editors/sed.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * sed.c - very minimalist version of sed
   4  *
   5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   7  * Copyright (C) 2002  Matt Kraai
   8  * Copyright (C) 2003 by Glenn McGrath <bug1@iinet.net.au>
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * MAINTAINER: Rob Landley <rob@landley.net>
  12  *
  13  * Licensed under GPL version 2, see file LICENSE in this tarball for details.
  14  */
  15
  16 /* Code overview.
  17
  18   Files are laid out to avoid unnecessary function declarations.  So for
  19   example, every function add_cmd calls occurs before add_cmd in this file.
  20
  21   add_cmd() is called on each line of sed command text (from a file or from
  22   the command line).  It calls get_address() and parse_cmd_args().  The
  23   resulting sed_cmd_t structures are appended to a linked list
  24   (bbg.sed_cmd_head/bbg.sed_cmd_tail).
  25
  26   add_input_file() adds a FILE * to the list of input files.  We need to
  27   know all input sources ahead of time to find the last line for the $ match.
  28
  29   process_files() does actual sedding, reading data lines from each input FILE *
  30   (which could be stdin) and applying the sed command list (sed_cmd_head) to
  31   each of the resulting lines.
  32
  33   sed_main() is where external code calls into this, with a command line.
  34 */
  35
  36
  37 /*
  38         Supported features and commands in this version of sed:
  39
  40          - comments ('#')
  41          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  42          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  43          - edit commands: (a)ppend, (i)nsert, (c)hange
  44          - file commands: (r)ead
  45          - backreferences in substitution expressions (\0, \1, \2...\9)
  46          - grouped commands: {cmd1;cmd2}
  47          - transliteration (y/source-chars/dest-chars/)
  48          - pattern space hold space storing / swapping (g, h, x)
  49          - labels / branching (: label, b, t, T)
  50
  51          (Note: Specifying an address (range) to match is *optional*; commands
  52          default to the whole pattern space if no specific address match was
  53          requested.)
  54
  55         Todo:
  56          - Create a wrapper around regex to make libc's regex conform with sed
  57
  58         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  59 */
  60
  61 #include "busybox.h"
  62 #include "xregex.h"
  63
  64 /* Each sed command turns into one of these structures. */
  65 typedef struct sed_cmd_s {
  66     /* Ordered by alignment requirements: currently 36 bytes on x86 */
  67
  68     /* address storage */
  69     regex_t *beg_match; /* sed -e '/match/cmd' */
  70     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
  71     regex_t *sub_match; /* For 's/sub_match/string/' */
  72     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
  73     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  74
  75     FILE *file;                 /* File (sw) command writes to, -1 for none. */
  76     char *string;               /* Data string for (saicytb) commands. */
  77
  78     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
  79
  80     /* Bitfields (gcc won't group them if we don't) */
  81     unsigned int invert:1;                      /* the '!' after the address */
  82     unsigned int in_match:1;            /* Next line also included in match? */
  83     unsigned int no_newline:1;          /* Last line written by (sw) had no '\n' */
  84     unsigned int sub_p:1;                       /* (s) print option */
  85
  86     /* GENERAL FIELDS */
  87     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
  88     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
  89 } sed_cmd_t;
  90
  91 static const char *const semicolon_whitespace = "; \n\r\t\v";
  92
  93 struct sed_globals
  94 {
  95         /* options */
  96         int be_quiet, in_place, regex_type;
  97         FILE *nonstdout;
  98         char *outname, *hold_space;
  99
 100         /* List of input files */
 101         int input_file_count,current_input_file;
 102         FILE **input_file_list;
 103
 104         regmatch_t regmatch[10];
 105         regex_t *previous_regex_ptr;
 106
 107         /* linked list of sed commands */
 108         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 109
 110         /* Linked list of append lines */
 111         llist_t *append_head;
 112
 113         char *add_cmd_line;
 114
 115         struct pipeline {
 116                 char *buf;      /* Space to hold string */
 117                 int idx;        /* Space used */
 118                 int len;        /* Space allocated */
 119         } pipeline;
 120 } bbg;
 121
 122
 123 void sed_free_and_close_stuff(void);
 124 #if ENABLE_FEATURE_CLEAN_UP
 125 void sed_free_and_close_stuff(void)
 126 {
 127         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 128
 129         llist_free(bbg.append_head, free);
 130
 131         while (sed_cmd) {
 132                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 133
 134                 if(sed_cmd->file)
 135                         xprint_and_close_file(sed_cmd->file);
 136
 137                 if (sed_cmd->beg_match) {
 138                         regfree(sed_cmd->beg_match);
 139                         free(sed_cmd->beg_match);
 140                 }
 141                 if (sed_cmd->end_match) {
 142                         regfree(sed_cmd->end_match);
 143                         free(sed_cmd->end_match);
 144                 }
 145                 if (sed_cmd->sub_match) {
 146                         regfree(sed_cmd->sub_match);
 147                         free(sed_cmd->sub_match);
 148                 }
 149                 free(sed_cmd->string);
 150                 free(sed_cmd);
 151                 sed_cmd = sed_cmd_next;
 152         }
 153
 154         if(bbg.hold_space) free(bbg.hold_space);
 155
 156     while(bbg.current_input_file<bbg.input_file_count)
 157                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 158 }
 159 #endif
 160
 161 /* If something bad happens during -i operation, delete temp file */
 162
 163 static void cleanup_outname(void)
 164 {
 165   if(bbg.outname) unlink(bbg.outname);
 166 }
 167
 168 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 169
 170 static void parse_escapes(char *dest, char *string, int len, char from, char to)
 171 {
 172         int i=0;
 173
 174         while(i<len) {
 175                 if(string[i] == '\\') {
 176                         if(!to || string[i+1] == from) {
 177                                 *(dest++) = to ? to : string[i+1];
 178                                 i+=2;
 179                                 continue;
 180                         } else *(dest++)=string[i++];
 181                 }
 182                 *(dest++) = string[i++];
 183         }
 184         *dest=0;
 185 }
 186
 187 static char *copy_parsing_escapes(char *string, int len)
 188 {
 189         char *dest=xmalloc(len+1);
 190
 191         parse_escapes(dest,string,len,'n','\n');
 192         return dest;
 193 }
 194
 195
 196 /*
 197  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 198  * beginning at a specified index and returns the index of the next regular
 199  * expression delimiter (typically a forward * slash ('/')) not preceded by
 200  * a backslash ('\').  A negative delimiter disables square bracket checking.
 201  */
 202 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
 203 {
 204         int bracket = -1;
 205         int escaped = 0;
 206         int idx = 0;
 207         char ch;
 208
 209         if (delimiter < 0) {
 210                 bracket--;
 211                 delimiter *= -1;
 212         }
 213
 214         for (; (ch = str[idx]); idx++) {
 215                 if (bracket >= 0) {
 216                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 217                                         && str[idx - 1] == '^')))
 218                                 bracket = -1;
 219                 } else if (escaped)
 220                         escaped = 0;
 221                 else if (ch == '\\')
 222                         escaped = 1;
 223                 else if (bracket == -1 && ch == '[')
 224                         bracket = idx;
 225                 else if (ch == delimiter)
 226                         return idx;
 227         }
 228
 229         /* if we make it to here, we've hit the end of the string */
 230         bb_error_msg_and_die("unmatched '%c'",delimiter);
 231 }
 232
 233 /*
 234  *  Returns the index of the third delimiter
 235  */
 236 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
 237 {
 238         char *cmdstr_ptr = cmdstr;
 239         char delimiter;
 240         int idx = 0;
 241
 242         /* verify that the 's' or 'y' is followed by something.  That something
 243          * (typically a 'slash') is now our regexp delimiter... */
 244         if (*cmdstr == '\0')
 245                 bb_error_msg_and_die("bad format in substitution expression");
 246         delimiter = *(cmdstr_ptr++);
 247
 248         /* save the match string */
 249         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 250         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 251
 252         /* save the replacement string */
 253         cmdstr_ptr += idx + 1;
 254         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 255         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 256
 257         return ((cmdstr_ptr - cmdstr) + idx);
 258 }
 259
 260 /*
 261  * returns the index in the string just past where the address ends.
 262  */
 263 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 264 {
 265         char *pos = my_str;
 266
 267         if (isdigit(*my_str)) {
 268                 *linenum = strtol(my_str, &pos, 10);
 269                 /* endstr shouldnt ever equal NULL */
 270         } else if (*my_str == '$') {
 271                 *linenum = -1;
 272                 pos++;
 273         } else if (*my_str == '/' || *my_str == '\\') {
 274                 int next;
 275                 char delimiter;
 276                 char *temp;
 277
 278                 if (*my_str == '\\') delimiter = *(++pos);
 279                 else delimiter = '/';
 280                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 281                 temp = copy_parsing_escapes(pos,next);
 282                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 283                 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
 284                 free(temp);
 285                 /* Move position to next character after last delimiter */
 286                 pos += (next+1);
 287         }
 288         return pos - my_str;
 289 }
 290
 291 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 292 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
 293 {
 294         int start = 0, idx, hack=0;
 295
 296         /* Skip whitespace, then grab filename to end of line */
 297         while (isspace(filecmdstr[start])) start++;
 298         idx=start;
 299         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 300         /* If lines glued together, put backslash back. */
 301         if(filecmdstr[idx]=='\n') hack=1;
 302         if(idx==start) bb_error_msg_and_die("Empty filename");
 303         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 304         if(hack) *(idx+*retval)='\\';
 305
 306         return idx;
 307 }
 308
 309 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
 310 {
 311         int cflags = bbg.regex_type;
 312         char *match;
 313         int idx = 0;
 314
 315         /*
 316          * A substitution command should look something like this:
 317          *    s/match/replace/ #gIpw
 318          *    ||     |        |||
 319          *    mandatory       optional
 320          */
 321         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 322
 323         /* determine the number of back references in the match string */
 324         /* Note: we compute this here rather than in the do_subst_command()
 325          * function to save processor time, at the expense of a little more memory
 326          * (4 bits) per sed_cmd */
 327
 328         /* process the flags */
 329
 330         sed_cmd->which_match=1;
 331         while (substr[++idx]) {
 332                 /* Parse match number */
 333                 if(isdigit(substr[idx])) {
 334                         if(match[0]!='^') {
 335                                 /* Match 0 treated as all, multiple matches we take the last one. */
 336                                 char *pos=substr+idx;
 337                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 338                                 idx=pos-substr;
 339                         }
 340                         continue;
 341                 }
 342                 /* Skip spaces */
 343                 if(isspace(substr[idx])) continue;
 344
 345                 switch (substr[idx]) {
 346                         /* Replace all occurrences */
 347                         case 'g':
 348                                 if (match[0] != '^') sed_cmd->which_match = 0;
 349                                 break;
 350                         /* Print pattern space */
 351                         case 'p':
 352                                 sed_cmd->sub_p = 1;
 353                                 break;
 354                         /* Write to file */
 355                         case 'w':
 356                         {
 357                                 char *temp;
 358                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 359
 360                                 break;
 361                         }
 362                         /* Ignore case (gnu exension) */
 363                         case 'I':
 364                                 cflags |= REG_ICASE;
 365                                 break;
 366                         /* Comment */
 367                         case '#':
 368                                 while(substr[++idx]);
 369                                 /* Fall through */
 370                         /* End of command */
 371                         case ';':
 372                         case '}':
 373                                 goto out;
 374                         default:
 375                                 bb_error_msg_and_die("bad option in substitution expression");
 376                 }
 377         }
 378 out:
 379         /* compile the match string into a regex */
 380         if (*match != '\0') {
 381                 /* If match is empty, we use last regex used at runtime */
 382                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 383                 xregcomp(sed_cmd->sub_match, match, cflags);
 384         }
 385         free(match);
 386
 387         return idx;
 388 }
 389
 390 /*
 391  *  Process the commands arguments
 392  */
 393 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 394 {
 395         /* handle (s)ubstitution command */
 396         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 397         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 398         else if (strchr("aic", sed_cmd->cmd)) {
 399                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 400                         bb_error_msg_and_die
 401                                 ("only a beginning address can be specified for edit commands");
 402                 for(;;) {
 403                         if(*cmdstr=='\n' || *cmdstr=='\\') {
 404                                 cmdstr++;
 405                                 break;
 406                         } else if(isspace(*cmdstr)) cmdstr++;
 407                         else break;
 408                 }
 409                 sed_cmd->string = xstrdup(cmdstr);
 410                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 411                 cmdstr += strlen(cmdstr);
 412         /* handle file cmds: (r)ead */
 413         } else if(strchr("rw", sed_cmd->cmd)) {
 414                 if (sed_cmd->end_line || sed_cmd->end_match)
 415                         bb_error_msg_and_die("Command only uses one address");
 416                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 417                 if(sed_cmd->cmd=='w')
 418                         sed_cmd->file=xfopen(sed_cmd->string,"w");
 419         /* handle branch commands */
 420         } else if (strchr(":btT", sed_cmd->cmd)) {
 421                 int length;
 422
 423                 while(isspace(*cmdstr)) cmdstr++;
 424                 length = strcspn(cmdstr, semicolon_whitespace);
 425                 if (length) {
 426                         sed_cmd->string = xstrndup(cmdstr, length);
 427                         cmdstr += length;
 428                 }
 429         }
 430         /* translation command */
 431         else if (sed_cmd->cmd == 'y') {
 432                 char *match, *replace;
 433                 int i=cmdstr[0];
 434
 435                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 436                 /* \n already parsed, but \delimiter needs unescaping. */
 437                 parse_escapes(match,match,strlen(match),i,i);
 438                 parse_escapes(replace,replace,strlen(replace),i,i);
 439
 440                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 441                 for (i = 0; match[i] && replace[i]; i++) {
 442                         sed_cmd->string[i * 2] = match[i];
 443                         sed_cmd->string[(i * 2) + 1] = replace[i];
 444                 }
 445                 free(match);
 446                 free(replace);
 447         }
 448         /* if it wasnt a single-letter command that takes no arguments
 449          * then it must be an invalid command.
 450          */
 451         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 452                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 453         }
 454
 455         /* give back whatever's left over */
 456         return (cmdstr);
 457 }
 458
 459
 460 /* Parse address+command sets, skipping comment lines. */
 461
 462 static void add_cmd(char *cmdstr)
 463 {
 464         sed_cmd_t *sed_cmd;
 465         int temp;
 466
 467         /* Append this line to any unfinished line from last time. */
 468         if (bbg.add_cmd_line) {
 469                 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 470                 free(bbg.add_cmd_line);
 471                 bbg.add_cmd_line = cmdstr;
 472         }
 473
 474         /* If this line ends with backslash, request next line. */
 475         temp=strlen(cmdstr);
 476         if(temp && cmdstr[temp-1]=='\\') {
 477                 if (!bbg.add_cmd_line) bbg.add_cmd_line = xstrdup(cmdstr);
 478                 bbg.add_cmd_line[temp-1] = 0;
 479                 return;
 480         }
 481
 482         /* Loop parsing all commands in this line. */
 483         while(*cmdstr) {
 484                 /* Skip leading whitespace and semicolons */
 485                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 486
 487                 /* If no more commands, exit. */
 488                 if(!*cmdstr) break;
 489
 490                 /* if this is a comment, jump past it and keep going */
 491                 if (*cmdstr == '#') {
 492                         /* "#n" is the same as using -n on the command line */
 493                         if (cmdstr[1] == 'n') bbg.be_quiet++;
 494                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 495                         continue;
 496                 }
 497
 498                 /* parse the command
 499                  * format is: [addr][,addr][!]cmd
 500                  *            |----||-----||-|
 501                  *            part1 part2  part3
 502                  */
 503
 504                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 505
 506                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 507                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 508
 509                 /* second part (if present) will begin with a comma */
 510                 if (*cmdstr == ',') {
 511                         int idx;
 512
 513                         cmdstr++;
 514                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 515                         if (!idx) bb_error_msg_and_die("no address after comma");
 516                         cmdstr += idx;
 517                 }
 518
 519                 /* skip whitespace before the command */
 520                 while (isspace(*cmdstr)) cmdstr++;
 521
 522                 /* Check for inversion flag */
 523                 if (*cmdstr == '!') {
 524                         sed_cmd->invert = 1;
 525                         cmdstr++;
 526
 527                         /* skip whitespace before the command */
 528                         while (isspace(*cmdstr)) cmdstr++;
 529                 }
 530
 531                 /* last part (mandatory) will be a command */
 532                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 533                 sed_cmd->cmd = *(cmdstr++);
 534                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 535
 536                 /* Add the command to the command array */
 537                 bbg.sed_cmd_tail->next = sed_cmd;
 538                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 539         }
 540
 541         /* If we glued multiple lines together, free the memory. */
 542         free(bbg.add_cmd_line);
 543         bbg.add_cmd_line = NULL;
 544 }
 545
 546 /* Append to a string, reallocating memory as necessary. */
 547
 548 #define PIPE_GROW 64
 549
 550 static void pipe_putc(char c)
 551 {
 552         if(bbg.pipeline.idx==bbg.pipeline.len) {
 553                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 554                                                                 bbg.pipeline.len + PIPE_GROW);
 555                 bbg.pipeline.len+=PIPE_GROW;
 556         }
 557         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 558 }
 559
 560 static void do_subst_w_backrefs(char *line, char *replace)
 561 {
 562         int i,j;
 563
 564         /* go through the replacement string */
 565         for (i = 0; replace[i]; i++) {
 566                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 567                 if (replace[i] == '\\' && replace[i+1]>='0' && replace[i+1]<='9') {
 568                         int backref=replace[++i]-'0';
 569
 570                         /* print out the text held in bbg.regmatch[backref] */
 571                         if(bbg.regmatch[backref].rm_so != -1)
 572                                 for (j = bbg.regmatch[backref].rm_so;
 573                                         j < bbg.regmatch[backref].rm_eo; j++) pipe_putc(line[j]);
 574                 }
 575
 576                 /* if we find a backslash escaped character, print the character */
 577                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 578
 579                 /* if we find an unescaped '&' print out the whole matched text. */
 580                 else if (replace[i] == '&')
 581                         for (j = bbg.regmatch[0].rm_so; j < bbg.regmatch[0].rm_eo; j++)
 582                                 pipe_putc(line[j]);
 583                 /* Otherwise just output the character. */
 584                 else pipe_putc(replace[i]);
 585         }
 586 }
 587
 588 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 589 {
 590         char *oldline = *line;
 591         int altered = 0;
 592         int match_count=0;
 593         regex_t *current_regex;
 594
 595         /* Handle empty regex. */
 596         if (sed_cmd->sub_match == NULL) {
 597                 current_regex = bbg.previous_regex_ptr;
 598                 if(!current_regex)
 599                         bb_error_msg_and_die("No previous regexp.");
 600         } else bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 601
 602         /* Find the first match */
 603         if(REG_NOMATCH==regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 604                 return 0;
 605
 606         /* Initialize temporary output buffer. */
 607         bbg.pipeline.buf=xmalloc(PIPE_GROW);
 608         bbg.pipeline.len=PIPE_GROW;
 609         bbg.pipeline.idx=0;
 610
 611         /* Now loop through, substituting for matches */
 612         do {
 613                 int i;
 614
 615                 /* Work around bug in glibc regexec, demonstrated by:
 616                    echo " a.b" | busybox sed 's [^ .]* x g'
 617                    The match_count check is so not to break
 618                    echo "hi" | busybox sed 's/^/!/g' */
 619                 if(!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
 620                         pipe_putc(*(oldline++));
 621                         continue;
 622                 }
 623
 624                 match_count++;
 625
 626                 /* If we aren't interested in this match, output old line to
 627                    end of match and continue */
 628                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 629                         for(i=0;i<bbg.regmatch[0].rm_eo;i++)
 630                                 pipe_putc(*(oldline++));
 631                         continue;
 632                 }
 633
 634                 /* print everything before the match */
 635                 for (i = 0; i < bbg.regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 636
 637                 /* then print the substitution string */
 638                 do_subst_w_backrefs(oldline, sed_cmd->string);
 639
 640                 /* advance past the match */
 641                 oldline += bbg.regmatch[0].rm_eo;
 642                 /* flag that something has changed */
 643                 altered++;
 644
 645                 /* if we're not doing this globally, get out now */
 646                 if (sed_cmd->which_match) break;
 647         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 648
 649         /* Copy rest of string into output pipeline */
 650
 651         while(*oldline) pipe_putc(*(oldline++));
 652         pipe_putc(0);
 653
 654         free(*line);
 655         *line = bbg.pipeline.buf;
 656         return altered;
 657 }
 658
 659 /* Set command pointer to point to this label.  (Does not handle null label.) */
 660 static sed_cmd_t *branch_to(char *label)
 661 {
 662         sed_cmd_t *sed_cmd;
 663
 664         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 665                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 666                         return (sed_cmd);
 667                 }
 668         }
 669         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 670 }
 671
 672 static void append(char *s)
 673 {
 674         llist_add_to_end(&bbg.append_head, xstrdup(s));
 675 }
 676
 677 static void flush_append(void)
 678 {
 679         char *data;
 680
 681         /* Output appended lines. */
 682         while((data = (char *)llist_pop(&bbg.append_head))) {
 683                 fprintf(bbg.nonstdout,"%s\n",data);
 684                 free(data);
 685         }
 686 }
 687
 688 static void add_input_file(FILE *file)
 689 {
 690         bbg.input_file_list=xrealloc(bbg.input_file_list,
 691                         (bbg.input_file_count + 1) * sizeof(FILE *));
 692         bbg.input_file_list[bbg.input_file_count++] = file;
 693 }
 694
 695 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 696  * noting if we ran out of files without a newline on the last line we read.
 697  */
 698 static char *get_next_line(int *no_newline)
 699 {
 700         char *temp=NULL;
 701         int len;
 702
 703         flush_append();
 704         while (bbg.current_input_file<bbg.input_file_count) {
 705                 temp = bb_get_chunk_from_file(bbg.input_file_list[bbg.current_input_file],&len);
 706                 if (temp) {
 707                         *no_newline = !(len && temp[len-1]=='\n');
 708                         if (!*no_newline) temp[len-1] = 0;
 709                         break;
 710                 // Close this file and advance to next one
 711                 } else fclose(bbg.input_file_list[bbg.current_input_file++]);
 712         }
 713
 714         return temp;
 715 }
 716
 717 /* Output line of text.  missing_newline means the last line output did not
 718    end with a newline.  no_newline means this line does not end with a
 719    newline. */
 720
 721 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 722 {
 723         if(missing_newline) fputc('\n',file);
 724         fputs(s,file);
 725         if(!no_newline) fputc('\n',file);
 726
 727     if(ferror(file)) {
 728                 bb_default_error_retval = 4;  /* It's what gnu sed exits with... */
 729                 bb_error_msg_and_die(bb_msg_write_error);
 730         }
 731
 732         return no_newline;
 733 }
 734
 735 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,bbg.nonstdout,missing_newline,n)
 736
 737 /* Process all the lines in all the files */
 738
 739 static void process_files(void)
 740 {
 741         char *pattern_space, *next_line;
 742         int linenum = 0, missing_newline=0;
 743         int no_newline,next_no_newline=0;
 744
 745         /* Prime the pump */
 746         next_line = get_next_line(&next_no_newline);
 747
 748         /* go through every line in each file */
 749         for(;;) {
 750                 sed_cmd_t *sed_cmd;
 751                 int substituted=0;
 752
 753                 /* Advance to next line.  Stop if out of lines. */
 754                 if(!(pattern_space=next_line)) break;
 755                 no_newline=next_no_newline;
 756
 757                 /* Read one line in advance so we can act on the last line,
 758                  * the '$' address */
 759                 next_line = get_next_line(&next_no_newline);
 760                 linenum++;
 761 restart:
 762                 /* for every line, go through all the commands */
 763                 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next)
 764                 {
 765                         int old_matched, matched;
 766
 767                         old_matched = sed_cmd->in_match;
 768
 769                         /* Determine if this command matches this line: */
 770
 771                         /* Are we continuing a previous multi-line match? */
 772
 773                         sed_cmd->in_match = sed_cmd->in_match
 774
 775                         /* Or is no range necessary? */
 776                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 777                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 778
 779                         /* Or did we match the start of a numerical range? */
 780                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 781
 782                         /* Or does this line match our begin address regex? */
 783                                 || (sed_cmd->beg_match &&
 784                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 785
 786                         /* Or did we match last line of input? */
 787                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 788
 789                         /* Snapshot the value */
 790
 791                         matched = sed_cmd->in_match;
 792
 793                         /* Is this line the end of the current match? */
 794
 795                         if(matched) {
 796                                 sed_cmd->in_match = !(
 797                                         /* has the ending line come, or is this a single address command? */
 798                                         (sed_cmd->end_line ?
 799                                                 sed_cmd->end_line==-1 ?
 800                                                         !next_line
 801                                                         : sed_cmd->end_line<=linenum
 802                                                 : !sed_cmd->end_match)
 803                                         /* or does this line matches our last address regex */
 804                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 805                                 );
 806                         }
 807
 808                         /* Skip blocks of commands we didn't match. */
 809                         if (sed_cmd->cmd == '{') {
 810                                 if(sed_cmd->invert ? matched : !matched)
 811                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 812                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 813                                 continue;
 814                         }
 815
 816                         /* Okay, so did this line match? */
 817                         if (sed_cmd->invert ? !matched : matched) {
 818                                 /* Update last used regex in case a blank substitute BRE is found */
 819                                 if (sed_cmd->beg_match) {
 820                                         bbg.previous_regex_ptr = sed_cmd->beg_match;
 821                                 }
 822
 823                                 /* actual sedding */
 824                                 switch (sed_cmd->cmd) {
 825
 826                                         /* Print line number */
 827                                         case '=':
 828                                                 fprintf(bbg.nonstdout,"%d\n", linenum);
 829                                                 break;
 830
 831                                         /* Write the current pattern space up to the first newline */
 832                                         case 'P':
 833                                         {
 834                                                 char *tmp = strchr(pattern_space, '\n');
 835
 836                                                 if (tmp) {
 837                                                         *tmp = '\0';
 838                                                         sed_puts(pattern_space,1);
 839                                                         *tmp = '\n';
 840                                                         break;
 841                                                 }
 842                                                 /* Fall Through */
 843                                         }
 844
 845                                         /* Write the current pattern space to output */
 846                                         case 'p':
 847                                                 sed_puts(pattern_space,no_newline);
 848                                                 break;
 849                                         /* Delete up through first newline */
 850                                         case 'D':
 851                                         {
 852                                                 char *tmp = strchr(pattern_space,'\n');
 853
 854                                                 if(tmp) {
 855                                                         tmp=xstrdup(tmp+1);
 856                                                         free(pattern_space);
 857                                                         pattern_space=tmp;
 858                                                         goto restart;
 859                                                 }
 860                                         }
 861                                         /* discard this line. */
 862                                         case 'd':
 863                                                 goto discard_line;
 864
 865                                         /* Substitute with regex */
 866                                         case 's':
 867                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 868                                                         substituted|=1;
 869
 870                                                         /* handle p option */
 871                                                         if(sed_cmd->sub_p)
 872                                                                 sed_puts(pattern_space,no_newline);
 873                                                         /* handle w option */
 874                                                         if(sed_cmd->file)
 875                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 876
 877                                                 }
 878                                                 break;
 879
 880                                         /* Append line to linked list to be printed later */
 881                                         case 'a':
 882                                         {
 883                                                 append(sed_cmd->string);
 884                                                 break;
 885                                         }
 886
 887                                         /* Insert text before this line */
 888                                         case 'i':
 889                                                 sed_puts(sed_cmd->string,1);
 890                                                 break;
 891
 892                                         /* Cut and paste text (replace) */
 893                                         case 'c':
 894                                                 /* Only triggers on last line of a matching range. */
 895                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,0);
 896                                                 goto discard_line;
 897
 898                                         /* Read file, append contents to output */
 899                                         case 'r':
 900                                         {
 901                                                 FILE *rfile;
 902
 903                                                 rfile = fopen(sed_cmd->string, "r");
 904                                                 if (rfile) {
 905                                                         char *line;
 906
 907                                                         while ((line = bb_get_chomped_line_from_file(rfile))
 908                                                                         != NULL)
 909                                                                 append(line);
 910                                                         xprint_and_close_file(rfile);
 911                                                 }
 912
 913                                                 break;
 914                                         }
 915
 916                                         /* Write pattern space to file. */
 917                                         case 'w':
 918                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 919                                                 break;
 920
 921                                         /* Read next line from input */
 922                                         case 'n':
 923                                                 if (!bbg.be_quiet)
 924                                                         sed_puts(pattern_space,no_newline);
 925                                                 if (next_line) {
 926                                                         free(pattern_space);
 927                                                         pattern_space = next_line;
 928                                                         no_newline=next_no_newline;
 929                                                         next_line = get_next_line(&next_no_newline);
 930                                                         linenum++;
 931                                                         break;
 932                                                 }
 933                                                 /* fall through */
 934
 935                                         /* Quit.  End of script, end of input. */
 936                                         case 'q':
 937                                                 /* Exit the outer while loop */
 938                                                 free(next_line);
 939                                                 next_line = NULL;
 940                                                 goto discard_commands;
 941
 942                                         /* Append the next line to the current line */
 943                                         case 'N':
 944                                         {
 945                                                 /* If no next line, jump to end of script and exit. */
 946                                                 if (next_line == NULL) {
 947                                                         /* Jump to end of script and exit */
 948                                                         free(next_line);
 949                                                         next_line = NULL;
 950                                                         goto discard_line;
 951                                                 /* append next_line, read new next_line. */
 952                                                 } else {
 953                                                         int len=strlen(pattern_space);
 954
 955                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 956                                                         pattern_space[len]='\n';
 957                                                         strcpy(pattern_space+len+1, next_line);
 958                                                         no_newline=next_no_newline;
 959                                                         next_line = get_next_line(&next_no_newline);
 960                                                         linenum++;
 961                                                 }
 962                                                 break;
 963                                         }
 964
 965                                         /* Test/branch if substitution occurred */
 966                                         case 't':
 967                                                 if(!substituted) break;
 968                                                 substituted=0;
 969                                                 /* Fall through */
 970                                         /* Test/branch if substitution didn't occur */
 971                                         case 'T':
 972                                                 if (substituted) break;
 973                                                 /* Fall through */
 974                                         /* Branch to label */
 975                                         case 'b':
 976                                                 if (!sed_cmd->string) goto discard_commands;
 977                                                 else sed_cmd = branch_to(sed_cmd->string);
 978                                                 break;
 979                                         /* Transliterate characters */
 980                                         case 'y':
 981                                         {
 982                                                 int i;
 983
 984                                                 for (i = 0; pattern_space[i]; i++) {
 985                                                         int j;
 986
 987                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
 988                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
 989                                                                         pattern_space[i] = sed_cmd->string[j + 1];
 990                                                                         break;
 991                                                                 }
 992                                                         }
 993                                                 }
 994
 995                                                 break;
 996                                         }
 997                                         case 'g':       /* Replace pattern space with hold space */
 998                                                 free(pattern_space);
 999                                                 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1000                                                 break;
1001                                         case 'G':       /* Append newline and hold space to pattern space */
1002                                         {
1003                                                 int pattern_space_size = 2;
1004                                                 int hold_space_size = 0;
1005
1006                                                 if (pattern_space)
1007                                                         pattern_space_size += strlen(pattern_space);
1008                                                 if (bbg.hold_space)
1009                                                         hold_space_size = strlen(bbg.hold_space);
1010                                                 pattern_space = xrealloc(pattern_space,
1011                                                                 pattern_space_size + hold_space_size);
1012                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1013                                                 strcat(pattern_space, "\n");
1014                                                 if (bbg.hold_space)
1015                                                         strcat(pattern_space, bbg.hold_space);
1016                                                 no_newline=0;
1017
1018                                                 break;
1019                                         }
1020                                         case 'h':       /* Replace hold space with pattern space */
1021                                                 free(bbg.hold_space);
1022                                                 bbg.hold_space = xstrdup(pattern_space);
1023                                                 break;
1024                                         case 'H':       /* Append newline and pattern space to hold space */
1025                                         {
1026                                                 int hold_space_size = 2;
1027                                                 int pattern_space_size = 0;
1028
1029                                                 if (bbg.hold_space)
1030                                                         hold_space_size += strlen(bbg.hold_space);
1031                                                 if (pattern_space)
1032                                                         pattern_space_size = strlen(pattern_space);
1033                                                 bbg.hold_space = xrealloc(bbg.hold_space,
1034                                                                                         hold_space_size + pattern_space_size);
1035
1036                                                 if (hold_space_size == 2) *bbg.hold_space=0;
1037                                                 strcat(bbg.hold_space, "\n");
1038                                                 if (pattern_space) strcat(bbg.hold_space, pattern_space);
1039
1040                                                 break;
1041                                         }
1042                                         case 'x': /* Exchange hold and pattern space */
1043                                         {
1044                                                 char *tmp = pattern_space;
1045                                                 pattern_space = bbg.hold_space ? : xzalloc(1);
1046                                                 no_newline=0;
1047                                                 bbg.hold_space = tmp;
1048                                                 break;
1049                                         }
1050                                 }
1051                         }
1052                 }
1053
1054                 /*
1055                  * exit point from sedding...
1056                  */
1057 discard_commands:
1058                 /* we will print the line unless we were told to be quiet ('-n')
1059                    or if the line was suppressed (ala 'd'elete) */
1060                 if (!bbg.be_quiet) sed_puts(pattern_space,no_newline);
1061
1062                 /* Delete and such jump here. */
1063 discard_line:
1064                 flush_append();
1065                 free(pattern_space);
1066         }
1067 }
1068
1069 /* It is possible to have a command line argument with embedded
1070    newlines.  This counts as multiple command lines. */
1071
1072 static void add_cmd_block(char *cmdstr)
1073 {
1074         int go=1;
1075         char *temp=xstrdup(cmdstr),*temp2=temp;
1076
1077         while(go) {
1078                 int len=strcspn(temp2,"\n");
1079                 if(!temp2[len]) go=0;
1080                 else temp2[len]=0;
1081                 add_cmd(temp2);
1082                 temp2+=len+1;
1083         }
1084         free(temp);
1085 }
1086
1087 int sed_main(int argc, char **argv)
1088 {
1089         unsigned long opt;
1090         llist_t *opt_e, *opt_f;
1091         int status = EXIT_SUCCESS;
1092
1093         bbg.sed_cmd_tail=&bbg.sed_cmd_head;
1094
1095         /* destroy command strings on exit */
1096         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1097
1098         /* Lie to autoconf when it starts asking stupid questions. */
1099         if(argc==2 && !strcmp(argv[1],"--version")) {
1100                 printf("This is not GNU sed version 4.0\n");
1101                 exit(0);
1102         }
1103
1104         /* do normal option parsing */
1105         opt_e = opt_f = NULL;
1106         bb_opt_complementally = "e::f::"; /* can occur multiple times */
1107         opt = bb_getopt_ulflags(argc, argv, "irne:f:", &opt_e, &opt_f);
1108         if (opt & 0x1) { // -i
1109                 bbg.in_place++;
1110                 atexit(cleanup_outname);
1111         }
1112         if (opt & 0x2) bbg.regex_type|=REG_EXTENDED; // -r
1113         if (opt & 0x4) bbg.be_quiet++; // -n
1114         if (opt & 0x8) { // -e
1115                 while (opt_e) {
1116                         llist_t *cur = opt_e;
1117                         add_cmd_block(cur->data);
1118                         opt_e = cur->link;
1119                         free(cur);
1120                 }
1121         }
1122         if (opt & 0x10) { // -f
1123                 while (opt_f) {
1124                         llist_t *cur = opt_f;
1125                         FILE *cmdfile;
1126                         char *line;
1127                         cmdfile = xfopen(cur->data, "r");
1128                         while ((line = bb_get_chomped_line_from_file(cmdfile)) != NULL) {
1129                                 add_cmd(line);
1130                                 free(line);
1131                         }
1132                         xprint_and_close_file(cmdfile);
1133                         opt_f = cur->link;
1134                         free(cur);
1135                 }
1136         }
1137         /* if we didn't get a pattern from -e or -f, use argv[optind] */
1138         if(!(opt & 0x18)) {
1139                 if (argv[optind] == NULL)
1140                         bb_show_usage();
1141                 else
1142                         add_cmd_block(argv[optind++]);
1143         }
1144         /* Flush any unfinished commands. */
1145         add_cmd("");
1146
1147         /* By default, we write to stdout */
1148         bbg.nonstdout = stdout;
1149
1150         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1151          * files were specified or '-' was specified, take input from stdin.
1152          * Otherwise, we process all the files specified. */
1153         if (argv[optind] == NULL) {
1154                 if(bbg.in_place) bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1155                 add_input_file(stdin);
1156                 process_files();
1157         } else {
1158                 int i;
1159                 FILE *file;
1160
1161                 for (i = optind; i < argc; i++) {
1162                         if(!strcmp(argv[i], "-") && !bbg.in_place) {
1163                                 add_input_file(stdin);
1164                                 process_files();
1165                         } else {
1166                                 file = bb_wfopen(argv[i], "r");
1167                                 if (file) {
1168                                         if(bbg.in_place) {
1169                                                 struct stat statbuf;
1170                                                 int nonstdoutfd;
1171
1172                                                 bbg.outname=xstrndup(argv[i],strlen(argv[i])+6);
1173                                                 strcat(bbg.outname,"XXXXXX");
1174                                                 if(-1==(nonstdoutfd=mkstemp(bbg.outname)))
1175                                                         bb_error_msg_and_die("no temp file");
1176                                                 bbg.nonstdout=fdopen(nonstdoutfd,"w");
1177
1178                                                 /* Set permissions of output file */
1179
1180                                                 fstat(fileno(file),&statbuf);
1181                                                 fchmod(nonstdoutfd,statbuf.st_mode);
1182                                                 add_input_file(file);
1183                                                 process_files();
1184                                                 fclose(bbg.nonstdout);
1185
1186                                                 bbg.nonstdout=stdout;
1187                                                 unlink(argv[i]);
1188                                                 rename(bbg.outname,argv[i]);
1189                                                 free(bbg.outname);
1190                                                 bbg.outname=0;
1191                                         } else add_input_file(file);
1192                                 } else {
1193                                         status = EXIT_FAILURE;
1194                                 }
1195                         }
1196                 }
1197                 if(bbg.input_file_count>bbg.current_input_file) process_files();
1198         }
1199
1200         return status;
1201 }