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         int last_char;                  /* Last line written by (sw) had no '\n' */
  85
  86         /* GENERAL FIELDS */
  87         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
  88         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
  89 } sed_cmd_t;
  90
  91 static const char *const semicolon_whitespace = "; \n\r\t\v";
  92
  93 struct sed_globals
  94 {
  95         /* options */
  96         int be_quiet, in_place, regex_type;
  97         FILE *nonstdout;
  98         char *outname, *hold_space;
  99
 100         /* List of input files */
 101         int input_file_count,current_input_file;
 102         FILE **input_file_list;
 103
 104         regmatch_t regmatch[10];
 105         regex_t *previous_regex_ptr;
 106
 107         /* linked list of sed commands */
 108         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 109
 110         /* Linked list of append lines */
 111         llist_t *append_head;
 112
 113         char *add_cmd_line;
 114
 115         struct pipeline {
 116                 char *buf;      /* Space to hold string */
 117                 int idx;        /* Space used */
 118                 int len;        /* Space allocated */
 119         } pipeline;
 120 } bbg;
 121
 122
 123 void sed_free_and_close_stuff(void);
 124 #if ENABLE_FEATURE_CLEAN_UP
 125 void sed_free_and_close_stuff(void)
 126 {
 127         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 128
 129         llist_free(bbg.append_head, free);
 130
 131         while (sed_cmd) {
 132                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 133
 134                 if (sed_cmd->file)
 135                         xprint_and_close_file(sed_cmd->file);
 136
 137                 if (sed_cmd->beg_match) {
 138                         regfree(sed_cmd->beg_match);
 139                         free(sed_cmd->beg_match);
 140                 }
 141                 if (sed_cmd->end_match) {
 142                         regfree(sed_cmd->end_match);
 143                         free(sed_cmd->end_match);
 144                 }
 145                 if (sed_cmd->sub_match) {
 146                         regfree(sed_cmd->sub_match);
 147                         free(sed_cmd->sub_match);
 148                 }
 149                 free(sed_cmd->string);
 150                 free(sed_cmd);
 151                 sed_cmd = sed_cmd_next;
 152         }
 153
 154         if (bbg.hold_space) free(bbg.hold_space);
 155
 156         while (bbg.current_input_file < bbg.input_file_count)
 157                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 158 }
 159 #endif
 160
 161 /* If something bad happens during -i operation, delete temp file */
 162
 163 static void cleanup_outname(void)
 164 {
 165         if (bbg.outname) unlink(bbg.outname);
 166 }
 167
 168 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 169
 170 static void parse_escapes(char *dest, char *string, int len, char from, char to)
 171 {
 172         int i = 0;
 173
 174         while (i < len) {
 175                 if (string[i] == '\\') {
 176                         if (!to || string[i+1] == from) {
 177                                 *(dest++) = to ? to : string[i+1];
 178                                 i += 2;
 179                                 continue;
 180                         } else *(dest++) = string[i++];
 181                 }
 182                 *(dest++) = string[i++];
 183         }
 184         *dest = 0;
 185 }
 186
 187 static char *copy_parsing_escapes(char *string, int len)
 188 {
 189         char *dest = xmalloc(len+1);
 190
 191         parse_escapes(dest,string,len,'n','\n');
 192         return dest;
 193 }
 194
 195
 196 /*
 197  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 198  * beginning at a specified index and returns the index of the next regular
 199  * expression delimiter (typically a forward * slash ('/')) not preceded by
 200  * a backslash ('\').  A negative delimiter disables square bracket checking.
 201  */
 202 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
 203 {
 204         int bracket = -1;
 205         int escaped = 0;
 206         int idx = 0;
 207         char ch;
 208
 209         if (delimiter < 0) {
 210                 bracket--;
 211                 delimiter *= -1;
 212         }
 213
 214         for (; (ch = str[idx]); idx++) {
 215                 if (bracket >= 0) {
 216                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 217                                         && str[idx - 1] == '^')))
 218                                 bracket = -1;
 219                 } else if (escaped)
 220                         escaped = 0;
 221                 else if (ch == '\\')
 222                         escaped = 1;
 223                 else if (bracket == -1 && ch == '[')
 224                         bracket = idx;
 225                 else if (ch == delimiter)
 226                         return idx;
 227         }
 228
 229         /* if we make it to here, we've hit the end of the string */
 230         bb_error_msg_and_die("unmatched '%c'",delimiter);
 231 }
 232
 233 /*
 234  *  Returns the index of the third delimiter
 235  */
 236 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
 237 {
 238         char *cmdstr_ptr = cmdstr;
 239         char delimiter;
 240         int idx = 0;
 241
 242         /* verify that the 's' or 'y' is followed by something.  That something
 243          * (typically a 'slash') is now our regexp delimiter... */
 244         if (*cmdstr == '\0')
 245                 bb_error_msg_and_die("bad format in substitution expression");
 246         delimiter = *cmdstr_ptr++;
 247
 248         /* save the match string */
 249         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 250         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 251
 252         /* save the replacement string */
 253         cmdstr_ptr += idx + 1;
 254         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 255         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 256
 257         return ((cmdstr_ptr - cmdstr) + idx);
 258 }
 259
 260 /*
 261  * returns the index in the string just past where the address ends.
 262  */
 263 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 264 {
 265         char *pos = my_str;
 266
 267         if (isdigit(*my_str)) {
 268                 *linenum = strtol(my_str, &pos, 10);
 269                 /* endstr shouldnt ever equal NULL */
 270         } else if (*my_str == '$') {
 271                 *linenum = -1;
 272                 pos++;
 273         } else if (*my_str == '/' || *my_str == '\\') {
 274                 int next;
 275                 char delimiter;
 276                 char *temp;
 277
 278                 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                                 sed_cmd->which_match = (unsigned short)strtol(substr+idx,&pos,10);
 340                                 idx = pos-substr;
 341                         }
 342                         continue;
 343                 }
 344                 /* Skip spaces */
 345                 if (isspace(substr[idx])) continue;
 346
 347                 switch (substr[idx]) {
 348                 /* Replace all occurrences */
 349                 case 'g':
 350                         if (match[0] != '^') sed_cmd->which_match = 0;
 351                         break;
 352                 /* Print pattern space */
 353                 case 'p':
 354                         sed_cmd->sub_p = 1;
 355                         break;
 356                 /* Write to file */
 357                 case 'w':
 358                 {
 359                         char *temp;
 360                         idx += parse_file_cmd(sed_cmd,substr+idx,&temp);
 361
 362                         break;
 363                 }
 364                 /* Ignore case (gnu exension) */
 365                 case 'I':
 366                         cflags |= REG_ICASE;
 367                         break;
 368                 /* Comment */
 369                 case '#':
 370                         while (substr[++idx]) /*skip all*/;
 371                         /* Fall through */
 372                 /* End of command */
 373                 case ';':
 374                 case '}':
 375                         goto out;
 376                 default:
 377                         bb_error_msg_and_die("bad option in substitution expression");
 378                 }
 379         }
 380 out:
 381         /* compile the match string into a regex */
 382         if (*match != '\0') {
 383                 /* If match is empty, we use last regex used at runtime */
 384                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 385                 xregcomp(sed_cmd->sub_match, match, cflags);
 386         }
 387         free(match);
 388
 389         return idx;
 390 }
 391
 392 /*
 393  *  Process the commands arguments
 394  */
 395 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 396 {
 397         /* handle (s)ubstitution command */
 398         if (sed_cmd->cmd == 's')
 399                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 400         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 401         else if (strchr("aic", sed_cmd->cmd)) {
 402                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 403                         bb_error_msg_and_die
 404                                 ("only a beginning address can be specified for edit commands");
 405                 for (;;) {
 406                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 407                                 cmdstr++;
 408                                 break;
 409                         } else if (isspace(*cmdstr))
 410                                 cmdstr++;
 411                         else
 412                                 break;
 413                 }
 414                 sed_cmd->string = xstrdup(cmdstr);
 415                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 416                 cmdstr += strlen(cmdstr);
 417         /* handle file cmds: (r)ead */
 418         } else if (strchr("rw", sed_cmd->cmd)) {
 419                 if (sed_cmd->end_line || sed_cmd->end_match)
 420                         bb_error_msg_and_die("command only uses one address");
 421                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 422                 if (sed_cmd->cmd == 'w')
 423                         sed_cmd->file = xfopen(sed_cmd->string,"w");
 424         /* handle branch commands */
 425         } else if (strchr(":btT", sed_cmd->cmd)) {
 426                 int length;
 427
 428                 cmdstr = skip_whitespace(cmdstr);
 429                 length = strcspn(cmdstr, semicolon_whitespace);
 430                 if (length) {
 431                         sed_cmd->string = xstrndup(cmdstr, length);
 432                         cmdstr += length;
 433                 }
 434         }
 435         /* translation command */
 436         else if (sed_cmd->cmd == 'y') {
 437                 char *match, *replace;
 438                 int i = cmdstr[0];
 439
 440                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 441                 /* \n already parsed, but \delimiter needs unescaping. */
 442                 parse_escapes(match,match,strlen(match),i,i);
 443                 parse_escapes(replace,replace,strlen(replace),i,i);
 444
 445                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 446                 for (i = 0; match[i] && replace[i]; i++) {
 447                         sed_cmd->string[i*2] = match[i];
 448                         sed_cmd->string[i*2+1] = replace[i];
 449                 }
 450                 free(match);
 451                 free(replace);
 452         }
 453         /* if it wasnt a single-letter command that takes no arguments
 454          * then it must be an invalid command.
 455          */
 456         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 457                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 458         }
 459
 460         /* give back whatever's left over */
 461         return cmdstr;
 462 }
 463
 464
 465 /* Parse address+command sets, skipping comment lines. */
 466
 467 static void add_cmd(char *cmdstr)
 468 {
 469         sed_cmd_t *sed_cmd;
 470         int temp;
 471
 472         /* Append this line to any unfinished line from last time. */
 473         if (bbg.add_cmd_line) {
 474                 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 475                 free(bbg.add_cmd_line);
 476                 bbg.add_cmd_line = cmdstr;
 477         }
 478
 479         /* If this line ends with backslash, request next line. */
 480         temp = strlen(cmdstr);
 481         if (temp && cmdstr[temp-1] == '\\') {
 482                 if (!bbg.add_cmd_line)
 483                         bbg.add_cmd_line = xstrdup(cmdstr);
 484                 bbg.add_cmd_line[temp-1] = 0;
 485                 return;
 486         }
 487
 488         /* Loop parsing all commands in this line. */
 489         while (*cmdstr) {
 490                 /* Skip leading whitespace and semicolons */
 491                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 492
 493                 /* If no more commands, exit. */
 494                 if (!*cmdstr) break;
 495
 496                 /* if this is a comment, jump past it and keep going */
 497                 if (*cmdstr == '#') {
 498                         /* "#n" is the same as using -n on the command line */
 499                         if (cmdstr[1] == 'n')
 500                                 bbg.be_quiet++;
 501                         cmdstr = strpbrk(cmdstr, "\n\r");
 502                         if (!cmdstr) break;
 503                         continue;
 504                 }
 505
 506                 /* parse the command
 507                  * format is: [addr][,addr][!]cmd
 508                  *            |----||-----||-|
 509                  *            part1 part2  part3
 510                  */
 511
 512                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 513
 514                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 515                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 516
 517                 /* second part (if present) will begin with a comma */
 518                 if (*cmdstr == ',') {
 519                         int idx;
 520
 521                         cmdstr++;
 522                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 523                         if (!idx)
 524                                 bb_error_msg_and_die("no address after comma");
 525                         cmdstr += idx;
 526                 }
 527
 528                 /* skip whitespace before the command */
 529                 cmdstr = skip_whitespace(cmdstr);
 530
 531                 /* Check for inversion flag */
 532                 if (*cmdstr == '!') {
 533                         sed_cmd->invert = 1;
 534                         cmdstr++;
 535
 536                         /* skip whitespace before the command */
 537                         cmdstr = skip_whitespace(cmdstr);
 538                 }
 539
 540                 /* last part (mandatory) will be a command */
 541                 if (!*cmdstr)
 542                         bb_error_msg_and_die("missing command");
 543                 sed_cmd->cmd = *(cmdstr++);
 544                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 545
 546                 /* Add the command to the command array */
 547                 bbg.sed_cmd_tail->next = sed_cmd;
 548                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 549         }
 550
 551         /* If we glued multiple lines together, free the memory. */
 552         free(bbg.add_cmd_line);
 553         bbg.add_cmd_line = NULL;
 554 }
 555
 556 /* Append to a string, reallocating memory as necessary. */
 557
 558 #define PIPE_GROW 64
 559
 560 static void pipe_putc(char c)
 561 {
 562         if (bbg.pipeline.idx == bbg.pipeline.len) {
 563                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 564                                 bbg.pipeline.len + PIPE_GROW);
 565                 bbg.pipeline.len += PIPE_GROW;
 566         }
 567         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 568 }
 569
 570 static void do_subst_w_backrefs(char *line, char *replace)
 571 {
 572         int i,j;
 573
 574         /* go through the replacement string */
 575         for (i = 0; replace[i]; i++) {
 576                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 577                 if (replace[i] == '\\' && replace[i+1] >= '0' && replace[i+1] <= '9') {
 578                         int backref = replace[++i]-'0';
 579
 580                         /* print out the text held in bbg.regmatch[backref] */
 581                         if (bbg.regmatch[backref].rm_so != -1) {
 582                                 j = bbg.regmatch[backref].rm_so;
 583                                 while (j < bbg.regmatch[backref].rm_eo)
 584                                         pipe_putc(line[j++]);
 585                         }
 586                 }
 587
 588                 /* if we find a backslash escaped character, print the character */
 589                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 590
 591                 /* if we find an unescaped '&' print out the whole matched text. */
 592                 else if (replace[i] == '&') {
 593                         j = bbg.regmatch[0].rm_so;
 594                         while (j < bbg.regmatch[0].rm_eo)
 595                                 pipe_putc(line[j++]);
 596                 }
 597                 /* Otherwise just output the character. */
 598                 else pipe_putc(replace[i]);
 599         }
 600 }
 601
 602 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 603 {
 604         char *oldline = *line;
 605         int altered = 0;
 606         int match_count = 0;
 607         regex_t *current_regex;
 608
 609         /* Handle empty regex. */
 610         if (sed_cmd->sub_match == NULL) {
 611                 current_regex = bbg.previous_regex_ptr;
 612                 if (!current_regex)
 613                         bb_error_msg_and_die("no previous regexp");
 614         } else
 615                 bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 616
 617         /* Find the first match */
 618         if (REG_NOMATCH == regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 619                 return 0;
 620
 621         /* Initialize temporary output buffer. */
 622         bbg.pipeline.buf = xmalloc(PIPE_GROW);
 623         bbg.pipeline.len = PIPE_GROW;
 624         bbg.pipeline.idx = 0;
 625
 626         /* Now loop through, substituting for matches */
 627         do {
 628                 int i;
 629
 630                 /* Work around bug in glibc regexec, demonstrated by:
 631                    echo " a.b" | busybox sed 's [^ .]* x g'
 632                    The match_count check is so not to break
 633                    echo "hi" | busybox sed 's/^/!/g' */
 634                 if (!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
 635                         pipe_putc(*oldline++);
 636                         continue;
 637                 }
 638
 639                 match_count++;
 640
 641                 /* If we aren't interested in this match, output old line to
 642                    end of match and continue */
 643                 if (sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 644                         for (i = 0; i < bbg.regmatch[0].rm_eo; i++)
 645                                 pipe_putc(*oldline++);
 646                         continue;
 647                 }
 648
 649                 /* print everything before the match */
 650                 for (i = 0; i < bbg.regmatch[0].rm_so; i++)
 651                         pipe_putc(oldline[i]);
 652
 653                 /* then print the substitution string */
 654                 do_subst_w_backrefs(oldline, sed_cmd->string);
 655
 656                 /* advance past the match */
 657                 oldline += bbg.regmatch[0].rm_eo;
 658                 /* flag that something has changed */
 659                 altered++;
 660
 661                 /* if we're not doing this globally, get out now */
 662                 if (sed_cmd->which_match) break;
 663         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 664
 665         /* Copy rest of string into output pipeline */
 666
 667         while (*oldline)
 668                 pipe_putc(*oldline++);
 669         pipe_putc(0);
 670
 671         free(*line);
 672         *line = bbg.pipeline.buf;
 673         return altered;
 674 }
 675
 676 /* Set command pointer to point to this label.  (Does not handle null label.) */
 677 static sed_cmd_t *branch_to(char *label)
 678 {
 679         sed_cmd_t *sed_cmd;
 680
 681         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 682                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 683                         return sed_cmd;
 684                 }
 685         }
 686         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 687 }
 688
 689 static void append(char *s)
 690 {
 691         llist_add_to_end(&bbg.append_head, xstrdup(s));
 692 }
 693
 694 static void flush_append(void)
 695 {
 696         char *data;
 697
 698         /* Output appended lines. */
 699         while ((data = (char *)llist_pop(&bbg.append_head))) {
 700                 fprintf(bbg.nonstdout,"%s\n",data);
 701                 free(data);
 702         }
 703 }
 704
 705 static void add_input_file(FILE *file)
 706 {
 707         bbg.input_file_list = xrealloc(bbg.input_file_list,
 708                         (bbg.input_file_count + 1) * sizeof(FILE *));
 709         bbg.input_file_list[bbg.input_file_count++] = file;
 710 }
 711
 712 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 713  * noting if we ran out of files without a newline on the last line we read.
 714  */
 715 static char *get_next_line(int *last_char)
 716 {
 717         char *temp = NULL;
 718         int len;
 719
 720         /* will be returned if last line in the file
 721          * doesn't end with either '\n' or '\0' */
 722         *last_char = 0x100;
 723
 724         flush_append();
 725         while (bbg.current_input_file < bbg.input_file_count) {
 726                 temp = bb_get_chunk_from_file(
 727                         bbg.input_file_list[bbg.current_input_file], &len);
 728                 if (temp) {
 729                         /* len > 0 here, it's ok to do temp[len-1] */
 730                         char c = temp[len-1];
 731                         if (c == '\n' || c == '\0') {
 732                                 temp[len-1] = '\0';
 733                                 *last_char = (unsigned char)c;
 734                         }
 735                         break;
 736                 }
 737                 /* Close this file and advance to next one */
 738                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 739         }
 740         return temp;
 741 }
 742
 743 /* Output line of text. */
 744 /* Note:
 745  * echo -n thingy >z1
 746  * echo -n again >z2
 747  * >znull
 748  * sed "s/i/z/" z1 z2 znull | hexdump -vC output:
 749  * gnu sed 4.1.5:
 750  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 751  * bbox:
 752  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 753  * I am not sure that bbox is wrong here...
 754  */
 755
 756 static int puts_maybe_newline(char *s, FILE *file, int prev_last_char, int last_char)
 757 {
 758         fputs(s, file);
 759         if (last_char < 0x100) fputc(last_char, file);
 760
 761         if (ferror(file)) {
 762                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 763                 bb_error_msg_and_die(bb_msg_write_error);
 764         }
 765
 766         return last_char;
 767 }
 768
 769 #define sed_puts(s, n) \
 770         (prev_last_char = puts_maybe_newline(s, bbg.nonstdout, prev_last_char, n))
 771
 772 /* Process all the lines in all the files */
 773
 774 static void process_files(void)
 775 {
 776         char *pattern_space, *next_line;
 777         int linenum = 0, prev_last_char = 0;
 778         int last_char, next_last_char = 0;
 779
 780         /* Prime the pump */
 781         next_line = get_next_line(&next_last_char);
 782
 783         /* go through every line in each file */
 784         for (;;) {
 785                 sed_cmd_t *sed_cmd;
 786                 int substituted = 0;
 787
 788                 /* Advance to next line.  Stop if out of lines. */
 789                 pattern_space = next_line;
 790                 if (!pattern_space) break;
 791                 last_char = next_last_char;
 792
 793                 /* Read one line in advance so we can act on the last line,
 794                  * the '$' address */
 795                 next_line = get_next_line(&next_last_char);
 796                 linenum++;
 797 restart:
 798                 /* for every line, go through all the commands */
 799                 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 800                         int old_matched, matched;
 801
 802                         old_matched = sed_cmd->in_match;
 803
 804                         /* Determine if this command matches this line: */
 805
 806                         /* Are we continuing a previous multi-line match? */
 807                         sed_cmd->in_match = sed_cmd->in_match
 808                                 /* Or is no range necessary? */
 809                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 810                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 811                                 /* Or did we match the start of a numerical range? */
 812                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 813                                 /* Or does this line match our begin address regex? */
 814                                 || (sed_cmd->beg_match &&
 815                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 816                                 /* Or did we match last line of input? */
 817                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 818
 819                         /* Snapshot the value */
 820
 821                         matched = sed_cmd->in_match;
 822
 823                         /* Is this line the end of the current match? */
 824
 825                         if (matched) {
 826                                 sed_cmd->in_match = !(
 827                                         /* has the ending line come, or is this a single address command? */
 828                                         (sed_cmd->end_line ?
 829                                                 sed_cmd->end_line==-1 ?
 830                                                         !next_line
 831                                                         : sed_cmd->end_line<=linenum
 832                                                 : !sed_cmd->end_match)
 833                                         /* or does this line matches our last address regex */
 834                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 835                                 );
 836                         }
 837
 838                         /* Skip blocks of commands we didn't match. */
 839                         if (sed_cmd->cmd == '{') {
 840                                 if (sed_cmd->invert ? matched : !matched)
 841                                         while (sed_cmd && sed_cmd->cmd != '}')
 842                                                 sed_cmd = sed_cmd->next;
 843                                 if (!sed_cmd) bb_error_msg_and_die("unterminated {");
 844                                 continue;
 845                         }
 846
 847                         /* Okay, so did this line match? */
 848                         if (sed_cmd->invert ? !matched : matched) {
 849                                 /* Update last used regex in case a blank substitute BRE is found */
 850                                 if (sed_cmd->beg_match) {
 851                                         bbg.previous_regex_ptr = sed_cmd->beg_match;
 852                                 }
 853
 854                                 /* actual sedding */
 855                                 switch (sed_cmd->cmd) {
 856
 857                                 /* Print line number */
 858                                 case '=':
 859                                         fprintf(bbg.nonstdout, "%d\n", linenum);
 860                                         break;
 861
 862                                 /* Write the current pattern space up to the first newline */
 863                                 case 'P':
 864                                 {
 865                                         char *tmp = strchr(pattern_space, '\n');
 866
 867                                         if (tmp) {
 868                                                 *tmp = '\0';
 869                                                 sed_puts(pattern_space,1);
 870                                                 *tmp = '\n';
 871                                                 break;
 872                                         }
 873                                         /* Fall Through */
 874                                 }
 875
 876                                 /* Write the current pattern space to output */
 877                                 case 'p':
 878                                         sed_puts(pattern_space, last_char);
 879                                         break;
 880                                 /* Delete up through first newline */
 881                                 case 'D':
 882                                 {
 883                                         char *tmp = strchr(pattern_space,'\n');
 884
 885                                         if (tmp) {
 886                                                 tmp = xstrdup(tmp+1);
 887                                                 free(pattern_space);
 888                                                 pattern_space = tmp;
 889                                                 goto restart;
 890                                         }
 891                                 }
 892                                 /* discard this line. */
 893                                 case 'd':
 894                                         goto discard_line;
 895
 896                                 /* Substitute with regex */
 897                                 case 's':
 898                                         if (!do_subst_command(sed_cmd, &pattern_space))
 899                                                 break;
 900                                         substituted |= 1;
 901
 902                                         /* handle p option */
 903                                         if (sed_cmd->sub_p)
 904                                                 sed_puts(pattern_space, last_char);
 905                                         /* handle w option */
 906                                         if (sed_cmd->file)
 907                                                 sed_cmd->last_char = puts_maybe_newline(
 908                                                         pattern_space, sed_cmd->file,
 909                                                         sed_cmd->last_char, last_char);
 910                                         break;
 911
 912                                 /* Append line to linked list to be printed later */
 913                                 case 'a':
 914                                         append(sed_cmd->string);
 915                                         break;
 916
 917                                 /* Insert text before this line */
 918                                 case 'i':
 919                                         sed_puts(sed_cmd->string,1);
 920                                         break;
 921
 922                                 /* Cut and paste text (replace) */
 923                                 case 'c':
 924                                         /* Only triggers on last line of a matching range. */
 925                                         if (!sed_cmd->in_match)
 926                                                 sed_puts(sed_cmd->string,0);
 927                                         goto discard_line;
 928
 929                                 /* Read file, append contents to output */
 930                                 case 'r':
 931                                 {
 932                                         FILE *rfile;
 933
 934                                         rfile = fopen(sed_cmd->string, "r");
 935                                         if (rfile) {
 936                                                 char *line;
 937
 938                                                 while ((line = xmalloc_getline(rfile))
 939                                                                 != NULL)
 940                                                         append(line);
 941                                                 xprint_and_close_file(rfile);
 942                                         }
 943
 944                                         break;
 945                                 }
 946
 947                                 /* Write pattern space to file. */
 948                                 case 'w':
 949                                         sed_cmd->last_char = puts_maybe_newline(
 950                                                 pattern_space,sed_cmd->file,
 951                                                 sed_cmd->last_char, last_char);
 952                                         break;
 953
 954                                 /* Read next line from input */
 955                                 case 'n':
 956                                         if (!bbg.be_quiet)
 957                                                 sed_puts(pattern_space, last_char);
 958                                         if (next_line) {
 959                                                 free(pattern_space);
 960                                                 pattern_space = next_line;
 961                                                 last_char = next_last_char;
 962                                                 next_line = get_next_line(&next_last_char);
 963                                                 linenum++;
 964                                                 break;
 965                                         }
 966                                         /* fall through */
 967
 968                                 /* Quit.  End of script, end of input. */
 969                                 case 'q':
 970                                         /* Exit the outer while loop */
 971                                         free(next_line);
 972                                         next_line = NULL;
 973                                         goto discard_commands;
 974
 975                                 /* Append the next line to the current line */
 976                                 case 'N':
 977                                 {
 978                                         /* If no next line, jump to end of script and exit. */
 979                                         if (next_line == NULL) {
 980                                                 /* Jump to end of script and exit */
 981                                                 free(next_line);
 982                                                 next_line = NULL;
 983                                                 goto discard_line;
 984                                         /* append next_line, read new next_line. */
 985                                         } else {
 986                                                 int len = strlen(pattern_space);
 987
 988                                                 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 989                                                 pattern_space[len] = '\n';
 990                                                 strcpy(pattern_space + len+1, next_line);
 991                                                 last_char = next_last_char;
 992                                                 next_line = get_next_line(&next_last_char);
 993                                                 linenum++;
 994                                         }
 995                                         break;
 996                                 }
 997
 998                                 /* Test/branch if substitution occurred */
 999                                 case 't':
1000                                         if (!substituted) break;
1001                                         substituted = 0;
1002                                         /* Fall through */
1003                                 /* Test/branch if substitution didn't occur */
1004                                 case 'T':
1005                                         if (substituted) break;
1006                                         /* Fall through */
1007                                 /* Branch to label */
1008                                 case 'b':
1009                                         if (!sed_cmd->string) goto discard_commands;
1010                                         else sed_cmd = branch_to(sed_cmd->string);
1011                                         break;
1012                                 /* Transliterate characters */
1013                                 case 'y':
1014                                 {
1015                                         int i;
1016
1017                                         for (i = 0; pattern_space[i]; i++) {
1018                                                 int j;
1019
1020                                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1021                                                         if (pattern_space[i] == sed_cmd->string[j]) {
1022                                                                 pattern_space[i] = sed_cmd->string[j + 1];
1023                                                                 break;
1024                                                         }
1025                                                 }
1026                                         }
1027
1028                                         break;
1029                                 }
1030                                 case 'g':       /* Replace pattern space with hold space */
1031                                         free(pattern_space);
1032                                         pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1033                                         break;
1034                                 case 'G':       /* Append newline and hold space to pattern space */
1035                                 {
1036                                         int pattern_space_size = 2;
1037                                         int hold_space_size = 0;
1038
1039                                         if (pattern_space)
1040                                                 pattern_space_size += strlen(pattern_space);
1041                                         if (bbg.hold_space)
1042                                                 hold_space_size = strlen(bbg.hold_space);
1043                                         pattern_space = xrealloc(pattern_space,
1044                                                         pattern_space_size + hold_space_size);
1045                                         if (pattern_space_size == 2)
1046                                                 pattern_space[0] = 0;
1047                                         strcat(pattern_space, "\n");
1048                                         if (bbg.hold_space)
1049                                                 strcat(pattern_space, bbg.hold_space);
1050                                         last_char = 0x100;
1051
1052                                         break;
1053                                 }
1054                                 case 'h':       /* Replace hold space with pattern space */
1055                                         free(bbg.hold_space);
1056                                         bbg.hold_space = xstrdup(pattern_space);
1057                                         break;
1058                                 case 'H':       /* Append newline and pattern space to hold space */
1059                                 {
1060                                         int hold_space_size = 2;
1061                                         int pattern_space_size = 0;
1062
1063                                         if (bbg.hold_space)
1064                                                 hold_space_size += strlen(bbg.hold_space);
1065                                         if (pattern_space)
1066                                                 pattern_space_size = strlen(pattern_space);
1067                                         bbg.hold_space = xrealloc(bbg.hold_space,
1068                                                         hold_space_size + pattern_space_size);
1069
1070                                         if (hold_space_size == 2)
1071                                                 *bbg.hold_space = 0;
1072                                         strcat(bbg.hold_space, "\n");
1073                                         if (pattern_space)
1074                                                 strcat(bbg.hold_space, pattern_space);
1075
1076                                         break;
1077                                 }
1078                                 case 'x': /* Exchange hold and pattern space */
1079                                 {
1080                                         char *tmp = pattern_space;
1081                                         pattern_space = bbg.hold_space ? : xzalloc(1);
1082                                         last_char = 0x100;
1083                                         bbg.hold_space = tmp;
1084                                         break;
1085                                 }
1086                                 }
1087                         }
1088                 }
1089
1090                 /*
1091                  * exit point from sedding...
1092                  */
1093 discard_commands:
1094                 /* we will print the line unless we were told to be quiet ('-n')
1095                    or if the line was suppressed (ala 'd'elete) */
1096                 if (!bbg.be_quiet) sed_puts(pattern_space, last_char);
1097
1098                 /* Delete and such jump here. */
1099 discard_line:
1100                 flush_append();
1101                 free(pattern_space);
1102         }
1103 }
1104
1105 /* It is possible to have a command line argument with embedded
1106    newlines.  This counts as multiple command lines. */
1107
1108 static void add_cmd_block(char *cmdstr)
1109 {
1110         int go = 1;
1111         char *temp = xstrdup(cmdstr), *temp2 = temp;
1112
1113         while (go) {
1114                 int len = strcspn(temp2,"\n");
1115                 if (!temp2[len]) go = 0;
1116                 else temp2[len] = 0;
1117                 add_cmd(temp2);
1118                 temp2 += len+1;
1119         }
1120         free(temp);
1121 }
1122
1123 static void add_cmds_link(llist_t *opt_e)
1124 {
1125         if (!opt_e) return;
1126         add_cmds_link(opt_e->link);
1127         add_cmd_block(opt_e->data);
1128         free(opt_e);
1129 }
1130
1131 static void add_files_link(llist_t *opt_f)
1132 {
1133         char *line;
1134         FILE *cmdfile;
1135         if (!opt_f) return;
1136         add_files_link(opt_f->link);
1137         cmdfile = xfopen(opt_f->data, "r");
1138         while ((line = xmalloc_getline(cmdfile)) != NULL) {
1139                 add_cmd(line);
1140                 free(line);
1141         }
1142         xprint_and_close_file(cmdfile);
1143         free(opt_f);
1144 }
1145
1146 int sed_main(int argc, char **argv)
1147 {
1148         unsigned opt;
1149         llist_t *opt_e, *opt_f;
1150         int status = EXIT_SUCCESS;
1151
1152         bbg.sed_cmd_tail=&bbg.sed_cmd_head;
1153
1154         /* destroy command strings on exit */
1155         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1156
1157         /* Lie to autoconf when it starts asking stupid questions. */
1158         if (argc==2 && !strcmp(argv[1],"--version")) {
1159                 printf("This is not GNU sed version 4.0\n");
1160                 exit(0);
1161         }
1162
1163         /* do normal option parsing */
1164         opt_e = opt_f = NULL;
1165         opt_complementary = "e::f::"; /* can occur multiple times */
1166         opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f);
1167         if (opt & 0x1) { // -i
1168                 bbg.in_place++;
1169                 atexit(cleanup_outname);
1170         }
1171         if (opt & 0x2) bbg.regex_type|=REG_EXTENDED; // -r
1172         if (opt & 0x4) bbg.be_quiet++; // -n
1173         if (opt & 0x8) { // -e
1174                 /* getopt32 reverses order of arguments, handle it */
1175                 add_cmds_link(opt_e);
1176         }
1177         if (opt & 0x10) { // -f
1178                 /* getopt32 reverses order of arguments, handle it */
1179                 add_files_link(opt_f);
1180         }
1181         /* if we didn't get a pattern from -e or -f, use argv[optind] */
1182         if (!(opt & 0x18)) {
1183                 if (argv[optind] == NULL)
1184                         bb_show_usage();
1185                 else
1186                         add_cmd_block(argv[optind++]);
1187         }
1188         /* Flush any unfinished commands. */
1189         add_cmd("");
1190
1191         /* By default, we write to stdout */
1192         bbg.nonstdout = stdout;
1193
1194         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1195          * files were specified or '-' was specified, take input from stdin.
1196          * Otherwise, we process all the files specified. */
1197         if (argv[optind] == NULL) {
1198                 if (bbg.in_place)
1199                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1200                 add_input_file(stdin);
1201                 process_files();
1202         } else {
1203                 int i;
1204                 FILE *file;
1205
1206                 for (i = optind; i < argc; i++) {
1207                         struct stat statbuf;
1208                         int nonstdoutfd;
1209
1210                         if (!strcmp(argv[i], "-") && !bbg.in_place) {
1211                                 add_input_file(stdin);
1212                                 process_files();
1213                                 continue;
1214                         }
1215                         file = fopen_or_warn(argv[i], "r");
1216                         if (!file) {
1217                                 status = EXIT_FAILURE;
1218                                 continue;
1219                         }
1220                         if (!bbg.in_place) {
1221                                 add_input_file(file);
1222                                 continue;
1223                         }
1224
1225                         bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1226                         nonstdoutfd = mkstemp(bbg.outname);
1227                         if (-1 == nonstdoutfd)
1228                                 bb_error_msg_and_die("no temp file");
1229                         bbg.nonstdout = fdopen(nonstdoutfd,"w");
1230
1231                         /* Set permissions of output file */
1232
1233                         fstat(fileno(file),&statbuf);
1234                         fchmod(nonstdoutfd,statbuf.st_mode);
1235                         add_input_file(file);
1236                         process_files();
1237                         fclose(bbg.nonstdout);
1238
1239                         bbg.nonstdout = stdout;
1240                         /* unlink(argv[i]); */
1241                         // FIXME: error check / message?
1242                         rename(bbg.outname,argv[i]);
1243                         free(bbg.outname);
1244                         bbg.outname = 0;
1245                 }
1246                 if (bbg.input_file_count > bbg.current_input_file)
1247                         process_files();
1248         }
1249
1250         return status;
1251 }