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;
 115
 116 static const char bad_format_in_subst[] =
 117         "bad format in substitution expression";
 118 const char *const semicolon_whitespace = "; \n\r\t\v";
 119
 120 regmatch_t regmatch[10];
 121 static regex_t *previous_regex_ptr = NULL;
 122
 123 /* linked list of sed commands */
 124 static sed_cmd_t sed_cmd_head;
 125 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
 126
 127 /* Linked list of append lines */
 128 struct append_list {
 129         char *string;
 130         struct append_list *next;
 131 };
 132 struct append_list *append_head=NULL, *append_tail=NULL;
 133
 134 #ifdef CONFIG_FEATURE_CLEAN_UP
 135 static void free_and_close_stuff(void)
 136 {
 137         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 138
 139         while(append_head) {
 140                 append_tail=append_head->next;
 141                 free(append_head->string);
 142                 free(append_head);
 143                 append_head=append_tail;
 144         }
 145
 146         while (sed_cmd) {
 147                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 148
 149                 if(sed_cmd->file)
 150                         bb_xprint_and_close_file(sed_cmd->file);
 151
 152                 if (sed_cmd->beg_match) {
 153                         regfree(sed_cmd->beg_match);
 154                         free(sed_cmd->beg_match);
 155                 }
 156                 if (sed_cmd->end_match) {
 157                         regfree(sed_cmd->end_match);
 158                         free(sed_cmd->end_match);
 159                 }
 160                 if (sed_cmd->sub_match) {
 161                         regfree(sed_cmd->sub_match);
 162                         free(sed_cmd->sub_match);
 163                 }
 164                 free(sed_cmd->string);
 165                 free(sed_cmd);
 166                 sed_cmd = sed_cmd_next;
 167         }
 168 }
 169 #endif
 170
 171 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 172
 173 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 174 {
 175         int i=0;
 176
 177         while(i<len) {
 178                 if(string[i] == '\\') {
 179                         if(!to || string[i+1] == from) {
 180                                 *(dest++) = to ? to : string[i+1];
 181                                 i+=2;
 182                                 continue;
 183                         } else *(dest++)=string[i++];
 184                 }
 185                 *(dest++) = string[i++];
 186         }
 187         *dest=0;
 188 }
 189
 190 static char *copy_parsing_slashn(const char *string, int len)
 191 {
 192         char *dest=xmalloc(len+1);
 193
 194         parse_escapes(dest,string,len,'n','\n');
 195         return dest;
 196 }
 197
 198
 199 /*
 200  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 201  * beginning at a specified index and returns the index of the next regular
 202  * expression delimiter (typically a forward * slash ('/')) not preceeded by
 203  * a backslash ('\').
 204  */
 205 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 206         const char *str)
 207 {
 208         int bracket = -1;
 209         int escaped = 0;
 210         int idx = 0;
 211         char ch;
 212
 213         for (; (ch = str[idx]); idx++) {
 214                 if (bracket != -1) {
 215                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 216                                         && str[idx - 1] == '^')))
 217                                 bracket = -1;
 218                 } else if (escaped)
 219                         escaped = 0;
 220                 else if (ch == '\\')
 221                         escaped = 1;
 222                 else if (ch == '[')
 223                         bracket = idx;
 224                 else if (ch == delimiter)
 225                         return idx;
 226         }
 227
 228         /* if we make it to here, we've hit the end of the string */
 229         return -1;
 230 }
 231
 232 /*
 233  *  Returns the index of the third delimiter
 234  */
 235 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 236 {
 237         const char *cmdstr_ptr = cmdstr;
 238         char delimiter;
 239         int idx = 0;
 240
 241         /* verify that the 's' or 'y' is followed by something.  That something
 242          * (typically a 'slash') is now our regexp delimiter... */
 243         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
 244         delimiter = *(cmdstr_ptr++);
 245
 246         /* save the match string */
 247         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 248         if (idx == -1) {
 249                 bb_error_msg_and_die(bad_format_in_subst);
 250         }
 251         *match = copy_parsing_slashn(cmdstr_ptr, idx);
 252
 253         /* save the replacement string */
 254         cmdstr_ptr += idx + 1;
 255         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 256         if (idx == -1) {
 257                 bb_error_msg_and_die(bad_format_in_subst);
 258         }
 259         *replace = copy_parsing_slashn(cmdstr_ptr, idx);
 260
 261         return ((cmdstr_ptr - cmdstr) + idx);
 262 }
 263
 264 /*
 265  * returns the index in the string just past where the address ends.
 266  */
 267 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 268 {
 269         char *pos = my_str;
 270
 271         if (isdigit(*my_str)) {
 272                 *linenum = strtol(my_str, &pos, 10);
 273                 /* endstr shouldnt ever equal NULL */
 274         } else if (*my_str == '$') {
 275                 *linenum = -1;
 276                 pos++;
 277         } else if (*my_str == '/' || *my_str == '\\') {
 278                 int next;
 279                 char delimiter;
 280                 char *temp;
 281
 282                 if (*my_str == '\\') delimiter = *(++pos);
 283                 else delimiter = '/';
 284                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 285                 if (next == -1)
 286                         bb_error_msg_and_die("unterminated match expression");
 287
 288                 temp=copy_parsing_slashn(pos,next);
 289                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 290                 xregcomp(*regex, temp, REG_NEWLINE);
 291                 free(temp);
 292                 /* Move position to next character after last delimiter */
 293                 pos+=(next+1);
 294         }
 295         return pos - my_str;
 296 }
 297
 298 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 299 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
 300 {
 301         int start = 0, idx, hack=0;
 302
 303         /* Skip whitespace, then grab filename to end of line */
 304         while (isspace(filecmdstr[start])) start++;
 305         idx=start;
 306         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 307         /* If lines glued together, put backslash back. */
 308         if(filecmdstr[idx]=='\n') hack=1;
 309         if(idx==start) bb_error_msg_and_die("Empty filename");
 310         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
 311         if(hack) *(idx+*retval)='\\';
 312
 313         return idx;
 314 }
 315
 316 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
 317 {
 318         int cflags = 0;
 319         char *match;
 320         int idx = 0;
 321
 322         /*
 323          * A substitution command should look something like this:
 324          *    s/match/replace/ #gIpw
 325          *    ||     |        |||
 326          *    mandatory       optional
 327          */
 328         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 329
 330         /* determine the number of back references in the match string */
 331         /* Note: we compute this here rather than in the do_subst_command()
 332          * function to save processor time, at the expense of a little more memory
 333          * (4 bits) per sed_cmd */
 334
 335         /* process the flags */
 336
 337         sed_cmd->which_match=1;
 338         while (substr[++idx]) {
 339                 /* Parse match number */
 340                 if(isdigit(substr[idx])) {
 341                         if(match[0]!='^') {
 342                                 /* Match 0 treated as all, multiple matches we take the last one. */
 343                                 char *pos=substr+idx;
 344                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 345                                 idx=pos-substr;
 346                         }
 347                         continue;
 348                 }
 349                 /* Skip spaces */
 350                 if(isspace(substr[idx])) continue;
 351
 352                 switch (substr[idx]) {
 353                         /* Replace all occurrences */
 354                         case 'g':
 355                                 if (match[0] != '^') sed_cmd->which_match = 0;
 356                                 break;
 357                         /* Print pattern space */
 358                         case 'p':
 359                                 sed_cmd->sub_p = 1;
 360                                 break;
 361                         case 'w':
 362                         {
 363                                 char *temp;
 364                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 365
 366                                 break;
 367                         }
 368                         /* Ignore case (gnu exension) */
 369                         case 'I':
 370                                 cflags |= REG_ICASE;
 371                                 break;
 372                         case ';':
 373                         case '}':
 374                                 goto out;
 375                         default:
 376                                 bb_error_msg_and_die("bad option in substitution expression");
 377                 }
 378         }
 379 out:
 380         /* compile the match string into a regex */
 381         if (*match != '\0') {
 382                 /* If match is empty, we use last regex used at runtime */
 383                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 384                 xregcomp(sed_cmd->sub_match, match, cflags);
 385         }
 386         free(match);
 387
 388         return idx;
 389 }
 390
 391 /*
 392  *  Process the commands arguments
 393  */
 394 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 395 {
 396         /* handle (s)ubstitution command */
 397         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 398         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 399         else if (strchr("aic", sed_cmd->cmd)) {
 400                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 401                         bb_error_msg_and_die
 402                                 ("only a beginning address can be specified for edit commands");
 403                 while(isspace(*cmdstr)) cmdstr++;
 404                 sed_cmd->string = bb_xstrdup(cmdstr);
 405                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 406                 cmdstr += strlen(cmdstr);
 407         /* handle file cmds: (r)ead */
 408         } else if(strchr("rw", sed_cmd->cmd)) {
 409                 if (sed_cmd->end_line || sed_cmd->end_match)
 410                         bb_error_msg_and_die("Command only uses one address");
 411                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 412                 if(sed_cmd->cmd=='w')
 413                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
 414         /* handle branch commands */
 415         } else if (strchr(":bt", sed_cmd->cmd)) {
 416                 int length;
 417
 418                 while(isspace(*cmdstr)) cmdstr++;
 419                 length = strcspn(cmdstr, semicolon_whitespace);
 420                 if (length) {
 421                         sed_cmd->string = strndup(cmdstr, length);
 422                         cmdstr += length;
 423                 }
 424         }
 425         /* translation command */
 426         else if (sed_cmd->cmd == 'y') {
 427                 char *match, *replace;
 428                 int i=cmdstr[0];
 429
 430                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 431                 /* \n already parsed, but \delimiter needs unescaping. */
 432                 parse_escapes(match,match,strlen(match),i,i);
 433                 parse_escapes(replace,replace,strlen(replace),i,i);
 434
 435                 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
 436                 for (i = 0; match[i] && replace[i]; i++) {
 437                         sed_cmd->string[i * 2] = match[i];
 438                         sed_cmd->string[(i * 2) + 1] = replace[i];
 439                 }
 440                 free(match);
 441                 free(replace);
 442         }
 443         /* if it wasnt a single-letter command that takes no arguments
 444          * then it must be an invalid command.
 445          */
 446         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 447                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 448         }
 449
 450         /* give back whatever's left over */
 451         return (cmdstr);
 452 }
 453
 454
 455 /* Parse address+command sets, skipping comment lines. */
 456
 457 void add_cmd(char *cmdstr)
 458 {
 459         static char *add_cmd_line=NULL;
 460         sed_cmd_t *sed_cmd;
 461         int temp;
 462
 463         /* Append this line to any unfinished line from last time. */
 464         if(add_cmd_line) {
 465                 int lastlen=strlen(add_cmd_line);
 466                 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
 467
 468                 memcpy(tmp,add_cmd_line,lastlen);
 469                 tmp[lastlen]='\n';
 470                 strcpy(tmp+lastlen+1,cmdstr);
 471                 free(add_cmd_line);
 472                 cmdstr=add_cmd_line=tmp;
 473         } else add_cmd_line=NULL;
 474
 475         /* If this line ends with backslash, request next line. */
 476         temp=strlen(cmdstr);
 477         if(temp && cmdstr[temp-1]=='\\') {
 478                 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
 479                 add_cmd_line[temp-1]=0;
 480                 return;
 481         }
 482
 483         /* Loop parsing all commands in this line. */
 484         while(*cmdstr) {
 485                 /* Skip leading whitespace and semicolons */
 486                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 487
 488                 /* If no more commands, exit. */
 489                 if(!*cmdstr) break;
 490
 491                 /* if this is a comment, jump past it and keep going */
 492                 if (*cmdstr == '#') {
 493                         /* "#n" is the same as using -n on the command line */
 494                         if (cmdstr[1] == 'n') be_quiet++;
 495                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 496                         continue;
 497                 }
 498
 499                 /* parse the command
 500                  * format is: [addr][,addr][!]cmd
 501                  *            |----||-----||-|
 502                  *            part1 part2  part3
 503                  */
 504
 505                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 506
 507                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 508                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 509
 510                 /* second part (if present) will begin with a comma */
 511                 if (*cmdstr == ',') {
 512                         int idx;
 513
 514                         cmdstr++;
 515                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 516                         if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
 517                         cmdstr += idx;
 518                 }
 519
 520                 /* skip whitespace before the command */
 521                 while (isspace(*cmdstr)) cmdstr++;
 522
 523                 /* Check for inversion flag */
 524                 if (*cmdstr == '!') {
 525                         sed_cmd->invert = 1;
 526                         cmdstr++;
 527
 528                         /* skip whitespace before the command */
 529                         while (isspace(*cmdstr)) cmdstr++;
 530                 }
 531
 532                 /* last part (mandatory) will be a command */
 533                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 534                 sed_cmd->cmd = *(cmdstr++);
 535                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 536
 537                 /* Add the command to the command array */
 538                 sed_cmd_tail->next = sed_cmd;
 539                 sed_cmd_tail = sed_cmd_tail->next;
 540         }
 541
 542         /* If we glued multiple lines together, free the memory. */
 543         if(add_cmd_line) {
 544                 free(add_cmd_line);
 545                 add_cmd_line=NULL;
 546         }
 547 }
 548
 549 struct pipeline {
 550         char *buf;      /* Space to hold string */
 551         int idx;        /* Space used */
 552         int len;        /* Space allocated */
 553 } pipeline;
 554
 555 #define PIPE_GROW 64
 556
 557 void pipe_putc(char c)
 558 {
 559         if(pipeline.idx==pipeline.len) {
 560                 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
 561                 pipeline.len+=PIPE_GROW;
 562         }
 563         pipeline.buf[pipeline.idx++] = (c);
 564 }
 565
 566 static void do_subst_w_backrefs(const char *line, const char *replace)
 567 {
 568         int i,j;
 569
 570         /* go through the replacement string */
 571         for (i = 0; replace[i]; i++) {
 572                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 573                 if (replace[i] == '\\' && replace[i+1]>'0' && replace[i+1]<='9') {
 574                         int backref=replace[++i]-'0';
 575
 576                         /* print out the text held in regmatch[backref] */
 577                         if(regmatch[backref].rm_so != -1)
 578                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
 579                                         pipe_putc(line[j]);
 580                 }
 581
 582                 /* if we find a backslash escaped character, print the character */
 583                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 584
 585                 /* if we find an unescaped '&' print out the whole matched text. */
 586                 else if (replace[i] == '&')
 587                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 588                                 pipe_putc(line[j]);
 589                 /* Otherwise just output the character. */
 590                 else pipe_putc(replace[i]);
 591         }
 592 }
 593
 594 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 595 {
 596         char *oldline = *line;
 597         int altered = 0;
 598         int match_count=0;
 599         regex_t *current_regex;
 600
 601         /* Handle empty regex. */
 602         if (sed_cmd->sub_match == NULL) {
 603                 current_regex = previous_regex_ptr;
 604                 if(!current_regex)
 605                         bb_error_msg_and_die("No previous regexp.");
 606         } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
 607
 608         /* Find the first match */
 609         if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
 610                 return 0;
 611
 612         /* Initialize temporary output buffer. */
 613         pipeline.buf=xmalloc(PIPE_GROW);
 614         pipeline.len=PIPE_GROW;
 615         pipeline.idx=0;
 616
 617         /* Now loop through, substituting for matches */
 618         do {
 619                 int i;
 620
 621                 /* Work around bug in glibc regexec, demonstrated by:
 622                    echo " a.b" | busybox sed 's [^ .]* x g'
 623                    The match_count check is so not to break
 624                    echo "hi" | busybox sed 's/^/!/g' */
 625                 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
 626                         pipe_putc(*(oldline++));
 627                         continue;
 628                 }
 629
 630                 match_count++;
 631
 632                 /* If we aren't interested in this match, output old line to
 633                    end of match and continue */
 634                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 635                         for(i=0;i<regmatch[0].rm_eo;i++)
 636                                 pipe_putc(oldline[i]);
 637                         continue;
 638                 }
 639
 640                 /* print everything before the match */
 641                 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 642
 643                 /* then print the substitution string */
 644                 do_subst_w_backrefs(oldline, sed_cmd->string);
 645
 646                 /* advance past the match */
 647                 oldline += regmatch[0].rm_eo;
 648                 /* flag that something has changed */
 649                 altered++;
 650
 651                 /* if we're not doing this globally, get out now */
 652                 if (sed_cmd->which_match) break;
 653         } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
 654
 655         /* Copy rest of string into output pipeline */
 656
 657         while(*oldline) pipe_putc(*(oldline++));
 658         pipe_putc(0);
 659
 660         free(*line);
 661         *line = pipeline.buf;
 662         return altered;
 663 }
 664
 665 /* Set command pointer to point to this label.  (Does not handle null label.) */
 666 static sed_cmd_t *branch_to(const char *label)
 667 {
 668         sed_cmd_t *sed_cmd;
 669
 670         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 671                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 672                         return (sed_cmd);
 673                 }
 674         }
 675         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 676 }
 677
 678 /* Append copy of string to append buffer */
 679 static void append(char *s)
 680 {
 681         struct append_list *temp=calloc(1,sizeof(struct append_list));
 682
 683         if(append_head)
 684                 append_tail=(append_tail->next=temp);
 685         else append_head=append_tail=temp;
 686         temp->string=strdup(s);
 687 }
 688
 689 static void flush_append(void)
 690 {
 691         /* Output appended lines. */
 692         while(append_head) {
 693                 puts(append_head->string);
 694                 append_tail=append_head->next;
 695                 free(append_head->string);
 696                 free(append_head);
 697                 append_head=append_tail;
 698         }
 699         append_head=append_tail=NULL;
 700 }
 701
 702 /* Get next line of input, flushing append buffer and noting if we hit EOF
 703  * without a newline on the last line.
 704  */
 705 static char *get_next_line(FILE * file, int *no_newline)
 706 {
 707         char *temp;
 708         int len;
 709
 710         flush_append();
 711         temp=bb_get_line_from_file(file);
 712         if(temp) {
 713                 len=strlen(temp);
 714                 if(len && temp[len-1]=='\n') temp[len-1]=0;
 715                 else *no_newline=1;
 716         }
 717
 718         return temp;
 719 }
 720
 721 /* Output line of text.  missing_newline means the last line output did not
 722    end with a newline.  no_newline means this line does not end with a
 723    newline. */
 724
 725 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 726 {
 727         if(missing_newline) fputc('\n',file);
 728         fputs(s,file);
 729         if(!no_newline) fputc('\n',file);
 730
 731         return no_newline;
 732 }
 733
 734 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,stdout,missing_newline,n)
 735
 736 static void process_file(FILE * file)
 737 {
 738         char *pattern_space, *next_line, *hold_space=NULL;
 739         static int linenum = 0, missing_newline=0;
 740         int no_newline,next_no_newline=0;
 741
 742         next_line = get_next_line(file,&next_no_newline);
 743
 744         /* go through every line in the file */
 745         for(;;) {
 746                 sed_cmd_t *sed_cmd;
 747                 int substituted=0;
 748
 749                 /* Advance to next line.  Stop if out of lines. */
 750                 if(!(pattern_space=next_line)) break;
 751                 no_newline=next_no_newline;
 752
 753                 /* Read one line in advance so we can act on the last line, the '$' address */
 754                 next_line = get_next_line(file,&next_no_newline);
 755                 linenum++;
 756 restart:
 757                 /* for every line, go through all the commands */
 758                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 759                         int old_matched, matched;
 760
 761                         old_matched = sed_cmd->in_match;
 762
 763                         /* Determine if this command matches this line: */
 764
 765                         /* Are we continuing a previous multi-line match? */
 766
 767                         sed_cmd->in_match = sed_cmd->in_match
 768
 769                         /* Or is no range necessary? */
 770                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 771                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 772
 773                         /* Or did we match the start of a numerical range? */
 774                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 775
 776                         /* Or does this line match our begin address regex? */
 777                                 || (sed_cmd->beg_match &&
 778                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 779
 780                         /* Or did we match last line of input? */
 781                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 782
 783                         /* Snapshot the value */
 784
 785                         matched = sed_cmd->in_match;
 786
 787                         /* Is this line the end of the current match? */
 788
 789                         if(matched) {
 790                                 sed_cmd->in_match = !(
 791                                         /* has the ending line come, or is this a single address command? */
 792                                         (sed_cmd->end_line ?
 793                                                 sed_cmd->end_line==-1 ?
 794                                                         !next_line
 795                                                         : sed_cmd->end_line<=linenum
 796                                                 : !sed_cmd->end_match)
 797                                         /* or does this line matches our last address regex */
 798                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 799                                 );
 800                         }
 801
 802                         /* Skip blocks of commands we didn't match. */
 803                         if (sed_cmd->cmd == '{') {
 804                                 if(sed_cmd->invert ? matched : !matched)
 805                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 806                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 807                                 continue;
 808                         }
 809
 810                         /* Okay, so did this line match? */
 811                         if (sed_cmd->invert ? !matched : matched) {
 812                                 /* Update last used regex in case a blank substitute BRE is found */
 813                                 if (sed_cmd->beg_match) {
 814                                         previous_regex_ptr = sed_cmd->beg_match;
 815                                 }
 816
 817                                 /* actual sedding */
 818                                 switch (sed_cmd->cmd) {
 819
 820                                         /* Print line number */
 821                                         case '=':
 822                                                 printf("%d\n", linenum);
 823                                                 break;
 824
 825                                         /* Write the current pattern space up to the first newline */
 826                                         case 'P':
 827                                         {
 828                                                 char *tmp = strchr(pattern_space, '\n');
 829
 830                                                 if (tmp) {
 831                                                         *tmp = '\0';
 832                                                         sed_puts(pattern_space,1);
 833                                                         *tmp = '\n';
 834                                                         break;
 835                                                 }
 836                                                 /* Fall Through */
 837                                         }
 838
 839                                         /* Write the current pattern space to output */
 840                                         case 'p':
 841                                                 sed_puts(pattern_space,no_newline);
 842                                                 break;
 843                                         /* Delete up through first newline */
 844                                         case 'D':
 845                                         {
 846                                                 char *tmp = strchr(pattern_space,'\n');
 847
 848                                                 if(tmp) {
 849                                                         tmp=bb_xstrdup(tmp+1);
 850                                                         free(pattern_space);
 851                                                         pattern_space=tmp;
 852                                                         goto restart;
 853                                                 }
 854                                         }
 855                                         /* discard this line. */
 856                                         case 'd':
 857                                                 goto discard_line;
 858
 859                                         /* Substitute with regex */
 860                                         case 's':
 861                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 862                                                         substituted|=1;
 863
 864                                                         /* handle p option */
 865                                                         if(sed_cmd->sub_p)
 866                                                                 sed_puts(pattern_space,no_newline);
 867                                                         /* handle w option */
 868                                                         if(sed_cmd->file)
 869                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 870
 871                                                 }
 872                                                 break;
 873
 874                                         /* Append line to linked list to be printed later */
 875                                         case 'a':
 876                                         {
 877                                                 append(sed_cmd->string);
 878                                                 break;
 879                                         }
 880
 881                                         /* Insert text before this line */
 882                                         case 'i':
 883                                                 sed_puts(sed_cmd->string,1);
 884                                                 break;
 885
 886                                         /* Cut and paste text (replace) */
 887                                         case 'c':
 888                                                 /* Only triggers on last line of a matching range. */
 889                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,1);
 890                                                 goto discard_line;
 891
 892                                         /* Read file, append contents to output */
 893                                         case 'r':
 894                                         {
 895                                                 FILE *outfile;
 896
 897                                                 outfile = fopen(sed_cmd->string, "r");
 898                                                 if (outfile) {
 899                                                         char *line;
 900
 901                                                         while ((line = bb_get_chomped_line_from_file(outfile))
 902                                                                         != NULL)
 903                                                                 append(line);
 904                                                         bb_xprint_and_close_file(outfile);
 905                                                 }
 906
 907                                                 break;
 908                                         }
 909
 910                                         /* Write pattern space to file. */
 911                                         case 'w':
 912                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 913                                                 break;
 914
 915                                         /* Read next line from input */
 916                                         case 'n':
 917                                                 if (!be_quiet)
 918                                                         sed_puts(pattern_space,no_newline);
 919                                                 if (next_line) {
 920                                                         free(pattern_space);
 921                                                         pattern_space = next_line;
 922                                                         no_newline=next_no_newline;
 923                                                         next_line = get_next_line(file,&next_no_newline);
 924                                                         linenum++;
 925                                                         break;
 926                                                 }
 927                                                 /* fall through */
 928
 929                                         /* Quit.  End of script, end of input. */
 930                                         case 'q':
 931                                                 /* Exit the outer while loop */
 932                                                 free(next_line);
 933                                                 next_line = NULL;
 934                                                 goto discard_commands;
 935
 936                                         /* Append the next line to the current line */
 937                                         case 'N':
 938                                         {
 939                                                 /* If no next line, jump to end of script and exit. */
 940                                                 if (next_line == NULL) {
 941                                                         /* Jump to end of script and exit */
 942                                                         free(next_line);
 943                                                         next_line = NULL;
 944                                                         goto discard_line;
 945                                                 /* append next_line, read new next_line. */
 946                                                 } else {
 947                                                         int len=strlen(pattern_space);
 948
 949                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 950                                                         pattern_space[len]='\n';
 951                                                         strcpy(pattern_space+len+1, next_line);
 952                                                         no_newline=next_no_newline;
 953                                                         next_line = get_next_line(file,&next_no_newline);
 954                                                         linenum++;
 955                                                 }
 956                                                 break;
 957                                         }
 958
 959                                         /* Test if substition worked, branch if so. */
 960                                         case 't':
 961                                                 if (!substituted) break;
 962                                                 substituted=0;
 963                                                         /* Fall through */
 964                                         /* Branch to label */
 965                                         case 'b':
 966                                                 if (!sed_cmd->string) goto discard_commands;
 967                                                 else sed_cmd = branch_to(sed_cmd->string);
 968                                                 break;
 969                                         /* Transliterate characters */
 970                                         case 'y':
 971                                         {
 972                                                 int i;
 973
 974                                                 for (i = 0; pattern_space[i]; i++) {
 975                                                         int j;
 976
 977                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
 978                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
 979                                                                         pattern_space[i] = sed_cmd->string[j + 1];
 980                                                                 }
 981                                                         }
 982                                                 }
 983
 984                                                 break;
 985                                         }
 986                                         case 'g':       /* Replace pattern space with hold space */
 987                                                 free(pattern_space);
 988                                                 if (hold_space) {
 989                                                         pattern_space = strdup(hold_space);
 990                                                         no_newline=0;
 991                                                 }
 992                                                 break;
 993                                         case 'G':       /* Append newline and hold space to pattern space */
 994                                         {
 995                                                 int pattern_space_size = 2;
 996                                                 int hold_space_size = 0;
 997
 998                                                 if (pattern_space)
 999                                                         pattern_space_size += strlen(pattern_space);
1000                                                 if (hold_space) hold_space_size = strlen(hold_space);
1001                                                 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1002                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1003                                                 strcat(pattern_space, "\n");
1004                                                 if (hold_space) strcat(pattern_space, hold_space);
1005                                                 no_newline=0;
1006
1007                                                 break;
1008                                         }
1009                                         case 'h':       /* Replace hold space with pattern space */
1010                                                 free(hold_space);
1011                                                 hold_space = strdup(pattern_space);
1012                                                 break;
1013                                         case 'H':       /* Append newline and pattern space to hold space */
1014                                         {
1015                                                 int hold_space_size = 2;
1016                                                 int pattern_space_size = 0;
1017
1018                                                 if (hold_space) hold_space_size += strlen(hold_space);
1019                                                 if (pattern_space)
1020                                                         pattern_space_size = strlen(pattern_space);
1021                                                 hold_space = xrealloc(hold_space,
1022                                                                                         hold_space_size + pattern_space_size);
1023
1024                                                 if (hold_space_size == 2) hold_space[0]=0;
1025                                                 strcat(hold_space, "\n");
1026                                                 if (pattern_space) strcat(hold_space, pattern_space);
1027
1028                                                 break;
1029                                         }
1030                                         case 'x': /* Exchange hold and pattern space */
1031                                         {
1032                                                 char *tmp = pattern_space;
1033                                                 pattern_space = hold_space;
1034                                                 no_newline=0;
1035                                                 hold_space = tmp;
1036                                                 break;
1037                                         }
1038                                 }
1039                         }
1040                 }
1041
1042                 /*
1043                  * exit point from sedding...
1044                  */
1045 discard_commands:
1046                 /* we will print the line unless we were told to be quiet ('-n')
1047                    or if the line was suppressed (ala 'd'elete) */
1048                 if (!be_quiet) sed_puts(pattern_space,no_newline);
1049
1050                 /* Delete and such jump here. */
1051 discard_line:
1052                 flush_append();
1053                 free(pattern_space);
1054         }
1055 }
1056
1057 /* It is possible to have a command line argument with embedded
1058    newlines.  This counts as multiple command lines. */
1059
1060 static void add_cmd_block(char *cmdstr)
1061 {
1062         int go=1;
1063         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1064
1065         while(go) {
1066                 int len=strcspn(temp2,"\n");
1067                 if(!temp2[len]) go=0;
1068                 else temp2[len]=0;
1069                 add_cmd(temp2);
1070                 temp2+=len+1;
1071         }
1072         free(temp);
1073 }
1074
1075 extern int sed_main(int argc, char **argv)
1076 {
1077         int opt, status = EXIT_SUCCESS;
1078
1079 #ifdef CONFIG_FEATURE_CLEAN_UP
1080         /* destroy command strings on exit */
1081         if (atexit(free_and_close_stuff) == -1)
1082                 bb_perror_msg_and_die("atexit");
1083 #endif
1084
1085 #define LIE_TO_AUTOCONF
1086 #ifdef LIE_TO_AUTOCONF
1087         if(argc==2 && !strcmp(argv[1],"--version")) {
1088                 printf("This is not GNU sed version 4.0\n");
1089                 exit(0);
1090         }
1091 #endif
1092
1093         /* do normal option parsing */
1094         while ((opt = getopt(argc, argv, "ne:f:")) > 0) {
1095                 switch (opt) {
1096                 case 'n':
1097                         be_quiet++;
1098                         break;
1099                 case 'e':
1100                         add_cmd_block(optarg);
1101                         break;
1102                 case 'f':
1103                 {
1104                         FILE *cmdfile;
1105                         char *line;
1106
1107                         cmdfile = bb_xfopen(optarg, "r");
1108
1109                         while ((line = bb_get_chomped_line_from_file(cmdfile))
1110                                  != NULL) {
1111                                 add_cmd(line);
1112                                 free(line);
1113                         }
1114                         bb_xprint_and_close_file(cmdfile);
1115
1116                         break;
1117                 }
1118                 default:
1119                         bb_show_usage();
1120                 }
1121         }
1122
1123         /* if we didn't get a pattern from a -e and no command file was specified,
1124          * argv[optind] should be the pattern. no pattern, no worky */
1125         if (sed_cmd_head.next == NULL) {
1126                 if (argv[optind] == NULL)
1127                         bb_show_usage();
1128                 else
1129                         add_cmd_block(argv[optind++]);
1130         }
1131         /* Flush any unfinished commands. */
1132         add_cmd("");
1133
1134         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1135          * files were specified or '-' was specified, take input from stdin.
1136          * Otherwise, we process all the files specified. */
1137         if (argv[optind] == NULL) {
1138                 process_file(stdin);
1139         } else {
1140                 int i;
1141                 FILE *file;
1142
1143                 for (i = optind; i < argc; i++) {
1144                         if(!strcmp(argv[i], "-")) {
1145                                 process_file(stdin);
1146                         } else {
1147                                 file = bb_wfopen(argv[i], "r");
1148                                 if (file) {
1149                                         process_file(file);
1150                                         fclose(file);
1151                                 } else {
1152                                         status = EXIT_FAILURE;
1153                                 }
1154                         }
1155                 }
1156         }
1157
1158         return status;
1159 }