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 {
  96         /* options */
  97         int be_quiet, in_place, regex_type;
  98         FILE *nonstdout;
  99         char *outname, *hold_space;
 100
 101         /* List of input files */
 102         int input_file_count,current_input_file;
 103         FILE **input_file_list;
 104
 105         regmatch_t regmatch[10];
 106         regex_t *previous_regex_ptr;
 107
 108         /* linked list of sed commands */
 109         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 110
 111         /* Linked list of append lines */
 112         llist_t *append_head;
 113
 114         char *add_cmd_line;
 115
 116         struct pipeline {
 117                 char *buf;      /* Space to hold string */
 118                 int idx;        /* Space used */
 119                 int len;        /* Space allocated */
 120         } pipeline;
 121 } bbg;
 122
 123
 124 void sed_free_and_close_stuff(void);
 125 #if ENABLE_FEATURE_CLEAN_UP
 126 void sed_free_and_close_stuff(void)
 127 {
 128         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 129
 130         llist_free(bbg.append_head, free);
 131
 132         while (sed_cmd) {
 133                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 134
 135                 if (sed_cmd->file)
 136                         xprint_and_close_file(sed_cmd->file);
 137
 138                 if (sed_cmd->beg_match) {
 139                         regfree(sed_cmd->beg_match);
 140                         free(sed_cmd->beg_match);
 141                 }
 142                 if (sed_cmd->end_match) {
 143                         regfree(sed_cmd->end_match);
 144                         free(sed_cmd->end_match);
 145                 }
 146                 if (sed_cmd->sub_match) {
 147                         regfree(sed_cmd->sub_match);
 148                         free(sed_cmd->sub_match);
 149                 }
 150                 free(sed_cmd->string);
 151                 free(sed_cmd);
 152                 sed_cmd = sed_cmd_next;
 153         }
 154
 155         if (bbg.hold_space) free(bbg.hold_space);
 156
 157         while (bbg.current_input_file < bbg.input_file_count)
 158                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 159 }
 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                         } else *(dest++) = string[i++];
 182                 }
 183                 *(dest++) = string[i++];
 184         }
 185         *dest = 0;
 186 }
 187
 188 static char *copy_parsing_escapes(char *string, int len)
 189 {
 190         char *dest = xmalloc(len+1);
 191
 192         parse_escapes(dest,string,len,'n','\n');
 193         return dest;
 194 }
 195
 196
 197 /*
 198  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 199  * beginning at a specified index and returns the index of the next regular
 200  * expression delimiter (typically a forward * slash ('/')) not preceded by
 201  * a backslash ('\').  A negative delimiter disables square bracket checking.
 202  */
 203 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
 204 {
 205         int bracket = -1;
 206         int escaped = 0;
 207         int idx = 0;
 208         char ch;
 209
 210         if (delimiter < 0) {
 211                 bracket--;
 212                 delimiter *= -1;
 213         }
 214
 215         for (; (ch = str[idx]); idx++) {
 216                 if (bracket >= 0) {
 217                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 218                                         && str[idx - 1] == '^')))
 219                                 bracket = -1;
 220                 } else if (escaped)
 221                         escaped = 0;
 222                 else if (ch == '\\')
 223                         escaped = 1;
 224                 else if (bracket == -1 && ch == '[')
 225                         bracket = idx;
 226                 else if (ch == delimiter)
 227                         return idx;
 228         }
 229
 230         /* if we make it to here, we've hit the end of the string */
 231         bb_error_msg_and_die("unmatched '%c'",delimiter);
 232 }
 233
 234 /*
 235  *  Returns the index of the third delimiter
 236  */
 237 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
 238 {
 239         char *cmdstr_ptr = cmdstr;
 240         char delimiter;
 241         int idx = 0;
 242
 243         /* verify that the 's' or 'y' is followed by something.  That something
 244          * (typically a 'slash') is now our regexp delimiter... */
 245         if (*cmdstr == '\0')
 246                 bb_error_msg_and_die("bad format in substitution expression");
 247         delimiter = *cmdstr_ptr++;
 248
 249         /* save the match string */
 250         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 251         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 252
 253         /* save the replacement string */
 254         cmdstr_ptr += idx + 1;
 255         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 256         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 257
 258         return ((cmdstr_ptr - cmdstr) + idx);
 259 }
 260
 261 /*
 262  * returns the index in the string just past where the address ends.
 263  */
 264 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 265 {
 266         char *pos = my_str;
 267
 268         if (isdigit(*my_str)) {
 269                 *linenum = strtol(my_str, &pos, 10);
 270                 /* endstr shouldnt ever equal NULL */
 271         } else if (*my_str == '$') {
 272                 *linenum = -1;
 273                 pos++;
 274         } else if (*my_str == '/' || *my_str == '\\') {
 275                 int next;
 276                 char delimiter;
 277                 char *temp;
 278
 279                 delimiter = '/';
 280                 if (*my_str == '\\') delimiter = *++pos;
 281                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 282                 temp = copy_parsing_escapes(pos,next);
 283                 *regex = xmalloc(sizeof(regex_t));
 284                 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
 285                 free(temp);
 286                 /* Move position to next character after last delimiter */
 287                 pos += (next+1);
 288         }
 289         return pos - my_str;
 290 }
 291
 292 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 293 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
 294 {
 295         int start = 0, idx, hack = 0;
 296
 297         /* Skip whitespace, then grab filename to end of line */
 298         while (isspace(filecmdstr[start])) start++;
 299         idx = start;
 300         while (filecmdstr[idx] && filecmdstr[idx] != '\n') idx++;
 301
 302         /* If lines glued together, put backslash back. */
 303         if (filecmdstr[idx] == '\n') hack = 1;
 304         if (idx == start)
 305                 bb_error_msg_and_die("empty filename");
 306         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 307         if (hack) (*retval)[idx] = '\\';
 308
 309         return idx;
 310 }
 311
 312 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
 313 {
 314         int cflags = bbg.regex_type;
 315         char *match;
 316         int idx = 0;
 317
 318         /*
 319          * A substitution command should look something like this:
 320          *    s/match/replace/ #gIpw
 321          *    ||     |        |||
 322          *    mandatory       optional
 323          */
 324         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 325
 326         /* determine the number of back references in the match string */
 327         /* Note: we compute this here rather than in the do_subst_command()
 328          * function to save processor time, at the expense of a little more memory
 329          * (4 bits) per sed_cmd */
 330
 331         /* process the flags */
 332
 333         sed_cmd->which_match = 1;
 334         while (substr[++idx]) {
 335                 /* Parse match number */
 336                 if (isdigit(substr[idx])) {
 337                         if (match[0] != '^') {
 338                                 /* Match 0 treated as all, multiple matches we take the last one. */
 339                                 char *pos = substr+idx;
 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
 800         /* Prime the pump */
 801         next_line = get_next_line(&next_last_char);
 802
 803         /* go through every line in each file */
 804         for (;;) {
 805                 sed_cmd_t *sed_cmd;
 806                 int substituted = 0;
 807
 808                 /* Advance to next line.  Stop if out of lines. */
 809                 pattern_space = next_line;
 810                 if (!pattern_space) break;
 811                 last_char = next_last_char;
 812
 813                 /* Read one line in advance so we can act on the last line,
 814                  * the '$' address */
 815                 next_line = get_next_line(&next_last_char);
 816                 linenum++;
 817 restart:
 818                 /* for every line, go through all the commands */
 819                 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 820                         int old_matched, matched;
 821
 822                         old_matched = sed_cmd->in_match;
 823
 824                         /* Determine if this command matches this line: */
 825
 826                         /* Are we continuing a previous multi-line match? */
 827                         sed_cmd->in_match = sed_cmd->in_match
 828                                 /* Or is no range necessary? */
 829                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 830                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 831                                 /* Or did we match the start of a numerical range? */
 832                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 833                                 /* Or does this line match our begin address regex? */
 834                                 || (sed_cmd->beg_match &&
 835                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 836                                 /* Or did we match last line of input? */
 837                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 838
 839                         /* Snapshot the value */
 840
 841                         matched = sed_cmd->in_match;
 842
 843                         /* Is this line the end of the current match? */
 844
 845                         if (matched) {
 846                                 sed_cmd->in_match = !(
 847                                         /* has the ending line come, or is this a single address command? */
 848                                         (sed_cmd->end_line ?
 849                                                 sed_cmd->end_line==-1 ?
 850                                                         !next_line
 851                                                         : sed_cmd->end_line<=linenum
 852                                                 : !sed_cmd->end_match)
 853                                         /* or does this line matches our last address regex */
 854                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 855                                 );
 856                         }
 857
 858                         /* Skip blocks of commands we didn't match. */
 859                         if (sed_cmd->cmd == '{') {
 860                                 if (sed_cmd->invert ? matched : !matched)
 861                                         while (sed_cmd && sed_cmd->cmd != '}')
 862                                                 sed_cmd = sed_cmd->next;
 863                                 if (!sed_cmd) bb_error_msg_and_die("unterminated {");
 864                                 continue;
 865                         }
 866
 867                         /* Okay, so did this line match? */
 868                         if (sed_cmd->invert ? !matched : matched) {
 869                                 /* Update last used regex in case a blank substitute BRE is found */
 870                                 if (sed_cmd->beg_match) {
 871                                         bbg.previous_regex_ptr = sed_cmd->beg_match;
 872                                 }
 873
 874                                 /* actual sedding */
 875                                 switch (sed_cmd->cmd) {
 876
 877                                 /* Print line number */
 878                                 case '=':
 879                                         fprintf(bbg.nonstdout, "%d\n", linenum);
 880                                         break;
 881
 882                                 /* Write the current pattern space up to the first newline */
 883                                 case 'P':
 884                                 {
 885                                         char *tmp = strchr(pattern_space, '\n');
 886
 887                                         if (tmp) {
 888                                                 *tmp = '\0';
 889                                                 sed_puts(pattern_space,1);
 890                                                 *tmp = '\n';
 891                                                 break;
 892                                         }
 893                                         /* Fall Through */
 894                                 }
 895
 896                                 /* Write the current pattern space to output */
 897                                 case 'p':
 898                                         sed_puts(pattern_space, last_char);
 899                                         break;
 900                                 /* Delete up through first newline */
 901                                 case 'D':
 902                                 {
 903                                         char *tmp = strchr(pattern_space,'\n');
 904
 905                                         if (tmp) {
 906                                                 tmp = xstrdup(tmp+1);
 907                                                 free(pattern_space);
 908                                                 pattern_space = tmp;
 909                                                 goto restart;
 910                                         }
 911                                 }
 912                                 /* discard this line. */
 913                                 case 'd':
 914                                         goto discard_line;
 915
 916                                 /* Substitute with regex */
 917                                 case 's':
 918                                         if (!do_subst_command(sed_cmd, &pattern_space))
 919                                                 break;
 920                                         substituted |= 1;
 921
 922                                         /* handle p option */
 923                                         if (sed_cmd->sub_p)
 924                                                 sed_puts(pattern_space, last_char);
 925                                         /* handle w option */
 926                                         if (sed_cmd->file)
 927                                                 sed_cmd->last_char = puts_maybe_newline(
 928                                                         pattern_space, sed_cmd->file,
 929                                                         sed_cmd->last_char, last_char);
 930                                         break;
 931
 932                                 /* Append line to linked list to be printed later */
 933                                 case 'a':
 934                                         append(sed_cmd->string);
 935                                         break;
 936
 937                                 /* Insert text before this line */
 938                                 case 'i':
 939                                         sed_puts(sed_cmd->string,1);
 940                                         break;
 941
 942                                 /* Cut and paste text (replace) */
 943                                 case 'c':
 944                                         /* Only triggers on last line of a matching range. */
 945                                         if (!sed_cmd->in_match)
 946                                                 sed_puts(sed_cmd->string,0);
 947                                         goto discard_line;
 948
 949                                 /* Read file, append contents to output */
 950                                 case 'r':
 951                                 {
 952                                         FILE *rfile;
 953
 954                                         rfile = fopen(sed_cmd->string, "r");
 955                                         if (rfile) {
 956                                                 char *line;
 957
 958                                                 while ((line = xmalloc_getline(rfile))
 959                                                                 != NULL)
 960                                                         append(line);
 961                                                 xprint_and_close_file(rfile);
 962                                         }
 963
 964                                         break;
 965                                 }
 966
 967                                 /* Write pattern space to file. */
 968                                 case 'w':
 969                                         sed_cmd->last_char = puts_maybe_newline(
 970                                                 pattern_space,sed_cmd->file,
 971                                                 sed_cmd->last_char, last_char);
 972                                         break;
 973
 974                                 /* Read next line from input */
 975                                 case 'n':
 976                                         if (!bbg.be_quiet)
 977                                                 sed_puts(pattern_space, last_char);
 978                                         if (next_line) {
 979                                                 free(pattern_space);
 980                                                 pattern_space = next_line;
 981                                                 last_char = next_last_char;
 982                                                 next_line = get_next_line(&next_last_char);
 983                                                 linenum++;
 984                                                 break;
 985                                         }
 986                                         /* fall through */
 987
 988                                 /* Quit.  End of script, end of input. */
 989                                 case 'q':
 990                                         /* Exit the outer while loop */
 991                                         free(next_line);
 992                                         next_line = NULL;
 993                                         goto discard_commands;
 994
 995                                 /* Append the next line to the current line */
 996                                 case 'N':
 997                                 {
 998                                         /* If no next line, jump to end of script and exit. */
 999                                         if (next_line == NULL) {
1000                                                 /* Jump to end of script and exit */
1001                                                 free(next_line);
1002                                                 next_line = NULL;
1003                                                 goto discard_line;
1004                                         /* append next_line, read new next_line. */
1005                                         } else {
1006                                                 int len = strlen(pattern_space);
1007
1008                                                 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1009                                                 pattern_space[len] = '\n';
1010                                                 strcpy(pattern_space + len+1, next_line);
1011                                                 last_char = next_last_char;
1012                                                 next_line = get_next_line(&next_last_char);
1013                                                 linenum++;
1014                                         }
1015                                         break;
1016                                 }
1017
1018                                 /* Test/branch if substitution occurred */
1019                                 case 't':
1020                                         if (!substituted) break;
1021                                         substituted = 0;
1022                                         /* Fall through */
1023                                 /* Test/branch if substitution didn't occur */
1024                                 case 'T':
1025                                         if (substituted) break;
1026                                         /* Fall through */
1027                                 /* Branch to label */
1028                                 case 'b':
1029                                         if (!sed_cmd->string) goto discard_commands;
1030                                         else sed_cmd = branch_to(sed_cmd->string);
1031                                         break;
1032                                 /* Transliterate characters */
1033                                 case 'y':
1034                                 {
1035                                         int i;
1036
1037                                         for (i = 0; pattern_space[i]; i++) {
1038                                                 int j;
1039
1040                                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1041                                                         if (pattern_space[i] == sed_cmd->string[j]) {
1042                                                                 pattern_space[i] = sed_cmd->string[j + 1];
1043                                                                 break;
1044                                                         }
1045                                                 }
1046                                         }
1047
1048                                         break;
1049                                 }
1050                                 case 'g':       /* Replace pattern space with hold space */
1051                                         free(pattern_space);
1052                                         pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1053                                         break;
1054                                 case 'G':       /* Append newline and hold space to pattern space */
1055                                 {
1056                                         int pattern_space_size = 2;
1057                                         int hold_space_size = 0;
1058
1059                                         if (pattern_space)
1060                                                 pattern_space_size += strlen(pattern_space);
1061                                         if (bbg.hold_space)
1062                                                 hold_space_size = strlen(bbg.hold_space);
1063                                         pattern_space = xrealloc(pattern_space,
1064                                                         pattern_space_size + hold_space_size);
1065                                         if (pattern_space_size == 2)
1066                                                 pattern_space[0] = 0;
1067                                         strcat(pattern_space, "\n");
1068                                         if (bbg.hold_space)
1069                                                 strcat(pattern_space, bbg.hold_space);
1070                                         last_char = '\n';
1071
1072                                         break;
1073                                 }
1074                                 case 'h':       /* Replace hold space with pattern space */
1075                                         free(bbg.hold_space);
1076                                         bbg.hold_space = xstrdup(pattern_space);
1077                                         break;
1078                                 case 'H':       /* Append newline and pattern space to hold space */
1079                                 {
1080                                         int hold_space_size = 2;
1081                                         int pattern_space_size = 0;
1082
1083                                         if (bbg.hold_space)
1084                                                 hold_space_size += strlen(bbg.hold_space);
1085                                         if (pattern_space)
1086                                                 pattern_space_size = strlen(pattern_space);
1087                                         bbg.hold_space = xrealloc(bbg.hold_space,
1088                                                         hold_space_size + pattern_space_size);
1089
1090                                         if (hold_space_size == 2)
1091                                                 *bbg.hold_space = 0;
1092                                         strcat(bbg.hold_space, "\n");
1093                                         if (pattern_space)
1094                                                 strcat(bbg.hold_space, pattern_space);
1095
1096                                         break;
1097                                 }
1098                                 case 'x': /* Exchange hold and pattern space */
1099                                 {
1100                                         char *tmp = pattern_space;
1101                                         pattern_space = bbg.hold_space ? : xzalloc(1);
1102                                         last_char = '\n';
1103                                         bbg.hold_space = tmp;
1104                                         break;
1105                                 }
1106                                 }
1107                         }
1108                 }
1109
1110                 /*
1111                  * exit point from sedding...
1112                  */
1113 discard_commands:
1114                 /* we will print the line unless we were told to be quiet ('-n')
1115                    or if the line was suppressed (ala 'd'elete) */
1116                 if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1117
1118                 /* Delete and such jump here. */
1119 discard_line:
1120                 flush_append();
1121                 free(pattern_space);
1122         }
1123 }
1124
1125 /* It is possible to have a command line argument with embedded
1126    newlines.  This counts as multiple command lines. */
1127
1128 static void add_cmd_block(char *cmdstr)
1129 {
1130         int go = 1;
1131         char *temp = xstrdup(cmdstr), *temp2 = temp;
1132
1133         while (go) {
1134                 int len = strcspn(temp2,"\n");
1135                 if (!temp2[len]) go = 0;
1136                 else temp2[len] = 0;
1137                 add_cmd(temp2);
1138                 temp2 += len+1;
1139         }
1140         free(temp);
1141 }
1142
1143 static void add_cmds_link(llist_t *opt_e)
1144 {
1145         if (!opt_e) return;
1146         add_cmds_link(opt_e->link);
1147         add_cmd_block(opt_e->data);
1148         free(opt_e);
1149 }
1150
1151 static void add_files_link(llist_t *opt_f)
1152 {
1153         char *line;
1154         FILE *cmdfile;
1155         if (!opt_f) return;
1156         add_files_link(opt_f->link);
1157         cmdfile = xfopen(opt_f->data, "r");
1158         while ((line = xmalloc_getline(cmdfile)) != NULL) {
1159                 add_cmd(line);
1160                 free(line);
1161         }
1162         xprint_and_close_file(cmdfile);
1163         free(opt_f);
1164 }
1165
1166 int sed_main(int argc, char **argv)
1167 {
1168         unsigned opt;
1169         llist_t *opt_e, *opt_f;
1170         int status = EXIT_SUCCESS;
1171
1172         bbg.sed_cmd_tail=&bbg.sed_cmd_head;
1173
1174         /* destroy command strings on exit */
1175         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1176
1177         /* Lie to autoconf when it starts asking stupid questions. */
1178         if (argc==2 && !strcmp(argv[1],"--version")) {
1179                 printf("This is not GNU sed version 4.0\n");
1180                 exit(0);
1181         }
1182
1183         /* do normal option parsing */
1184         opt_e = opt_f = NULL;
1185         opt_complementary = "e::f::"; /* can occur multiple times */
1186         opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f);
1187         if (opt & 0x1) { // -i
1188                 bbg.in_place++;
1189                 atexit(cleanup_outname);
1190         }
1191         if (opt & 0x2) bbg.regex_type|=REG_EXTENDED; // -r
1192         if (opt & 0x4) bbg.be_quiet++; // -n
1193         if (opt & 0x8) { // -e
1194                 /* getopt32 reverses order of arguments, handle it */
1195                 add_cmds_link(opt_e);
1196         }
1197         if (opt & 0x10) { // -f
1198                 /* getopt32 reverses order of arguments, handle it */
1199                 add_files_link(opt_f);
1200         }
1201         /* if we didn't get a pattern from -e or -f, use argv[optind] */
1202         if (!(opt & 0x18)) {
1203                 if (argv[optind] == NULL)
1204                         bb_show_usage();
1205                 else
1206                         add_cmd_block(argv[optind++]);
1207         }
1208         /* Flush any unfinished commands. */
1209         add_cmd("");
1210
1211         /* By default, we write to stdout */
1212         bbg.nonstdout = stdout;
1213
1214         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1215          * files were specified or '-' was specified, take input from stdin.
1216          * Otherwise, we process all the files specified. */
1217         if (argv[optind] == NULL) {
1218                 if (bbg.in_place)
1219                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1220                 add_input_file(stdin);
1221                 process_files();
1222         } else {
1223                 int i;
1224                 FILE *file;
1225
1226                 for (i = optind; i < argc; i++) {
1227                         struct stat statbuf;
1228                         int nonstdoutfd;
1229
1230                         if (!strcmp(argv[i], "-") && !bbg.in_place) {
1231                                 add_input_file(stdin);
1232                                 process_files();
1233                                 continue;
1234                         }
1235                         file = fopen_or_warn(argv[i], "r");
1236                         if (!file) {
1237                                 status = EXIT_FAILURE;
1238                                 continue;
1239                         }
1240                         if (!bbg.in_place) {
1241                                 add_input_file(file);
1242                                 continue;
1243                         }
1244
1245                         bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1246                         nonstdoutfd = mkstemp(bbg.outname);
1247                         if (-1 == nonstdoutfd)
1248                                 bb_error_msg_and_die("no temp file");
1249                         bbg.nonstdout = fdopen(nonstdoutfd,"w");
1250
1251                         /* Set permissions of output file */
1252
1253                         fstat(fileno(file),&statbuf);
1254                         fchmod(nonstdoutfd,statbuf.st_mode);
1255                         add_input_file(file);
1256                         process_files();
1257                         fclose(bbg.nonstdout);
1258
1259                         bbg.nonstdout = stdout;
1260                         /* unlink(argv[i]); */
1261                         // FIXME: error check / message?
1262                         rename(bbg.outname,argv[i]);
1263                         free(bbg.outname);
1264                         bbg.outname = 0;
1265                 }
1266                 if (bbg.input_file_count > bbg.current_input_file)
1267                         process_files();
1268         }
1269
1270         return status;
1271 }