
 853
 854                                         /* Write the current pattern space to output */
 855                                         case 'p':
 856                                                 sed_puts(pattern_space,no_newline);
 857                                                 break;
 858                                         /* Delete up through first newline */
 859                                         case 'D':
 860                                         {
 861                                                 char *tmp = strchr(pattern_space,'\n');
 862
 863                                                 if(tmp) {
 864                                                         tmp=bb_xstrdup(tmp+1);
 865                                                         free(pattern_space);
 866                                                         pattern_space=tmp;
 867                                                         goto restart;
 868                                                 }
 869                                         }
 870                                         /* discard this line. */
 871                                         case 'd':
 872                                                 goto discard_line;
 873
 874                                         /* Substitute with regex */
 875                                         case 's':
 876                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 877                                                         substituted|=1;
 878
 879                                                         /* handle p option */
 880                                                         if(sed_cmd->sub_p)
 881                                                                 sed_puts(pattern_space,no_newline);
 882                                                         /* handle w option */
 883                                                         if(sed_cmd->file)
 884                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 885
 886                                                 }
 887                                                 break;
 888
 889                                         /* Append line to linked list to be printed later */
 890                                         case 'a':
 891                                         {
 892                                                 append(sed_cmd->string);
 893                                                 break;
 894                                         }
 895
 896                                         /* Insert text before this line */
 897                                         case 'i':
 898                                                 sed_puts(sed_cmd->string,1);
 899                                                 break;
 900
 901                                         /* Cut and paste text (replace) */
 902                                         case 'c':
 903                                                 /* Only triggers on last line of a matching range. */
 904                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,1);
 905                                                 goto discard_line;
 906
 907                                         /* Read file, append contents to output */
 908                                         case 'r':
 909                                         {
 910                                                 FILE *outfile;
 911
 912                                                 outfile = fopen(sed_cmd->string, "r");
 913                                                 if (outfile) {
 914                                                         char *line;
 915
 916                                                         while ((line = bb_get_chomped_line_from_file(outfile))
 917                                                                         != NULL)
 918                                                                 append(line);
 919                                                         bb_xprint_and_close_file(outfile);
 920                                                 }
 921
 922                                                 break;
 923                                         }
 924
 925                                         /* Write pattern space to file. */
 926                                         case 'w':
 927                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 928                                                 break;
 929
 930                                         /* Read next line from input */
 931                                         case 'n':
 932                                                 if (!be_quiet)
 933                                                         sed_puts(pattern_space,no_newline);
 934                                                 if (next_line) {
 935                                                         free(pattern_space);
 936                                                         pattern_space = next_line;
 937                                                         no_newline=next_no_newline;
 938                                                         next_line = get_next_line(file,&next_no_newline);
 939                                                         linenum++;
 940                                                         break;
 941                                                 }
 942                                                 /* fall through */
 943
 944                                         /* Quit.  End of script, end of input. */
 945                                         case 'q':
 946                                                 /* Exit the outer while loop */
 947                                                 free(next_line);
 948                                                 next_line = NULL;
 949                                                 goto discard_commands;
 950
 951                                         /* Append the next line to the current line */
 952                                         case 'N':
 953                                         {
 954                                                 /* If no next line, jump to end of script and exit. */
 955                                                 if (next_line == NULL) {
 956                                                         /* Jump to end of script and exit */
 957                                                         free(next_line);
 958                                                         next_line = NULL;
 959                                                         goto discard_line;
 960                                                 /* append next_line, read new next_line. */
 961                                                 } else {
 962                                                         int len=strlen(pattern_space);
 963
 964                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 965                                                         pattern_space[len]='\n';
 966                                                         strcpy(pattern_space+len+1, next_line);
 967                                                         no_newline=next_no_newline;
 968                                                         next_line = get_next_line(file,&next_no_newline);
 969                                                         linenum++;
 970                                                 }
 971                                                 break;
 972                                         }
 973
 974                                         /* Test if substition worked, branch if so. */
 975                                         case 't':
 976                                                 if (!substituted) break;
 977                                                 substituted=0;
 978                                                         /* Fall through */
 979                                         /* Branch to label */
 980                                         case 'b':
 981                                                 if (!sed_cmd->string) goto discard_commands;
 982                                                 else sed_cmd = branch_to(sed_cmd->string);
 983                                                 break;
 984                                         /* Transliterate characters */
 985                                         case 'y':
 986                                         {
 987                                                 int i;
 988
 989                                                 for (i = 0; pattern_space[i]; i++) {
 990                                                         int j;
 991
 992                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
 993                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
 994                                                                         pattern_space[i] = sed_cmd->string[j + 1];
 995                                                                 }
 996                                                         }
 997                                                 }
 998
 999                                                 break;
1000                                         }
1001                                         case 'g':       /* Replace pattern space with hold space */
1002                                                 free(pattern_space);
1003                                                 if (hold_space) {
1004                                                         pattern_space = strdup(hold_space);
1005                                                         no_newline=0;
1006                                                 }
1007                                                 break;
1008                                         case 'G':       /* Append newline and hold space to pattern space */
1009                                         {
1010                                                 int pattern_space_size = 2;
1011                                                 int hold_space_size = 0;
1012
1013                                                 if (pattern_space)
1014                                                         pattern_space_size += strlen(pattern_space);
1015                                                 if (hold_space) hold_space_size = strlen(hold_space);
1016                                                 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1017                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1018                                                 strcat(pattern_space, "\n");
1019                                                 if (hold_space) strcat(pattern_space, hold_space);
1020                                                 no_newline=0;
1021
1022                                                 break;
1023                                         }
1024                                         case 'h':       /* Replace hold space with pattern space */
1025                                                 free(hold_space);
1026                                                 hold_space = strdup(pattern_space);
1027                                                 break;
1028                                         case 'H':       /* Append newline and pattern space to hold space */
1029                                         {
1030                                                 int hold_space_size = 2;
1031                                                 int pattern_space_size = 0;
1032
1033                                                 if (hold_space) hold_space_size += strlen(hold_space);
1034                                                 if (pattern_space)
1035                                                         pattern_space_size = strlen(pattern_space);
1036                                                 hold_space = xrealloc(hold_space,
1037                                                                                         hold_space_size + pattern_space_size);
1038
1039                                                 if (hold_space_size == 2) hold_space[0]=0;
1040                                                 strcat(hold_space, "\n");
1041                                                 if (pattern_space) strcat(hold_space, pattern_space);
1042
1043                                                 break;
1044                                         }
1045                                         case 'x': /* Exchange hold and pattern space */
1046                                         {
1047                                                 char *tmp = pattern_space;
1048                                                 pattern_space = hold_space;
1049                                                 no_newline=0;
1050                                                 hold_space = tmp;
1051                                                 break;
1052                                         }
1053                                 }
1054                         }
1055                 }
1056
1057                 /*
1058                  * exit point from sedding...
1059                  */
1060 discard_commands:
1061                 /* we will print the line unless we were told to be quiet ('-n')
1062                    or if the line was suppressed (ala 'd'elete) */
1063                 if (!be_quiet) sed_puts(pattern_space,no_newline);
1064
1065                 /* Delete and such jump here. */
1066 discard_line:
1067                 flush_append();
1068                 free(pattern_space);
1069         }
1070 }
1071
1072 /* It is possible to have a command line argument with embedded
1073    newlines.  This counts as multiple command lines. */
1074
1075 static void add_cmd_block(char *cmdstr)
1076 {
1077         int go=1;
1078         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1079
1080         while(go) {
1081                 int len=strcspn(temp2,"\n");
1082                 if(!temp2[len]) go=0;
1083                 else temp2[len]=0;
1084                 add_cmd(temp2);
1085                 temp2+=len+1;
1086         }
1087         free(temp);
1088 }
1089
1090 extern int sed_main(int argc, char **argv)
1091 {
1092         int opt, status = EXIT_SUCCESS;
1093
1094 #ifdef CONFIG_FEATURE_CLEAN_UP
1095         /* destroy command strings on exit */
1096         if (atexit(free_and_close_stuff) == -1)
1097                 bb_perror_msg_and_die("atexit");
1098 #endif
1099
1100 #define LIE_TO_AUTOCONF
1101 #ifdef LIE_TO_AUTOCONF
1102         if(argc==2 && !strcmp(argv[1],"--version")) {
1103                 printf("This is not GNU sed version 4.0\n");
1104                 exit(0);
1105         }
1106 #endif
1107
1108         /* do normal option parsing */
1109         while ((opt = getopt(argc, argv, "ine:f:")) > 0) {
1110                 switch (opt) {
1111                 case 'i':
1112                         in_place++;
1113                         atexit(cleanup_outname);
1114                         break;
1115                 case 'n':
1116                         be_quiet++;
1117                         break;
1118                 case 'e':
1119                         add_cmd_block(optarg);
1120                         break;
1121                 case 'f':
1122                 {
1123                         FILE *cmdfile;
1124                         char *line;
1125
1126                         cmdfile = bb_xfopen(optarg, "r");
1127
1128                         while ((line = bb_get_chomped_line_from_file(cmdfile))
1129                                  != NULL) {
1130                                 add_cmd(line);
1131                                 free(line);
1132                         }
1133                         bb_xprint_and_close_file(cmdfile);
1134
1135                         break;
1136                 }
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_block(argv[optind++]);
1149         }
1150         /* Flush any unfinished commands. */
1151         add_cmd("");
1152
1153         /* By default, we write to stdout */
1154         nonstdout=stdout;
1155
1156         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1157          * files were specified or '-' was specified, take input from stdin.
1158          * Otherwise, we process all the files specified. */
1159         if (argv[optind] == NULL) {
1160                 if(in_place) {
1161                         fprintf(stderr,"sed: Filename required for -i\n");
1162                         exit(1);
1163                 }
1164                 process_file(stdin);
1165         } else {
1166                 int i;
1167                 FILE *file;
1168
1169                 for (i = optind; i < argc; i++) {
1170                         if(!strcmp(argv[i], "-") && !in_place) {
1171                                 process_file(stdin);
1172                         } else {
1173                                 file = bb_wfopen(argv[i], "r");
1174                                 if (file) {
1175                                         if(in_place) {
1176                                                 struct stat statbuf;
1177                                                 outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1178                                                 strcat(outname,"XXXXXX");
1179                                                 /* Set permissions of output file */
1180                                                 fstat(fileno(file),&statbuf);
1181                                                 mkstemp(outname);
1182                                                 nonstdout=bb_wfopen(outname,"w");
1183                                                 /* Set permissions of output file */
1184                                                 fstat(fileno(file),&statbuf);
1185                                                 fchmod(fileno(file),statbuf.st_mode);
1186                                                 atexit(cleanup_outname);
1187                                         }
1188                                         process_file(file);
1189                                         fclose(file);
1190                                         if(in_place) {
1191                                                 fclose(nonstdout);
1192                                                 nonstdout=stdout;
1193                                                 unlink(argv[i]);
1194                                                 rename(outname,argv[i]);
1195                                                 free(outname);
1196                                                 outname=0;
1197                                         }
1198                                 } else {
1199                                         status = EXIT_FAILURE;
1200                                 }
1201                         }
1202                 }
1203         }
1204
1205         return status;
1206 }