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