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 void sed_free_and_close_stuff(void);
 124 #if ENABLE_FEATURE_CLEAN_UP
 125 static 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 = -delimiter;
 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                 delimiter = '/';
 279                 if (*my_str == '\\') delimiter = *++pos;
 280                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 281                 temp = copy_parsing_escapes(pos, next);
 282                 *regex = 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
 301         /* If lines glued together, put backslash back. */
 302         if (filecmdstr[idx] == '\n') hack = 1;
 303         if (idx == start)
 304                 bb_error_msg_and_die("empty filename");
 305         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 306         if (hack) (*retval)[idx] = '\\';
 307
 308         return idx;
 309 }
 310
 311 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
 312 {
 313         int cflags = bbg.regex_type;
 314         char *match;
 315         int idx = 0;
 316
 317         /*
 318          * A substitution command should look something like this:
 319          *    s/match/replace/ #gIpw
 320          *    ||     |        |||
 321          *    mandatory       optional
 322          */
 323         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 324
 325         /* determine the number of back references in the match string */
 326         /* Note: we compute this here rather than in the do_subst_command()
 327          * function to save processor time, at the expense of a little more memory
 328          * (4 bits) per sed_cmd */
 329
 330         /* process the flags */
 331
 332         sed_cmd->which_match = 1;
 333         while (substr[++idx]) {
 334                 /* Parse match number */
 335                 if (isdigit(substr[idx])) {
 336                         if (match[0] != '^') {
 337                                 /* Match 0 treated as all, multiple matches we take the last one. */
 338                                 char *pos = substr + idx;
 339                                 /* FIXME: error check? */
 340                                 sed_cmd->which_match = (unsigned short)strtol(substr+idx, &pos, 10);
 341                                 idx = pos - substr;
 342                         }
 343                         continue;
 344                 }
 345                 /* Skip spaces */
 346                 if (isspace(substr[idx])) continue;
 347
 348                 switch (substr[idx]) {
 349                 /* Replace all occurrences */
 350                 case 'g':
 351                         if (match[0] != '^') sed_cmd->which_match = 0;
 352                         break;
 353                 /* Print pattern space */
 354                 case 'p':
 355                         sed_cmd->sub_p = 1;
 356                         break;
 357                 /* Write to file */
 358                 case 'w':
 359                 {
 360                         char *temp;
 361                         idx += parse_file_cmd(sed_cmd, substr+idx, &temp);
 362
 363                         break;
 364                 }
 365                 /* Ignore case (gnu exension) */
 366                 case 'I':
 367                         cflags |= REG_ICASE;
 368                         break;
 369                 /* Comment */
 370                 case '#':
 371                         while (substr[++idx]) /*skip all*/;
 372                         /* Fall through */
 373                 /* End of command */
 374                 case ';':
 375                 case '}':
 376                         goto out;
 377                 default:
 378                         bb_error_msg_and_die("bad option in substitution expression");
 379                 }
 380         }
 381 out:
 382         /* compile the match string into a regex */
 383         if (*match != '\0') {
 384                 /* If match is empty, we use last regex used at runtime */
 385                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 386                 xregcomp(sed_cmd->sub_match, match, cflags);
 387         }
 388         free(match);
 389
 390         return idx;
 391 }
 392
 393 /*
 394  *  Process the commands arguments
 395  */
 396 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 397 {
 398         /* handle (s)ubstitution command */
 399         if (sed_cmd->cmd == 's')
 400                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 401         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 402         else if (strchr("aic", sed_cmd->cmd)) {
 403                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 404                         bb_error_msg_and_die
 405                                 ("only a beginning address can be specified for edit commands");
 406                 for (;;) {
 407                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 408                                 cmdstr++;
 409                                 break;
 410                         } else if (isspace(*cmdstr))
 411                                 cmdstr++;
 412                         else
 413                                 break;
 414                 }
 415                 sed_cmd->string = xstrdup(cmdstr);
 416                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), 0, 0);
 417                 cmdstr += strlen(cmdstr);
 418         /* handle file cmds: (r)ead */
 419         } else if (strchr("rw", sed_cmd->cmd)) {
 420                 if (sed_cmd->end_line || sed_cmd->end_match)
 421                         bb_error_msg_and_die("command only uses one address");
 422                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 423                 if (sed_cmd->cmd == 'w')
 424                         sed_cmd->file = xfopen(sed_cmd->string, "w");
 425         /* handle branch commands */
 426         } else if (strchr(":btT", sed_cmd->cmd)) {
 427                 int length;
 428
 429                 cmdstr = skip_whitespace(cmdstr);
 430                 length = strcspn(cmdstr, semicolon_whitespace);
 431                 if (length) {
 432                         sed_cmd->string = xstrndup(cmdstr, length);
 433                         cmdstr += length;
 434                 }
 435         }
 436         /* translation command */
 437         else if (sed_cmd->cmd == 'y') {
 438                 char *match, *replace;
 439                 int i = cmdstr[0];
 440
 441                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 442                 /* \n already parsed, but \delimiter needs unescaping. */
 443                 parse_escapes(match, match, strlen(match), i, i);
 444                 parse_escapes(replace, replace, strlen(replace), i, i);
 445
 446                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 447                 for (i = 0; match[i] && replace[i]; i++) {
 448                         sed_cmd->string[i*2] = match[i];
 449                         sed_cmd->string[i*2+1] = replace[i];
 450                 }
 451                 free(match);
 452                 free(replace);
 453         }
 454         /* if it wasnt a single-letter command that takes no arguments
 455          * then it must be an invalid command.
 456          */
 457         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 458                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 459         }
 460
 461         /* give back whatever's left over */
 462         return cmdstr;
 463 }
 464
 465
 466 /* Parse address+command sets, skipping comment lines. */
 467
 468 static void add_cmd(char *cmdstr)
 469 {
 470         sed_cmd_t *sed_cmd;
 471         int temp;
 472
 473         /* Append this line to any unfinished line from last time. */
 474         if (bbg.add_cmd_line) {
 475                 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 476                 free(bbg.add_cmd_line);
 477                 bbg.add_cmd_line = cmdstr;
 478         }
 479
 480         /* If this line ends with backslash, request next line. */
 481         temp = strlen(cmdstr);
 482         if (temp && cmdstr[temp-1] == '\\') {
 483                 if (!bbg.add_cmd_line)
 484                         bbg.add_cmd_line = xstrdup(cmdstr);
 485                 bbg.add_cmd_line[temp-1] = 0;
 486                 return;
 487         }
 488
 489         /* Loop parsing all commands in this line. */
 490         while (*cmdstr) {
 491                 /* Skip leading whitespace and semicolons */
 492                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 493
 494                 /* If no more commands, exit. */
 495                 if (!*cmdstr) break;
 496
 497                 /* if this is a comment, jump past it and keep going */
 498                 if (*cmdstr == '#') {
 499                         /* "#n" is the same as using -n on the command line */
 500                         if (cmdstr[1] == 'n')
 501                                 bbg.be_quiet++;
 502                         cmdstr = strpbrk(cmdstr, "\n\r");
 503                         if (!cmdstr) break;
 504                         continue;
 505                 }
 506
 507                 /* parse the command
 508                  * format is: [addr][,addr][!]cmd
 509                  *            |----||-----||-|
 510                  *            part1 part2  part3
 511                  */
 512
 513                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 514
 515                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 516                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 517
 518                 /* second part (if present) will begin with a comma */
 519                 if (*cmdstr == ',') {
 520                         int idx;
 521
 522                         cmdstr++;
 523                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 524                         if (!idx)
 525                                 bb_error_msg_and_die("no address after comma");
 526                         cmdstr += idx;
 527                 }
 528
 529                 /* skip whitespace before the command */
 530                 cmdstr = skip_whitespace(cmdstr);
 531
 532                 /* Check for inversion flag */
 533                 if (*cmdstr == '!') {
 534                         sed_cmd->invert = 1;
 535                         cmdstr++;
 536
 537                         /* skip whitespace before the command */
 538                         cmdstr = skip_whitespace(cmdstr);
 539                 }
 540
 541                 /* last part (mandatory) will be a command */
 542                 if (!*cmdstr)
 543                         bb_error_msg_and_die("missing command");
 544                 sed_cmd->cmd = *(cmdstr++);
 545                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 546
 547                 /* Add the command to the command array */
 548                 bbg.sed_cmd_tail->next = sed_cmd;
 549                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 550         }
 551
 552         /* If we glued multiple lines together, free the memory. */
 553         free(bbg.add_cmd_line);
 554         bbg.add_cmd_line = NULL;
 555 }
 556
 557 /* Append to a string, reallocating memory as necessary. */
 558
 559 #define PIPE_GROW 64
 560
 561 static void pipe_putc(char c)
 562 {
 563         if (bbg.pipeline.idx == bbg.pipeline.len) {
 564                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 565                                 bbg.pipeline.len + PIPE_GROW);
 566                 bbg.pipeline.len += PIPE_GROW;
 567         }
 568         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 569 }
 570
 571 static void do_subst_w_backrefs(char *line, char *replace)
 572 {
 573         int i,j;
 574
 575         /* go through the replacement string */
 576         for (i = 0; replace[i]; i++) {
 577                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 578                 if (replace[i] == '\\' && replace[i+1] >= '0' && replace[i+1] <= '9') {
 579                         int backref = replace[++i]-'0';
 580
 581                         /* print out the text held in bbg.regmatch[backref] */
 582                         if (bbg.regmatch[backref].rm_so != -1) {
 583                                 j = bbg.regmatch[backref].rm_so;
 584                                 while (j < bbg.regmatch[backref].rm_eo)
 585                                         pipe_putc(line[j++]);
 586                         }
 587                 }
 588
 589                 /* if we find a backslash escaped character, print the character */
 590                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 591
 592                 /* if we find an unescaped '&' print out the whole matched text. */
 593                 else if (replace[i] == '&') {
 594                         j = bbg.regmatch[0].rm_so;
 595                         while (j < bbg.regmatch[0].rm_eo)
 596                                 pipe_putc(line[j++]);
 597                 }
 598                 /* Otherwise just output the character. */
 599                 else pipe_putc(replace[i]);
 600         }
 601 }
 602
 603 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 604 {
 605         char *oldline = *line;
 606         int altered = 0;
 607         int match_count = 0;
 608         regex_t *current_regex;
 609
 610         /* Handle empty regex. */
 611         if (sed_cmd->sub_match == NULL) {
 612                 current_regex = bbg.previous_regex_ptr;
 613                 if (!current_regex)
 614                         bb_error_msg_and_die("no previous regexp");
 615         } else
 616                 bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 617
 618         /* Find the first match */
 619         if (REG_NOMATCH == regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 620                 return 0;
 621
 622         /* Initialize temporary output buffer. */
 623         bbg.pipeline.buf = xmalloc(PIPE_GROW);
 624         bbg.pipeline.len = PIPE_GROW;
 625         bbg.pipeline.idx = 0;
 626
 627         /* Now loop through, substituting for matches */
 628         do {
 629                 int i;
 630
 631                 /* Work around bug in glibc regexec, demonstrated by:
 632                    echo " a.b" | busybox sed 's [^ .]* x g'
 633                    The match_count check is so not to break
 634                    echo "hi" | busybox sed 's/^/!/g' */
 635                 if (!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
 636                         pipe_putc(*oldline++);
 637                         continue;
 638                 }
 639
 640                 match_count++;
 641
 642                 /* If we aren't interested in this match, output old line to
 643                    end of match and continue */
 644                 if (sed_cmd->which_match && sed_cmd->which_match != match_count) {
 645                         for (i = 0; i < bbg.regmatch[0].rm_eo; i++)
 646                                 pipe_putc(*oldline++);
 647                         continue;
 648                 }
 649
 650                 /* print everything before the match */
 651                 for (i = 0; i < bbg.regmatch[0].rm_so; i++)
 652                         pipe_putc(oldline[i]);
 653
 654                 /* then print the substitution string */
 655                 do_subst_w_backrefs(oldline, sed_cmd->string);
 656
 657                 /* advance past the match */
 658                 oldline += bbg.regmatch[0].rm_eo;
 659                 /* flag that something has changed */
 660                 altered++;
 661
 662                 /* if we're not doing this globally, get out now */
 663                 if (sed_cmd->which_match) break;
 664         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 665
 666         /* Copy rest of string into output pipeline */
 667
 668         while (*oldline)
 669                 pipe_putc(*oldline++);
 670         pipe_putc(0);
 671
 672         free(*line);
 673         *line = bbg.pipeline.buf;
 674         return altered;
 675 }
 676
 677 /* Set command pointer to point to this label.  (Does not handle null label.) */
 678 static sed_cmd_t *branch_to(char *label)
 679 {
 680         sed_cmd_t *sed_cmd;
 681
 682         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 683                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 684                         return sed_cmd;
 685                 }
 686         }
 687         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 688 }
 689
 690 static void append(char *s)
 691 {
 692         llist_add_to_end(&bbg.append_head, xstrdup(s));
 693 }
 694
 695 static void flush_append(void)
 696 {
 697         char *data;
 698
 699         /* Output appended lines. */
 700         while ((data = (char *)llist_pop(&bbg.append_head))) {
 701                 fprintf(bbg.nonstdout, "%s\n", data);
 702                 free(data);
 703         }
 704 }
 705
 706 static void add_input_file(FILE *file)
 707 {
 708         bbg.input_file_list = xrealloc(bbg.input_file_list,
 709                         (bbg.input_file_count + 1) * sizeof(FILE *));
 710         bbg.input_file_list[bbg.input_file_count++] = file;
 711 }
 712
 713 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 714  * noting if we ran out of files without a newline on the last line we read.
 715  */
 716 static char *get_next_line(int *last_char)
 717 {
 718         char *temp = NULL;
 719         int len, lc;
 720
 721         lc = 0;
 722         flush_append();
 723         while (bbg.current_input_file < bbg.input_file_count) {
 724                 temp = bb_get_chunk_from_file(
 725                         bbg.input_file_list[bbg.current_input_file], &len);
 726                 if (temp) {
 727                         /* len > 0 here, it's ok to do temp[len-1] */
 728                         char c = temp[len-1];
 729                         if (c == '\n' || c == '\0') {
 730                                 temp[len-1] = '\0';
 731                                 lc |= (unsigned char)c;
 732                                 break;
 733                         }
 734                         /* will be returned if last line in the file
 735                          * doesn't end with either '\n' or '\0' */
 736                         lc |= 0x100;
 737                         break;
 738                 }
 739                 /* Close this file and advance to next one */
 740                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 741                 /* "this is the first line from new input file" */
 742                 lc |= 0x200;
 743         }
 744         *last_char = lc;
 745         return temp;
 746 }
 747
 748 /* Output line of text. */
 749 /* Note:
 750  * The tricks with 0x200 and last_puts_char are there to emulate gnu sed.
 751  * Without them, we had this:
 752  * echo -n thingy >z1
 753  * echo -n again >z2
 754  * >znull
 755  * sed "s/i/z/" z1 z2 znull | hexdump -vC output:
 756  * gnu sed 4.1.5:
 757  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 758  * bbox:
 759  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 760  */
 761
 762 static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_char)
 763 {
 764         static char last_puts_char;
 765
 766         /* Is this a first line from new file
 767          * and old file didn't end with '\n'? */
 768         if ((last_char & 0x200) && last_puts_char != '\n') {
 769                 fputc('\n', file);
 770                 last_puts_char = '\n';
 771         }
 772         fputs(s, file);
 773         /* 'x': we don't care what is it, but we know it isn't '\n' */
 774         if (s[0]) last_puts_char = 'x';
 775         if (!(last_char & 0x100)) { /* had trailing '\n' or '\0'? */
 776                 last_char &= 0xff;
 777                 fputc(last_char, file);
 778                 last_puts_char = last_char;
 779         }
 780
 781         if (ferror(file)) {
 782                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 783                 bb_error_msg_and_die(bb_msg_write_error);
 784         }
 785
 786         return last_char;
 787 }
 788
 789 #define sed_puts(s, n) \
 790         (prev_last_char = puts_maybe_newline(s, bbg.nonstdout, prev_last_char, n))
 791
 792 /* Process all the lines in all the files */
 793
 794 static void process_files(void)
 795 {
 796         char *pattern_space, *next_line;
 797         int linenum = 0, prev_last_char = 0;
 798         int last_char, next_last_char = 0;
 799         sed_cmd_t *sed_cmd;
 800         int substituted;
 801
 802         /* Prime the pump */
 803         next_line = get_next_line(&next_last_char);
 804
 805         /* go through every line in each file */
 806 again:
 807         substituted = 0;
 808
 809         /* Advance to next line.  Stop if out of lines. */
 810         pattern_space = next_line;
 811         if (!pattern_space) return;
 812         last_char = next_last_char;
 813
 814         /* Read one line in advance so we can act on the last line,
 815          * the '$' address */
 816         next_line = get_next_line(&next_last_char);
 817         linenum++;
 818 restart:
 819         /* for every line, go through all the commands */
 820         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 821                 int old_matched, matched;
 822
 823                 old_matched = sed_cmd->in_match;
 824
 825                 /* Determine if this command matches this line: */
 826
 827                 /* Are we continuing a previous multi-line match? */
 828                 sed_cmd->in_match = sed_cmd->in_match
 829                         /* Or is no range necessary? */
 830                         || (!sed_cmd->beg_line && !sed_cmd->end_line
 831                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
 832                         /* Or did we match the start of a numerical range? */
 833                         || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 834                         /* Or does this line match our begin address regex? */
 835                         || (sed_cmd->beg_match &&
 836                             !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 837                         /* Or did we match last line of input? */
 838                         || (sed_cmd->beg_line == -1 && next_line == NULL);
 839
 840                 /* Snapshot the value */
 841
 842                 matched = sed_cmd->in_match;
 843
 844                 /* Is this line the end of the current match? */
 845
 846                 if (matched) {
 847                         sed_cmd->in_match = !(
 848                                 /* has the ending line come, or is this a single address command? */
 849                                 (sed_cmd->end_line ?
 850                                         sed_cmd->end_line == -1 ?
 851                                                 !next_line
 852                                                 : (sed_cmd->end_line <= linenum)
 853                                         : !sed_cmd->end_match
 854                                 )
 855                                 /* or does this line matches our last address regex */
 856                                 || (sed_cmd->end_match && old_matched
 857                                      && (regexec(sed_cmd->end_match,
 858                                                  pattern_space, 0, NULL, 0) == 0))
 859                         );
 860                 }
 861
 862                 /* Skip blocks of commands we didn't match. */
 863                 if (sed_cmd->cmd == '{') {
 864                         if (sed_cmd->invert ? matched : !matched)
 865                                 while (sed_cmd && sed_cmd->cmd != '}')
 866                                         sed_cmd = sed_cmd->next;
 867                         if (!sed_cmd) bb_error_msg_and_die("unterminated {");
 868                         continue;
 869                 }
 870
 871                 /* Okay, so did this line match? */
 872                 if (sed_cmd->invert ? !matched : matched) {
 873                         /* Update last used regex in case a blank substitute BRE is found */
 874                         if (sed_cmd->beg_match) {
 875                                 bbg.previous_regex_ptr = sed_cmd->beg_match;
 876                         }
 877
 878                         /* actual sedding */
 879                         switch (sed_cmd->cmd) {
 880
 881                         /* Print line number */
 882                         case '=':
 883                                 fprintf(bbg.nonstdout, "%d\n", linenum);
 884                                 break;
 885
 886                         /* Write the current pattern space up to the first newline */
 887                         case 'P':
 888                         {
 889                                 char *tmp = strchr(pattern_space, '\n');
 890
 891                                 if (tmp) {
 892                                         *tmp = '\0';
 893                                         sed_puts(pattern_space, 1);
 894                                         *tmp = '\n';
 895                                         break;
 896                                 }
 897                                 /* Fall Through */
 898                         }
 899
 900                         /* Write the current pattern space to output */
 901                         case 'p':
 902                                 sed_puts(pattern_space, last_char);
 903                                 break;
 904                         /* Delete up through first newline */
 905                         case 'D':
 906                         {
 907                                 char *tmp = strchr(pattern_space, '\n');
 908
 909                                 if (tmp) {
 910                                         tmp = xstrdup(tmp+1);
 911                                         free(pattern_space);
 912                                         pattern_space = tmp;
 913                                         goto restart;
 914                                 }
 915                         }
 916                         /* discard this line. */
 917                         case 'd':
 918                                 goto discard_line;
 919
 920                         /* Substitute with regex */
 921                         case 's':
 922                                 if (!do_subst_command(sed_cmd, &pattern_space))
 923                                         break;
 924                                 substituted |= 1;
 925
 926                                 /* handle p option */
 927                                 if (sed_cmd->sub_p)
 928                                         sed_puts(pattern_space, last_char);
 929                                 /* handle w option */
 930                                 if (sed_cmd->file)
 931                                         sed_cmd->last_char = puts_maybe_newline(
 932                                                 pattern_space, sed_cmd->file,
 933                                                 sed_cmd->last_char, last_char);
 934                                 break;
 935
 936                         /* Append line to linked list to be printed later */
 937                         case 'a':
 938                                 append(sed_cmd->string);
 939                                 break;
 940
 941                         /* Insert text before this line */
 942                         case 'i':
 943                                 sed_puts(sed_cmd->string, 1);
 944                                 break;
 945
 946                         /* Cut and paste text (replace) */
 947                         case 'c':
 948                                 /* Only triggers on last line of a matching range. */
 949                                 if (!sed_cmd->in_match)
 950                                         sed_puts(sed_cmd->string, 0);
 951                                 goto discard_line;
 952
 953                         /* Read file, append contents to output */
 954                         case 'r':
 955                         {
 956                                 FILE *rfile;
 957
 958                                 rfile = fopen(sed_cmd->string, "r");
 959                                 if (rfile) {
 960                                         char *line;
 961
 962                                         while ((line = xmalloc_getline(rfile))
 963                                                         != NULL)
 964                                                 append(line);
 965                                         xprint_and_close_file(rfile);
 966                                 }
 967
 968                                 break;
 969                         }
 970
 971                         /* Write pattern space to file. */
 972                         case 'w':
 973                                 sed_cmd->last_char = puts_maybe_newline(
 974                                         pattern_space, sed_cmd->file,
 975                                         sed_cmd->last_char, last_char);
 976                                 break;
 977
 978                         /* Read next line from input */
 979                         case 'n':
 980                                 if (!bbg.be_quiet)
 981                                         sed_puts(pattern_space, last_char);
 982                                 if (next_line) {
 983                                         free(pattern_space);
 984                                         pattern_space = next_line;
 985                                         last_char = next_last_char;
 986                                         next_line = get_next_line(&next_last_char);
 987                                         linenum++;
 988                                         break;
 989                                 }
 990                                 /* fall through */
 991
 992                         /* Quit.  End of script, end of input. */
 993                         case 'q':
 994                                 /* Exit the outer while loop */
 995                                 free(next_line);
 996                                 next_line = NULL;
 997                                 goto discard_commands;
 998
 999                         /* Append the next line to the current line */
1000                         case 'N':
1001                         {
1002                                 int len;
1003                                 /* If no next line, jump to end of script and exit. */
1004                                 if (next_line == NULL) {
1005                                         /* Jump to end of script and exit */
1006                                         free(next_line);
1007                                         next_line = NULL;
1008                                         goto discard_line;
1009                                 /* append next_line, read new next_line. */
1010                                 }
1011                                 len = strlen(pattern_space);
1012                                 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1013                                 pattern_space[len] = '\n';
1014                                 strcpy(pattern_space + len+1, next_line);
1015                                 last_char = next_last_char;
1016                                 next_line = get_next_line(&next_last_char);
1017                                 linenum++;
1018                                 break;
1019                         }
1020
1021                         /* Test/branch if substitution occurred */
1022                         case 't':
1023                                 if (!substituted) break;
1024                                 substituted = 0;
1025                                 /* Fall through */
1026                         /* Test/branch if substitution didn't occur */
1027                         case 'T':
1028                                 if (substituted) break;
1029                                 /* Fall through */
1030                         /* Branch to label */
1031                         case 'b':
1032                                 if (!sed_cmd->string) goto discard_commands;
1033                                 else sed_cmd = branch_to(sed_cmd->string);
1034                                 break;
1035                         /* Transliterate characters */
1036                         case 'y':
1037                         {
1038                                 int i, j;
1039
1040                                 for (i = 0; pattern_space[i]; i++) {
1041                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1042                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1043                                                         pattern_space[i] = sed_cmd->string[j + 1];
1044                                                         break;
1045                                                 }
1046                                         }
1047                                 }
1048
1049                                 break;
1050                         }
1051                         case 'g':       /* Replace pattern space with hold space */
1052                                 free(pattern_space);
1053                                 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1054                                 break;
1055                         case 'G':       /* Append newline and hold space to pattern space */
1056                         {
1057                                 int pattern_space_size = 2;
1058                                 int hold_space_size = 0;
1059
1060                                 if (pattern_space)
1061                                         pattern_space_size += strlen(pattern_space);
1062                                 if (bbg.hold_space)
1063                                         hold_space_size = strlen(bbg.hold_space);
1064                                 pattern_space = xrealloc(pattern_space,
1065                                                 pattern_space_size + hold_space_size);
1066                                 if (pattern_space_size == 2)
1067                                         pattern_space[0] = 0;
1068                                 strcat(pattern_space, "\n");
1069                                 if (bbg.hold_space)
1070                                         strcat(pattern_space, bbg.hold_space);
1071                                 last_char = '\n';
1072
1073                                 break;
1074                         }
1075                         case 'h':       /* Replace hold space with pattern space */
1076                                 free(bbg.hold_space);
1077                                 bbg.hold_space = xstrdup(pattern_space);
1078                                 break;
1079                         case 'H':       /* Append newline and pattern space to hold space */
1080                         {
1081                                 int hold_space_size = 2;
1082                                 int pattern_space_size = 0;
1083
1084                                 if (bbg.hold_space)
1085                                         hold_space_size += strlen(bbg.hold_space);
1086                                 if (pattern_space)
1087                                         pattern_space_size = strlen(pattern_space);
1088                                 bbg.hold_space = xrealloc(bbg.hold_space,
1089                                                 hold_space_size + pattern_space_size);
1090
1091                                 if (hold_space_size == 2)
1092                                         *bbg.hold_space = 0;
1093                                 strcat(bbg.hold_space, "\n");
1094                                 if (pattern_space)
1095                                         strcat(bbg.hold_space, pattern_space);
1096
1097                                 break;
1098                         }
1099                         case 'x': /* Exchange hold and pattern space */
1100                         {
1101                                 char *tmp = pattern_space;
1102                                 pattern_space = bbg.hold_space ? : xzalloc(1);
1103                                 last_char = '\n';
1104                                 bbg.hold_space = tmp;
1105                                 break;
1106                         }
1107                         }
1108                 }
1109         }
1110
1111         /*
1112          * exit point from sedding...
1113          */
1114 discard_commands:
1115         /* we will print the line unless we were told to be quiet ('-n')
1116            or if the line was suppressed (ala 'd'elete) */
1117         if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1118
1119         /* Delete and such jump here. */
1120 discard_line:
1121         flush_append();
1122         free(pattern_space);
1123
1124         goto again;
1125 }
1126
1127 /* It is possible to have a command line argument with embedded
1128    newlines.  This counts as multiple command lines. */
1129
1130 static void add_cmd_block(char *cmdstr)
1131 {
1132         int go = 1;
1133         char *temp = xstrdup(cmdstr), *temp2 = temp;
1134
1135         while (go) {
1136                 int len = strcspn(temp2, "\n");
1137                 if (!temp2[len]) go = 0;
1138                 else temp2[len] = 0;
1139                 add_cmd(temp2);
1140                 temp2 += len+1;
1141         }
1142         free(temp);
1143 }
1144
1145 static void add_cmds_link(llist_t *opt_e)
1146 {
1147         if (!opt_e) return;
1148         add_cmds_link(opt_e->link);
1149         add_cmd_block(opt_e->data);
1150         free(opt_e);
1151 }
1152
1153 static void add_files_link(llist_t *opt_f)
1154 {
1155         char *line;
1156         FILE *cmdfile;
1157         if (!opt_f) return;
1158         add_files_link(opt_f->link);
1159         cmdfile = xfopen(opt_f->data, "r");
1160         while ((line = xmalloc_getline(cmdfile)) != NULL) {
1161                 add_cmd(line);
1162                 free(line);
1163         }
1164         xprint_and_close_file(cmdfile);
1165         free(opt_f);
1166 }
1167
1168 int sed_main(int argc, char **argv)
1169 {
1170         enum {
1171                 OPT_in_place = 1 << 0,
1172         };
1173         unsigned opt;
1174         llist_t *opt_e, *opt_f;
1175         int status = EXIT_SUCCESS;
1176
1177         bbg.sed_cmd_tail = &bbg.sed_cmd_head;
1178
1179         /* destroy command strings on exit */
1180         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1181
1182         /* Lie to autoconf when it starts asking stupid questions. */
1183         if (argc == 2 && !strcmp(argv[1], "--version")) {
1184                 puts("This is not GNU sed version 4.0");
1185                 return 0;
1186         }
1187
1188         /* do normal option parsing */
1189         opt_e = opt_f = NULL;
1190         opt_complementary = "e::f::" /* can occur multiple times */
1191                             "nn"; /* count -n */
1192         opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f,
1193                             &bbg.be_quiet); /* counter for -n */
1194         argc -= optind;
1195         argv += optind;
1196         if (opt & OPT_in_place) { // -i
1197                 atexit(cleanup_outname);
1198         }
1199         if (opt & 0x2) bbg.regex_type |= REG_EXTENDED; // -r
1200         //if (opt & 0x4) bbg.be_quiet++; // -n
1201         if (opt & 0x8) { // -e
1202                 /* getopt32 reverses order of arguments, handle it */
1203                 add_cmds_link(opt_e);
1204         }
1205         if (opt & 0x10) { // -f
1206                 /* getopt32 reverses order of arguments, handle it */
1207                 add_files_link(opt_f);
1208         }
1209         /* if we didn't get a pattern from -e or -f, use argv[0] */
1210         if (!(opt & 0x18)) {
1211                 if (!argc)
1212                         bb_show_usage();
1213                 add_cmd_block(*argv++);
1214                 argc--;
1215         }
1216         /* Flush any unfinished commands. */
1217         add_cmd("");
1218
1219         /* By default, we write to stdout */
1220         bbg.nonstdout = stdout;
1221
1222         /* argv[0..(argc-1)] should be names of file to process. If no
1223          * files were specified or '-' was specified, take input from stdin.
1224          * Otherwise, we process all the files specified. */
1225         if (argv[0] == NULL) {
1226                 if (opt & OPT_in_place)
1227                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1228                 add_input_file(stdin);
1229                 process_files();
1230         } else {
1231                 int i;
1232                 FILE *file;
1233
1234                 for (i = 0; i < argc; i++) {
1235                         struct stat statbuf;
1236                         int nonstdoutfd;
1237
1238                         if (argv[i][0] == '-' && !argv[i][1]
1239                          && !(opt & OPT_in_place)
1240                         ) {
1241                                 add_input_file(stdin);
1242                                 process_files();
1243                                 continue;
1244                         }
1245                         file = fopen_or_warn(argv[i], "r");
1246                         if (!file) {
1247                                 status = EXIT_FAILURE;
1248                                 continue;
1249                         }
1250                         if (!(opt & OPT_in_place)) {
1251                                 add_input_file(file);
1252                                 continue;
1253                         }
1254
1255                         bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1256                         nonstdoutfd = mkstemp(bbg.outname);
1257                         if (-1 == nonstdoutfd)
1258                                 bb_error_msg_and_die("no temp file");
1259                         bbg.nonstdout = fdopen(nonstdoutfd, "w");
1260
1261                         /* Set permissions of output file */
1262
1263                         fstat(fileno(file), &statbuf);
1264                         fchmod(nonstdoutfd, statbuf.st_mode);
1265                         add_input_file(file);
1266                         process_files();
1267                         fclose(bbg.nonstdout);
1268
1269                         bbg.nonstdout = stdout;
1270                         /* unlink(argv[i]); */
1271                         // FIXME: error check / message?
1272                         rename(bbg.outname, argv[i]);
1273                         free(bbg.outname);
1274                         bbg.outname = 0;
1275                 }
1276                 if (bbg.input_file_count > bbg.current_input_file)
1277                         process_files();
1278         }
1279
1280         return status;
1281 }