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
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * MAINTAINER: Rob Landley <rob@landley.net>
  12  *
  13  * Licensed under GPLv2, see file LICENSE in this source tree.
  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  * (G.sed_cmd_head/G.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 /* Supported features and commands in this version of sed:
  37  *
  38  * - comments ('#')
  39  * - address matching: num|/matchstr/[,num|/matchstr/|$]command
  40  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  41  * - edit commands: (a)ppend, (i)nsert, (c)hange
  42  * - file commands: (r)ead
  43  * - backreferences in substitution expressions (\0, \1, \2...\9)
  44  * - grouped commands: {cmd1;cmd2}
  45  * - transliteration (y/source-chars/dest-chars/)
  46  * - pattern space hold space storing / swapping (g, h, x)
  47  * - labels / branching (: label, b, t, T)
  48  *
  49  * (Note: Specifying an address (range) to match is *optional*; commands
  50  * default to the whole pattern space if no specific address match was
  51  * requested.)
  52  *
  53  * Todo:
  54  * - Create a wrapper around regex to make libc's regex conform with sed
  55  *
  56  * Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  57  */
  58
  59 //usage:#define sed_trivial_usage
  60 //usage:       "[-inr] [-f FILE]... [-e CMD]... [FILE]...\n"
  61 //usage:       "or: sed [-inr] CMD [FILE]..."
  62 //usage:#define sed_full_usage "\n\n"
  63 //usage:       "        -e CMD  Add CMD to sed commands to be executed"
  64 //usage:     "\n        -f FILE Add FILE contents to sed commands to be executed"
  65 //usage:     "\n        -i[SFX] Edit files in-place (otherwise sends to stdout)"
  66 //usage:     "\n                Optionally back files up, appending SFX"
  67 //usage:     "\n        -n      Suppress automatic printing of pattern space"
  68 //usage:     "\n        -r      Use extended regex syntax"
  69 //usage:     "\n"
  70 //usage:     "\nIf no -e or -f, the first non-option argument is the sed command string."
  71 //usage:     "\nRemaining arguments are input files (stdin if none)."
  72 //usage:
  73 //usage:#define sed_example_usage
  74 //usage:       "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
  75 //usage:       "bar\n"
  76
  77 #include "libbb.h"
  78 #include "xregex.h"
  79
  80 #if 0
  81 # define dbg(...) bb_error_msg(__VA_ARGS__)
  82 #else
  83 # define dbg(...) ((void)0)
  84 #endif
  85
  86
  87 enum {
  88         OPT_in_place = 1 << 0,
  89 };
  90
  91 /* Each sed command turns into one of these structures. */
  92 typedef struct sed_cmd_s {
  93         /* Ordered by alignment requirements: currently 36 bytes on x86 */
  94         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
  95
  96         /* address storage */
  97         regex_t *beg_match;     /* sed -e '/match/cmd' */
  98         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
  99         regex_t *sub_match;     /* For 's/sub_match/string/' */
 100         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
 101         int beg_line_orig;      /* copy of the above, needed for -i */
 102         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
 103
 104         FILE *sw_file;          /* File (sw) command writes to, -1 for none. */
 105         char *string;           /* Data string for (saicytb) commands. */
 106
 107         unsigned which_match;   /* (s) Which match to replace (0 for all) */
 108
 109         /* Bitfields (gcc won't group them if we don't) */
 110         unsigned invert:1;      /* the '!' after the address */
 111         unsigned in_match:1;    /* Next line also included in match? */
 112         unsigned sub_p:1;       /* (s) print option */
 113
 114         char sw_last_char;      /* Last line written by (sw) had no '\n' */
 115
 116         /* GENERAL FIELDS */
 117         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 118 } sed_cmd_t;
 119
 120 static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
 121
 122 struct globals {
 123         /* options */
 124         int be_quiet, regex_type;
 125         FILE *nonstdout;
 126         char *outname, *hold_space;
 127
 128         /* List of input files */
 129         int input_file_count, current_input_file;
 130         FILE **input_file_list;
 131
 132         regmatch_t regmatch[10];
 133         regex_t *previous_regex_ptr;
 134
 135         /* linked list of sed commands */
 136         sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
 137
 138         /* Linked list of append lines */
 139         llist_t *append_head;
 140
 141         char *add_cmd_line;
 142
 143         struct pipeline {
 144                 char *buf;  /* Space to hold string */
 145                 int idx;    /* Space used */
 146                 int len;    /* Space allocated */
 147         } pipeline;
 148 } FIX_ALIASING;
 149 #define G (*(struct globals*)&bb_common_bufsiz1)
 150 struct BUG_G_too_big {
 151         char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
 152 };
 153 #define INIT_G() do { \
 154         G.sed_cmd_tail = &G.sed_cmd_head; \
 155 } while (0)
 156
 157
 158 #if ENABLE_FEATURE_CLEAN_UP
 159 static void sed_free_and_close_stuff(void)
 160 {
 161         sed_cmd_t *sed_cmd = G.sed_cmd_head;
 162
 163         llist_free(G.append_head, free);
 164
 165         while (sed_cmd) {
 166                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 167
 168                 if (sed_cmd->sw_file)
 169                         xprint_and_close_file(sed_cmd->sw_file);
 170
 171                 if (sed_cmd->beg_match) {
 172                         regfree(sed_cmd->beg_match);
 173                         free(sed_cmd->beg_match);
 174                 }
 175                 if (sed_cmd->end_match) {
 176                         regfree(sed_cmd->end_match);
 177                         free(sed_cmd->end_match);
 178                 }
 179                 if (sed_cmd->sub_match) {
 180                         regfree(sed_cmd->sub_match);
 181                         free(sed_cmd->sub_match);
 182                 }
 183                 free(sed_cmd->string);
 184                 free(sed_cmd);
 185                 sed_cmd = sed_cmd_next;
 186         }
 187
 188         free(G.hold_space);
 189
 190         while (G.current_input_file < G.input_file_count)
 191                 fclose(G.input_file_list[G.current_input_file++]);
 192 }
 193 #else
 194 void sed_free_and_close_stuff(void);
 195 #endif
 196
 197 /* If something bad happens during -i operation, delete temp file */
 198
 199 static void cleanup_outname(void)
 200 {
 201         if (G.outname) unlink(G.outname);
 202 }
 203
 204 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
 205
 206 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 207 {
 208         int i = 0;
 209
 210         while (i < len) {
 211                 if (string[i] == '\\') {
 212                         if (!to || string[i+1] == from) {
 213                                 *dest++ = to ? to : string[i+1];
 214                                 i += 2;
 215                                 continue;
 216                         }
 217                         *dest++ = string[i++];
 218                 }
 219                 /* TODO: is it safe wrt a string with trailing '\\' ? */
 220                 *dest++ = string[i++];
 221         }
 222         *dest = '\0';
 223 }
 224
 225 static char *copy_parsing_escapes(const char *string, int len)
 226 {
 227         const char *s;
 228         char *dest = xmalloc(len + 1);
 229
 230         /* sed recognizes \n */
 231         /* GNU sed also recognizes \t and \r */
 232         for (s = "\nn\tt\rr"; *s; s += 2) {
 233                 parse_escapes(dest, string, len, s[1], s[0]);
 234                 string = dest;
 235                 len = strlen(dest);
 236         }
 237         return dest;
 238 }
 239
 240
 241 /*
 242  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 243  * beginning at a specified index and returns the index of the next regular
 244  * expression delimiter (typically a forward slash ('/')) not preceded by
 245  * a backslash ('\').  A negative delimiter disables square bracket checking.
 246  */
 247 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
 248 {
 249         int bracket = -1;
 250         int escaped = 0;
 251         int idx = 0;
 252         char ch;
 253
 254         if (delimiter < 0) {
 255                 bracket--;
 256                 delimiter = -delimiter;
 257         }
 258
 259         for (; (ch = str[idx]) != '\0'; idx++) {
 260                 if (bracket >= 0) {
 261                         if (ch == ']'
 262                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
 263                         ) {
 264                                 bracket = -1;
 265                         }
 266                 } else if (escaped)
 267                         escaped = 0;
 268                 else if (ch == '\\')
 269                         escaped = 1;
 270                 else if (bracket == -1 && ch == '[')
 271                         bracket = idx;
 272                 else if (ch == delimiter)
 273                         return idx;
 274         }
 275
 276         /* if we make it to here, we've hit the end of the string */
 277         bb_error_msg_and_die("unmatched '%c'", delimiter);
 278 }
 279
 280 /*
 281  *  Returns the index of the third delimiter
 282  */
 283 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 284 {
 285         const char *cmdstr_ptr = cmdstr;
 286         unsigned char delimiter;
 287         int idx = 0;
 288
 289         /* verify that the 's' or 'y' is followed by something.  That something
 290          * (typically a 'slash') is now our regexp delimiter... */
 291         if (*cmdstr == '\0')
 292                 bb_error_msg_and_die("bad format in substitution expression");
 293         delimiter = *cmdstr_ptr++;
 294
 295         /* save the match string */
 296         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 297         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 298
 299         /* save the replacement string */
 300         cmdstr_ptr += idx + 1;
 301         idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
 302         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 303
 304         return ((cmdstr_ptr - cmdstr) + idx);
 305 }
 306
 307 /*
 308  * returns the index in the string just past where the address ends.
 309  */
 310 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
 311 {
 312         const char *pos = my_str;
 313
 314         if (isdigit(*my_str)) {
 315                 *linenum = strtol(my_str, (char**)&pos, 10);
 316                 /* endstr shouldnt ever equal NULL */
 317         } else if (*my_str == '$') {
 318                 *linenum = -1;
 319                 pos++;
 320         } else if (*my_str == '/' || *my_str == '\\') {
 321                 int next;
 322                 char delimiter;
 323                 char *temp;
 324
 325                 delimiter = '/';
 326                 if (*my_str == '\\')
 327                         delimiter = *++pos;
 328                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 329                 temp = copy_parsing_escapes(pos, next);
 330                 *regex = xzalloc(sizeof(regex_t));
 331                 xregcomp(*regex, temp, G.regex_type|REG_NEWLINE);
 332                 free(temp);
 333                 /* Move position to next character after last delimiter */
 334                 pos += (next+1);
 335         }
 336         return pos - my_str;
 337 }
 338
 339 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 340 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
 341 {
 342         int start = 0, idx, hack = 0;
 343
 344         /* Skip whitespace, then grab filename to end of line */
 345         while (isspace(filecmdstr[start]))
 346                 start++;
 347         idx = start;
 348         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
 349                 idx++;
 350
 351         /* If lines glued together, put backslash back. */
 352         if (filecmdstr[idx] == '\n')
 353                 hack = 1;
 354         if (idx == start)
 355                 bb_error_msg_and_die("empty filename");
 356         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 357         if (hack)
 358                 (*retval)[idx] = '\\';
 359
 360         return idx;
 361 }
 362
 363 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
 364 {
 365         int cflags = G.regex_type;
 366         char *match;
 367         int idx;
 368
 369         /*
 370          * A substitution command should look something like this:
 371          *    s/match/replace/ #gIpw
 372          *    ||     |        |||
 373          *    mandatory       optional
 374          */
 375         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 376
 377         /* determine the number of back references in the match string */
 378         /* Note: we compute this here rather than in the do_subst_command()
 379          * function to save processor time, at the expense of a little more memory
 380          * (4 bits) per sed_cmd */
 381
 382         /* process the flags */
 383
 384         sed_cmd->which_match = 1;
 385         while (substr[++idx]) {
 386                 /* Parse match number */
 387                 if (isdigit(substr[idx])) {
 388                         if (match[0] != '^') {
 389                                 /* Match 0 treated as all, multiple matches we take the last one. */
 390                                 const char *pos = substr + idx;
 391 /* FIXME: error check? */
 392                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
 393                                 idx = pos - substr;
 394                         }
 395                         continue;
 396                 }
 397                 /* Skip spaces */
 398                 if (isspace(substr[idx]))
 399                         continue;
 400
 401                 switch (substr[idx]) {
 402                 /* Replace all occurrences */
 403                 case 'g':
 404                         if (match[0] != '^')
 405                                 sed_cmd->which_match = 0;
 406                         break;
 407                 /* Print pattern space */
 408                 case 'p':
 409                         sed_cmd->sub_p = 1;
 410                         break;
 411                 /* Write to file */
 412                 case 'w':
 413                 {
 414                         char *temp;
 415                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
 416                         break;
 417                 }
 418                 /* Ignore case (gnu exension) */
 419                 case 'I':
 420                         cflags |= REG_ICASE;
 421                         break;
 422                 /* Comment */
 423                 case '#':
 424                         // while (substr[++idx]) continue;
 425                         idx += strlen(substr + idx); // same
 426                         /* Fall through */
 427                 /* End of command */
 428                 case ';':
 429                 case '}':
 430                         goto out;
 431                 default:
 432                         bb_error_msg_and_die("bad option in substitution expression");
 433                 }
 434         }
 435  out:
 436         /* compile the match string into a regex */
 437         if (*match != '\0') {
 438                 /* If match is empty, we use last regex used at runtime */
 439                 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
 440                 dbg("xregcomp('%s',%x)", match, cflags);
 441                 xregcomp(sed_cmd->sub_match, match, cflags);
 442                 dbg("regcomp ok");
 443         }
 444         free(match);
 445
 446         return idx;
 447 }
 448
 449 /*
 450  *  Process the commands arguments
 451  */
 452 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
 453 {
 454         static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
 455         enum {
 456                 IDX_s = 0,
 457                 IDX_a,
 458                 IDX_i,
 459                 IDX_c,
 460                 IDX_r,
 461                 IDX_w,
 462                 IDX_colon,
 463                 IDX_b,
 464                 IDX_t,
 465                 IDX_T,
 466                 IDX_y,
 467                 IDX_d,
 468                 IDX_D,
 469                 IDX_g,
 470                 IDX_G,
 471                 IDX_h,
 472                 IDX_H,
 473                 IDX_l,
 474                 IDX_n,
 475                 IDX_N,
 476                 IDX_p,
 477                 IDX_P,
 478                 IDX_q,
 479                 IDX_x,
 480                 IDX_equal,
 481                 IDX_lbrace,
 482                 IDX_rbrace,
 483                 IDX_nul
 484         };
 485         struct chk { char chk[sizeof(cmd_letters)-1 == IDX_nul ? 1 : -1]; };
 486
 487         unsigned idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
 488
 489         /* handle (s)ubstitution command */
 490         if (idx == IDX_s) {
 491                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 492         }
 493         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 494         else if (idx <= IDX_c) { /* a,i,c */
 495                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 496                         bb_error_msg_and_die("only a beginning address can be specified for edit commands");
 497                 for (;;) {
 498                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 499                                 cmdstr++;
 500                                 break;
 501                         }
 502                         if (!isspace(*cmdstr))
 503                                 break;
 504                         cmdstr++;
 505                 }
 506                 sed_cmd->string = xstrdup(cmdstr);
 507                 /* "\anychar" -> "anychar" */
 508                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
 509                 cmdstr += strlen(cmdstr);
 510         }
 511         /* handle file cmds: (r)ead */
 512         else if (idx <= IDX_w) { /* r,w */
 513                 if (sed_cmd->end_line || sed_cmd->end_match)
 514                         bb_error_msg_and_die("command only uses one address");
 515                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
 516                 if (sed_cmd->cmd == 'w') {
 517                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
 518                         sed_cmd->sw_last_char = '\n';
 519                 }
 520         }
 521         /* handle branch commands */
 522         else if (idx <= IDX_T) { /* :,b,t,T */
 523                 int length;
 524
 525                 cmdstr = skip_whitespace(cmdstr);
 526                 length = strcspn(cmdstr, semicolon_whitespace);
 527                 if (length) {
 528                         sed_cmd->string = xstrndup(cmdstr, length);
 529                         cmdstr += length;
 530                 }
 531         }
 532         /* translation command */
 533         else if (idx == IDX_y) {
 534                 char *match, *replace;
 535                 int i = cmdstr[0];
 536
 537                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 538                 /* \n already parsed, but \delimiter needs unescaping. */
 539                 parse_escapes(match, match, strlen(match), i, i);
 540                 parse_escapes(replace, replace, strlen(replace), i, i);
 541
 542                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 543                 for (i = 0; match[i] && replace[i]; i++) {
 544                         sed_cmd->string[i*2] = match[i];
 545                         sed_cmd->string[i*2+1] = replace[i];
 546                 }
 547                 free(match);
 548                 free(replace);
 549         }
 550         /* if it wasnt a single-letter command that takes no arguments
 551          * then it must be an invalid command.
 552          */
 553         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
 554                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 555         }
 556
 557         /* give back whatever's left over */
 558         return cmdstr;
 559 }
 560
 561
 562 /* Parse address+command sets, skipping comment lines. */
 563
 564 static void add_cmd(const char *cmdstr)
 565 {
 566         sed_cmd_t *sed_cmd;
 567         unsigned len, n;
 568
 569         /* Append this line to any unfinished line from last time. */
 570         if (G.add_cmd_line) {
 571                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
 572                 free(G.add_cmd_line);
 573                 cmdstr = G.add_cmd_line = tp;
 574         }
 575
 576         /* If this line ends with unescaped backslash, request next line. */
 577         n = len = strlen(cmdstr);
 578         while (n && cmdstr[n-1] == '\\')
 579                 n--;
 580         if ((len - n) & 1) { /* if odd number of trailing backslashes */
 581                 if (!G.add_cmd_line)
 582                         G.add_cmd_line = xstrdup(cmdstr);
 583                 G.add_cmd_line[len-1] = '\0';
 584                 return;
 585         }
 586
 587         /* Loop parsing all commands in this line. */
 588         while (*cmdstr) {
 589                 /* Skip leading whitespace and semicolons */
 590                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 591
 592                 /* If no more commands, exit. */
 593                 if (!*cmdstr) break;
 594
 595                 /* if this is a comment, jump past it and keep going */
 596                 if (*cmdstr == '#') {
 597                         /* "#n" is the same as using -n on the command line */
 598                         if (cmdstr[1] == 'n')
 599                                 G.be_quiet++;
 600                         cmdstr = strpbrk(cmdstr, "\n\r");
 601                         if (!cmdstr) break;
 602                         continue;
 603                 }
 604
 605                 /* parse the command
 606                  * format is: [addr][,addr][!]cmd
 607                  *            |----||-----||-|
 608                  *            part1 part2  part3
 609                  */
 610
 611                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 612
 613                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 614                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 615                 sed_cmd->beg_line_orig = sed_cmd->beg_line;
 616
 617                 /* second part (if present) will begin with a comma */
 618                 if (*cmdstr == ',') {
 619                         int idx;
 620
 621                         cmdstr++;
 622                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 623                         if (!idx)
 624                                 bb_error_msg_and_die("no address after comma");
 625                         cmdstr += idx;
 626                 }
 627
 628                 /* skip whitespace before the command */
 629                 cmdstr = skip_whitespace(cmdstr);
 630
 631                 /* Check for inversion flag */
 632                 if (*cmdstr == '!') {
 633                         sed_cmd->invert = 1;
 634                         cmdstr++;
 635
 636                         /* skip whitespace before the command */
 637                         cmdstr = skip_whitespace(cmdstr);
 638                 }
 639
 640                 /* last part (mandatory) will be a command */
 641                 if (!*cmdstr)
 642                         bb_error_msg_and_die("missing command");
 643                 sed_cmd->cmd = *cmdstr++;
 644                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 645
 646                 /* Add the command to the command array */
 647                 *G.sed_cmd_tail = sed_cmd;
 648                 G.sed_cmd_tail = &sed_cmd->next;
 649         }
 650
 651         /* If we glued multiple lines together, free the memory. */
 652         free(G.add_cmd_line);
 653         G.add_cmd_line = NULL;
 654 }
 655
 656 /* Append to a string, reallocating memory as necessary. */
 657
 658 #define PIPE_GROW 64
 659
 660 static void pipe_putc(char c)
 661 {
 662         if (G.pipeline.idx == G.pipeline.len) {
 663                 G.pipeline.buf = xrealloc(G.pipeline.buf,
 664                                 G.pipeline.len + PIPE_GROW);
 665                 G.pipeline.len += PIPE_GROW;
 666         }
 667         G.pipeline.buf[G.pipeline.idx++] = c;
 668 }
 669
 670 static void do_subst_w_backrefs(char *line, char *replace)
 671 {
 672         int i, j;
 673
 674         /* go through the replacement string */
 675         for (i = 0; replace[i]; i++) {
 676                 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
 677                 if (replace[i] == '\\') {
 678                         unsigned backref = replace[++i] - '0';
 679                         if (backref <= 9) {
 680                                 /* print out the text held in G.regmatch[backref] */
 681                                 if (G.regmatch[backref].rm_so != -1) {
 682                                         j = G.regmatch[backref].rm_so;
 683                                         while (j < G.regmatch[backref].rm_eo)
 684                                                 pipe_putc(line[j++]);
 685                                 }
 686                                 continue;
 687                         }
 688                         /* I _think_ it is impossible to get '\' to be
 689                          * the last char in replace string. Thus we dont check
 690                          * for replace[i] == NUL. (counterexample anyone?) */
 691                         /* if we find a backslash escaped character, print the character */
 692                         pipe_putc(replace[i]);
 693                         continue;
 694                 }
 695                 /* if we find an unescaped '&' print out the whole matched text. */
 696                 if (replace[i] == '&') {
 697                         j = G.regmatch[0].rm_so;
 698                         while (j < G.regmatch[0].rm_eo)
 699                                 pipe_putc(line[j++]);
 700                         continue;
 701                 }
 702                 /* Otherwise just output the character. */
 703                 pipe_putc(replace[i]);
 704         }
 705 }
 706
 707 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
 708 {
 709         char *line = *line_p;
 710         unsigned match_count = 0;
 711         bool altered = 0;
 712         bool prev_match_empty = 1;
 713         bool tried_at_eol = 0;
 714         regex_t *current_regex;
 715
 716         current_regex = sed_cmd->sub_match;
 717         /* Handle empty regex. */
 718         if (!current_regex) {
 719                 current_regex = G.previous_regex_ptr;
 720                 if (!current_regex)
 721                         bb_error_msg_and_die("no previous regexp");
 722         }
 723         G.previous_regex_ptr = current_regex;
 724
 725         /* Find the first match */
 726         dbg("matching '%s'", line);
 727         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
 728                 dbg("no match");
 729                 return 0;
 730         }
 731         dbg("match");
 732
 733         /* Initialize temporary output buffer. */
 734         G.pipeline.buf = xmalloc(PIPE_GROW);
 735         G.pipeline.len = PIPE_GROW;
 736         G.pipeline.idx = 0;
 737
 738         /* Now loop through, substituting for matches */
 739         do {
 740                 int start = G.regmatch[0].rm_so;
 741                 int end = G.regmatch[0].rm_eo;
 742                 int i;
 743
 744                 match_count++;
 745
 746                 /* If we aren't interested in this match, output old line to
 747                  * end of match and continue */
 748                 if (sed_cmd->which_match
 749                  && (sed_cmd->which_match != match_count)
 750                 ) {
 751                         for (i = 0; i < end; i++)
 752                                 pipe_putc(*line++);
 753                         /* Null match? Print one more char */
 754                         if (start == end && *line)
 755                                 pipe_putc(*line++);
 756                         goto next;
 757                 }
 758
 759                 /* Print everything before the match */
 760                 for (i = 0; i < start; i++)
 761                         pipe_putc(line[i]);
 762
 763                 /* Then print the substitution string,
 764                  * unless we just matched empty string after non-empty one.
 765                  * Example: string "cccd", pattern "c*", repl "R":
 766                  * result is "RdR", not "RRdR": first match "ccc",
 767                  * second is "" before "d", third is "" after "d".
 768                  * Second match is NOT replaced!
 769                  */
 770                 if (prev_match_empty || start != 0 || start != end) {
 771                         //dbg("%d %d %d", prev_match_empty, start, end);
 772                         dbg("inserting replacement at %d in '%s'", start, line);
 773                         do_subst_w_backrefs(line, sed_cmd->string);
 774                         /* Flag that something has changed */
 775                         altered = 1;
 776                 } else {
 777                         dbg("NOT inserting replacement at %d in '%s'", start, line);
 778                 }
 779
 780                 /* If matched string is empty (f.e. "c*" pattern),
 781                  * copy verbatim one char after it before attempting more matches
 782                  */
 783                 prev_match_empty = (start == end);
 784                 if (prev_match_empty) {
 785                         if (!line[end]) {
 786                                 tried_at_eol = 1;
 787                         } else {
 788                                 pipe_putc(line[end]);
 789                                 end++;
 790                         }
 791                 }
 792
 793                 /* Advance past the match */
 794                 dbg("line += %d", end);
 795                 line += end;
 796
 797                 /* if we're not doing this globally, get out now */
 798                 if (sed_cmd->which_match != 0)
 799                         break;
 800  next:
 801                 /* Exit if we are at EOL and already tried matching at it */
 802                 if (*line == '\0') {
 803                         if (tried_at_eol)
 804                                 break;
 805                         tried_at_eol = 1;
 806                 }
 807
 808 //maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
 809         } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
 810
 811         /* Copy rest of string into output pipeline */
 812         while (1) {
 813                 char c = *line++;
 814                 pipe_putc(c);
 815                 if (c == '\0')
 816                         break;
 817         }
 818
 819         free(*line_p);
 820         *line_p = G.pipeline.buf;
 821         return altered;
 822 }
 823
 824 /* Set command pointer to point to this label.  (Does not handle null label.) */
 825 static sed_cmd_t *branch_to(char *label)
 826 {
 827         sed_cmd_t *sed_cmd;
 828
 829         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
 830                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 831                         return sed_cmd;
 832                 }
 833         }
 834         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 835 }
 836
 837 static void append(char *s)
 838 {
 839         llist_add_to_end(&G.append_head, xstrdup(s));
 840 }
 841
 842 static void flush_append(void)
 843 {
 844         char *data;
 845
 846         /* Output appended lines. */
 847         while ((data = (char *)llist_pop(&G.append_head))) {
 848                 fprintf(G.nonstdout, "%s\n", data);
 849                 free(data);
 850         }
 851 }
 852
 853 static void add_input_file(FILE *file)
 854 {
 855         G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count);
 856         G.input_file_list[G.input_file_count++] = file;
 857 }
 858
 859 /* Get next line of input from G.input_file_list, flushing append buffer and
 860  * noting if we ran out of files without a newline on the last line we read.
 861  */
 862 enum {
 863         NO_EOL_CHAR = 1,
 864         LAST_IS_NUL = 2,
 865 };
 866 static char *get_next_line(char *gets_char)
 867 {
 868         char *temp = NULL;
 869         int len;
 870         char gc;
 871
 872         flush_append();
 873
 874         /* will be returned if last line in the file
 875          * doesn't end with either '\n' or '\0' */
 876         gc = NO_EOL_CHAR;
 877         while (G.current_input_file < G.input_file_count) {
 878                 FILE *fp = G.input_file_list[G.current_input_file];
 879                 /* Read line up to a newline or NUL byte, inclusive,
 880                  * return malloc'ed char[]. length of the chunk read
 881                  * is stored in len. NULL if EOF/error */
 882                 temp = bb_get_chunk_from_file(fp, &len);
 883                 if (temp) {
 884                         /* len > 0 here, it's ok to do temp[len-1] */
 885                         char c = temp[len-1];
 886                         if (c == '\n' || c == '\0') {
 887                                 temp[len-1] = '\0';
 888                                 gc = c;
 889                                 if (c == '\0') {
 890                                         int ch = fgetc(fp);
 891                                         if (ch != EOF)
 892                                                 ungetc(ch, fp);
 893                                         else
 894                                                 gc = LAST_IS_NUL;
 895                                 }
 896                         }
 897                         /* else we put NO_EOL_CHAR into *gets_char */
 898                         break;
 899
 900                 /* NB: I had the idea of peeking next file(s) and returning
 901                  * NO_EOL_CHAR only if it is the *last* non-empty
 902                  * input file. But there is a case where this won't work:
 903                  * file1: "a woo\nb woo"
 904                  * file2: "c no\nd no"
 905                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
 906                  * (note: *no* newline after "b bang"!) */
 907                 }
 908                 /* Close this file and advance to next one */
 909                 fclose(fp);
 910                 G.current_input_file++;
 911         }
 912         *gets_char = gc;
 913         return temp;
 914 }
 915
 916 /* Output line of text. */
 917 /* Note:
 918  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 919  * Without them, we had this:
 920  * echo -n thingy >z1
 921  * echo -n again >z2
 922  * >znull
 923  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 924  * output:
 925  * gnu sed 4.1.5:
 926  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 927  * bbox:
 928  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 929  */
 930 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 931 {
 932         char lpc = *last_puts_char;
 933
 934         /* Need to insert a '\n' between two files because first file's
 935          * last line wasn't terminated? */
 936         if (lpc != '\n' && lpc != '\0') {
 937                 fputc('\n', file);
 938                 lpc = '\n';
 939         }
 940         fputs(s, file);
 941
 942         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 943         if (s[0])
 944                 lpc = 'x';
 945
 946         /* had trailing '\0' and it was last char of file? */
 947         if (last_gets_char == LAST_IS_NUL) {
 948                 fputc('\0', file);
 949                 lpc = 'x'; /* */
 950         } else
 951         /* had trailing '\n' or '\0'? */
 952         if (last_gets_char != NO_EOL_CHAR) {
 953                 fputc(last_gets_char, file);
 954                 lpc = last_gets_char;
 955         }
 956
 957         if (ferror(file)) {
 958                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 959                 bb_error_msg_and_die(bb_msg_write_error);
 960         }
 961         *last_puts_char = lpc;
 962 }
 963
 964 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
 965
 966 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
 967 {
 968         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
 969         if (retval)
 970                 G.previous_regex_ptr = sed_cmd->beg_match;
 971         return retval;
 972 }
 973
 974 /* Process all the lines in all the files */
 975
 976 static void process_files(void)
 977 {
 978         char *pattern_space, *next_line;
 979         int linenum = 0;
 980         char last_puts_char = '\n';
 981         char last_gets_char, next_gets_char;
 982         sed_cmd_t *sed_cmd;
 983         int substituted;
 984
 985         /* Prime the pump */
 986         next_line = get_next_line(&next_gets_char);
 987
 988         /* Go through every line in each file */
 989  again:
 990         substituted = 0;
 991
 992         /* Advance to next line.  Stop if out of lines. */
 993         pattern_space = next_line;
 994         if (!pattern_space)
 995                 return;
 996         last_gets_char = next_gets_char;
 997
 998         /* Read one line in advance so we can act on the last line,
 999          * the '$' address */
1000         next_line = get_next_line(&next_gets_char);
1001         linenum++;
1002
1003         /* For every line, go through all the commands */
1004  restart:
1005         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1006                 int old_matched, matched;
1007
1008                 old_matched = sed_cmd->in_match;
1009
1010                 /* Determine if this command matches this line: */
1011
1012                 dbg("match1:%d", sed_cmd->in_match);
1013                 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1014                                 && !sed_cmd->beg_match && !sed_cmd->end_match));
1015                 dbg("match3:%d", (sed_cmd->beg_line > 0
1016                         && (sed_cmd->end_line || sed_cmd->end_match
1017                             ? (sed_cmd->beg_line <= linenum)
1018                             : (sed_cmd->beg_line == linenum)
1019                             )
1020                         ));
1021                 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1022                 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
1023
1024                 /* Are we continuing a previous multi-line match? */
1025                 sed_cmd->in_match = sed_cmd->in_match
1026                         /* Or is no range necessary? */
1027                         || (!sed_cmd->beg_line && !sed_cmd->end_line
1028                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
1029                         /* Or did we match the start of a numerical range? */
1030                         || (sed_cmd->beg_line > 0
1031                             && (sed_cmd->end_line || sed_cmd->end_match
1032                                   /* note: even if end is numeric and is < linenum too,
1033                                    * GNU sed matches! We match too, therefore we don't
1034                                    * check here that linenum <= end.
1035                                    * Example:
1036                                    * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1037                                    * first three input lines are deleted;
1038                                    * 4th line is matched and printed
1039                                    * by "2,3" (!) and by "4" ranges
1040                                    */
1041                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
1042                                 : (sed_cmd->beg_line == linenum)    /* N */
1043                                 )
1044                             )
1045                         /* Or does this line match our begin address regex? */
1046                         || (beg_match(sed_cmd, pattern_space))
1047                         /* Or did we match last line of input? */
1048                         || (sed_cmd->beg_line == -1 && next_line == NULL);
1049
1050                 /* Snapshot the value */
1051                 matched = sed_cmd->in_match;
1052
1053                 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1054                         sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
1055
1056                 /* Is this line the end of the current match? */
1057
1058                 if (matched) {
1059                         /* once matched, "n,xxx" range is dead, disabling it */
1060                         if (sed_cmd->beg_line > 0) {
1061                                 sed_cmd->beg_line = -2;
1062                         }
1063                         sed_cmd->in_match = !(
1064                                 /* has the ending line come, or is this a single address command? */
1065                                 (sed_cmd->end_line
1066                                         ? sed_cmd->end_line == -1
1067                                                 ? !next_line
1068                                                 : (sed_cmd->end_line <= linenum)
1069                                         : !sed_cmd->end_match
1070                                 )
1071                                 /* or does this line matches our last address regex */
1072                                 || (sed_cmd->end_match && old_matched
1073                                      && (regexec(sed_cmd->end_match,
1074                                                  pattern_space, 0, NULL, 0) == 0)
1075                                 )
1076                         );
1077                 }
1078
1079                 /* Skip blocks of commands we didn't match */
1080                 if (sed_cmd->cmd == '{') {
1081                         if (sed_cmd->invert ? matched : !matched) {
1082                                 unsigned nest_cnt = 0;
1083                                 while (1) {
1084                                         if (sed_cmd->cmd == '{')
1085                                                 nest_cnt++;
1086                                         if (sed_cmd->cmd == '}') {
1087                                                 nest_cnt--;
1088                                                 if (nest_cnt == 0)
1089                                                         break;
1090                                         }
1091                                         sed_cmd = sed_cmd->next;
1092                                         if (!sed_cmd)
1093                                                 bb_error_msg_and_die("unterminated {");
1094                                 }
1095                         }
1096                         continue;
1097                 }
1098
1099                 /* Okay, so did this line match? */
1100                 if (sed_cmd->invert ? matched : !matched)
1101                         continue; /* no */
1102
1103                 /* Update last used regex in case a blank substitute BRE is found */
1104                 if (sed_cmd->beg_match) {
1105                         G.previous_regex_ptr = sed_cmd->beg_match;
1106                 }
1107
1108                 /* actual sedding */
1109                 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1110                                 pattern_space, next_line, sed_cmd->cmd);
1111                 switch (sed_cmd->cmd) {
1112
1113                 /* Print line number */
1114                 case '=':
1115                         fprintf(G.nonstdout, "%d\n", linenum);
1116                         break;
1117
1118                 /* Write the current pattern space up to the first newline */
1119                 case 'P':
1120                 {
1121                         char *tmp = strchr(pattern_space, '\n');
1122                         if (tmp) {
1123                                 *tmp = '\0';
1124                                 /* TODO: explain why '\n' below */
1125                                 sed_puts(pattern_space, '\n');
1126                                 *tmp = '\n';
1127                                 break;
1128                         }
1129                         /* Fall Through */
1130                 }
1131
1132                 /* Write the current pattern space to output */
1133                 case 'p':
1134                         /* NB: we print this _before_ the last line
1135                          * (of current file) is printed. Even if
1136                          * that line is nonterminated, we print
1137                          * '\n' here (gnu sed does the same) */
1138                         sed_puts(pattern_space, '\n');
1139                         break;
1140                 /* Delete up through first newline */
1141                 case 'D':
1142                 {
1143                         char *tmp = strchr(pattern_space, '\n');
1144                         if (tmp) {
1145                                 overlapping_strcpy(pattern_space, tmp + 1);
1146                                 goto restart;
1147                         }
1148                 }
1149                 /* discard this line. */
1150                 case 'd':
1151                         goto discard_line;
1152
1153                 /* Substitute with regex */
1154                 case 's':
1155                         if (!do_subst_command(sed_cmd, &pattern_space))
1156                                 break;
1157                         dbg("do_subst_command succeeded:'%s'", pattern_space);
1158                         substituted |= 1;
1159
1160                         /* handle p option */
1161                         if (sed_cmd->sub_p)
1162                                 sed_puts(pattern_space, last_gets_char);
1163                         /* handle w option */
1164                         if (sed_cmd->sw_file)
1165                                 puts_maybe_newline(
1166                                         pattern_space, sed_cmd->sw_file,
1167                                         &sed_cmd->sw_last_char, last_gets_char);
1168                         break;
1169
1170                 /* Append line to linked list to be printed later */
1171                 case 'a':
1172                         append(sed_cmd->string);
1173                         break;
1174
1175                 /* Insert text before this line */
1176                 case 'i':
1177                         sed_puts(sed_cmd->string, '\n');
1178                         break;
1179
1180                 /* Cut and paste text (replace) */
1181                 case 'c':
1182                         /* Only triggers on last line of a matching range. */
1183                         if (!sed_cmd->in_match)
1184                                 sed_puts(sed_cmd->string, '\n');
1185                         goto discard_line;
1186
1187                 /* Read file, append contents to output */
1188                 case 'r':
1189                 {
1190                         FILE *rfile;
1191                         rfile = fopen_for_read(sed_cmd->string);
1192                         if (rfile) {
1193                                 char *line;
1194
1195                                 while ((line = xmalloc_fgetline(rfile))
1196                                                 != NULL)
1197                                         append(line);
1198                                 xprint_and_close_file(rfile);
1199                         }
1200
1201                         break;
1202                 }
1203
1204                 /* Write pattern space to file. */
1205                 case 'w':
1206                         puts_maybe_newline(
1207                                 pattern_space, sed_cmd->sw_file,
1208                                 &sed_cmd->sw_last_char, last_gets_char);
1209                         break;
1210
1211                 /* Read next line from input */
1212                 case 'n':
1213                         if (!G.be_quiet)
1214                                 sed_puts(pattern_space, last_gets_char);
1215                         if (next_line) {
1216                                 free(pattern_space);
1217                                 pattern_space = next_line;
1218                                 last_gets_char = next_gets_char;
1219                                 next_line = get_next_line(&next_gets_char);
1220                                 substituted = 0;
1221                                 linenum++;
1222                                 break;
1223                         }
1224                         /* fall through */
1225
1226                 /* Quit.  End of script, end of input. */
1227                 case 'q':
1228                         /* Exit the outer while loop */
1229                         free(next_line);
1230                         next_line = NULL;
1231                         goto discard_commands;
1232
1233                 /* Append the next line to the current line */
1234                 case 'N':
1235                 {
1236                         int len;
1237                         /* If no next line, jump to end of script and exit. */
1238                         /* http://www.gnu.org/software/sed/manual/sed.html:
1239                          * "Most versions of sed exit without printing anything
1240                          * when the N command is issued on the last line of
1241                          * a file. GNU sed prints pattern space before exiting
1242                          * unless of course the -n command switch has been
1243                          * specified. This choice is by design."
1244                          */
1245                         if (next_line == NULL) {
1246                                 //goto discard_line;
1247                                 goto discard_commands; /* GNU behavior */
1248                         }
1249                         /* Append next_line, read new next_line. */
1250                         len = strlen(pattern_space);
1251                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1252                         pattern_space[len] = '\n';
1253                         strcpy(pattern_space + len+1, next_line);
1254                         last_gets_char = next_gets_char;
1255                         next_line = get_next_line(&next_gets_char);
1256                         linenum++;
1257                         break;
1258                 }
1259
1260                 /* Test/branch if substitution occurred */
1261                 case 't':
1262                         if (!substituted) break;
1263                         substituted = 0;
1264                         /* Fall through */
1265                 /* Test/branch if substitution didn't occur */
1266                 case 'T':
1267                         if (substituted) break;
1268                         /* Fall through */
1269                 /* Branch to label */
1270                 case 'b':
1271                         if (!sed_cmd->string) goto discard_commands;
1272                         else sed_cmd = branch_to(sed_cmd->string);
1273                         break;
1274                 /* Transliterate characters */
1275                 case 'y':
1276                 {
1277                         int i, j;
1278                         for (i = 0; pattern_space[i]; i++) {
1279                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1280                                         if (pattern_space[i] == sed_cmd->string[j]) {
1281                                                 pattern_space[i] = sed_cmd->string[j + 1];
1282                                                 break;
1283                                         }
1284                                 }
1285                         }
1286
1287                         break;
1288                 }
1289                 case 'g':       /* Replace pattern space with hold space */
1290                         free(pattern_space);
1291                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1292                         break;
1293                 case 'G':       /* Append newline and hold space to pattern space */
1294                 {
1295                         int pattern_space_size = 2;
1296                         int hold_space_size = 0;
1297
1298                         if (pattern_space)
1299                                 pattern_space_size += strlen(pattern_space);
1300                         if (G.hold_space)
1301                                 hold_space_size = strlen(G.hold_space);
1302                         pattern_space = xrealloc(pattern_space,
1303                                         pattern_space_size + hold_space_size);
1304                         if (pattern_space_size == 2)
1305                                 pattern_space[0] = 0;
1306                         strcat(pattern_space, "\n");
1307                         if (G.hold_space)
1308                                 strcat(pattern_space, G.hold_space);
1309                         last_gets_char = '\n';
1310
1311                         break;
1312                 }
1313                 case 'h':       /* Replace hold space with pattern space */
1314                         free(G.hold_space);
1315                         G.hold_space = xstrdup(pattern_space);
1316                         break;
1317                 case 'H':       /* Append newline and pattern space to hold space */
1318                 {
1319                         int hold_space_size = 2;
1320                         int pattern_space_size = 0;
1321
1322                         if (G.hold_space)
1323                                 hold_space_size += strlen(G.hold_space);
1324                         if (pattern_space)
1325                                 pattern_space_size = strlen(pattern_space);
1326                         G.hold_space = xrealloc(G.hold_space,
1327                                         hold_space_size + pattern_space_size);
1328
1329                         if (hold_space_size == 2)
1330                                 *G.hold_space = 0;
1331                         strcat(G.hold_space, "\n");
1332                         if (pattern_space)
1333                                 strcat(G.hold_space, pattern_space);
1334
1335                         break;
1336                 }
1337                 case 'x': /* Exchange hold and pattern space */
1338                 {
1339                         char *tmp = pattern_space;
1340                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1341                         last_gets_char = '\n';
1342                         G.hold_space = tmp;
1343                         break;
1344                 }
1345                 } /* switch */
1346         } /* for each cmd */
1347
1348         /*
1349          * Exit point from sedding...
1350          */
1351  discard_commands:
1352         /* we will print the line unless we were told to be quiet ('-n')
1353            or if the line was suppressed (ala 'd'elete) */
1354         if (!G.be_quiet)
1355                 sed_puts(pattern_space, last_gets_char);
1356
1357         /* Delete and such jump here. */
1358  discard_line:
1359         flush_append();
1360         free(pattern_space);
1361
1362         goto again;
1363 }
1364
1365 /* It is possible to have a command line argument with embedded
1366  * newlines.  This counts as multiple command lines.
1367  * However, newline can be escaped: 's/e/z\<newline>z/'
1368  * We check for this.
1369  */
1370
1371 static void add_cmd_block(char *cmdstr)
1372 {
1373         char *sv, *eol;
1374
1375         cmdstr = sv = xstrdup(cmdstr);
1376         do {
1377                 eol = strchr(cmdstr, '\n');
1378  next:
1379                 if (eol) {
1380                         /* Count preceding slashes */
1381                         int slashes = 0;
1382                         char *sl = eol;
1383
1384                         while (sl != cmdstr && *--sl == '\\')
1385                                 slashes++;
1386                         /* Odd number of preceding slashes - newline is escaped */
1387                         if (slashes & 1) {
1388                                 overlapping_strcpy(eol - 1, eol);
1389                                 eol = strchr(eol, '\n');
1390                                 goto next;
1391                         }
1392                         *eol = '\0';
1393                 }
1394                 add_cmd(cmdstr);
1395                 cmdstr = eol + 1;
1396         } while (eol);
1397         free(sv);
1398 }
1399
1400 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1401 int sed_main(int argc UNUSED_PARAM, char **argv)
1402 {
1403         unsigned opt;
1404         llist_t *opt_e, *opt_f;
1405         char *opt_i;
1406
1407 #if ENABLE_LONG_OPTS
1408         static const char sed_longopts[] ALIGN1 =
1409                 /* name             has_arg             short */
1410                 "in-place\0"        Optional_argument   "i"
1411                 "regexp-extended\0" No_argument         "r"
1412                 "quiet\0"           No_argument         "n"
1413                 "silent\0"          No_argument         "n"
1414                 "expression\0"      Required_argument   "e"
1415                 "file\0"            Required_argument   "f";
1416 #endif
1417
1418         int status = EXIT_SUCCESS;
1419
1420         INIT_G();
1421
1422         /* destroy command strings on exit */
1423         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1424
1425         /* Lie to autoconf when it starts asking stupid questions. */
1426         if (argv[1] && strcmp(argv[1], "--version") == 0) {
1427                 puts("This is not GNU sed version 4.0");
1428                 return 0;
1429         }
1430
1431         /* do normal option parsing */
1432         opt_e = opt_f = NULL;
1433         opt_i = NULL;
1434         opt_complementary = "e::f::" /* can occur multiple times */
1435                             "nn"; /* count -n */
1436
1437         IF_LONG_OPTS(applet_long_options = sed_longopts);
1438
1439         /* -i must be first, to match OPT_in_place definition */
1440         opt = getopt32(argv, "i::rne:f:", &opt_i, &opt_e, &opt_f,
1441                             &G.be_quiet); /* counter for -n */
1442         //argc -= optind;
1443         argv += optind;
1444         if (opt & OPT_in_place) { // -i
1445                 atexit(cleanup_outname);
1446         }
1447         if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r
1448         //if (opt & 0x4) G.be_quiet++; // -n
1449         while (opt_e) { // -e
1450                 add_cmd_block(llist_pop(&opt_e));
1451         }
1452         while (opt_f) { // -f
1453                 char *line;
1454                 FILE *cmdfile;
1455                 cmdfile = xfopen_for_read(llist_pop(&opt_f));
1456                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1457                         add_cmd(line);
1458                         free(line);
1459                 }
1460                 fclose(cmdfile);
1461         }
1462         /* if we didn't get a pattern from -e or -f, use argv[0] */
1463         if (!(opt & 0x18)) {
1464                 if (!*argv)
1465                         bb_show_usage();
1466                 add_cmd_block(*argv++);
1467         }
1468         /* Flush any unfinished commands. */
1469         add_cmd("");
1470
1471         /* By default, we write to stdout */
1472         G.nonstdout = stdout;
1473
1474         /* argv[0..(argc-1)] should be names of file to process. If no
1475          * files were specified or '-' was specified, take input from stdin.
1476          * Otherwise, we process all the files specified. */
1477         if (argv[0] == NULL) {
1478                 if (opt & OPT_in_place)
1479                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1480                 add_input_file(stdin);
1481         } else {
1482                 int i;
1483
1484                 for (i = 0; argv[i]; i++) {
1485                         struct stat statbuf;
1486                         int nonstdoutfd;
1487                         FILE *file;
1488                         sed_cmd_t *sed_cmd;
1489
1490                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1491                                 add_input_file(stdin);
1492                                 process_files();
1493                                 continue;
1494                         }
1495                         file = fopen_or_warn(argv[i], "r");
1496                         if (!file) {
1497                                 status = EXIT_FAILURE;
1498                                 continue;
1499                         }
1500                         add_input_file(file);
1501                         if (!(opt & OPT_in_place)) {
1502                                 continue;
1503                         }
1504
1505                         /* -i: process each FILE separately: */
1506
1507                         G.outname = xasprintf("%sXXXXXX", argv[i]);
1508                         nonstdoutfd = xmkstemp(G.outname);
1509                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1510
1511                         /* Set permissions/owner of output file */
1512                         fstat(fileno(file), &statbuf);
1513                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1514                          * but GNU sed 4.2.1 does not preserve them either */
1515                         fchmod(nonstdoutfd, statbuf.st_mode);
1516                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1517
1518                         process_files();
1519                         fclose(G.nonstdout);
1520                         G.nonstdout = stdout;
1521
1522                         if (opt_i) {
1523                                 char *backupname = xasprintf("%s%s", argv[i], opt_i);
1524                                 xrename(argv[i], backupname);
1525                                 free(backupname);
1526                         }
1527                         /* else unlink(argv[i]); - rename below does this */
1528                         xrename(G.outname, argv[i]); //TODO: rollback backup on error?
1529                         free(G.outname);
1530                         G.outname = NULL;
1531
1532                         /* Re-enable disabled range matches */
1533                         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1534                                 sed_cmd->beg_line = sed_cmd->beg_line_orig;
1535                         }
1536                 }
1537                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1538                  * if (G.current_input_file >= G.input_file_count)
1539                  *      return status;
1540                  * but it's not needed since process_files() works correctly
1541                  * in this case too. */
1542         }
1543         process_files();
1544
1545         return status;
1546 }