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  * process_files() does actual sedding, reading data lines from each input FILE*
  27  * (which could be stdin) and applying the sed command list (sed_cmd_head) to
  28  * each of the resulting lines.
  29  *
  30  * sed_main() is where external code calls into this, with a command line.
  31  */
  32
  33 /* Supported features and commands in this version of sed:
  34  *
  35  * - comments ('#')
  36  * - address matching: num|/matchstr/[,num|/matchstr/|$]command
  37  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  38  * - edit commands: (a)ppend, (i)nsert, (c)hange
  39  * - file commands: (r)ead
  40  * - backreferences in substitution expressions (\0, \1, \2...\9)
  41  * - grouped commands: {cmd1;cmd2}
  42  * - transliteration (y/source-chars/dest-chars/)
  43  * - pattern space hold space storing / swapping (g, h, x)
  44  * - labels / branching (: label, b, t, T)
  45  *
  46  * (Note: Specifying an address (range) to match is *optional*; commands
  47  * default to the whole pattern space if no specific address match was
  48  * requested.)
  49  *
  50  * Todo:
  51  * - Create a wrapper around regex to make libc's regex conform with sed
  52  *
  53  * Reference
  54  * http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  55  * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html
  56  * http://sed.sourceforge.net/sedfaq3.html
  57  */
  58
  59 //config:config SED
  60 //config:       bool "sed"
  61 //config:       default y
  62 //config:       help
  63 //config:         sed is used to perform text transformations on a file
  64 //config:         or input from a pipeline.
  65
  66 //kbuild:lib-$(CONFIG_SED) += sed.o
  67
  68 //applet:IF_SED(APPLET(sed, BB_DIR_BIN, BB_SUID_DROP))
  69
  70 //usage:#define sed_trivial_usage
  71 //usage:       "[-inrE] [-f FILE]... [-e CMD]... [FILE]...\n"
  72 //usage:       "or: sed [-inrE] CMD [FILE]..."
  73 //usage:#define sed_full_usage "\n\n"
  74 //usage:       "        -e CMD  Add CMD to sed commands to be executed"
  75 //usage:     "\n        -f FILE Add FILE contents to sed commands to be executed"
  76 //usage:     "\n        -i[SFX] Edit files in-place (otherwise sends to stdout)"
  77 //usage:     "\n                Optionally back files up, appending SFX"
  78 //usage:     "\n        -n      Suppress automatic printing of pattern space"
  79 //usage:     "\n        -r,-E   Use extended regex syntax"
  80 //usage:     "\n"
  81 //usage:     "\nIf no -e or -f, the first non-option argument is the sed command string."
  82 //usage:     "\nRemaining arguments are input files (stdin if none)."
  83 //usage:
  84 //usage:#define sed_example_usage
  85 //usage:       "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
  86 //usage:       "bar\n"
  87
  88 #include "libbb.h"
  89 #include "common_bufsiz.h"
  90 #include "xregex.h"
  91
  92 #if 0
  93 # define dbg(...) bb_error_msg(__VA_ARGS__)
  94 #else
  95 # define dbg(...) ((void)0)
  96 #endif
  97
  98
  99 enum {
 100         OPT_in_place = 1 << 0,
 101 };
 102
 103 /* Each sed command turns into one of these structures. */
 104 typedef struct sed_cmd_s {
 105         /* Ordered by alignment requirements: currently 36 bytes on x86 */
 106         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
 107
 108         /* address storage */
 109         regex_t *beg_match;     /* sed -e '/match/cmd' */
 110         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
 111         regex_t *sub_match;     /* For 's/sub_match/string/' */
 112         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
 113         int beg_line_orig;      /* copy of the above, needed for -i */
 114         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($). -2-N = +N */
 115         int end_line_orig;
 116
 117         FILE *sw_file;          /* File (sw) command writes to, NULL for none. */
 118         char *string;           /* Data string for (saicytb) commands. */
 119
 120         unsigned which_match;   /* (s) Which match to replace (0 for all) */
 121
 122         /* Bitfields (gcc won't group them if we don't) */
 123         unsigned invert:1;      /* the '!' after the address */
 124         unsigned in_match:1;    /* Next line also included in match? */
 125         unsigned sub_p:1;       /* (s) print option */
 126
 127         char sw_last_char;      /* Last line written by (sw) had no '\n' */
 128
 129         /* GENERAL FIELDS */
 130         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 131 } sed_cmd_t;
 132
 133 static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
 134
 135 struct globals {
 136         /* options */
 137         int be_quiet, regex_type;
 138
 139         FILE *nonstdout;
 140         char *outname, *hold_space;
 141         smallint exitcode;
 142
 143         /* list of input files */
 144         int current_input_file, last_input_file;
 145         char **input_file_list;
 146         FILE *current_fp;
 147
 148         regmatch_t regmatch[10];
 149         regex_t *previous_regex_ptr;
 150
 151         /* linked list of sed commands */
 152         sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
 153
 154         /* linked list of append lines */
 155         llist_t *append_head;
 156
 157         char *add_cmd_line;
 158
 159         struct pipeline {
 160                 char *buf;  /* Space to hold string */
 161                 int idx;    /* Space used */
 162                 int len;    /* Space allocated */
 163         } pipeline;
 164 } FIX_ALIASING;
 165 #define G (*(struct globals*)bb_common_bufsiz1)
 166 #define INIT_G() do { \
 167         setup_common_bufsiz(); \
 168         BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 169         G.sed_cmd_tail = &G.sed_cmd_head; \
 170 } while (0)
 171
 172
 173 #if ENABLE_FEATURE_CLEAN_UP
 174 static void sed_free_and_close_stuff(void)
 175 {
 176         sed_cmd_t *sed_cmd = G.sed_cmd_head;
 177
 178         llist_free(G.append_head, free);
 179
 180         while (sed_cmd) {
 181                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 182
 183                 if (sed_cmd->sw_file)
 184                         fclose(sed_cmd->sw_file);
 185
 186                 if (sed_cmd->beg_match) {
 187                         regfree(sed_cmd->beg_match);
 188                         free(sed_cmd->beg_match);
 189                 }
 190                 if (sed_cmd->end_match) {
 191                         regfree(sed_cmd->end_match);
 192                         free(sed_cmd->end_match);
 193                 }
 194                 if (sed_cmd->sub_match) {
 195                         regfree(sed_cmd->sub_match);
 196                         free(sed_cmd->sub_match);
 197                 }
 198                 free(sed_cmd->string);
 199                 free(sed_cmd);
 200                 sed_cmd = sed_cmd_next;
 201         }
 202
 203         free(G.hold_space);
 204
 205         if (G.current_fp)
 206                 fclose(G.current_fp);
 207 }
 208 #else
 209 void sed_free_and_close_stuff(void);
 210 #endif
 211
 212 /* If something bad happens during -i operation, delete temp file */
 213
 214 static void cleanup_outname(void)
 215 {
 216         if (G.outname) unlink(G.outname);
 217 }
 218
 219 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
 220
 221 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 222 {
 223         int i = 0;
 224
 225         while (i < len) {
 226                 if (string[i] == '\\') {
 227                         if (!to || string[i+1] == from) {
 228                                 *dest++ = to ? to : string[i+1];
 229                                 i += 2;
 230                                 continue;
 231                         }
 232                         *dest++ = string[i++];
 233                 }
 234                 /* TODO: is it safe wrt a string with trailing '\\' ? */
 235                 *dest++ = string[i++];
 236         }
 237         *dest = '\0';
 238 }
 239
 240 static char *copy_parsing_escapes(const char *string, int len)
 241 {
 242         const char *s;
 243         char *dest = xmalloc(len + 1);
 244
 245         /* sed recognizes \n */
 246         /* GNU sed also recognizes \t and \r */
 247         for (s = "\nn\tt\rr"; *s; s += 2) {
 248                 parse_escapes(dest, string, len, s[1], s[0]);
 249                 string = dest;
 250                 len = strlen(dest);
 251         }
 252         return dest;
 253 }
 254
 255
 256 /*
 257  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 258  * beginning at a specified index and returns the index of the next regular
 259  * expression delimiter (typically a forward slash ('/')) not preceded by
 260  * a backslash ('\').  A negative delimiter disables square bracket checking.
 261  */
 262 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
 263 {
 264         int bracket = -1;
 265         int escaped = 0;
 266         int idx = 0;
 267         char ch;
 268
 269         if (delimiter < 0) {
 270                 bracket--;
 271                 delimiter = -delimiter;
 272         }
 273
 274         for (; (ch = str[idx]) != '\0'; idx++) {
 275                 if (bracket >= 0) {
 276                         if (ch == ']'
 277                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
 278                         ) {
 279                                 bracket = -1;
 280                         }
 281                 } else if (escaped)
 282                         escaped = 0;
 283                 else if (ch == '\\')
 284                         escaped = 1;
 285                 else if (bracket == -1 && ch == '[')
 286                         bracket = idx;
 287                 else if (ch == delimiter)
 288                         return idx;
 289         }
 290
 291         /* if we make it to here, we've hit the end of the string */
 292         bb_error_msg_and_die("unmatched '%c'", delimiter);
 293 }
 294
 295 /*
 296  *  Returns the index of the third delimiter
 297  */
 298 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 299 {
 300         const char *cmdstr_ptr = cmdstr;
 301         unsigned char delimiter;
 302         int idx = 0;
 303
 304         /* verify that the 's' or 'y' is followed by something.  That something
 305          * (typically a 'slash') is now our regexp delimiter... */
 306         if (*cmdstr == '\0')
 307                 bb_error_msg_and_die("bad format in substitution expression");
 308         delimiter = *cmdstr_ptr++;
 309
 310         /* save the match string */
 311         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 312         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 313
 314         /* save the replacement string */
 315         cmdstr_ptr += idx + 1;
 316         idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
 317         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 318
 319         return ((cmdstr_ptr - cmdstr) + idx);
 320 }
 321
 322 /*
 323  * returns the index in the string just past where the address ends.
 324  */
 325 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
 326 {
 327         const char *pos = my_str;
 328
 329         if (isdigit(*my_str)) {
 330                 *linenum = strtol(my_str, (char**)&pos, 10);
 331                 /* endstr shouldnt ever equal NULL */
 332         } else if (*my_str == '$') {
 333                 *linenum = -1;
 334                 pos++;
 335         } else if (*my_str == '/' || *my_str == '\\') {
 336                 int next;
 337                 char delimiter;
 338                 char *temp;
 339
 340                 delimiter = '/';
 341                 if (*my_str == '\\')
 342                         delimiter = *++pos;
 343                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 344                 temp = copy_parsing_escapes(pos, next);
 345                 *regex = xzalloc(sizeof(regex_t));
 346                 xregcomp(*regex, temp, G.regex_type);
 347                 free(temp);
 348                 /* Move position to next character after last delimiter */
 349                 pos += (next+1);
 350         }
 351         return pos - my_str;
 352 }
 353
 354 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 355 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
 356 {
 357         int start = 0, idx, hack = 0;
 358
 359         /* Skip whitespace, then grab filename to end of line */
 360         while (isspace(filecmdstr[start]))
 361                 start++;
 362         idx = start;
 363         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
 364                 idx++;
 365
 366         /* If lines glued together, put backslash back. */
 367         if (filecmdstr[idx] == '\n')
 368                 hack = 1;
 369         if (idx == start)
 370                 bb_error_msg_and_die("empty filename");
 371         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 372         if (hack)
 373                 (*retval)[idx] = '\\';
 374
 375         return idx;
 376 }
 377
 378 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
 379 {
 380         int cflags = G.regex_type;
 381         char *match;
 382         int idx;
 383
 384         /*
 385          * A substitution command should look something like this:
 386          *    s/match/replace/ #giIpw
 387          *    ||     |        |||
 388          *    mandatory       optional
 389          */
 390         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 391
 392         /* determine the number of back references in the match string */
 393         /* Note: we compute this here rather than in the do_subst_command()
 394          * function to save processor time, at the expense of a little more memory
 395          * (4 bits) per sed_cmd */
 396
 397         /* process the flags */
 398
 399         sed_cmd->which_match = 1;
 400         dbg("s flags:'%s'", substr + idx + 1);
 401         while (substr[++idx]) {
 402                 dbg("s flag:'%c'", substr[idx]);
 403                 /* Parse match number */
 404                 if (isdigit(substr[idx])) {
 405                         if (match[0] != '^') {
 406                                 /* Match 0 treated as all, multiple matches we take the last one. */
 407                                 const char *pos = substr + idx;
 408 /* FIXME: error check? */
 409                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
 410                                 idx = pos - substr - 1;
 411                         }
 412                         continue;
 413                 }
 414                 /* Skip spaces */
 415                 if (isspace(substr[idx]))
 416                         continue;
 417
 418                 switch (substr[idx]) {
 419                 /* Replace all occurrences */
 420                 case 'g':
 421                         if (match[0] != '^')
 422                                 sed_cmd->which_match = 0;
 423                         break;
 424                 /* Print pattern space */
 425                 case 'p':
 426                         sed_cmd->sub_p = 1;
 427                         break;
 428                 /* Write to file */
 429                 case 'w':
 430                 {
 431                         char *fname;
 432                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname);
 433                         sed_cmd->sw_file = xfopen_for_write(fname);
 434                         sed_cmd->sw_last_char = '\n';
 435                         free(fname);
 436                         break;
 437                 }
 438                 /* Ignore case (gnu exension) */
 439                 case 'i':
 440                 case 'I':
 441                         cflags |= REG_ICASE;
 442                         break;
 443                 /* Comment */
 444                 case '#':
 445                         // while (substr[++idx]) continue;
 446                         idx += strlen(substr + idx); // same
 447                         /* Fall through */
 448                 /* End of command */
 449                 case ';':
 450                 case '}':
 451                         goto out;
 452                 default:
 453                         dbg("s bad flags:'%s'", substr + idx);
 454                         bb_error_msg_and_die("bad option in substitution expression");
 455                 }
 456         }
 457  out:
 458         /* compile the match string into a regex */
 459         if (*match != '\0') {
 460                 /* If match is empty, we use last regex used at runtime */
 461                 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
 462                 dbg("xregcomp('%s',%x)", match, cflags);
 463                 xregcomp(sed_cmd->sub_match, match, cflags);
 464                 dbg("regcomp ok");
 465         }
 466         free(match);
 467
 468         return idx;
 469 }
 470
 471 /*
 472  *  Process the commands arguments
 473  */
 474 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
 475 {
 476         static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
 477         enum {
 478                 IDX_s = 0,
 479                 IDX_a,
 480                 IDX_i,
 481                 IDX_c,
 482                 IDX_r,
 483                 IDX_w,
 484                 IDX_colon,
 485                 IDX_b,
 486                 IDX_t,
 487                 IDX_T,
 488                 IDX_y,
 489                 IDX_d,
 490                 IDX_D,
 491                 IDX_g,
 492                 IDX_G,
 493                 IDX_h,
 494                 IDX_H,
 495                 IDX_l,
 496                 IDX_n,
 497                 IDX_N,
 498                 IDX_p,
 499                 IDX_P,
 500                 IDX_q,
 501                 IDX_x,
 502                 IDX_equal,
 503                 IDX_lbrace,
 504                 IDX_rbrace,
 505                 IDX_nul
 506         };
 507         unsigned idx;
 508
 509         BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul);
 510
 511         idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
 512
 513         /* handle (s)ubstitution command */
 514         if (idx == IDX_s) {
 515                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 516         }
 517         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 518         else if (idx <= IDX_c) { /* a,i,c */
 519                 if (idx < IDX_c) { /* a,i */
 520                         if (sed_cmd->end_line || sed_cmd->end_match)
 521                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 522                 }
 523                 for (;;) {
 524                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 525                                 cmdstr++;
 526                                 break;
 527                         }
 528                         if (!isspace(*cmdstr))
 529                                 break;
 530                         cmdstr++;
 531                 }
 532                 sed_cmd->string = xstrdup(cmdstr);
 533                 /* "\anychar" -> "anychar" */
 534                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
 535                 cmdstr += strlen(cmdstr);
 536         }
 537         /* handle file cmds: (r)ead */
 538         else if (idx <= IDX_w) { /* r,w */
 539                 if (idx < IDX_w) { /* r */
 540                         if (sed_cmd->end_line || sed_cmd->end_match)
 541                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 542                 }
 543                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
 544                 if (sed_cmd->cmd == 'w') {
 545                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
 546                         sed_cmd->sw_last_char = '\n';
 547                 }
 548         }
 549         /* handle branch commands */
 550         else if (idx <= IDX_T) { /* :,b,t,T */
 551                 int length;
 552
 553                 cmdstr = skip_whitespace(cmdstr);
 554                 length = strcspn(cmdstr, semicolon_whitespace);
 555                 if (length) {
 556                         sed_cmd->string = xstrndup(cmdstr, length);
 557                         cmdstr += length;
 558                 }
 559         }
 560         /* translation command */
 561         else if (idx == IDX_y) {
 562                 char *match, *replace;
 563                 int i = cmdstr[0];
 564
 565                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 566                 /* \n already parsed, but \delimiter needs unescaping. */
 567                 parse_escapes(match, match, strlen(match), i, i);
 568                 parse_escapes(replace, replace, strlen(replace), i, i);
 569
 570                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 571                 for (i = 0; match[i] && replace[i]; i++) {
 572                         sed_cmd->string[i*2] = match[i];
 573                         sed_cmd->string[i*2+1] = replace[i];
 574                 }
 575                 free(match);
 576                 free(replace);
 577         }
 578         /* if it wasnt a single-letter command that takes no arguments
 579          * then it must be an invalid command.
 580          */
 581         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
 582                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 583         }
 584
 585         /* give back whatever's left over */
 586         return cmdstr;
 587 }
 588
 589
 590 /* Parse address+command sets, skipping comment lines. */
 591
 592 static void add_cmd(const char *cmdstr)
 593 {
 594         sed_cmd_t *sed_cmd;
 595         unsigned len, n;
 596
 597         /* Append this line to any unfinished line from last time. */
 598         if (G.add_cmd_line) {
 599                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
 600                 free(G.add_cmd_line);
 601                 cmdstr = G.add_cmd_line = tp;
 602         }
 603
 604         /* If this line ends with unescaped backslash, request next line. */
 605         n = len = strlen(cmdstr);
 606         while (n && cmdstr[n-1] == '\\')
 607                 n--;
 608         if ((len - n) & 1) { /* if odd number of trailing backslashes */
 609                 if (!G.add_cmd_line)
 610                         G.add_cmd_line = xstrdup(cmdstr);
 611                 G.add_cmd_line[len-1] = '\0';
 612                 return;
 613         }
 614
 615         /* Loop parsing all commands in this line. */
 616         while (*cmdstr) {
 617                 /* Skip leading whitespace and semicolons */
 618                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 619
 620                 /* If no more commands, exit. */
 621                 if (!*cmdstr) break;
 622
 623                 /* if this is a comment, jump past it and keep going */
 624                 if (*cmdstr == '#') {
 625                         /* "#n" is the same as using -n on the command line */
 626                         if (cmdstr[1] == 'n')
 627                                 G.be_quiet++;
 628                         cmdstr = strpbrk(cmdstr, "\n\r");
 629                         if (!cmdstr) break;
 630                         continue;
 631                 }
 632
 633                 /* parse the command
 634                  * format is: [addr][,addr][!]cmd
 635                  *            |----||-----||-|
 636                  *            part1 part2  part3
 637                  */
 638
 639                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 640
 641                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 642                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 643                 sed_cmd->beg_line_orig = sed_cmd->beg_line;
 644
 645                 /* second part (if present) will begin with a comma */
 646                 if (*cmdstr == ',') {
 647                         int idx;
 648
 649                         cmdstr++;
 650                         if (*cmdstr == '+' && isdigit(cmdstr[1])) {
 651                                 /* http://sed.sourceforge.net/sedfaq3.html#s3.3
 652                                  * Under GNU sed 3.02+, ssed, and sed15+, <address2>
 653                                  * may also be a notation of the form +num,
 654                                  * indicating the next num lines after <address1> is
 655                                  * matched.
 656                                  * GNU sed 4.2.1 accepts even "+" (meaning "+0").
 657                                  * We don't (we check for isdigit, see above), think
 658                                  * about the "+-3" case.
 659                                  */
 660                                 char *end;
 661                                 /* code is smaller compared to using &cmdstr here: */
 662                                 idx = strtol(cmdstr+1, &end, 10);
 663                                 sed_cmd->end_line = -2 - idx;
 664                                 cmdstr = end;
 665                         } else {
 666                                 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 667                                 cmdstr += idx;
 668                                 idx--; /* if 0, trigger error check below */
 669                         }
 670                         if (idx < 0)
 671                                 bb_error_msg_and_die("no address after comma");
 672                         sed_cmd->end_line_orig = sed_cmd->end_line;
 673                 }
 674
 675                 /* skip whitespace before the command */
 676                 cmdstr = skip_whitespace(cmdstr);
 677
 678                 /* Check for inversion flag */
 679                 if (*cmdstr == '!') {
 680                         sed_cmd->invert = 1;
 681                         cmdstr++;
 682
 683                         /* skip whitespace before the command */
 684                         cmdstr = skip_whitespace(cmdstr);
 685                 }
 686
 687                 /* last part (mandatory) will be a command */
 688                 if (!*cmdstr)
 689                         bb_error_msg_and_die("missing command");
 690                 sed_cmd->cmd = *cmdstr++;
 691                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 692
 693                 /* cmdstr now points past args.
 694                  * GNU sed requires a separator, if there are more commands,
 695                  * else it complains "char N: extra characters after command".
 696                  * Example: "sed 'p;d'". We also allow "sed 'pd'".
 697                  */
 698
 699                 /* Add the command to the command array */
 700                 *G.sed_cmd_tail = sed_cmd;
 701                 G.sed_cmd_tail = &sed_cmd->next;
 702         }
 703
 704         /* If we glued multiple lines together, free the memory. */
 705         free(G.add_cmd_line);
 706         G.add_cmd_line = NULL;
 707 }
 708
 709 /* Append to a string, reallocating memory as necessary. */
 710
 711 #define PIPE_GROW 64
 712
 713 static void pipe_putc(char c)
 714 {
 715         if (G.pipeline.idx == G.pipeline.len) {
 716                 G.pipeline.buf = xrealloc(G.pipeline.buf,
 717                                 G.pipeline.len + PIPE_GROW);
 718                 G.pipeline.len += PIPE_GROW;
 719         }
 720         G.pipeline.buf[G.pipeline.idx++] = c;
 721 }
 722
 723 static void do_subst_w_backrefs(char *line, char *replace)
 724 {
 725         int i, j;
 726
 727         /* go through the replacement string */
 728         for (i = 0; replace[i]; i++) {
 729                 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
 730                 if (replace[i] == '\\') {
 731                         unsigned backref = replace[++i] - '0';
 732                         if (backref <= 9) {
 733                                 /* print out the text held in G.regmatch[backref] */
 734                                 if (G.regmatch[backref].rm_so != -1) {
 735                                         j = G.regmatch[backref].rm_so;
 736                                         while (j < G.regmatch[backref].rm_eo)
 737                                                 pipe_putc(line[j++]);
 738                                 }
 739                                 continue;
 740                         }
 741                         /* I _think_ it is impossible to get '\' to be
 742                          * the last char in replace string. Thus we dont check
 743                          * for replace[i] == NUL. (counterexample anyone?) */
 744                         /* if we find a backslash escaped character, print the character */
 745                         pipe_putc(replace[i]);
 746                         continue;
 747                 }
 748                 /* if we find an unescaped '&' print out the whole matched text. */
 749                 if (replace[i] == '&') {
 750                         j = G.regmatch[0].rm_so;
 751                         while (j < G.regmatch[0].rm_eo)
 752                                 pipe_putc(line[j++]);
 753                         continue;
 754                 }
 755                 /* Otherwise just output the character. */
 756                 pipe_putc(replace[i]);
 757         }
 758 }
 759
 760 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
 761 {
 762         char *line = *line_p;
 763         unsigned match_count = 0;
 764         bool altered = 0;
 765         bool prev_match_empty = 1;
 766         bool tried_at_eol = 0;
 767         regex_t *current_regex;
 768
 769         current_regex = sed_cmd->sub_match;
 770         /* Handle empty regex. */
 771         if (!current_regex) {
 772                 current_regex = G.previous_regex_ptr;
 773                 if (!current_regex)
 774                         bb_error_msg_and_die("no previous regexp");
 775         }
 776         G.previous_regex_ptr = current_regex;
 777
 778         /* Find the first match */
 779         dbg("matching '%s'", line);
 780         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
 781                 dbg("no match");
 782                 return 0;
 783         }
 784         dbg("match");
 785
 786         /* Initialize temporary output buffer. */
 787         G.pipeline.buf = xmalloc(PIPE_GROW);
 788         G.pipeline.len = PIPE_GROW;
 789         G.pipeline.idx = 0;
 790
 791         /* Now loop through, substituting for matches */
 792         do {
 793                 int start = G.regmatch[0].rm_so;
 794                 int end = G.regmatch[0].rm_eo;
 795                 int i;
 796
 797                 match_count++;
 798
 799                 /* If we aren't interested in this match, output old line to
 800                  * end of match and continue */
 801                 if (sed_cmd->which_match
 802                  && (sed_cmd->which_match != match_count)
 803                 ) {
 804                         for (i = 0; i < end; i++)
 805                                 pipe_putc(*line++);
 806                         /* Null match? Print one more char */
 807                         if (start == end && *line)
 808                                 pipe_putc(*line++);
 809                         goto next;
 810                 }
 811
 812                 /* Print everything before the match */
 813                 for (i = 0; i < start; i++)
 814                         pipe_putc(line[i]);
 815
 816                 /* Then print the substitution string,
 817                  * unless we just matched empty string after non-empty one.
 818                  * Example: string "cccd", pattern "c*", repl "R":
 819                  * result is "RdR", not "RRdR": first match "ccc",
 820                  * second is "" before "d", third is "" after "d".
 821                  * Second match is NOT replaced!
 822                  */
 823                 if (prev_match_empty || start != 0 || start != end) {
 824                         //dbg("%d %d %d", prev_match_empty, start, end);
 825                         dbg("inserting replacement at %d in '%s'", start, line);
 826                         do_subst_w_backrefs(line, sed_cmd->string);
 827                         /* Flag that something has changed */
 828                         altered = 1;
 829                 } else {
 830                         dbg("NOT inserting replacement at %d in '%s'", start, line);
 831                 }
 832
 833                 /* If matched string is empty (f.e. "c*" pattern),
 834                  * copy verbatim one char after it before attempting more matches
 835                  */
 836                 prev_match_empty = (start == end);
 837                 if (prev_match_empty) {
 838                         if (!line[end]) {
 839                                 tried_at_eol = 1;
 840                         } else {
 841                                 pipe_putc(line[end]);
 842                                 end++;
 843                         }
 844                 }
 845
 846                 /* Advance past the match */
 847                 dbg("line += %d", end);
 848                 line += end;
 849
 850                 /* if we're not doing this globally, get out now */
 851                 if (sed_cmd->which_match != 0)
 852                         break;
 853  next:
 854                 /* Exit if we are at EOL and already tried matching at it */
 855                 if (*line == '\0') {
 856                         if (tried_at_eol)
 857                                 break;
 858                         tried_at_eol = 1;
 859                 }
 860
 861 //maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
 862         } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
 863
 864         /* Copy rest of string into output pipeline */
 865         while (1) {
 866                 char c = *line++;
 867                 pipe_putc(c);
 868                 if (c == '\0')
 869                         break;
 870         }
 871
 872         free(*line_p);
 873         *line_p = G.pipeline.buf;
 874         return altered;
 875 }
 876
 877 /* Set command pointer to point to this label.  (Does not handle null label.) */
 878 static sed_cmd_t *branch_to(char *label)
 879 {
 880         sed_cmd_t *sed_cmd;
 881
 882         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
 883                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 884                         return sed_cmd;
 885                 }
 886         }
 887         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 888 }
 889
 890 static void append(char *s)
 891 {
 892         llist_add_to_end(&G.append_head, s);
 893 }
 894
 895 /* Output line of text. */
 896 /* Note:
 897  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 898  * Without them, we had this:
 899  * echo -n thingy >z1
 900  * echo -n again >z2
 901  * >znull
 902  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 903  * output:
 904  * gnu sed 4.1.5:
 905  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 906  * bbox:
 907  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 908  */
 909 enum {
 910         NO_EOL_CHAR = 1,
 911         LAST_IS_NUL = 2,
 912 };
 913 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 914 {
 915         char lpc = *last_puts_char;
 916
 917         /* Need to insert a '\n' between two files because first file's
 918          * last line wasn't terminated? */
 919         if (lpc != '\n' && lpc != '\0') {
 920                 fputc('\n', file);
 921                 lpc = '\n';
 922         }
 923         fputs(s, file);
 924
 925         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 926         if (s[0])
 927                 lpc = 'x';
 928
 929         /* had trailing '\0' and it was last char of file? */
 930         if (last_gets_char == LAST_IS_NUL) {
 931                 fputc('\0', file);
 932                 lpc = 'x'; /* */
 933         } else
 934         /* had trailing '\n' or '\0'? */
 935         if (last_gets_char != NO_EOL_CHAR) {
 936                 fputc(last_gets_char, file);
 937                 lpc = last_gets_char;
 938         }
 939
 940         if (ferror(file)) {
 941                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 942                 bb_error_msg_and_die(bb_msg_write_error);
 943         }
 944         *last_puts_char = lpc;
 945 }
 946
 947 static void flush_append(char *last_puts_char, char last_gets_char)
 948 {
 949         char *data;
 950
 951         /* Output appended lines. */
 952         while ((data = (char *)llist_pop(&G.append_head))) {
 953                 puts_maybe_newline(data, G.nonstdout, last_puts_char, last_gets_char);
 954                 free(data);
 955         }
 956 }
 957
 958 /* Get next line of input from G.input_file_list, flushing append buffer and
 959  * noting if we ran out of files without a newline on the last line we read.
 960  */
 961 static char *get_next_line(char *gets_char, char *last_puts_char, char last_gets_char)
 962 {
 963         char *temp = NULL;
 964         int len;
 965         char gc;
 966
 967         flush_append(last_puts_char, last_gets_char);
 968
 969         /* will be returned if last line in the file
 970          * doesn't end with either '\n' or '\0' */
 971         gc = NO_EOL_CHAR;
 972         for (; G.current_input_file <= G.last_input_file; G.current_input_file++) {
 973                 FILE *fp = G.current_fp;
 974                 if (!fp) {
 975                         const char *path = G.input_file_list[G.current_input_file];
 976                         fp = stdin;
 977                         if (path != bb_msg_standard_input) {
 978                                 fp = fopen_or_warn(path, "r");
 979                                 if (!fp) {
 980                                         G.exitcode = EXIT_FAILURE;
 981                                         continue;
 982                                 }
 983                         }
 984                         G.current_fp = fp;
 985                 }
 986                 /* Read line up to a newline or NUL byte, inclusive,
 987                  * return malloc'ed char[]. length of the chunk read
 988                  * is stored in len. NULL if EOF/error */
 989                 temp = bb_get_chunk_from_file(fp, &len);
 990                 if (temp) {
 991                         /* len > 0 here, it's ok to do temp[len-1] */
 992                         char c = temp[len-1];
 993                         if (c == '\n' || c == '\0') {
 994                                 temp[len-1] = '\0';
 995                                 gc = c;
 996                                 if (c == '\0') {
 997                                         int ch = fgetc(fp);
 998                                         if (ch != EOF)
 999                                                 ungetc(ch, fp);
1000                                         else
1001                                                 gc = LAST_IS_NUL;
1002                                 }
1003                         }
1004                         /* else we put NO_EOL_CHAR into *gets_char */
1005                         break;
1006
1007                 /* NB: I had the idea of peeking next file(s) and returning
1008                  * NO_EOL_CHAR only if it is the *last* non-empty
1009                  * input file. But there is a case where this won't work:
1010                  * file1: "a woo\nb woo"
1011                  * file2: "c no\nd no"
1012                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
1013                  * (note: *no* newline after "b bang"!) */
1014                 }
1015                 /* Close this file and advance to next one */
1016                 fclose_if_not_stdin(fp);
1017                 G.current_fp = NULL;
1018         }
1019         *gets_char = gc;
1020         return temp;
1021 }
1022
1023 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
1024
1025 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
1026 {
1027         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
1028         if (retval)
1029                 G.previous_regex_ptr = sed_cmd->beg_match;
1030         return retval;
1031 }
1032
1033 /* Process all the lines in all the files */
1034
1035 static void process_files(void)
1036 {
1037         char *pattern_space, *next_line;
1038         int linenum = 0;
1039         char last_puts_char = '\n';
1040         char last_gets_char, next_gets_char;
1041         sed_cmd_t *sed_cmd;
1042         int substituted;
1043
1044         /* Prime the pump */
1045         next_line = get_next_line(&next_gets_char, &last_puts_char, '\n' /*last_gets_char*/);
1046
1047         /* Go through every line in each file */
1048  again:
1049         substituted = 0;
1050
1051         /* Advance to next line.  Stop if out of lines. */
1052         pattern_space = next_line;
1053         if (!pattern_space)
1054                 return;
1055         last_gets_char = next_gets_char;
1056
1057         /* Read one line in advance so we can act on the last line,
1058          * the '$' address */
1059         next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1060         linenum++;
1061
1062         /* For every line, go through all the commands */
1063  restart:
1064         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1065                 int old_matched, matched;
1066
1067                 old_matched = sed_cmd->in_match;
1068
1069                 /* Determine if this command matches this line: */
1070
1071                 dbg("match1:%d", sed_cmd->in_match);
1072                 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1073                                 && !sed_cmd->beg_match && !sed_cmd->end_match));
1074                 dbg("match3:%d", (sed_cmd->beg_line > 0
1075                         && (sed_cmd->end_line || sed_cmd->end_match
1076                             ? (sed_cmd->beg_line <= linenum)
1077                             : (sed_cmd->beg_line == linenum)
1078                             )
1079                         ));
1080                 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1081                 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
1082
1083                 /* Are we continuing a previous multi-line match? */
1084                 sed_cmd->in_match = sed_cmd->in_match
1085                         /* Or is no range necessary? */
1086                         || (!sed_cmd->beg_line && !sed_cmd->end_line
1087                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
1088                         /* Or did we match the start of a numerical range? */
1089                         || (sed_cmd->beg_line > 0
1090                             && (sed_cmd->end_line || sed_cmd->end_match
1091                                   /* note: even if end is numeric and is < linenum too,
1092                                    * GNU sed matches! We match too, therefore we don't
1093                                    * check here that linenum <= end.
1094                                    * Example:
1095                                    * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1096                                    * first three input lines are deleted;
1097                                    * 4th line is matched and printed
1098                                    * by "2,3" (!) and by "4" ranges
1099                                    */
1100                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
1101                                 : (sed_cmd->beg_line == linenum)    /* N */
1102                                 )
1103                             )
1104                         /* Or does this line match our begin address regex? */
1105                         || (beg_match(sed_cmd, pattern_space))
1106                         /* Or did we match last line of input? */
1107                         || (sed_cmd->beg_line == -1 && next_line == NULL);
1108
1109                 /* Snapshot the value */
1110                 matched = sed_cmd->in_match;
1111
1112                 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1113                         sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
1114
1115                 /* Is this line the end of the current match? */
1116
1117                 if (matched) {
1118                         if (sed_cmd->end_line <= -2) {
1119                                 /* address2 is +N, i.e. N lines from beg_line */
1120                                 sed_cmd->end_line = linenum + (-sed_cmd->end_line - 2);
1121                         }
1122                         /* once matched, "n,xxx" range is dead, disabling it */
1123                         if (sed_cmd->beg_line > 0) {
1124                                 sed_cmd->beg_line = -2;
1125                         }
1126                         dbg("end1:%d", sed_cmd->end_line ? sed_cmd->end_line == -1
1127                                                 ? !next_line : (sed_cmd->end_line <= linenum)
1128                                         : !sed_cmd->end_match);
1129                         dbg("end2:%d", sed_cmd->end_match && old_matched
1130                                         && !regexec(sed_cmd->end_match,pattern_space, 0, NULL, 0));
1131                         sed_cmd->in_match = !(
1132                                 /* has the ending line come, or is this a single address command? */
1133                                 (sed_cmd->end_line
1134                                         ? sed_cmd->end_line == -1
1135                                                 ? !next_line
1136                                                 : (sed_cmd->end_line <= linenum)
1137                                         : !sed_cmd->end_match
1138                                 )
1139                                 /* or does this line matches our last address regex */
1140                                 || (sed_cmd->end_match && old_matched
1141                                      && (regexec(sed_cmd->end_match,
1142                                                 pattern_space, 0, NULL, 0) == 0)
1143                                 )
1144                         );
1145                 }
1146
1147                 /* Skip blocks of commands we didn't match */
1148                 if (sed_cmd->cmd == '{') {
1149                         if (sed_cmd->invert ? matched : !matched) {
1150                                 unsigned nest_cnt = 0;
1151                                 while (1) {
1152                                         if (sed_cmd->cmd == '{')
1153                                                 nest_cnt++;
1154                                         if (sed_cmd->cmd == '}') {
1155                                                 nest_cnt--;
1156                                                 if (nest_cnt == 0)
1157                                                         break;
1158                                         }
1159                                         sed_cmd = sed_cmd->next;
1160                                         if (!sed_cmd)
1161                                                 bb_error_msg_and_die("unterminated {");
1162                                 }
1163                         }
1164                         continue;
1165                 }
1166
1167                 /* Okay, so did this line match? */
1168                 if (sed_cmd->invert ? matched : !matched)
1169                         continue; /* no */
1170
1171                 /* Update last used regex in case a blank substitute BRE is found */
1172                 if (sed_cmd->beg_match) {
1173                         G.previous_regex_ptr = sed_cmd->beg_match;
1174                 }
1175
1176                 /* actual sedding */
1177                 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1178                                 pattern_space, next_line, sed_cmd->cmd);
1179                 switch (sed_cmd->cmd) {
1180
1181                 /* Print line number */
1182                 case '=':
1183                         fprintf(G.nonstdout, "%d\n", linenum);
1184                         break;
1185
1186                 /* Write the current pattern space up to the first newline */
1187                 case 'P':
1188                 {
1189                         char *tmp = strchr(pattern_space, '\n');
1190                         if (tmp) {
1191                                 *tmp = '\0';
1192                                 /* TODO: explain why '\n' below */
1193                                 sed_puts(pattern_space, '\n');
1194                                 *tmp = '\n';
1195                                 break;
1196                         }
1197                         /* Fall Through */
1198                 }
1199
1200                 /* Write the current pattern space to output */
1201                 case 'p':
1202                         /* NB: we print this _before_ the last line
1203                          * (of current file) is printed. Even if
1204                          * that line is nonterminated, we print
1205                          * '\n' here (gnu sed does the same) */
1206                         sed_puts(pattern_space, '\n');
1207                         break;
1208                 /* Delete up through first newline */
1209                 case 'D':
1210                 {
1211                         char *tmp = strchr(pattern_space, '\n');
1212                         if (tmp) {
1213                                 overlapping_strcpy(pattern_space, tmp + 1);
1214                                 goto restart;
1215                         }
1216                 }
1217                 /* discard this line. */
1218                 case 'd':
1219                         goto discard_line;
1220
1221                 /* Substitute with regex */
1222                 case 's':
1223                         if (!do_subst_command(sed_cmd, &pattern_space))
1224                                 break;
1225                         dbg("do_subst_command succeeded:'%s'", pattern_space);
1226                         substituted |= 1;
1227
1228                         /* handle p option */
1229                         if (sed_cmd->sub_p)
1230                                 sed_puts(pattern_space, last_gets_char);
1231                         /* handle w option */
1232                         if (sed_cmd->sw_file)
1233                                 puts_maybe_newline(
1234                                         pattern_space, sed_cmd->sw_file,
1235                                         &sed_cmd->sw_last_char, last_gets_char);
1236                         break;
1237
1238                 /* Append line to linked list to be printed later */
1239                 case 'a':
1240                         append(xstrdup(sed_cmd->string));
1241                         break;
1242
1243                 /* Insert text before this line */
1244                 case 'i':
1245                         sed_puts(sed_cmd->string, '\n');
1246                         break;
1247
1248                 /* Cut and paste text (replace) */
1249                 case 'c':
1250                         /* Only triggers on last line of a matching range. */
1251                         if (!sed_cmd->in_match)
1252                                 sed_puts(sed_cmd->string, '\n');
1253                         goto discard_line;
1254
1255                 /* Read file, append contents to output */
1256                 case 'r':
1257                 {
1258                         FILE *rfile;
1259                         rfile = fopen_for_read(sed_cmd->string);
1260                         if (rfile) {
1261                                 char *line;
1262                                 while ((line = xmalloc_fgetline(rfile))
1263                                                 != NULL)
1264                                         append(line);
1265                                 fclose(rfile);
1266                         }
1267
1268                         break;
1269                 }
1270
1271                 /* Write pattern space to file. */
1272                 case 'w':
1273                         puts_maybe_newline(
1274                                 pattern_space, sed_cmd->sw_file,
1275                                 &sed_cmd->sw_last_char, last_gets_char);
1276                         break;
1277
1278                 /* Read next line from input */
1279                 case 'n':
1280                         if (!G.be_quiet)
1281                                 sed_puts(pattern_space, last_gets_char);
1282                         if (next_line) {
1283                                 free(pattern_space);
1284                                 pattern_space = next_line;
1285                                 last_gets_char = next_gets_char;
1286                                 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1287                                 substituted = 0;
1288                                 linenum++;
1289                                 break;
1290                         }
1291                         /* fall through */
1292
1293                 /* Quit.  End of script, end of input. */
1294                 case 'q':
1295                         /* Exit the outer while loop */
1296                         free(next_line);
1297                         next_line = NULL;
1298                         goto discard_commands;
1299
1300                 /* Append the next line to the current line */
1301                 case 'N':
1302                 {
1303                         int len;
1304                         /* If no next line, jump to end of script and exit. */
1305                         /* http://www.gnu.org/software/sed/manual/sed.html:
1306                          * "Most versions of sed exit without printing anything
1307                          * when the N command is issued on the last line of
1308                          * a file. GNU sed prints pattern space before exiting
1309                          * unless of course the -n command switch has been
1310                          * specified. This choice is by design."
1311                          */
1312                         if (next_line == NULL) {
1313                                 //goto discard_line;
1314                                 goto discard_commands; /* GNU behavior */
1315                         }
1316                         /* Append next_line, read new next_line. */
1317                         len = strlen(pattern_space);
1318                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1319                         pattern_space[len] = '\n';
1320                         strcpy(pattern_space + len+1, next_line);
1321                         last_gets_char = next_gets_char;
1322                         next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1323                         linenum++;
1324                         break;
1325                 }
1326
1327                 /* Test/branch if substitution occurred */
1328                 case 't':
1329                         if (!substituted) break;
1330                         substituted = 0;
1331                         /* Fall through */
1332                 /* Test/branch if substitution didn't occur */
1333                 case 'T':
1334                         if (substituted) break;
1335                         /* Fall through */
1336                 /* Branch to label */
1337                 case 'b':
1338                         if (!sed_cmd->string) goto discard_commands;
1339                         else sed_cmd = branch_to(sed_cmd->string);
1340                         break;
1341                 /* Transliterate characters */
1342                 case 'y':
1343                 {
1344                         int i, j;
1345                         for (i = 0; pattern_space[i]; i++) {
1346                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1347                                         if (pattern_space[i] == sed_cmd->string[j]) {
1348                                                 pattern_space[i] = sed_cmd->string[j + 1];
1349                                                 break;
1350                                         }
1351                                 }
1352                         }
1353
1354                         break;
1355                 }
1356                 case 'g':       /* Replace pattern space with hold space */
1357                         free(pattern_space);
1358                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1359                         break;
1360                 case 'G':       /* Append newline and hold space to pattern space */
1361                 {
1362                         int pattern_space_size = 2;
1363                         int hold_space_size = 0;
1364
1365                         if (pattern_space)
1366                                 pattern_space_size += strlen(pattern_space);
1367                         if (G.hold_space)
1368                                 hold_space_size = strlen(G.hold_space);
1369                         pattern_space = xrealloc(pattern_space,
1370                                         pattern_space_size + hold_space_size);
1371                         if (pattern_space_size == 2)
1372                                 pattern_space[0] = 0;
1373                         strcat(pattern_space, "\n");
1374                         if (G.hold_space)
1375                                 strcat(pattern_space, G.hold_space);
1376                         last_gets_char = '\n';
1377
1378                         break;
1379                 }
1380                 case 'h':       /* Replace hold space with pattern space */
1381                         free(G.hold_space);
1382                         G.hold_space = xstrdup(pattern_space);
1383                         break;
1384                 case 'H':       /* Append newline and pattern space to hold space */
1385                 {
1386                         int hold_space_size = 2;
1387                         int pattern_space_size = 0;
1388
1389                         if (G.hold_space)
1390                                 hold_space_size += strlen(G.hold_space);
1391                         if (pattern_space)
1392                                 pattern_space_size = strlen(pattern_space);
1393                         G.hold_space = xrealloc(G.hold_space,
1394                                         hold_space_size + pattern_space_size);
1395
1396                         if (hold_space_size == 2)
1397                                 *G.hold_space = 0;
1398                         strcat(G.hold_space, "\n");
1399                         if (pattern_space)
1400                                 strcat(G.hold_space, pattern_space);
1401
1402                         break;
1403                 }
1404                 case 'x': /* Exchange hold and pattern space */
1405                 {
1406                         char *tmp = pattern_space;
1407                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1408                         last_gets_char = '\n';
1409                         G.hold_space = tmp;
1410                         break;
1411                 }
1412                 } /* switch */
1413         } /* for each cmd */
1414
1415         /*
1416          * Exit point from sedding...
1417          */
1418  discard_commands:
1419         /* we will print the line unless we were told to be quiet ('-n')
1420            or if the line was suppressed (ala 'd'elete) */
1421         if (!G.be_quiet)
1422                 sed_puts(pattern_space, last_gets_char);
1423
1424         /* Delete and such jump here. */
1425  discard_line:
1426         flush_append(&last_puts_char, last_gets_char);
1427         free(pattern_space);
1428
1429         goto again;
1430 }
1431
1432 /* It is possible to have a command line argument with embedded
1433  * newlines.  This counts as multiple command lines.
1434  * However, newline can be escaped: 's/e/z\<newline>z/'
1435  * add_cmd() handles this.
1436  */
1437
1438 static void add_cmd_block(char *cmdstr)
1439 {
1440         char *sv, *eol;
1441
1442         cmdstr = sv = xstrdup(cmdstr);
1443         do {
1444                 eol = strchr(cmdstr, '\n');
1445                 if (eol)
1446                         *eol = '\0';
1447                 add_cmd(cmdstr);
1448                 cmdstr = eol + 1;
1449         } while (eol);
1450         free(sv);
1451 }
1452
1453 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1454 int sed_main(int argc UNUSED_PARAM, char **argv)
1455 {
1456         unsigned opt;
1457         llist_t *opt_e, *opt_f;
1458         char *opt_i;
1459
1460 #if ENABLE_LONG_OPTS
1461         static const char sed_longopts[] ALIGN1 =
1462                 /* name             has_arg             short */
1463                 "in-place\0"        Optional_argument   "i"
1464                 "regexp-extended\0" No_argument         "r"
1465                 "quiet\0"           No_argument         "n"
1466                 "silent\0"          No_argument         "n"
1467                 "expression\0"      Required_argument   "e"
1468                 "file\0"            Required_argument   "f";
1469 #endif
1470
1471         INIT_G();
1472
1473         /* destroy command strings on exit */
1474         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1475
1476         /* Lie to autoconf when it starts asking stupid questions. */
1477         if (argv[1] && strcmp(argv[1], "--version") == 0) {
1478                 puts("This is not GNU sed version 4.0");
1479                 return 0;
1480         }
1481
1482         /* do normal option parsing */
1483         opt_e = opt_f = NULL;
1484         opt_i = NULL;
1485         opt_complementary = "e::f::" /* can occur multiple times */
1486                             "nn"; /* count -n */
1487
1488         IF_LONG_OPTS(applet_long_options = sed_longopts);
1489
1490         /* -i must be first, to match OPT_in_place definition */
1491         /* -E is a synonym of -r:
1492          * GNU sed 4.2.1 mentions it in neither --help
1493          * nor manpage, but does recognize it.
1494          */
1495         opt = getopt32(argv, "i::rEne:f:", &opt_i, &opt_e, &opt_f,
1496                             &G.be_quiet); /* counter for -n */
1497         //argc -= optind;
1498         argv += optind;
1499         if (opt & OPT_in_place) { // -i
1500                 atexit(cleanup_outname);
1501         }
1502         if (opt & (2|4))
1503                 G.regex_type |= REG_EXTENDED; // -r or -E
1504         //if (opt & 8)
1505         //      G.be_quiet++; // -n (implemented with a counter instead)
1506         while (opt_e) { // -e
1507                 add_cmd_block(llist_pop(&opt_e));
1508         }
1509         while (opt_f) { // -f
1510                 char *line;
1511                 FILE *cmdfile;
1512                 cmdfile = xfopen_stdin(llist_pop(&opt_f));
1513                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1514                         add_cmd(line);
1515                         free(line);
1516                 }
1517                 fclose_if_not_stdin(cmdfile);
1518         }
1519         /* if we didn't get a pattern from -e or -f, use argv[0] */
1520         if (!(opt & 0x30)) {
1521                 if (!*argv)
1522                         bb_show_usage();
1523                 add_cmd_block(*argv++);
1524         }
1525         /* Flush any unfinished commands. */
1526         add_cmd("");
1527
1528         /* By default, we write to stdout */
1529         G.nonstdout = stdout;
1530
1531         /* argv[0..(argc-1)] should be names of file to process. If no
1532          * files were specified or '-' was specified, take input from stdin.
1533          * Otherwise, we process all the files specified. */
1534         G.input_file_list = argv;
1535         if (!argv[0]) {
1536                 if (opt & OPT_in_place)
1537                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1538                 argv[0] = (char*)bb_msg_standard_input;
1539                 /* G.last_input_file = 0; - already is */
1540         } else {
1541                 goto start;
1542
1543                 for (; *argv; argv++) {
1544                         struct stat statbuf;
1545                         int nonstdoutfd;
1546                         sed_cmd_t *sed_cmd;
1547
1548                         G.last_input_file++;
1549  start:
1550                         if (!(opt & OPT_in_place)) {
1551                                 if (LONE_DASH(*argv)) {
1552                                         *argv = (char*)bb_msg_standard_input;
1553                                         process_files();
1554                                 }
1555                                 continue;
1556                         }
1557
1558                         /* -i: process each FILE separately: */
1559
1560                         if (stat(*argv, &statbuf) != 0) {
1561                                 bb_simple_perror_msg(*argv);
1562                                 G.exitcode = EXIT_FAILURE;
1563                                 G.current_input_file++;
1564                                 continue;
1565                         }
1566                         G.outname = xasprintf("%sXXXXXX", *argv);
1567                         nonstdoutfd = xmkstemp(G.outname);
1568                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1569                         /* Set permissions/owner of output file */
1570                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1571                          * but GNU sed 4.2.1 does not preserve them either */
1572                         fchmod(nonstdoutfd, statbuf.st_mode);
1573                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1574
1575                         process_files();
1576                         fclose(G.nonstdout);
1577                         G.nonstdout = stdout;
1578
1579                         if (opt_i) {
1580                                 char *backupname = xasprintf("%s%s", *argv, opt_i);
1581                                 xrename(*argv, backupname);
1582                                 free(backupname);
1583                         }
1584                         /* else unlink(*argv); - rename below does this */
1585                         xrename(G.outname, *argv); //TODO: rollback backup on error?
1586                         free(G.outname);
1587                         G.outname = NULL;
1588
1589                         /* Fix disabled range matches and mangled ",+N" ranges */
1590                         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1591                                 sed_cmd->beg_line = sed_cmd->beg_line_orig;
1592                                 sed_cmd->end_line = sed_cmd->end_line_orig;
1593                         }
1594                 }
1595                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1596                  * if (G.current_input_file[G.current_input_file] == NULL)
1597                  *      return G.exitcode;
1598                  * but it's not needed since process_files() works correctly
1599                  * in this case too. */
1600         }
1601
1602         process_files();
1603
1604         return G.exitcode;
1605 }