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  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * This program is free software; you can redistribute it and/or modify
  12  * it under the terms of the GNU General Public License as published by
  13  * the Free Software Foundation; either version 2 of the License, or
  14  * (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  19  * General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program; if not, write to the Free Software
  23  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  24  *
  25  */
  26
  27 /* Code overview.
  28
  29   Files are laid out to avoid unnecessary function declarations.  So for
  30   example, every function add_cmd calls occurs before add_cmd in this file.
  31
  32   add_cmd() is called on each line of sed command text (from a file or from
  33   the command line).  It calls get_address() and parse_cmd_args().  The
  34   resulting sed_cmd_t structures are appended to a linked list
  35   (sed_cmd_head/sed_cmd_tail).
  36
  37   process_file() does actual sedding, reading data lines from an input FILE *
  38   (which could be stdin) and applying the sed command list (sed_cmd_head) to
  39   each of the resulting lines.
  40
  41   sed_main() is where external code calls into this, with a command line.
  42 */
  43
  44
  45 /*
  46         Supported features and commands in this version of sed:
  47
  48          - comments ('#')
  49          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  50          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  51          - edit commands: (a)ppend, (i)nsert, (c)hange
  52          - file commands: (r)ead
  53          - backreferences in substitution expressions (\1, \2...\9)
  54          - grouped commands: {cmd1;cmd2}
  55          - transliteration (y/source-chars/dest-chars/)
  56          - pattern space hold space storing / swapping (g, h, x)
  57          - labels / branching (: label, b, t)
  58
  59          (Note: Specifying an address (range) to match is *optional*; commands
  60          default to the whole pattern space if no specific address match was
  61          requested.)
  62
  63         Unsupported features:
  64
  65          - GNU extensions
  66          - and more.
  67
  68         Todo:
  69
  70          - Create a wrapper around regex to make libc's regex conform with sed
  71          - Fix bugs
  72
  73
  74         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  75 */
  76
  77 #include <stdio.h>
  78 #include <unistd.h>             /* for getopt() */
  79 #include <regex.h>
  80 #include <string.h>             /* for strdup() */
  81 #include <errno.h>
  82 #include <ctype.h>              /* for isspace() */
  83 #include <stdlib.h>
  84 #include "busybox.h"
  85
  86 typedef struct sed_cmd_s {
  87     /* Ordered by alignment requirements: currently 36 bytes on x86 */
  88
  89     /* address storage */
  90     regex_t *beg_match; /* sed -e '/match/cmd' */
  91     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
  92     regex_t *sub_match; /* For 's/sub_match/string/' */
  93     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
  94     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  95
  96     FILE *file;                 /* File (sr) command writes to, -1 for none. */
  97     char *string;               /* Data string for (saicytb) commands. */
  98
  99     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
 100
 101     /* Bitfields (gcc won't group them if we don't) */
 102     unsigned int invert:1;                      /* the '!' after the address */
 103     unsigned int in_match:1;            /* Next line also included in match? */
 104     unsigned int no_newline:1;          /* Last line written by (sr) had no '\n' */
 105     unsigned int sub_p:1;                       /* (s) print option */
 106
 107
 108     /* GENERAL FIELDS */
 109     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 110     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
 111 } sed_cmd_t;
 112
 113 /* globals */
 114 /* options */
 115 static int be_quiet = 0, in_place=0;
 116 FILE *nonstdout;
 117 char *outname;
 118
 119
 120 static const char bad_format_in_subst[] =
 121         "bad format in substitution expression";
 122 const char *const semicolon_whitespace = "; \n\r\t\v";
 123
 124 regmatch_t regmatch[10];
 125 static regex_t *previous_regex_ptr = NULL;
 126
 127 /* linked list of sed commands */
 128 static sed_cmd_t sed_cmd_head;
 129 static sed_cmd_t *sed_cmd_tail = &sed_cmd_head;
 130
 131 /* Linked list of append lines */
 132 struct append_list {
 133         char *string;
 134         struct append_list *next;
 135 };
 136 struct append_list *append_head=NULL, *append_tail=NULL;
 137
 138 #ifdef CONFIG_FEATURE_CLEAN_UP
 139 static void free_and_close_stuff(void)
 140 {
 141         sed_cmd_t *sed_cmd = sed_cmd_head.next;
 142
 143         while(append_head) {
 144                 append_tail=append_head->next;
 145                 free(append_head->string);
 146                 free(append_head);
 147                 append_head=append_tail;
 148         }
 149
 150         while (sed_cmd) {
 151                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 152
 153                 if(sed_cmd->file)
 154                         bb_xprint_and_close_file(sed_cmd->file);
 155
 156                 if (sed_cmd->beg_match) {
 157                         regfree(sed_cmd->beg_match);
 158                         free(sed_cmd->beg_match);
 159                 }
 160                 if (sed_cmd->end_match) {
 161                         regfree(sed_cmd->end_match);
 162                         free(sed_cmd->end_match);
 163                 }
 164                 if (sed_cmd->sub_match) {
 165                         regfree(sed_cmd->sub_match);
 166                         free(sed_cmd->sub_match);
 167                 }
 168                 free(sed_cmd->string);
 169                 free(sed_cmd);
 170                 sed_cmd = sed_cmd_next;
 171         }
 172 }
 173 #endif
 174
 175 /* If something bad happens during -i operation, delete temp file */
 176
 177 static void cleanup_outname(void)
 178 {
 179   if(outname) unlink(outname);
 180 }
 181
 182 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 183
 184 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 185 {
 186         int i=0;
 187
 188         while(i<len) {
 189                 if(string[i] == '\\') {
 190                         if(!to || string[i+1] == from) {
 191                                 *(dest++) = to ? to : string[i+1];
 192                                 i+=2;
 193                                 continue;
 194                         } else *(dest++)=string[i++];
 195                 }
 196                 *(dest++) = string[i++];
 197         }
 198         *dest=0;
 199 }
 200
 201 static char *copy_parsing_slashn(const char *string, int len)
 202 {
 203         char *dest=xmalloc(len+1);
 204
 205         parse_escapes(dest,string,len,'n','\n');
 206         return dest;
 207 }
 208
 209
 210 /*
 211  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 212  * beginning at a specified index and returns the index of the next regular
 213  * expression delimiter (typically a forward * slash ('/')) not preceeded by
 214  * a backslash ('\').
 215  */
 216 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 217         const char *str)
 218 {
 219         int bracket = -1;
 220         int escaped = 0;
 221         int idx = 0;
 222         char ch;
 223
 224         for (; (ch = str[idx]); idx++) {
 225                 if (bracket != -1) {
 226                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 227                                         && str[idx - 1] == '^')))
 228                                 bracket = -1;
 229                 } else if (escaped)
 230                         escaped = 0;
 231                 else if (ch == '\\')
 232                         escaped = 1;
 233                 else if (ch == '[')
 234                         bracket = idx;
 235                 else if (ch == delimiter)
 236                         return idx;
 237         }
 238
 239         /* if we make it to here, we've hit the end of the string */
 240         return -1;
 241 }
 242
 243 /*
 244  *  Returns the index of the third delimiter
 245  */
 246 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 247 {
 248         const char *cmdstr_ptr = cmdstr;
 249         char delimiter;
 250         int idx = 0;
 251
 252         /* verify that the 's' or 'y' is followed by something.  That something
 253          * (typically a 'slash') is now our regexp delimiter... */
 254         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
 255         delimiter = *(cmdstr_ptr++);
 256
 257         /* save the match string */
 258         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 259         if (idx == -1) {
 260                 bb_error_msg_and_die(bad_format_in_subst);
 261         }
 262         *match = copy_parsing_slashn(cmdstr_ptr, idx);
 263
 264         /* save the replacement string */
 265         cmdstr_ptr += idx + 1;
 266         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 267         if (idx == -1) {
 268                 bb_error_msg_and_die(bad_format_in_subst);
 269         }
 270         *replace = copy_parsing_slashn(cmdstr_ptr, idx);
 271
 272         return ((cmdstr_ptr - cmdstr) + idx);
 273 }
 274
 275 /*
 276  * returns the index in the string just past where the address ends.
 277  */
 278 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 279 {
 280         char *pos = my_str;
 281
 282         if (isdigit(*my_str)) {
 283                 *linenum = strtol(my_str, &pos, 10);
 284                 /* endstr shouldnt ever equal NULL */
 285         } else if (*my_str == '$') {
 286                 *linenum = -1;
 287                 pos++;
 288         } else if (*my_str == '/' || *my_str == '\\') {
 289                 int next;
 290                 char delimiter;
 291                 char *temp;
 292
 293                 if (*my_str == '\\') delimiter = *(++pos);
 294                 else delimiter = '/';
 295                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 296                 if (next == -1)
 297                         bb_error_msg_and_die("unterminated match expression");
 298
 299                 temp=copy_parsing_slashn(pos,next);
 300                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 301                 xregcomp(*regex, temp, REG_NEWLINE);
 302                 free(temp);
 303                 /* Move position to next character after last delimiter */
 304                 pos+=(next+1);
 305         }
 306         return pos - my_str;
 307 }
 308
 309 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 310 static int parse_file_cmd(sed_cmd_t * sed_cmd, const char *filecmdstr, char **retval)
 311 {
 312         int start = 0, idx, hack=0;
 313
 314         /* Skip whitespace, then grab filename to end of line */
 315         while (isspace(filecmdstr[start])) start++;
 316         idx=start;
 317         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 318         /* If lines glued together, put backslash back. */
 319         if(filecmdstr[idx]=='\n') hack=1;
 320         if(idx==start) bb_error_msg_and_die("Empty filename");
 321         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
 322         if(hack) *(idx+*retval)='\\';
 323
 324         return idx;
 325 }
 326
 327 static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
 328 {
 329         int cflags = 0;
 330         char *match;
 331         int idx = 0;
 332
 333         /*
 334          * A substitution command should look something like this:
 335          *    s/match/replace/ #gIpw
 336          *    ||     |        |||
 337          *    mandatory       optional
 338          */
 339         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 340
 341         /* determine the number of back references in the match string */
 342         /* Note: we compute this here rather than in the do_subst_command()
 343          * function to save processor time, at the expense of a little more memory
 344          * (4 bits) per sed_cmd */
 345
 346         /* process the flags */
 347
 348         sed_cmd->which_match=1;
 349         while (substr[++idx]) {
 350                 /* Parse match number */
 351                 if(isdigit(substr[idx])) {
 352                         if(match[0]!='^') {
 353                                 /* Match 0 treated as all, multiple matches we take the last one. */
 354                                 char *pos=substr+idx;
 355                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 356                                 idx=pos-substr;
 357                         }
 358                         continue;
 359                 }
 360                 /* Skip spaces */
 361                 if(isspace(substr[idx])) continue;
 362
 363                 switch (substr[idx]) {
 364                         /* Replace all occurrences */
 365                         case 'g':
 366                                 if (match[0] != '^') sed_cmd->which_match = 0;
 367                                 break;
 368                         /* Print pattern space */
 369                         case 'p':
 370                                 sed_cmd->sub_p = 1;
 371                                 break;
 372                         case 'w':
 373                         {
 374                                 char *temp;
 375                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 376
 377                                 break;
 378                         }
 379                         /* Ignore case (gnu exension) */
 380                         case 'I':
 381                                 cflags |= REG_ICASE;
 382                                 break;
 383                         case ';':
 384                         case '}':
 385                                 goto out;
 386                         default:
 387                                 bb_error_msg_and_die("bad option in substitution expression");
 388                 }
 389         }
 390 out:
 391         /* compile the match string into a regex */
 392         if (*match != '\0') {
 393                 /* If match is empty, we use last regex used at runtime */
 394                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 395                 xregcomp(sed_cmd->sub_match, match, cflags);
 396         }
 397         free(match);
 398
 399         return idx;
 400 }
 401
 402 /*
 403  *  Process the commands arguments
 404  */
 405 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 406 {
 407         /* handle (s)ubstitution command */
 408         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 409         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 410         else if (strchr("aic", sed_cmd->cmd)) {
 411                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 412                         bb_error_msg_and_die
 413                                 ("only a beginning address can be specified for edit commands");
 414                 for(;;) {
 415                         if(*cmdstr=='\n' || *cmdstr=='\\') {
 416                                 cmdstr++;
 417                                 break;
 418                         } else if(isspace(*cmdstr)) cmdstr++;
 419                         else break;
 420                 }
 421                 sed_cmd->string = bb_xstrdup(cmdstr);
 422                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 423                 cmdstr += strlen(cmdstr);
 424         /* handle file cmds: (r)ead */
 425         } else if(strchr("rw", sed_cmd->cmd)) {
 426                 if (sed_cmd->end_line || sed_cmd->end_match)
 427                         bb_error_msg_and_die("Command only uses one address");
 428                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 429                 if(sed_cmd->cmd=='w')
 430                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
 431         /* handle branch commands */
 432         } else if (strchr(":bt", sed_cmd->cmd)) {
 433                 int length;
 434
 435                 while(isspace(*cmdstr)) cmdstr++;
 436                 length = strcspn(cmdstr, semicolon_whitespace);
 437                 if (length) {
 438                         sed_cmd->string = strndup(cmdstr, length);
 439                         cmdstr += length;
 440                 }
 441         }
 442         /* translation command */
 443         else if (sed_cmd->cmd == 'y') {
 444                 char *match, *replace;
 445                 int i=cmdstr[0];
 446
 447                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 448                 /* \n already parsed, but \delimiter needs unescaping. */
 449                 parse_escapes(match,match,strlen(match),i,i);
 450                 parse_escapes(replace,replace,strlen(replace),i,i);
 451
 452                 sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
 453                 for (i = 0; match[i] && replace[i]; i++) {
 454                         sed_cmd->string[i * 2] = match[i];
 455                         sed_cmd->string[(i * 2) + 1] = replace[i];
 456                 }
 457                 free(match);
 458                 free(replace);
 459         }
 460         /* if it wasnt a single-letter command that takes no arguments
 461          * then it must be an invalid command.
 462          */
 463         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 464                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 465         }
 466
 467         /* give back whatever's left over */
 468         return (cmdstr);
 469 }
 470
 471
 472 /* Parse address+command sets, skipping comment lines. */
 473
 474 void add_cmd(char *cmdstr)
 475 {
 476         static char *add_cmd_line=NULL;
 477         sed_cmd_t *sed_cmd;
 478         int temp;
 479
 480         /* Append this line to any unfinished line from last time. */
 481         if(add_cmd_line) {
 482                 int lastlen=strlen(add_cmd_line);
 483                 char *tmp=xmalloc(lastlen+strlen(cmdstr)+2);
 484
 485                 memcpy(tmp,add_cmd_line,lastlen);
 486                 tmp[lastlen]='\n';
 487                 strcpy(tmp+lastlen+1,cmdstr);
 488                 free(add_cmd_line);
 489                 cmdstr=add_cmd_line=tmp;
 490         } else add_cmd_line=NULL;
 491
 492         /* If this line ends with backslash, request next line. */
 493         temp=strlen(cmdstr);
 494         if(temp && cmdstr[temp-1]=='\\') {
 495                 if(!add_cmd_line) add_cmd_line=strdup(cmdstr);
 496                 add_cmd_line[temp-1]=0;
 497                 return;
 498         }
 499
 500         /* Loop parsing all commands in this line. */
 501         while(*cmdstr) {
 502                 /* Skip leading whitespace and semicolons */
 503                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 504
 505                 /* If no more commands, exit. */
 506                 if(!*cmdstr) break;
 507
 508                 /* if this is a comment, jump past it and keep going */
 509                 if (*cmdstr == '#') {
 510                         /* "#n" is the same as using -n on the command line */
 511                         if (cmdstr[1] == 'n') be_quiet++;
 512                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 513                         continue;
 514                 }
 515
 516                 /* parse the command
 517                  * format is: [addr][,addr][!]cmd
 518                  *            |----||-----||-|
 519                  *            part1 part2  part3
 520                  */
 521
 522                 sed_cmd = xcalloc(1, sizeof(sed_cmd_t));
 523
 524                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 525                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 526
 527                 /* second part (if present) will begin with a comma */
 528                 if (*cmdstr == ',') {
 529                         int idx;
 530
 531                         cmdstr++;
 532                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 533                         if (!idx) bb_error_msg_and_die("get_address: no address found in string\n");
 534                         cmdstr += idx;
 535                 }
 536
 537                 /* skip whitespace before the command */
 538                 while (isspace(*cmdstr)) cmdstr++;
 539
 540                 /* Check for inversion flag */
 541                 if (*cmdstr == '!') {
 542                         sed_cmd->invert = 1;
 543                         cmdstr++;
 544
 545                         /* skip whitespace before the command */
 546                         while (isspace(*cmdstr)) cmdstr++;
 547                 }
 548
 549                 /* last part (mandatory) will be a command */
 550                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 551                 sed_cmd->cmd = *(cmdstr++);
 552                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 553
 554                 /* Add the command to the command array */
 555                 sed_cmd_tail->next = sed_cmd;
 556                 sed_cmd_tail = sed_cmd_tail->next;
 557         }
 558
 559         /* If we glued multiple lines together, free the memory. */
 560         if(add_cmd_line) {
 561                 free(add_cmd_line);
 562                 add_cmd_line=NULL;
 563         }
 564 }
 565
 566 struct pipeline {
 567         char *buf;      /* Space to hold string */
 568         int idx;        /* Space used */
 569         int len;        /* Space allocated */
 570 } pipeline;
 571
 572 #define PIPE_GROW 64
 573
 574 void pipe_putc(char c)
 575 {
 576         if(pipeline.idx==pipeline.len) {
 577                 pipeline.buf = xrealloc(pipeline.buf, pipeline.len + PIPE_GROW);
 578                 pipeline.len+=PIPE_GROW;
 579         }
 580         pipeline.buf[pipeline.idx++] = (c);
 581 }
 582
 583 static void do_subst_w_backrefs(const char *line, const char *replace)
 584 {
 585         int i,j;
 586
 587         /* go through the replacement string */
 588         for (i = 0; replace[i]; i++) {
 589                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 590                 if (replace[i] == '\\' && replace[i+1]>'0' && replace[i+1]<='9') {
 591                         int backref=replace[++i]-'0';
 592
 593                         /* print out the text held in regmatch[backref] */
 594                         if(regmatch[backref].rm_so != -1)
 595                                 for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
 596                                         pipe_putc(line[j]);
 597                 }
 598
 599                 /* if we find a backslash escaped character, print the character */
 600                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 601
 602                 /* if we find an unescaped '&' print out the whole matched text. */
 603                 else if (replace[i] == '&')
 604                         for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
 605                                 pipe_putc(line[j]);
 606                 /* Otherwise just output the character. */
 607                 else pipe_putc(replace[i]);
 608         }
 609 }
 610
 611 static int do_subst_command(sed_cmd_t * sed_cmd, char **line)
 612 {
 613         char *oldline = *line;
 614         int altered = 0;
 615         int match_count=0;
 616         regex_t *current_regex;
 617
 618         /* Handle empty regex. */
 619         if (sed_cmd->sub_match == NULL) {
 620                 current_regex = previous_regex_ptr;
 621                 if(!current_regex)
 622                         bb_error_msg_and_die("No previous regexp.");
 623         } else previous_regex_ptr = current_regex = sed_cmd->sub_match;
 624
 625         /* Find the first match */
 626         if(REG_NOMATCH==regexec(current_regex, oldline, 10, regmatch, 0))
 627                 return 0;
 628
 629         /* Initialize temporary output buffer. */
 630         pipeline.buf=xmalloc(PIPE_GROW);
 631         pipeline.len=PIPE_GROW;
 632         pipeline.idx=0;
 633
 634         /* Now loop through, substituting for matches */
 635         do {
 636                 int i;
 637
 638                 /* Work around bug in glibc regexec, demonstrated by:
 639                    echo " a.b" | busybox sed 's [^ .]* x g'
 640                    The match_count check is so not to break
 641                    echo "hi" | busybox sed 's/^/!/g' */
 642                 if(!regmatch[0].rm_so && !regmatch[0].rm_eo && match_count) {
 643                         pipe_putc(*(oldline++));
 644                         continue;
 645                 }
 646
 647                 match_count++;
 648
 649                 /* If we aren't interested in this match, output old line to
 650                    end of match and continue */
 651                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 652                         for(i=0;i<regmatch[0].rm_eo;i++)
 653                                 pipe_putc(oldline[i]);
 654                         continue;
 655                 }
 656
 657                 /* print everything before the match */
 658                 for (i = 0; i < regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 659
 660                 /* then print the substitution string */
 661                 do_subst_w_backrefs(oldline, sed_cmd->string);
 662
 663                 /* advance past the match */
 664                 oldline += regmatch[0].rm_eo;
 665                 /* flag that something has changed */
 666                 altered++;
 667
 668                 /* if we're not doing this globally, get out now */
 669                 if (sed_cmd->which_match) break;
 670         } while (*oldline && (regexec(current_regex, oldline, 10, regmatch, 0) != REG_NOMATCH));
 671
 672         /* Copy rest of string into output pipeline */
 673
 674         while(*oldline) pipe_putc(*(oldline++));
 675         pipe_putc(0);
 676
 677         free(*line);
 678         *line = pipeline.buf;
 679         return altered;
 680 }
 681
 682 /* Set command pointer to point to this label.  (Does not handle null label.) */
 683 static sed_cmd_t *branch_to(const char *label)
 684 {
 685         sed_cmd_t *sed_cmd;
 686
 687         for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 688                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 689                         return (sed_cmd);
 690                 }
 691         }
 692         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 693 }
 694
 695 /* Append copy of string to append buffer */
 696 static void append(char *s)
 697 {
 698         struct append_list *temp=calloc(1,sizeof(struct append_list));
 699
 700         if(append_head)
 701                 append_tail=(append_tail->next=temp);
 702         else append_head=append_tail=temp;
 703         temp->string=strdup(s);
 704 }
 705
 706 static void flush_append(void)
 707 {
 708         /* Output appended lines. */
 709         while(append_head) {
 710                 fprintf(nonstdout,"%s\n",append_head->string);
 711                 append_tail=append_head->next;
 712                 free(append_head->string);
 713                 free(append_head);
 714                 append_head=append_tail;
 715         }
 716         append_head=append_tail=NULL;
 717 }
 718
 719 /* Get next line of input, flushing append buffer and noting if we hit EOF
 720  * without a newline on the last line.
 721  */
 722 static char *get_next_line(FILE * file, int *no_newline)
 723 {
 724         char *temp;
 725         int len;
 726
 727         flush_append();
 728         temp=bb_get_line_from_file(file);
 729         if(temp) {
 730                 len=strlen(temp);
 731                 if(len && temp[len-1]=='\n') temp[len-1]=0;
 732                 else *no_newline=1;
 733         }
 734
 735         return temp;
 736 }
 737
 738 /* Output line of text.  missing_newline means the last line output did not
 739    end with a newline.  no_newline means this line does not end with a
 740    newline. */
 741
 742 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 743 {
 744         if(missing_newline) fputc('\n',file);
 745         fputs(s,file);
 746         if(!no_newline) fputc('\n',file);
 747
 748     if(ferror(file)) {
 749                 fprintf(stderr,"Write failed.\n");
 750                 exit(4);  /* It's what gnu sed exits with... */
 751         }
 752
 753         return no_newline;
 754 }
 755
 756 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,nonstdout,missing_newline,n)
 757
 758 static void process_file(FILE *file)
 759 {
 760         char *pattern_space, *next_line, *hold_space=NULL;
 761         static int linenum = 0, missing_newline=0;
 762         int no_newline,next_no_newline=0;
 763
 764         next_line = get_next_line(file,&next_no_newline);
 765
 766         /* go through every line in the file */
 767         for(;;) {
 768                 sed_cmd_t *sed_cmd;
 769                 int substituted=0;
 770
 771                 /* Advance to next line.  Stop if out of lines. */
 772                 if(!(pattern_space=next_line)) break;
 773                 no_newline=next_no_newline;
 774
 775                 /* Read one line in advance so we can act on the last line, the '$' address */
 776                 next_line = get_next_line(file,&next_no_newline);
 777                 linenum++;
 778 restart:
 779                 /* for every line, go through all the commands */
 780                 for (sed_cmd = sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 781                         int old_matched, matched;
 782
 783                         old_matched = sed_cmd->in_match;
 784
 785                         /* Determine if this command matches this line: */
 786
 787                         /* Are we continuing a previous multi-line match? */
 788
 789                         sed_cmd->in_match = sed_cmd->in_match
 790
 791                         /* Or is no range necessary? */
 792                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 793                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 794
 795                         /* Or did we match the start of a numerical range? */
 796                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 797
 798                         /* Or does this line match our begin address regex? */
 799                                 || (sed_cmd->beg_match &&
 800                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 801
 802                         /* Or did we match last line of input? */
 803                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 804
 805                         /* Snapshot the value */
 806
 807                         matched = sed_cmd->in_match;
 808
 809                         /* Is this line the end of the current match? */
 810
 811                         if(matched) {
 812                                 sed_cmd->in_match = !(
 813                                         /* has the ending line come, or is this a single address command? */
 814                                         (sed_cmd->end_line ?
 815                                                 sed_cmd->end_line==-1 ?
 816                                                         !next_line
 817                                                         : sed_cmd->end_line<=linenum
 818                                                 : !sed_cmd->end_match)
 819                                         /* or does this line matches our last address regex */
 820                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 821                                 );
 822                         }
 823
 824                         /* Skip blocks of commands we didn't match. */
 825                         if (sed_cmd->cmd == '{') {
 826                                 if(sed_cmd->invert ? matched : !matched)
 827                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 828                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 829                                 continue;
 830                         }
 831
 832                         /* Okay, so did this line match? */
 833                         if (sed_cmd->invert ? !matched : matched) {
 834                                 /* Update last used regex in case a blank substitute BRE is found */
 835                                 if (sed_cmd->beg_match) {
 836                                         previous_regex_ptr = sed_cmd->beg_match;
 837                                 }
 838
 839                                 /* actual sedding */
 840                                 switch (sed_cmd->cmd) {
 841
 842                                         /* Print line number */
 843                                         case '=':
 844                                                 fprintf(nonstdout,"%d\n", linenum);
 845                                                 break;
 846
 847                                         /* Write the current pattern space up to the first newline */
 848                                         case 'P':
 849                                         {
 850                                                 char *tmp = strchr(pattern_space, '\n');
 851
 852                                                 if (tmp) {
 853                                                         *tmp = '\0';
 854                                                         sed_puts(pattern_space,1);
 855                                                         *tmp = '\n';
 856                                                         break;
 857                                                 }
 858                                                 /* Fall Through */
 859                                         }
 860
 861                                         /* Write the current pattern space to output */
 862                                         case 'p':
 863                                                 sed_puts(pattern_space,no_newline);
 864                                                 break;
 865                                         /* Delete up through first newline */
 866                                         case 'D':
 867                                         {
 868                                                 char *tmp = strchr(pattern_space,'\n');
 869
 870                                                 if(tmp) {
 871                                                         tmp=bb_xstrdup(tmp+1);
 872                                                         free(pattern_space);
 873                                                         pattern_space=tmp;
 874                                                         goto restart;
 875                                                 }
 876                                         }
 877                                         /* discard this line. */
 878                                         case 'd':
 879                                                 goto discard_line;
 880
 881                                         /* Substitute with regex */
 882                                         case 's':
 883                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 884                                                         substituted|=1;
 885
 886                                                         /* handle p option */
 887                                                         if(sed_cmd->sub_p)
 888                                                                 sed_puts(pattern_space,no_newline);
 889                                                         /* handle w option */
 890                                                         if(sed_cmd->file)
 891                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 892
 893                                                 }
 894                                                 break;
 895
 896                                         /* Append line to linked list to be printed later */
 897                                         case 'a':
 898                                         {
 899                                                 append(sed_cmd->string);
 900                                                 break;
 901                                         }
 902
 903                                         /* Insert text before this line */
 904                                         case 'i':
 905                                                 sed_puts(sed_cmd->string,1);
 906                                                 break;
 907
 908                                         /* Cut and paste text (replace) */
 909                                         case 'c':
 910                                                 /* Only triggers on last line of a matching range. */
 911                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,1);
 912                                                 goto discard_line;
 913
 914                                         /* Read file, append contents to output */
 915                                         case 'r':
 916                                         {
 917                                                 FILE *outfile;
 918
 919                                                 outfile = fopen(sed_cmd->string, "r");
 920                                                 if (outfile) {
 921                                                         char *line;
 922
 923                                                         while ((line = bb_get_chomped_line_from_file(outfile))
 924                                                                         != NULL)
 925                                                                 append(line);
 926                                                         bb_xprint_and_close_file(outfile);
 927                                                 }
 928
 929                                                 break;
 930                                         }
 931
 932                                         /* Write pattern space to file. */
 933                                         case 'w':
 934                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 935                                                 break;
 936
 937                                         /* Read next line from input */
 938                                         case 'n':
 939                                                 if (!be_quiet)
 940                                                         sed_puts(pattern_space,no_newline);
 941                                                 if (next_line) {
 942                                                         free(pattern_space);
 943                                                         pattern_space = next_line;
 944                                                         no_newline=next_no_newline;
 945                                                         next_line = get_next_line(file,&next_no_newline);
 946                                                         linenum++;
 947                                                         break;
 948                                                 }
 949                                                 /* fall through */
 950
 951                                         /* Quit.  End of script, end of input. */
 952                                         case 'q':
 953                                                 /* Exit the outer while loop */
 954                                                 free(next_line);
 955                                                 next_line = NULL;
 956                                                 goto discard_commands;
 957
 958                                         /* Append the next line to the current line */
 959                                         case 'N':
 960                                         {
 961                                                 /* If no next line, jump to end of script and exit. */
 962                                                 if (next_line == NULL) {
 963                                                         /* Jump to end of script and exit */
 964                                                         free(next_line);
 965                                                         next_line = NULL;
 966                                                         goto discard_line;
 967                                                 /* append next_line, read new next_line. */
 968                                                 } else {
 969                                                         int len=strlen(pattern_space);
 970
 971                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 972                                                         pattern_space[len]='\n';
 973                                                         strcpy(pattern_space+len+1, next_line);
 974                                                         no_newline=next_no_newline;
 975                                                         next_line = get_next_line(file,&next_no_newline);
 976                                                         linenum++;
 977                                                 }
 978                                                 break;
 979                                         }
 980
 981                                         /* Test if substition worked, branch if so. */
 982                                         case 't':
 983                                                 if (!substituted) break;
 984                                                 substituted=0;
 985                                                         /* Fall through */
 986                                         /* Branch to label */
 987                                         case 'b':
 988                                                 if (!sed_cmd->string) goto discard_commands;
 989                                                 else sed_cmd = branch_to(sed_cmd->string);
 990                                                 break;
 991                                         /* Transliterate characters */
 992                                         case 'y':
 993                                         {
 994                                                 int i;
 995
 996                                                 for (i = 0; pattern_space[i]; i++) {
 997                                                         int j;
 998
 999                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1000                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1001                                                                         pattern_space[i] = sed_cmd->string[j + 1];
1002                                                                 }
1003                                                         }
1004                                                 }
1005
1006                                                 break;
1007                                         }
1008                                         case 'g':       /* Replace pattern space with hold space */
1009                                                 free(pattern_space);
1010                                                 if (hold_space) {
1011                                                         pattern_space = strdup(hold_space);
1012                                                         no_newline=0;
1013                                                 }
1014                                                 break;
1015                                         case 'G':       /* Append newline and hold space to pattern space */
1016                                         {
1017                                                 int pattern_space_size = 2;
1018                                                 int hold_space_size = 0;
1019
1020                                                 if (pattern_space)
1021                                                         pattern_space_size += strlen(pattern_space);
1022                                                 if (hold_space) hold_space_size = strlen(hold_space);
1023                                                 pattern_space = xrealloc(pattern_space, pattern_space_size + hold_space_size);
1024                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1025                                                 strcat(pattern_space, "\n");
1026                                                 if (hold_space) strcat(pattern_space, hold_space);
1027                                                 no_newline=0;
1028
1029                                                 break;
1030                                         }
1031                                         case 'h':       /* Replace hold space with pattern space */
1032                                                 free(hold_space);
1033                                                 hold_space = strdup(pattern_space);
1034                                                 break;
1035                                         case 'H':       /* Append newline and pattern space to hold space */
1036                                         {
1037                                                 int hold_space_size = 2;
1038                                                 int pattern_space_size = 0;
1039
1040                                                 if (hold_space) hold_space_size += strlen(hold_space);
1041                                                 if (pattern_space)
1042                                                         pattern_space_size = strlen(pattern_space);
1043                                                 hold_space = xrealloc(hold_space,
1044                                                                                         hold_space_size + pattern_space_size);
1045
1046                                                 if (hold_space_size == 2) hold_space[0]=0;
1047                                                 strcat(hold_space, "\n");
1048                                                 if (pattern_space) strcat(hold_space, pattern_space);
1049
1050                                                 break;
1051                                         }
1052                                         case 'x': /* Exchange hold and pattern space */
1053                                         {
1054                                                 char *tmp = pattern_space;
1055                                                 pattern_space = hold_space;
1056                                                 no_newline=0;
1057                                                 hold_space = tmp;
1058                                                 break;
1059                                         }
1060                                 }
1061                         }
1062                 }
1063
1064                 /*
1065                  * exit point from sedding...
1066                  */
1067 discard_commands:
1068                 /* we will print the line unless we were told to be quiet ('-n')
1069                    or if the line was suppressed (ala 'd'elete) */
1070                 if (!be_quiet) sed_puts(pattern_space,no_newline);
1071
1072                 /* Delete and such jump here. */
1073 discard_line:
1074                 flush_append();
1075                 free(pattern_space);
1076         }
1077 }
1078
1079 /* It is possible to have a command line argument with embedded
1080    newlines.  This counts as multiple command lines. */
1081
1082 static void add_cmd_block(char *cmdstr)
1083 {
1084         int go=1;
1085         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1086
1087         while(go) {
1088                 int len=strcspn(temp2,"\n");
1089                 if(!temp2[len]) go=0;
1090                 else temp2[len]=0;
1091                 add_cmd(temp2);
1092                 temp2+=len+1;
1093         }
1094         free(temp);
1095 }
1096
1097 extern int sed_main(int argc, char **argv)
1098 {
1099         int opt, status = EXIT_SUCCESS;
1100
1101 #ifdef CONFIG_FEATURE_CLEAN_UP
1102         /* destroy command strings on exit */
1103         if (atexit(free_and_close_stuff) == -1)
1104                 bb_perror_msg_and_die("atexit");
1105 #endif
1106
1107 #define LIE_TO_AUTOCONF
1108 #ifdef LIE_TO_AUTOCONF
1109         if(argc==2 && !strcmp(argv[1],"--version")) {
1110                 printf("This is not GNU sed version 4.0\n");
1111                 exit(0);
1112         }
1113 #endif
1114
1115         /* do normal option parsing */
1116         while ((opt = getopt(argc, argv, "ine:f:")) > 0) {
1117                 switch (opt) {
1118                 case 'i':
1119                         in_place++;
1120                         atexit(cleanup_outname);
1121                         break;
1122                 case 'n':
1123                         be_quiet++;
1124                         break;
1125                 case 'e':
1126                         add_cmd_block(optarg);
1127                         break;
1128                 case 'f':
1129                 {
1130                         FILE *cmdfile;
1131                         char *line;
1132
1133                         cmdfile = bb_xfopen(optarg, "r");
1134
1135                         while ((line = bb_get_chomped_line_from_file(cmdfile))
1136                                  != NULL) {
1137                                 add_cmd(line);
1138                                 free(line);
1139                         }
1140                         bb_xprint_and_close_file(cmdfile);
1141
1142                         break;
1143                 }
1144                 default:
1145                         bb_show_usage();
1146                 }
1147         }
1148
1149         /* if we didn't get a pattern from a -e and no command file was specified,
1150          * argv[optind] should be the pattern. no pattern, no worky */
1151         if (sed_cmd_head.next == NULL) {
1152                 if (argv[optind] == NULL)
1153                         bb_show_usage();
1154                 else
1155                         add_cmd_block(argv[optind++]);
1156         }
1157         /* Flush any unfinished commands. */
1158         add_cmd("");
1159
1160         /* By default, we write to stdout */
1161         nonstdout=stdout;
1162
1163         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1164          * files were specified or '-' was specified, take input from stdin.
1165          * Otherwise, we process all the files specified. */
1166         if (argv[optind] == NULL) {
1167                 if(in_place) {
1168                         fprintf(stderr,"sed: Filename required for -i\n");
1169                         exit(1);
1170                 }
1171                 process_file(stdin);
1172         } else {
1173                 int i;
1174                 FILE *file;
1175
1176                 for (i = optind; i < argc; i++) {
1177                         if(!strcmp(argv[i], "-") && !in_place) {
1178                                 process_file(stdin);
1179                         } else {
1180                                 file = bb_wfopen(argv[i], "r");
1181                                 if (file) {
1182                                         if(in_place) {
1183                                                 struct stat statbuf;
1184                                                 outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1185                                                 strcat(outname,"XXXXXX");
1186                                                 /* Set permissions of output file */
1187                                                 fstat(fileno(file),&statbuf);
1188                                                 mkstemp(outname);
1189                                                 nonstdout=bb_wfopen(outname,"w");
1190                                                 /* Set permissions of output file */
1191                                                 fstat(fileno(file),&statbuf);
1192                                                 fchmod(fileno(file),statbuf.st_mode);
1193                                                 atexit(cleanup_outname);
1194                                         }
1195                                         process_file(file);
1196                                         fclose(file);
1197                                         if(in_place) {
1198                                                 fclose(nonstdout);
1199                                                 nonstdout=stdout;
1200                                                 unlink(argv[i]);
1201                                                 rename(outname,argv[i]);
1202                                                 free(outname);
1203                                                 outname=0;
1204                                         }
1205                                 } else {
1206                                         status = EXIT_FAILURE;
1207                                 }
1208                         }
1209                 }
1210         }
1211
1212         return status;
1213 }