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 sub_p:1;           /* (s) print option */
  84
  85         int last_char;                  /* Last line written by (sw) had no '\n' */
  86
  87         /* GENERAL FIELDS */
  88         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
  89         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
  90 } sed_cmd_t;
  91
  92 static const char *const semicolon_whitespace = "; \n\r\t\v";
  93
  94 struct sed_globals {
  95         /* options */
  96         int be_quiet, 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 #if ENABLE_FEATURE_CLEAN_UP
 124 static void sed_free_and_close_stuff(void)
 125 {
 126         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 127
 128         llist_free(bbg.append_head, free);
 129
 130         while (sed_cmd) {
 131                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 132
 133                 if (sed_cmd->file)
 134                         xprint_and_close_file(sed_cmd->file);
 135
 136                 if (sed_cmd->beg_match) {
 137                         regfree(sed_cmd->beg_match);
 138                         free(sed_cmd->beg_match);
 139                 }
 140                 if (sed_cmd->end_match) {
 141                         regfree(sed_cmd->end_match);
 142                         free(sed_cmd->end_match);
 143                 }
 144                 if (sed_cmd->sub_match) {
 145                         regfree(sed_cmd->sub_match);
 146                         free(sed_cmd->sub_match);
 147                 }
 148                 free(sed_cmd->string);
 149                 free(sed_cmd);
 150                 sed_cmd = sed_cmd_next;
 151         }
 152
 153         if (bbg.hold_space) free(bbg.hold_space);
 154
 155         while (bbg.current_input_file < bbg.input_file_count)
 156                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 157 }
 158 #else
 159 void sed_free_and_close_stuff(void);
 160 #endif
 161
 162 /* If something bad happens during -i operation, delete temp file */
 163
 164 static void cleanup_outname(void)
 165 {
 166         if (bbg.outname) unlink(bbg.outname);
 167 }
 168
 169 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 170
 171 static void parse_escapes(char *dest, char *string, int len, char from, char to)
 172 {
 173         int i = 0;
 174
 175         while (i < len) {
 176                 if (string[i] == '\\') {
 177                         if (!to || string[i+1] == from) {
 178                                 *dest++ = to ? to : string[i+1];
 179                                 i += 2;
 180                                 continue;
 181                         }
 182                         *dest++ = string[i++];
 183                 }
 184                 *dest++ = string[i++];
 185         }
 186         *dest = 0;
 187 }
 188
 189 static char *copy_parsing_escapes(char *string, int len)
 190 {
 191         char *dest = xmalloc(len + 1);
 192
 193         parse_escapes(dest, string, len, 'n', '\n');
 194         return dest;
 195 }
 196
 197
 198 /*
 199  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 200  * beginning at a specified index and returns the index of the next regular
 201  * expression delimiter (typically a forward * slash ('/')) not preceded by
 202  * a backslash ('\').  A negative delimiter disables square bracket checking.
 203  */
 204 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
 205 {
 206         int bracket = -1;
 207         int escaped = 0;
 208         int idx = 0;
 209         char ch;
 210
 211         if (delimiter < 0) {
 212                 bracket--;
 213                 delimiter = -delimiter;
 214         }
 215
 216         for (; (ch = str[idx]); idx++) {
 217                 if (bracket >= 0) {
 218                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 219                                         && str[idx - 1] == '^')))
 220                                 bracket = -1;
 221                 } else if (escaped)
 222                         escaped = 0;
 223                 else if (ch == '\\')
 224                         escaped = 1;
 225                 else if (bracket == -1 && ch == '[')
 226                         bracket = idx;
 227                 else if (ch == delimiter)
 228                         return idx;
 229         }
 230
 231         /* if we make it to here, we've hit the end of the string */
 232         bb_error_msg_and_die("unmatched '%c'", delimiter);
 233 }
 234
 235 /*
 236  *  Returns the index of the third delimiter
 237  */
 238 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
 239 {
 240         char *cmdstr_ptr = cmdstr;
 241         char delimiter;
 242         int idx = 0;
 243
 244         /* verify that the 's' or 'y' is followed by something.  That something
 245          * (typically a 'slash') is now our regexp delimiter... */
 246         if (*cmdstr == '\0')
 247                 bb_error_msg_and_die("bad format in substitution expression");
 248         delimiter = *cmdstr_ptr++;
 249
 250         /* save the match string */
 251         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 252         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 253
 254         /* save the replacement string */
 255         cmdstr_ptr += idx + 1;
 256         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 257         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 258
 259         return ((cmdstr_ptr - cmdstr) + idx);
 260 }
 261
 262 /*
 263  * returns the index in the string just past where the address ends.
 264  */
 265 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 266 {
 267         char *pos = my_str;
 268
 269         if (isdigit(*my_str)) {
 270                 *linenum = strtol(my_str, &pos, 10);
 271                 /* endstr shouldnt ever equal NULL */
 272         } else if (*my_str == '$') {
 273                 *linenum = -1;
 274                 pos++;
 275         } else if (*my_str == '/' || *my_str == '\\') {
 276                 int next;
 277                 char delimiter;
 278                 char *temp;
 279
 280                 delimiter = '/';
 281                 if (*my_str == '\\') delimiter = *++pos;
 282                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 283                 temp = copy_parsing_escapes(pos, next);
 284                 *regex = xmalloc(sizeof(regex_t));
 285                 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
 286                 free(temp);
 287                 /* Move position to next character after last delimiter */
 288                 pos += (next+1);
 289         }
 290         return pos - my_str;
 291 }
 292
 293 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 294 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
 295 {
 296         int start = 0, idx, hack = 0;
 297
 298         /* Skip whitespace, then grab filename to end of line */
 299         while (isspace(filecmdstr[start])) start++;
 300         idx = start;
 301         while (filecmdstr[idx] && filecmdstr[idx] != '\n') idx++;
 302
 303         /* If lines glued together, put backslash back. */
 304         if (filecmdstr[idx] == '\n') hack = 1;
 305         if (idx == start)
 306                 bb_error_msg_and_die("empty filename");
 307         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 308         if (hack) (*retval)[idx] = '\\';
 309
 310         return idx;
 311 }
 312
 313 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
 314 {
 315         int cflags = bbg.regex_type;
 316         char *match;
 317         int idx = 0;
 318
 319         /*
 320          * A substitution command should look something like this:
 321          *    s/match/replace/ #gIpw
 322          *    ||     |        |||
 323          *    mandatory       optional
 324          */
 325         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 326
 327         /* determine the number of back references in the match string */
 328         /* Note: we compute this here rather than in the do_subst_command()
 329          * function to save processor time, at the expense of a little more memory
 330          * (4 bits) per sed_cmd */
 331
 332         /* process the flags */
 333
 334         sed_cmd->which_match = 1;
 335         while (substr[++idx]) {
 336                 /* Parse match number */
 337                 if (isdigit(substr[idx])) {
 338                         if (match[0] != '^') {
 339                                 /* Match 0 treated as all, multiple matches we take the last one. */
 340                                 char *pos = substr + idx;
 341                                 /* FIXME: error check? */
 342                                 sed_cmd->which_match = (unsigned short)strtol(substr+idx, &pos, 10);
 343                                 idx = pos - substr;
 344                         }
 345                         continue;
 346                 }
 347                 /* Skip spaces */
 348                 if (isspace(substr[idx])) continue;
 349
 350                 switch (substr[idx]) {
 351                 /* Replace all occurrences */
 352                 case 'g':
 353                         if (match[0] != '^') sed_cmd->which_match = 0;
 354                         break;
 355                 /* Print pattern space */
 356                 case 'p':
 357                         sed_cmd->sub_p = 1;
 358                         break;
 359                 /* Write to file */
 360                 case 'w':
 361                 {
 362                         char *temp;
 363                         idx += parse_file_cmd(sed_cmd, substr+idx, &temp);
 364
 365                         break;
 366                 }
 367                 /* Ignore case (gnu exension) */
 368                 case 'I':
 369                         cflags |= REG_ICASE;
 370                         break;
 371                 /* Comment */
 372                 case '#':
 373                         while (substr[++idx]) /*skip all*/;
 374                         /* Fall through */
 375                 /* End of command */
 376                 case ';':
 377                 case '}':
 378                         goto out;
 379                 default:
 380                         bb_error_msg_and_die("bad option in substitution expression");
 381                 }
 382         }
 383 out:
 384         /* compile the match string into a regex */
 385         if (*match != '\0') {
 386                 /* If match is empty, we use last regex used at runtime */
 387                 sed_cmd->sub_match = xmalloc(sizeof(regex_t));
 388                 xregcomp(sed_cmd->sub_match, match, cflags);
 389         }
 390         free(match);
 391
 392         return idx;
 393 }
 394
 395 /*
 396  *  Process the commands arguments
 397  */
 398 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 399 {
 400         /* handle (s)ubstitution command */
 401         if (sed_cmd->cmd == 's')
 402                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 403         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 404         else if (strchr("aic", sed_cmd->cmd)) {
 405                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 406                         bb_error_msg_and_die
 407                                 ("only a beginning address can be specified for edit commands");
 408                 for (;;) {
 409                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 410                                 cmdstr++;
 411                                 break;
 412                         } else if (isspace(*cmdstr))
 413                                 cmdstr++;
 414                         else
 415                                 break;
 416                 }
 417                 sed_cmd->string = xstrdup(cmdstr);
 418                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), 0, 0);
 419                 cmdstr += strlen(cmdstr);
 420         /* handle file cmds: (r)ead */
 421         } else if (strchr("rw", sed_cmd->cmd)) {
 422                 if (sed_cmd->end_line || sed_cmd->end_match)
 423                         bb_error_msg_and_die("command only uses one address");
 424                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 425                 if (sed_cmd->cmd == 'w')
 426                         sed_cmd->file = xfopen(sed_cmd->string, "w");
 427         /* handle branch commands */
 428         } else if (strchr(":btT", sed_cmd->cmd)) {
 429                 int length;
 430
 431                 cmdstr = skip_whitespace(cmdstr);
 432                 length = strcspn(cmdstr, semicolon_whitespace);
 433                 if (length) {
 434                         sed_cmd->string = xstrndup(cmdstr, length);
 435                         cmdstr += length;
 436                 }
 437         }
 438         /* translation command */
 439         else if (sed_cmd->cmd == 'y') {
 440                 char *match, *replace;
 441                 int i = cmdstr[0];
 442
 443                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 444                 /* \n already parsed, but \delimiter needs unescaping. */
 445                 parse_escapes(match, match, strlen(match), i, i);
 446                 parse_escapes(replace, replace, strlen(replace), i, i);
 447
 448                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 449                 for (i = 0; match[i] && replace[i]; i++) {
 450                         sed_cmd->string[i*2] = match[i];
 451                         sed_cmd->string[i*2+1] = replace[i];
 452                 }
 453                 free(match);
 454                 free(replace);
 455         }
 456         /* if it wasnt a single-letter command that takes no arguments
 457          * then it must be an invalid command.
 458          */
 459         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 460                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 461         }
 462
 463         /* give back whatever's left over */
 464         return cmdstr;
 465 }
 466
 467
 468 /* Parse address+command sets, skipping comment lines. */
 469
 470 static void add_cmd(char *cmdstr)
 471 {
 472         sed_cmd_t *sed_cmd;
 473         int temp;
 474
 475         /* Append this line to any unfinished line from last time. */
 476         if (bbg.add_cmd_line) {
 477                 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 478                 free(bbg.add_cmd_line);
 479                 bbg.add_cmd_line = cmdstr;
 480         }
 481
 482         /* If this line ends with backslash, request next line. */
 483         temp = strlen(cmdstr);
 484         if (temp && cmdstr[temp-1] == '\\') {
 485                 if (!bbg.add_cmd_line)
 486                         bbg.add_cmd_line = xstrdup(cmdstr);
 487                 bbg.add_cmd_line[temp-1] = 0;
 488                 return;
 489         }
 490
 491         /* Loop parsing all commands in this line. */
 492         while (*cmdstr) {
 493                 /* Skip leading whitespace and semicolons */
 494                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 495
 496                 /* If no more commands, exit. */
 497                 if (!*cmdstr) break;
 498
 499                 /* if this is a comment, jump past it and keep going */
 500                 if (*cmdstr == '#') {
 501                         /* "#n" is the same as using -n on the command line */
 502                         if (cmdstr[1] == 'n')
 503                                 bbg.be_quiet++;
 504                         cmdstr = strpbrk(cmdstr, "\n\r");
 505                         if (!cmdstr) break;
 506                         continue;
 507                 }
 508
 509                 /* parse the command
 510                  * format is: [addr][,addr][!]cmd
 511                  *            |----||-----||-|
 512                  *            part1 part2  part3
 513                  */
 514
 515                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 516
 517                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 518                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 519
 520                 /* second part (if present) will begin with a comma */
 521                 if (*cmdstr == ',') {
 522                         int idx;
 523
 524                         cmdstr++;
 525                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 526                         if (!idx)
 527                                 bb_error_msg_and_die("no address after comma");
 528                         cmdstr += idx;
 529                 }
 530
 531                 /* skip whitespace before the command */
 532                 cmdstr = skip_whitespace(cmdstr);
 533
 534                 /* Check for inversion flag */
 535                 if (*cmdstr == '!') {
 536                         sed_cmd->invert = 1;
 537                         cmdstr++;
 538
 539                         /* skip whitespace before the command */
 540                         cmdstr = skip_whitespace(cmdstr);
 541                 }
 542
 543                 /* last part (mandatory) will be a command */
 544                 if (!*cmdstr)
 545                         bb_error_msg_and_die("missing command");
 546                 sed_cmd->cmd = *(cmdstr++);
 547                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 548
 549                 /* Add the command to the command array */
 550                 bbg.sed_cmd_tail->next = sed_cmd;
 551                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 552         }
 553
 554         /* If we glued multiple lines together, free the memory. */
 555         free(bbg.add_cmd_line);
 556         bbg.add_cmd_line = NULL;
 557 }
 558
 559 /* Append to a string, reallocating memory as necessary. */
 560
 561 #define PIPE_GROW 64
 562
 563 static void pipe_putc(char c)
 564 {
 565         if (bbg.pipeline.idx == bbg.pipeline.len) {
 566                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 567                                 bbg.pipeline.len + PIPE_GROW);
 568                 bbg.pipeline.len += PIPE_GROW;
 569         }
 570         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 571 }
 572
 573 static void do_subst_w_backrefs(char *line, char *replace)
 574 {
 575         int i,j;
 576
 577         /* go through the replacement string */
 578         for (i = 0; replace[i]; i++) {
 579                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 580                 if (replace[i] == '\\') {
 581                         unsigned backref = replace[++i] - '0';
 582                         if (backref <= 9) {
 583                                 /* print out the text held in bbg.regmatch[backref] */
 584                                 if (bbg.regmatch[backref].rm_so != -1) {
 585                                         j = bbg.regmatch[backref].rm_so;
 586                                         while (j < bbg.regmatch[backref].rm_eo)
 587                                                 pipe_putc(line[j++]);
 588                                 }
 589                                 continue;
 590                         }
 591                         /* I _think_ it is impossible to get '\' to be
 592                          * the last char in replace string. Thus we dont check
 593                          * for replace[i] == NUL. (counterexample anyone?) */
 594                         /* if we find a backslash escaped character, print the character */
 595                         pipe_putc(replace[i]);
 596                         continue;
 597                 }
 598                 /* if we find an unescaped '&' print out the whole matched text. */
 599                 if (replace[i] == '&') {
 600                         j = bbg.regmatch[0].rm_so;
 601                         while (j < bbg.regmatch[0].rm_eo)
 602                                 pipe_putc(line[j++]);
 603                         continue;
 604                 }
 605                 /* Otherwise just output the character. */
 606                 pipe_putc(replace[i]);
 607         }
 608 }
 609
 610 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 611 {
 612         char *oldline = *line;
 613         int altered = 0;
 614         int match_count = 0;
 615         regex_t *current_regex;
 616
 617         /* Handle empty regex. */
 618         if (sed_cmd->sub_match == NULL) {
 619                 current_regex = bbg.previous_regex_ptr;
 620                 if (!current_regex)
 621                         bb_error_msg_and_die("no previous regexp");
 622         } else
 623                 bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 624
 625         /* Find the first match */
 626         if (REG_NOMATCH == regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 627                 return 0;
 628
 629         /* Initialize temporary output buffer. */
 630         bbg.pipeline.buf = xmalloc(PIPE_GROW);
 631         bbg.pipeline.len = PIPE_GROW;
 632         bbg.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 (!bbg.regmatch[0].rm_so && !bbg.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 < bbg.regmatch[0].rm_eo; i++)
 653                                 pipe_putc(*oldline++);
 654                         continue;
 655                 }
 656
 657                 /* print everything before the match */
 658                 for (i = 0; i < bbg.regmatch[0].rm_so; i++)
 659                         pipe_putc(oldline[i]);
 660
 661                 /* then print the substitution string */
 662                 do_subst_w_backrefs(oldline, sed_cmd->string);
 663
 664                 /* advance past the match */
 665                 oldline += bbg.regmatch[0].rm_eo;
 666                 /* flag that something has changed */
 667                 altered++;
 668
 669                 /* if we're not doing this globally, get out now */
 670                 if (sed_cmd->which_match) break;
 671         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 672
 673         /* Copy rest of string into output pipeline */
 674
 675         while (*oldline)
 676                 pipe_putc(*oldline++);
 677         pipe_putc(0);
 678
 679         free(*line);
 680         *line = bbg.pipeline.buf;
 681         return altered;
 682 }
 683
 684 /* Set command pointer to point to this label.  (Does not handle null label.) */
 685 static sed_cmd_t *branch_to(char *label)
 686 {
 687         sed_cmd_t *sed_cmd;
 688
 689         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 690                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 691                         return sed_cmd;
 692                 }
 693         }
 694         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 695 }
 696
 697 static void append(char *s)
 698 {
 699         llist_add_to_end(&bbg.append_head, xstrdup(s));
 700 }
 701
 702 static void flush_append(void)
 703 {
 704         char *data;
 705
 706         /* Output appended lines. */
 707         while ((data = (char *)llist_pop(&bbg.append_head))) {
 708                 fprintf(bbg.nonstdout, "%s\n", data);
 709                 free(data);
 710         }
 711 }
 712
 713 static void add_input_file(FILE *file)
 714 {
 715         bbg.input_file_list = xrealloc(bbg.input_file_list,
 716                         (bbg.input_file_count + 1) * sizeof(FILE *));
 717         bbg.input_file_list[bbg.input_file_count++] = file;
 718 }
 719
 720 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 721  * noting if we ran out of files without a newline on the last line we read.
 722  */
 723 static char *get_next_line(int *last_char)
 724 {
 725         char *temp = NULL;
 726         int len, lc;
 727
 728         lc = 0;
 729         flush_append();
 730         while (bbg.current_input_file < bbg.input_file_count) {
 731                 /* Read line up to a newline or NUL byte, inclusive,
 732                  * return malloc'ed char[]. length of the chunk read
 733                  * is stored in len. NULL if EOF/error */
 734                 temp = bb_get_chunk_from_file(
 735                         bbg.input_file_list[bbg.current_input_file], &len);
 736                 if (temp) {
 737                         /* len > 0 here, it's ok to do temp[len-1] */
 738                         char c = temp[len-1];
 739                         if (c == '\n' || c == '\0') {
 740                                 temp[len-1] = '\0';
 741                                 lc |= (unsigned char)c;
 742                                 break;
 743                         }
 744                         /* will be returned if last line in the file
 745                          * doesn't end with either '\n' or '\0' */
 746                         lc |= 0x100;
 747                         break;
 748                 }
 749                 /* Close this file and advance to next one */
 750                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 751                 /* "this is the first line from new input file" */
 752                 lc |= 0x200;
 753         }
 754         *last_char = lc;
 755         return temp;
 756 }
 757
 758 /* Output line of text. */
 759 /* Note:
 760  * The tricks with 0x200 and last_puts_char are there to emulate gnu sed.
 761  * Without them, we had this:
 762  * echo -n thingy >z1
 763  * echo -n again >z2
 764  * >znull
 765  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 766  * output:
 767  * gnu sed 4.1.5:
 768  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 769  * bbox:
 770  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 771  */
 772
 773 static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_char)
 774 {
 775         static char last_puts_char;
 776
 777         /* Is this a first line from new file
 778          * and old file didn't end with '\n'? */
 779         if ((last_char & 0x200) && last_puts_char != '\n') {
 780                 fputc('\n', file);
 781                 last_puts_char = '\n';
 782         }
 783         fputs(s, file);
 784         /* why 'x'? - just something which is not '\n' */
 785         if (s[0])
 786                 last_puts_char = 'x';
 787         if (!(last_char & 0x100)) { /* had trailing '\n' or '\0'? */
 788                 last_char &= 0xff;
 789                 fputc(last_char, file);
 790                 last_puts_char = last_char;
 791         }
 792
 793         if (ferror(file)) {
 794                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 795                 bb_error_msg_and_die(bb_msg_write_error);
 796         }
 797
 798         return last_char;
 799 }
 800
 801 #define sed_puts(s, n) \
 802         (prev_last_char = puts_maybe_newline(s, bbg.nonstdout, prev_last_char, n))
 803
 804 /* Process all the lines in all the files */
 805
 806 static void process_files(void)
 807 {
 808         char *pattern_space, *next_line;
 809         int linenum = 0, prev_last_char = 0;
 810         int last_char, next_last_char = 0;
 811         sed_cmd_t *sed_cmd;
 812         int substituted;
 813
 814         /* Prime the pump */
 815         next_line = get_next_line(&next_last_char);
 816
 817         /* go through every line in each file */
 818 again:
 819         substituted = 0;
 820
 821         /* Advance to next line.  Stop if out of lines. */
 822         pattern_space = next_line;
 823         if (!pattern_space) return;
 824         last_char = next_last_char;
 825
 826         /* Read one line in advance so we can act on the last line,
 827          * the '$' address */
 828         next_line = get_next_line(&next_last_char);
 829         linenum++;
 830 restart:
 831         /* for every line, go through all the commands */
 832         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 833                 int old_matched, matched;
 834
 835                 old_matched = sed_cmd->in_match;
 836
 837                 /* Determine if this command matches this line: */
 838
 839                 /* Are we continuing a previous multi-line match? */
 840                 sed_cmd->in_match = sed_cmd->in_match
 841                         /* Or is no range necessary? */
 842                         || (!sed_cmd->beg_line && !sed_cmd->end_line
 843                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
 844                         /* Or did we match the start of a numerical range? */
 845                         || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 846                         /* Or does this line match our begin address regex? */
 847                         || (sed_cmd->beg_match &&
 848                             !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 849                         /* Or did we match last line of input? */
 850                         || (sed_cmd->beg_line == -1 && next_line == NULL);
 851
 852                 /* Snapshot the value */
 853
 854                 matched = sed_cmd->in_match;
 855
 856                 /* Is this line the end of the current match? */
 857
 858                 if (matched) {
 859                         sed_cmd->in_match = !(
 860                                 /* has the ending line come, or is this a single address command? */
 861                                 (sed_cmd->end_line ?
 862                                         sed_cmd->end_line == -1 ?
 863                                                 !next_line
 864                                                 : (sed_cmd->end_line <= linenum)
 865                                         : !sed_cmd->end_match
 866                                 )
 867                                 /* or does this line matches our last address regex */
 868                                 || (sed_cmd->end_match && old_matched
 869                                      && (regexec(sed_cmd->end_match,
 870                                                  pattern_space, 0, NULL, 0) == 0))
 871                         );
 872                 }
 873
 874                 /* Skip blocks of commands we didn't match. */
 875                 if (sed_cmd->cmd == '{') {
 876                         if (sed_cmd->invert ? matched : !matched)
 877                                 while (sed_cmd && sed_cmd->cmd != '}')
 878                                         sed_cmd = sed_cmd->next;
 879                         if (!sed_cmd) bb_error_msg_and_die("unterminated {");
 880                         continue;
 881                 }
 882
 883                 /* Okay, so did this line match? */
 884                 if (sed_cmd->invert ? !matched : matched) {
 885                         /* Update last used regex in case a blank substitute BRE is found */
 886                         if (sed_cmd->beg_match) {
 887                                 bbg.previous_regex_ptr = sed_cmd->beg_match;
 888                         }
 889
 890                         /* actual sedding */
 891                         switch (sed_cmd->cmd) {
 892
 893                         /* Print line number */
 894                         case '=':
 895                                 fprintf(bbg.nonstdout, "%d\n", linenum);
 896                                 break;
 897
 898                         /* Write the current pattern space up to the first newline */
 899                         case 'P':
 900                         {
 901                                 char *tmp = strchr(pattern_space, '\n');
 902
 903                                 if (tmp) {
 904                                         *tmp = '\0';
 905                                         sed_puts(pattern_space, 1);
 906                                         *tmp = '\n';
 907                                         break;
 908                                 }
 909                                 /* Fall Through */
 910                         }
 911
 912                         /* Write the current pattern space to output */
 913                         case 'p':
 914                                 sed_puts(pattern_space, last_char);
 915                                 break;
 916                         /* Delete up through first newline */
 917                         case 'D':
 918                         {
 919                                 char *tmp = strchr(pattern_space, '\n');
 920
 921                                 if (tmp) {
 922                                         tmp = xstrdup(tmp+1);
 923                                         free(pattern_space);
 924                                         pattern_space = tmp;
 925                                         goto restart;
 926                                 }
 927                         }
 928                         /* discard this line. */
 929                         case 'd':
 930                                 goto discard_line;
 931
 932                         /* Substitute with regex */
 933                         case 's':
 934                                 if (!do_subst_command(sed_cmd, &pattern_space))
 935                                         break;
 936                                 substituted |= 1;
 937
 938                                 /* handle p option */
 939                                 if (sed_cmd->sub_p)
 940                                         sed_puts(pattern_space, last_char);
 941                                 /* handle w option */
 942                                 if (sed_cmd->file)
 943                                         sed_cmd->last_char = puts_maybe_newline(
 944                                                 pattern_space, sed_cmd->file,
 945                                                 sed_cmd->last_char, last_char);
 946                                 break;
 947
 948                         /* Append line to linked list to be printed later */
 949                         case 'a':
 950                                 append(sed_cmd->string);
 951                                 break;
 952
 953                         /* Insert text before this line */
 954                         case 'i':
 955                                 sed_puts(sed_cmd->string, 1);
 956                                 break;
 957
 958                         /* Cut and paste text (replace) */
 959                         case 'c':
 960                                 /* Only triggers on last line of a matching range. */
 961                                 if (!sed_cmd->in_match)
 962                                         sed_puts(sed_cmd->string, 0);
 963                                 goto discard_line;
 964
 965                         /* Read file, append contents to output */
 966                         case 'r':
 967                         {
 968                                 FILE *rfile;
 969
 970                                 rfile = fopen(sed_cmd->string, "r");
 971                                 if (rfile) {
 972                                         char *line;
 973
 974                                         while ((line = xmalloc_getline(rfile))
 975                                                         != NULL)
 976                                                 append(line);
 977                                         xprint_and_close_file(rfile);
 978                                 }
 979
 980                                 break;
 981                         }
 982
 983                         /* Write pattern space to file. */
 984                         case 'w':
 985                                 sed_cmd->last_char = puts_maybe_newline(
 986                                         pattern_space, sed_cmd->file,
 987                                         sed_cmd->last_char, last_char);
 988                                 break;
 989
 990                         /* Read next line from input */
 991                         case 'n':
 992                                 if (!bbg.be_quiet)
 993                                         sed_puts(pattern_space, last_char);
 994                                 if (next_line) {
 995                                         free(pattern_space);
 996                                         pattern_space = next_line;
 997                                         last_char = next_last_char;
 998                                         next_line = get_next_line(&next_last_char);
 999                                         linenum++;
1000                                         break;
1001                                 }
1002                                 /* fall through */
1003
1004                         /* Quit.  End of script, end of input. */
1005                         case 'q':
1006                                 /* Exit the outer while loop */
1007                                 free(next_line);
1008                                 next_line = NULL;
1009                                 goto discard_commands;
1010
1011                         /* Append the next line to the current line */
1012                         case 'N':
1013                         {
1014                                 int len;
1015                                 /* If no next line, jump to end of script and exit. */
1016                                 if (next_line == NULL) {
1017                                         /* Jump to end of script and exit */
1018                                         free(next_line);
1019                                         next_line = NULL;
1020                                         goto discard_line;
1021                                 /* append next_line, read new next_line. */
1022                                 }
1023                                 len = strlen(pattern_space);
1024                                 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1025                                 pattern_space[len] = '\n';
1026                                 strcpy(pattern_space + len+1, next_line);
1027                                 last_char = next_last_char;
1028                                 next_line = get_next_line(&next_last_char);
1029                                 linenum++;
1030                                 break;
1031                         }
1032
1033                         /* Test/branch if substitution occurred */
1034                         case 't':
1035                                 if (!substituted) break;
1036                                 substituted = 0;
1037                                 /* Fall through */
1038                         /* Test/branch if substitution didn't occur */
1039                         case 'T':
1040                                 if (substituted) break;
1041                                 /* Fall through */
1042                         /* Branch to label */
1043                         case 'b':
1044                                 if (!sed_cmd->string) goto discard_commands;
1045                                 else sed_cmd = branch_to(sed_cmd->string);
1046                                 break;
1047                         /* Transliterate characters */
1048                         case 'y':
1049                         {
1050                                 int i, j;
1051
1052                                 for (i = 0; pattern_space[i]; i++) {
1053                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1054                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1055                                                         pattern_space[i] = sed_cmd->string[j + 1];
1056                                                         break;
1057                                                 }
1058                                         }
1059                                 }
1060
1061                                 break;
1062                         }
1063                         case 'g':       /* Replace pattern space with hold space */
1064                                 free(pattern_space);
1065                                 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1066                                 break;
1067                         case 'G':       /* Append newline and hold space to pattern space */
1068                         {
1069                                 int pattern_space_size = 2;
1070                                 int hold_space_size = 0;
1071
1072                                 if (pattern_space)
1073                                         pattern_space_size += strlen(pattern_space);
1074                                 if (bbg.hold_space)
1075                                         hold_space_size = strlen(bbg.hold_space);
1076                                 pattern_space = xrealloc(pattern_space,
1077                                                 pattern_space_size + hold_space_size);
1078                                 if (pattern_space_size == 2)
1079                                         pattern_space[0] = 0;
1080                                 strcat(pattern_space, "\n");
1081                                 if (bbg.hold_space)
1082                                         strcat(pattern_space, bbg.hold_space);
1083                                 last_char = '\n';
1084
1085                                 break;
1086                         }
1087                         case 'h':       /* Replace hold space with pattern space */
1088                                 free(bbg.hold_space);
1089                                 bbg.hold_space = xstrdup(pattern_space);
1090                                 break;
1091                         case 'H':       /* Append newline and pattern space to hold space */
1092                         {
1093                                 int hold_space_size = 2;
1094                                 int pattern_space_size = 0;
1095
1096                                 if (bbg.hold_space)
1097                                         hold_space_size += strlen(bbg.hold_space);
1098                                 if (pattern_space)
1099                                         pattern_space_size = strlen(pattern_space);
1100                                 bbg.hold_space = xrealloc(bbg.hold_space,
1101                                                 hold_space_size + pattern_space_size);
1102
1103                                 if (hold_space_size == 2)
1104                                         *bbg.hold_space = 0;
1105                                 strcat(bbg.hold_space, "\n");
1106                                 if (pattern_space)
1107                                         strcat(bbg.hold_space, pattern_space);
1108
1109                                 break;
1110                         }
1111                         case 'x': /* Exchange hold and pattern space */
1112                         {
1113                                 char *tmp = pattern_space;
1114                                 pattern_space = bbg.hold_space ? : xzalloc(1);
1115                                 last_char = '\n';
1116                                 bbg.hold_space = tmp;
1117                                 break;
1118                         }
1119                         }
1120                 }
1121         }
1122
1123         /*
1124          * exit point from sedding...
1125          */
1126 discard_commands:
1127         /* we will print the line unless we were told to be quiet ('-n')
1128            or if the line was suppressed (ala 'd'elete) */
1129         if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1130
1131         /* Delete and such jump here. */
1132 discard_line:
1133         flush_append();
1134         free(pattern_space);
1135
1136         goto again;
1137 }
1138
1139 /* It is possible to have a command line argument with embedded
1140    newlines.  This counts as multiple command lines. */
1141
1142 static void add_cmd_block(char *cmdstr)
1143 {
1144         int go = 1;
1145         char *temp = xstrdup(cmdstr), *temp2 = temp;
1146
1147         while (go) {
1148                 int len = strcspn(temp2, "\n");
1149                 if (!temp2[len]) go = 0;
1150                 else temp2[len] = 0;
1151                 add_cmd(temp2);
1152                 temp2 += len+1;
1153         }
1154         free(temp);
1155 }
1156
1157 static void add_cmds_link(llist_t *opt_e)
1158 {
1159         if (!opt_e) return;
1160         add_cmds_link(opt_e->link);
1161         add_cmd_block(opt_e->data);
1162         free(opt_e);
1163 }
1164
1165 static void add_files_link(llist_t *opt_f)
1166 {
1167         char *line;
1168         FILE *cmdfile;
1169         if (!opt_f) return;
1170         add_files_link(opt_f->link);
1171         cmdfile = xfopen(opt_f->data, "r");
1172         while ((line = xmalloc_getline(cmdfile)) != NULL) {
1173                 add_cmd(line);
1174                 free(line);
1175         }
1176         xprint_and_close_file(cmdfile);
1177         free(opt_f);
1178 }
1179
1180 int sed_main(int argc, char **argv)
1181 {
1182         enum {
1183                 OPT_in_place = 1 << 0,
1184         };
1185         unsigned opt;
1186         llist_t *opt_e, *opt_f;
1187         int status = EXIT_SUCCESS;
1188
1189         bbg.sed_cmd_tail = &bbg.sed_cmd_head;
1190
1191         /* destroy command strings on exit */
1192         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1193
1194         /* Lie to autoconf when it starts asking stupid questions. */
1195         if (argc == 2 && !strcmp(argv[1], "--version")) {
1196                 puts("This is not GNU sed version 4.0");
1197                 return 0;
1198         }
1199
1200         /* do normal option parsing */
1201         opt_e = opt_f = NULL;
1202         opt_complementary = "e::f::" /* can occur multiple times */
1203                             "nn"; /* count -n */
1204         opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f,
1205                             &bbg.be_quiet); /* counter for -n */
1206         argc -= optind;
1207         argv += optind;
1208         if (opt & OPT_in_place) { // -i
1209                 atexit(cleanup_outname);
1210         }
1211         if (opt & 0x2) bbg.regex_type |= REG_EXTENDED; // -r
1212         //if (opt & 0x4) bbg.be_quiet++; // -n
1213         if (opt & 0x8) { // -e
1214                 /* getopt32 reverses order of arguments, handle it */
1215                 add_cmds_link(opt_e);
1216         }
1217         if (opt & 0x10) { // -f
1218                 /* getopt32 reverses order of arguments, handle it */
1219                 add_files_link(opt_f);
1220         }
1221         /* if we didn't get a pattern from -e or -f, use argv[0] */
1222         if (!(opt & 0x18)) {
1223                 if (!argc)
1224                         bb_show_usage();
1225                 add_cmd_block(*argv++);
1226                 argc--;
1227         }
1228         /* Flush any unfinished commands. */
1229         add_cmd("");
1230
1231         /* By default, we write to stdout */
1232         bbg.nonstdout = stdout;
1233
1234         /* argv[0..(argc-1)] should be names of file to process. If no
1235          * files were specified or '-' was specified, take input from stdin.
1236          * Otherwise, we process all the files specified. */
1237         if (argv[0] == NULL) {
1238                 if (opt & OPT_in_place)
1239                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1240                 add_input_file(stdin);
1241                 process_files();
1242         } else {
1243                 int i;
1244                 FILE *file;
1245
1246                 for (i = 0; i < argc; i++) {
1247                         struct stat statbuf;
1248                         int nonstdoutfd;
1249
1250                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1251                                 add_input_file(stdin);
1252                                 process_files();
1253                                 continue;
1254                         }
1255                         file = fopen_or_warn(argv[i], "r");
1256                         if (!file) {
1257                                 status = EXIT_FAILURE;
1258                                 continue;
1259                         }
1260                         if (!(opt & OPT_in_place)) {
1261                                 add_input_file(file);
1262                                 continue;
1263                         }
1264
1265                         bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1266                         nonstdoutfd = mkstemp(bbg.outname);
1267                         if (-1 == nonstdoutfd)
1268                                 bb_error_msg_and_die("no temp file");
1269                         bbg.nonstdout = fdopen(nonstdoutfd, "w");
1270
1271                         /* Set permissions of output file */
1272
1273                         fstat(fileno(file), &statbuf);
1274                         fchmod(nonstdoutfd, statbuf.st_mode);
1275                         add_input_file(file);
1276                         process_files();
1277                         fclose(bbg.nonstdout);
1278
1279                         bbg.nonstdout = stdout;
1280                         /* unlink(argv[i]); */
1281                         // FIXME: error check / message?
1282                         rename(bbg.outname, argv[i]);
1283                         free(bbg.outname);
1284                         bbg.outname = 0;
1285                 }
1286                 if (bbg.input_file_count > bbg.current_input_file)
1287                         process_files();
1288         }
1289
1290         return status;
1291 }