editors/sed.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * sed.c - very minimalist version of sed
   4  *
   5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   7  * Copyright (C) 2002  Matt Kraai
   8  * Copyright (C) 2003 by Glenn McGrath
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * MAINTAINER: Rob Landley <rob@landley.net>
  12  *
  13  * Licensed under GPLv2, see file LICENSE in this source tree.
  14  */
  15
  16 /* Code overview.
  17  *
  18  * Files are laid out to avoid unnecessary function declarations.  So for
  19  * example, every function add_cmd calls occurs before add_cmd in this file.
  20  *
  21  * add_cmd() is called on each line of sed command text (from a file or from
  22  * the command line).  It calls get_address() and parse_cmd_args().  The
  23  * resulting sed_cmd_t structures are appended to a linked list
  24  * (G.sed_cmd_head/G.sed_cmd_tail).
  25  *
  26  * add_input_file() adds a FILE* to the list of input files.  We need to
  27  * know all input sources ahead of time to find the last line for the $ match.
  28  *
  29  * process_files() does actual sedding, reading data lines from each input FILE*
  30  * (which could be stdin) and applying the sed command list (sed_cmd_head) to
  31  * each of the resulting lines.
  32  *
  33  * sed_main() is where external code calls into this, with a command line.
  34  */
  35
  36 /* Supported features and commands in this version of sed:
  37  *
  38  * - comments ('#')
  39  * - address matching: num|/matchstr/[,num|/matchstr/|$]command
  40  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  41  * - edit commands: (a)ppend, (i)nsert, (c)hange
  42  * - file commands: (r)ead
  43  * - backreferences in substitution expressions (\0, \1, \2...\9)
  44  * - grouped commands: {cmd1;cmd2}
  45  * - transliteration (y/source-chars/dest-chars/)
  46  * - pattern space hold space storing / swapping (g, h, x)
  47  * - labels / branching (: label, b, t, T)
  48  *
  49  * (Note: Specifying an address (range) to match is *optional*; commands
  50  * default to the whole pattern space if no specific address match was
  51  * requested.)
  52  *
  53  * Todo:
  54  * - Create a wrapper around regex to make libc's regex conform with sed
  55  *
  56  * Reference
  57  * http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  58  * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html
  59  */
  60
  61 //config:config SED
  62 //config:       bool "sed"
  63 //config:       default y
  64 //config:       help
  65 //config:         sed is used to perform text transformations on a file
  66 //config:         or input from a pipeline.
  67
  68 //kbuild:lib-$(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         FILE *nonstdout;
 139         char *outname, *hold_space;
 140
 141         /* List of input files */
 142         int input_file_count, current_input_file;
 143         FILE **input_file_list;
 144
 145         regmatch_t regmatch[10];
 146         regex_t *previous_regex_ptr;
 147
 148         /* linked list of sed commands */
 149         sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
 150
 151         /* Linked list of append lines */
 152         llist_t *append_head;
 153
 154         char *add_cmd_line;
 155
 156         struct pipeline {
 157                 char *buf;  /* Space to hold string */
 158                 int idx;    /* Space used */
 159                 int len;    /* Space allocated */
 160         } pipeline;
 161 } FIX_ALIASING;
 162 #define G (*(struct globals*)&bb_common_bufsiz1)
 163 struct BUG_G_too_big {
 164         char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
 165 };
 166 #define INIT_G() do { \
 167         G.sed_cmd_tail = &G.sed_cmd_head; \
 168 } while (0)
 169
 170
 171 #if ENABLE_FEATURE_CLEAN_UP
 172 static void sed_free_and_close_stuff(void)
 173 {
 174         sed_cmd_t *sed_cmd = G.sed_cmd_head;
 175
 176         llist_free(G.append_head, free);
 177
 178         while (sed_cmd) {
 179                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 180
 181                 if (sed_cmd->sw_file)
 182                         xprint_and_close_file(sed_cmd->sw_file);
 183
 184                 if (sed_cmd->beg_match) {
 185                         regfree(sed_cmd->beg_match);
 186                         free(sed_cmd->beg_match);
 187                 }
 188                 if (sed_cmd->end_match) {
 189                         regfree(sed_cmd->end_match);
 190                         free(sed_cmd->end_match);
 191                 }
 192                 if (sed_cmd->sub_match) {
 193                         regfree(sed_cmd->sub_match);
 194                         free(sed_cmd->sub_match);
 195                 }
 196                 free(sed_cmd->string);
 197                 free(sed_cmd);
 198                 sed_cmd = sed_cmd_next;
 199         }
 200
 201         free(G.hold_space);
 202
 203         while (G.current_input_file < G.input_file_count)
 204                 fclose(G.input_file_list[G.current_input_file++]);
 205 }
 206 #else
 207 void sed_free_and_close_stuff(void);
 208 #endif
 209
 210 /* If something bad happens during -i operation, delete temp file */
 211
 212 static void cleanup_outname(void)
 213 {
 214         if (G.outname) unlink(G.outname);
 215 }
 216
 217 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
 218
 219 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 220 {
 221         int i = 0;
 222
 223         while (i < len) {
 224                 if (string[i] == '\\') {
 225                         if (!to || string[i+1] == from) {
 226                                 *dest++ = to ? to : string[i+1];
 227                                 i += 2;
 228                                 continue;
 229                         }
 230                         *dest++ = string[i++];
 231                 }
 232                 /* TODO: is it safe wrt a string with trailing '\\' ? */
 233                 *dest++ = string[i++];
 234         }
 235         *dest = '\0';
 236 }
 237
 238 static char *copy_parsing_escapes(const char *string, int len)
 239 {
 240         const char *s;
 241         char *dest = xmalloc(len + 1);
 242
 243         /* sed recognizes \n */
 244         /* GNU sed also recognizes \t and \r */
 245         for (s = "\nn\tt\rr"; *s; s += 2) {
 246                 parse_escapes(dest, string, len, s[1], s[0]);
 247                 string = dest;
 248                 len = strlen(dest);
 249         }
 250         return dest;
 251 }
 252
 253
 254 /*
 255  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 256  * beginning at a specified index and returns the index of the next regular
 257  * expression delimiter (typically a forward slash ('/')) not preceded by
 258  * a backslash ('\').  A negative delimiter disables square bracket checking.
 259  */
 260 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
 261 {
 262         int bracket = -1;
 263         int escaped = 0;
 264         int idx = 0;
 265         char ch;
 266
 267         if (delimiter < 0) {
 268                 bracket--;
 269                 delimiter = -delimiter;
 270         }
 271
 272         for (; (ch = str[idx]) != '\0'; idx++) {
 273                 if (bracket >= 0) {
 274                         if (ch == ']'
 275                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
 276                         ) {
 277                                 bracket = -1;
 278                         }
 279                 } else if (escaped)
 280                         escaped = 0;
 281                 else if (ch == '\\')
 282                         escaped = 1;
 283                 else if (bracket == -1 && ch == '[')
 284                         bracket = idx;
 285                 else if (ch == delimiter)
 286                         return idx;
 287         }
 288
 289         /* if we make it to here, we've hit the end of the string */
 290         bb_error_msg_and_die("unmatched '%c'", delimiter);
 291 }
 292
 293 /*
 294  *  Returns the index of the third delimiter
 295  */
 296 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 297 {
 298         const char *cmdstr_ptr = cmdstr;
 299         unsigned char delimiter;
 300         int idx = 0;
 301
 302         /* verify that the 's' or 'y' is followed by something.  That something
 303          * (typically a 'slash') is now our regexp delimiter... */
 304         if (*cmdstr == '\0')
 305                 bb_error_msg_and_die("bad format in substitution expression");
 306         delimiter = *cmdstr_ptr++;
 307
 308         /* save the match string */
 309         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 310         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 311
 312         /* save the replacement string */
 313         cmdstr_ptr += idx + 1;
 314         idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
 315         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 316
 317         return ((cmdstr_ptr - cmdstr) + idx);
 318 }
 319
 320 /*
 321  * returns the index in the string just past where the address ends.
 322  */
 323 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
 324 {
 325         const char *pos = my_str;
 326
 327         if (isdigit(*my_str)) {
 328                 *linenum = strtol(my_str, (char**)&pos, 10);
 329                 /* endstr shouldnt ever equal NULL */
 330         } else if (*my_str == '$') {
 331                 *linenum = -1;
 332                 pos++;
 333         } else if (*my_str == '/' || *my_str == '\\') {
 334                 int next;
 335                 char delimiter;
 336                 char *temp;
 337
 338                 delimiter = '/';
 339                 if (*my_str == '\\')
 340                         delimiter = *++pos;
 341                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 342                 temp = copy_parsing_escapes(pos, next);
 343                 *regex = xzalloc(sizeof(regex_t));
 344                 xregcomp(*regex, temp, G.regex_type);
 345                 free(temp);
 346                 /* Move position to next character after last delimiter */
 347                 pos += (next+1);
 348         }
 349         return pos - my_str;
 350 }
 351
 352 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 353 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
 354 {
 355         int start = 0, idx, hack = 0;
 356
 357         /* Skip whitespace, then grab filename to end of line */
 358         while (isspace(filecmdstr[start]))
 359                 start++;
 360         idx = start;
 361         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
 362                 idx++;
 363
 364         /* If lines glued together, put backslash back. */
 365         if (filecmdstr[idx] == '\n')
 366                 hack = 1;
 367         if (idx == start)
 368                 bb_error_msg_and_die("empty filename");
 369         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 370         if (hack)
 371                 (*retval)[idx] = '\\';
 372
 373         return idx;
 374 }
 375
 376 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
 377 {
 378         int cflags = G.regex_type;
 379         char *match;
 380         int idx;
 381
 382         /*
 383          * A substitution command should look something like this:
 384          *    s/match/replace/ #gIpw
 385          *    ||     |        |||
 386          *    mandatory       optional
 387          */
 388         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 389
 390         /* determine the number of back references in the match string */
 391         /* Note: we compute this here rather than in the do_subst_command()
 392          * function to save processor time, at the expense of a little more memory
 393          * (4 bits) per sed_cmd */
 394
 395         /* process the flags */
 396
 397         sed_cmd->which_match = 1;
 398         while (substr[++idx]) {
 399                 /* Parse match number */
 400                 if (isdigit(substr[idx])) {
 401                         if (match[0] != '^') {
 402                                 /* Match 0 treated as all, multiple matches we take the last one. */
 403                                 const char *pos = substr + idx;
 404 /* FIXME: error check? */
 405                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
 406                                 idx = pos - substr;
 407                         }
 408                         continue;
 409                 }
 410                 /* Skip spaces */
 411                 if (isspace(substr[idx]))
 412                         continue;
 413
 414                 switch (substr[idx]) {
 415                 /* Replace all occurrences */
 416                 case 'g':
 417                         if (match[0] != '^')
 418                                 sed_cmd->which_match = 0;
 419                         break;
 420                 /* Print pattern space */
 421                 case 'p':
 422                         sed_cmd->sub_p = 1;
 423                         break;
 424                 /* Write to file */
 425                 case 'w':
 426                 {
 427                         char *temp;
 428                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
 429                         break;
 430                 }
 431                 /* Ignore case (gnu exension) */
 432                 case 'I':
 433                         cflags |= REG_ICASE;
 434                         break;
 435                 /* Comment */
 436                 case '#':
 437                         // while (substr[++idx]) continue;
 438                         idx += strlen(substr + idx); // same
 439                         /* Fall through */
 440                 /* End of command */
 441                 case ';':
 442                 case '}':
 443                         goto out;
 444                 default:
 445                         bb_error_msg_and_die("bad option in substitution expression");
 446                 }
 447         }
 448  out:
 449         /* compile the match string into a regex */
 450         if (*match != '\0') {
 451                 /* If match is empty, we use last regex used at runtime */
 452                 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
 453                 dbg("xregcomp('%s',%x)", match, cflags);
 454                 xregcomp(sed_cmd->sub_match, match, cflags);
 455                 dbg("regcomp ok");
 456         }
 457         free(match);
 458
 459         return idx;
 460 }
 461
 462 /*
 463  *  Process the commands arguments
 464  */
 465 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
 466 {
 467         static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
 468         enum {
 469                 IDX_s = 0,
 470                 IDX_a,
 471                 IDX_i,
 472                 IDX_c,
 473                 IDX_r,
 474                 IDX_w,
 475                 IDX_colon,
 476                 IDX_b,
 477                 IDX_t,
 478                 IDX_T,
 479                 IDX_y,
 480                 IDX_d,
 481                 IDX_D,
 482                 IDX_g,
 483                 IDX_G,
 484                 IDX_h,
 485                 IDX_H,
 486                 IDX_l,
 487                 IDX_n,
 488                 IDX_N,
 489                 IDX_p,
 490                 IDX_P,
 491                 IDX_q,
 492                 IDX_x,
 493                 IDX_equal,
 494                 IDX_lbrace,
 495                 IDX_rbrace,
 496                 IDX_nul
 497         };
 498         struct chk { char chk[sizeof(cmd_letters)-1 == IDX_nul ? 1 : -1]; };
 499
 500         unsigned idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
 501
 502         /* handle (s)ubstitution command */
 503         if (idx == IDX_s) {
 504                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 505         }
 506         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 507         else if (idx <= IDX_c) { /* a,i,c */
 508                 if (idx < IDX_c) { /* a,i */
 509                         if (sed_cmd->end_line || sed_cmd->end_match)
 510                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 511                 }
 512                 for (;;) {
 513                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 514                                 cmdstr++;
 515                                 break;
 516                         }
 517                         if (!isspace(*cmdstr))
 518                                 break;
 519                         cmdstr++;
 520                 }
 521                 sed_cmd->string = xstrdup(cmdstr);
 522                 /* "\anychar" -> "anychar" */
 523                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
 524                 cmdstr += strlen(cmdstr);
 525         }
 526         /* handle file cmds: (r)ead */
 527         else if (idx <= IDX_w) { /* r,w */
 528                 if (idx < IDX_w) { /* r */
 529                         if (sed_cmd->end_line || sed_cmd->end_match)
 530                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 531                 }
 532                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
 533                 if (sed_cmd->cmd == 'w') {
 534                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
 535                         sed_cmd->sw_last_char = '\n';
 536                 }
 537         }
 538         /* handle branch commands */
 539         else if (idx <= IDX_T) { /* :,b,t,T */
 540                 int length;
 541
 542                 cmdstr = skip_whitespace(cmdstr);
 543                 length = strcspn(cmdstr, semicolon_whitespace);
 544                 if (length) {
 545                         sed_cmd->string = xstrndup(cmdstr, length);
 546                         cmdstr += length;
 547                 }
 548         }
 549         /* translation command */
 550         else if (idx == IDX_y) {
 551                 char *match, *replace;
 552                 int i = cmdstr[0];
 553
 554                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 555                 /* \n already parsed, but \delimiter needs unescaping. */
 556                 parse_escapes(match, match, strlen(match), i, i);
 557                 parse_escapes(replace, replace, strlen(replace), i, i);
 558
 559                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 560                 for (i = 0; match[i] && replace[i]; i++) {
 561                         sed_cmd->string[i*2] = match[i];
 562                         sed_cmd->string[i*2+1] = replace[i];
 563                 }
 564                 free(match);
 565                 free(replace);
 566         }
 567         /* if it wasnt a single-letter command that takes no arguments
 568          * then it must be an invalid command.
 569          */
 570         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
 571                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 572         }
 573
 574         /* give back whatever's left over */
 575         return cmdstr;
 576 }
 577
 578
 579 /* Parse address+command sets, skipping comment lines. */
 580
 581 static void add_cmd(const char *cmdstr)
 582 {
 583         sed_cmd_t *sed_cmd;
 584         unsigned len, n;
 585
 586         /* Append this line to any unfinished line from last time. */
 587         if (G.add_cmd_line) {
 588                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
 589                 free(G.add_cmd_line);
 590                 cmdstr = G.add_cmd_line = tp;
 591         }
 592
 593         /* If this line ends with unescaped backslash, request next line. */
 594         n = len = strlen(cmdstr);
 595         while (n && cmdstr[n-1] == '\\')
 596                 n--;
 597         if ((len - n) & 1) { /* if odd number of trailing backslashes */
 598                 if (!G.add_cmd_line)
 599                         G.add_cmd_line = xstrdup(cmdstr);
 600                 G.add_cmd_line[len-1] = '\0';
 601                 return;
 602         }
 603
 604         /* Loop parsing all commands in this line. */
 605         while (*cmdstr) {
 606                 /* Skip leading whitespace and semicolons */
 607                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 608
 609                 /* If no more commands, exit. */
 610                 if (!*cmdstr) break;
 611
 612                 /* if this is a comment, jump past it and keep going */
 613                 if (*cmdstr == '#') {
 614                         /* "#n" is the same as using -n on the command line */
 615                         if (cmdstr[1] == 'n')
 616                                 G.be_quiet++;
 617                         cmdstr = strpbrk(cmdstr, "\n\r");
 618                         if (!cmdstr) break;
 619                         continue;
 620                 }
 621
 622                 /* parse the command
 623                  * format is: [addr][,addr][!]cmd
 624                  *            |----||-----||-|
 625                  *            part1 part2  part3
 626                  */
 627
 628                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 629
 630                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 631                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 632                 sed_cmd->beg_line_orig = sed_cmd->beg_line;
 633
 634                 /* second part (if present) will begin with a comma */
 635                 if (*cmdstr == ',') {
 636                         int idx;
 637
 638                         cmdstr++;
 639                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 640                         if (!idx)
 641                                 bb_error_msg_and_die("no address after comma");
 642                         cmdstr += idx;
 643                 }
 644
 645                 /* skip whitespace before the command */
 646                 cmdstr = skip_whitespace(cmdstr);
 647
 648                 /* Check for inversion flag */
 649                 if (*cmdstr == '!') {
 650                         sed_cmd->invert = 1;
 651                         cmdstr++;
 652
 653                         /* skip whitespace before the command */
 654                         cmdstr = skip_whitespace(cmdstr);
 655                 }
 656
 657                 /* last part (mandatory) will be a command */
 658                 if (!*cmdstr)
 659                         bb_error_msg_and_die("missing command");
 660                 sed_cmd->cmd = *cmdstr++;
 661                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 662
 663                 /* cmdstr now points past args.
 664                  * GNU sed requires a separator, if there are more commands,
 665                  * else it complains "char N: extra characters after command".
 666                  * Example: "sed 'p;d'". We also allow "sed 'pd'".
 667                  */
 668
 669                 /* Add the command to the command array */
 670                 *G.sed_cmd_tail = sed_cmd;
 671                 G.sed_cmd_tail = &sed_cmd->next;
 672         }
 673
 674         /* If we glued multiple lines together, free the memory. */
 675         free(G.add_cmd_line);
 676         G.add_cmd_line = NULL;
 677 }
 678
 679 /* Append to a string, reallocating memory as necessary. */
 680
 681 #define PIPE_GROW 64
 682
 683 static void pipe_putc(char c)
 684 {
 685         if (G.pipeline.idx == G.pipeline.len) {
 686                 G.pipeline.buf = xrealloc(G.pipeline.buf,
 687                                 G.pipeline.len + PIPE_GROW);
 688                 G.pipeline.len += PIPE_GROW;
 689         }
 690         G.pipeline.buf[G.pipeline.idx++] = c;
 691 }
 692
 693 static void do_subst_w_backrefs(char *line, char *replace)
 694 {
 695         int i, j;
 696
 697         /* go through the replacement string */
 698         for (i = 0; replace[i]; i++) {
 699                 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
 700                 if (replace[i] == '\\') {
 701                         unsigned backref = replace[++i] - '0';
 702                         if (backref <= 9) {
 703                                 /* print out the text held in G.regmatch[backref] */
 704                                 if (G.regmatch[backref].rm_so != -1) {
 705                                         j = G.regmatch[backref].rm_so;
 706                                         while (j < G.regmatch[backref].rm_eo)
 707                                                 pipe_putc(line[j++]);
 708                                 }
 709                                 continue;
 710                         }
 711                         /* I _think_ it is impossible to get '\' to be
 712                          * the last char in replace string. Thus we dont check
 713                          * for replace[i] == NUL. (counterexample anyone?) */
 714                         /* if we find a backslash escaped character, print the character */
 715                         pipe_putc(replace[i]);
 716                         continue;
 717                 }
 718                 /* if we find an unescaped '&' print out the whole matched text. */
 719                 if (replace[i] == '&') {
 720                         j = G.regmatch[0].rm_so;
 721                         while (j < G.regmatch[0].rm_eo)
 722                                 pipe_putc(line[j++]);
 723                         continue;
 724                 }
 725                 /* Otherwise just output the character. */
 726                 pipe_putc(replace[i]);
 727         }
 728 }
 729
 730 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
 731 {
 732         char *line = *line_p;
 733         unsigned match_count = 0;
 734         bool altered = 0;
 735         bool prev_match_empty = 1;
 736         bool tried_at_eol = 0;
 737         regex_t *current_regex;
 738
 739         current_regex = sed_cmd->sub_match;
 740         /* Handle empty regex. */
 741         if (!current_regex) {
 742                 current_regex = G.previous_regex_ptr;
 743                 if (!current_regex)
 744                         bb_error_msg_and_die("no previous regexp");
 745         }
 746         G.previous_regex_ptr = current_regex;
 747
 748         /* Find the first match */
 749         dbg("matching '%s'", line);
 750         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
 751                 dbg("no match");
 752                 return 0;
 753         }
 754         dbg("match");
 755
 756         /* Initialize temporary output buffer. */
 757         G.pipeline.buf = xmalloc(PIPE_GROW);
 758         G.pipeline.len = PIPE_GROW;
 759         G.pipeline.idx = 0;
 760
 761         /* Now loop through, substituting for matches */
 762         do {
 763                 int start = G.regmatch[0].rm_so;
 764                 int end = G.regmatch[0].rm_eo;
 765                 int i;
 766
 767                 match_count++;
 768
 769                 /* If we aren't interested in this match, output old line to
 770                  * end of match and continue */
 771                 if (sed_cmd->which_match
 772                  && (sed_cmd->which_match != match_count)
 773                 ) {
 774                         for (i = 0; i < end; i++)
 775                                 pipe_putc(*line++);
 776                         /* Null match? Print one more char */
 777                         if (start == end && *line)
 778                                 pipe_putc(*line++);
 779                         goto next;
 780                 }
 781
 782                 /* Print everything before the match */
 783                 for (i = 0; i < start; i++)
 784                         pipe_putc(line[i]);
 785
 786                 /* Then print the substitution string,
 787                  * unless we just matched empty string after non-empty one.
 788                  * Example: string "cccd", pattern "c*", repl "R":
 789                  * result is "RdR", not "RRdR": first match "ccc",
 790                  * second is "" before "d", third is "" after "d".
 791                  * Second match is NOT replaced!
 792                  */
 793                 if (prev_match_empty || start != 0 || start != end) {
 794                         //dbg("%d %d %d", prev_match_empty, start, end);
 795                         dbg("inserting replacement at %d in '%s'", start, line);
 796                         do_subst_w_backrefs(line, sed_cmd->string);
 797                         /* Flag that something has changed */
 798                         altered = 1;
 799                 } else {
 800                         dbg("NOT inserting replacement at %d in '%s'", start, line);
 801                 }
 802
 803                 /* If matched string is empty (f.e. "c*" pattern),
 804                  * copy verbatim one char after it before attempting more matches
 805                  */
 806                 prev_match_empty = (start == end);
 807                 if (prev_match_empty) {
 808                         if (!line[end]) {
 809                                 tried_at_eol = 1;
 810                         } else {
 811                                 pipe_putc(line[end]);
 812                                 end++;
 813                         }
 814                 }
 815
 816                 /* Advance past the match */
 817                 dbg("line += %d", end);
 818                 line += end;
 819
 820                 /* if we're not doing this globally, get out now */
 821                 if (sed_cmd->which_match != 0)
 822                         break;
 823  next:
 824                 /* Exit if we are at EOL and already tried matching at it */
 825                 if (*line == '\0') {
 826                         if (tried_at_eol)
 827                                 break;
 828                         tried_at_eol = 1;
 829                 }
 830
 831 //maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
 832         } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
 833
 834         /* Copy rest of string into output pipeline */
 835         while (1) {
 836                 char c = *line++;
 837                 pipe_putc(c);
 838                 if (c == '\0')
 839                         break;
 840         }
 841
 842         free(*line_p);
 843         *line_p = G.pipeline.buf;
 844         return altered;
 845 }
 846
 847 /* Set command pointer to point to this label.  (Does not handle null label.) */
 848 static sed_cmd_t *branch_to(char *label)
 849 {
 850         sed_cmd_t *sed_cmd;
 851
 852         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
 853                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 854                         return sed_cmd;
 855                 }
 856         }
 857         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 858 }
 859
 860 static void append(char *s)
 861 {
 862         llist_add_to_end(&G.append_head, s);
 863 }
 864
 865 /* Output line of text. */
 866 /* Note:
 867  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 868  * Without them, we had this:
 869  * echo -n thingy >z1
 870  * echo -n again >z2
 871  * >znull
 872  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 873  * output:
 874  * gnu sed 4.1.5:
 875  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 876  * bbox:
 877  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 878  */
 879 enum {
 880         NO_EOL_CHAR = 1,
 881         LAST_IS_NUL = 2,
 882 };
 883 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 884 {
 885         char lpc = *last_puts_char;
 886
 887         /* Need to insert a '\n' between two files because first file's
 888          * last line wasn't terminated? */
 889         if (lpc != '\n' && lpc != '\0') {
 890                 fputc('\n', file);
 891                 lpc = '\n';
 892         }
 893         fputs(s, file);
 894
 895         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 896         if (s[0])
 897                 lpc = 'x';
 898
 899         /* had trailing '\0' and it was last char of file? */
 900         if (last_gets_char == LAST_IS_NUL) {
 901                 fputc('\0', file);
 902                 lpc = 'x'; /* */
 903         } else
 904         /* had trailing '\n' or '\0'? */
 905         if (last_gets_char != NO_EOL_CHAR) {
 906                 fputc(last_gets_char, file);
 907                 lpc = last_gets_char;
 908         }
 909
 910         if (ferror(file)) {
 911                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 912                 bb_error_msg_and_die(bb_msg_write_error);
 913         }
 914         *last_puts_char = lpc;
 915 }
 916
 917 static void flush_append(char *last_puts_char, char last_gets_char)
 918 {
 919         char *data;
 920
 921         /* Output appended lines. */
 922         while ((data = (char *)llist_pop(&G.append_head))) {
 923                 puts_maybe_newline(data, G.nonstdout, last_puts_char, last_gets_char);
 924                 free(data);
 925         }
 926 }
 927
 928 /* Get next line of input from G.input_file_list, flushing append buffer and
 929  * noting if we ran out of files without a newline on the last line we read.
 930  */
 931 static char *get_next_line(char *gets_char, char *last_puts_char, char last_gets_char)
 932 {
 933         char *temp = NULL;
 934         int len;
 935         char gc;
 936
 937         flush_append(last_puts_char, last_gets_char);
 938
 939         /* will be returned if last line in the file
 940          * doesn't end with either '\n' or '\0' */
 941         gc = NO_EOL_CHAR;
 942         while (G.current_input_file < G.input_file_count) {
 943                 FILE *fp = G.input_file_list[G.current_input_file];
 944                 /* Read line up to a newline or NUL byte, inclusive,
 945                  * return malloc'ed char[]. length of the chunk read
 946                  * is stored in len. NULL if EOF/error */
 947                 temp = bb_get_chunk_from_file(fp, &len);
 948                 if (temp) {
 949                         /* len > 0 here, it's ok to do temp[len-1] */
 950                         char c = temp[len-1];
 951                         if (c == '\n' || c == '\0') {
 952                                 temp[len-1] = '\0';
 953                                 gc = c;
 954                                 if (c == '\0') {
 955                                         int ch = fgetc(fp);
 956                                         if (ch != EOF)
 957                                                 ungetc(ch, fp);
 958                                         else
 959                                                 gc = LAST_IS_NUL;
 960                                 }
 961                         }
 962                         /* else we put NO_EOL_CHAR into *gets_char */
 963                         break;
 964
 965                 /* NB: I had the idea of peeking next file(s) and returning
 966                  * NO_EOL_CHAR only if it is the *last* non-empty
 967                  * input file. But there is a case where this won't work:
 968                  * file1: "a woo\nb woo"
 969                  * file2: "c no\nd no"
 970                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
 971                  * (note: *no* newline after "b bang"!) */
 972                 }
 973                 /* Close this file and advance to next one */
 974                 fclose(fp);
 975                 G.current_input_file++;
 976         }
 977         *gets_char = gc;
 978         return temp;
 979 }
 980
 981 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
 982
 983 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
 984 {
 985         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
 986         if (retval)
 987                 G.previous_regex_ptr = sed_cmd->beg_match;
 988         return retval;
 989 }
 990
 991 /* Process all the lines in all the files */
 992
 993 static void process_files(void)
 994 {
 995         char *pattern_space, *next_line;
 996         int linenum = 0;
 997         char last_puts_char = '\n';
 998         char last_gets_char, next_gets_char;
 999         sed_cmd_t *sed_cmd;
1000         int substituted;
1001
1002         /* Prime the pump */
1003         next_line = get_next_line(&next_gets_char, &last_puts_char, '\n' /*last_gets_char*/);
1004
1005         /* Go through every line in each file */
1006  again:
1007         substituted = 0;
1008
1009         /* Advance to next line.  Stop if out of lines. */
1010         pattern_space = next_line;
1011         if (!pattern_space)
1012                 return;
1013         last_gets_char = next_gets_char;
1014
1015         /* Read one line in advance so we can act on the last line,
1016          * the '$' address */
1017         next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1018         linenum++;
1019
1020         /* For every line, go through all the commands */
1021  restart:
1022         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1023                 int old_matched, matched;
1024
1025                 old_matched = sed_cmd->in_match;
1026
1027                 /* Determine if this command matches this line: */
1028
1029                 dbg("match1:%d", sed_cmd->in_match);
1030                 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1031                                 && !sed_cmd->beg_match && !sed_cmd->end_match));
1032                 dbg("match3:%d", (sed_cmd->beg_line > 0
1033                         && (sed_cmd->end_line || sed_cmd->end_match
1034                             ? (sed_cmd->beg_line <= linenum)
1035                             : (sed_cmd->beg_line == linenum)
1036                             )
1037                         ));
1038                 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1039                 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
1040
1041                 /* Are we continuing a previous multi-line match? */
1042                 sed_cmd->in_match = sed_cmd->in_match
1043                         /* Or is no range necessary? */
1044                         || (!sed_cmd->beg_line && !sed_cmd->end_line
1045                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
1046                         /* Or did we match the start of a numerical range? */
1047                         || (sed_cmd->beg_line > 0
1048                             && (sed_cmd->end_line || sed_cmd->end_match
1049                                   /* note: even if end is numeric and is < linenum too,
1050                                    * GNU sed matches! We match too, therefore we don't
1051                                    * check here that linenum <= end.
1052                                    * Example:
1053                                    * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1054                                    * first three input lines are deleted;
1055                                    * 4th line is matched and printed
1056                                    * by "2,3" (!) and by "4" ranges
1057                                    */
1058                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
1059                                 : (sed_cmd->beg_line == linenum)    /* N */
1060                                 )
1061                             )
1062                         /* Or does this line match our begin address regex? */
1063                         || (beg_match(sed_cmd, pattern_space))
1064                         /* Or did we match last line of input? */
1065                         || (sed_cmd->beg_line == -1 && next_line == NULL);
1066
1067                 /* Snapshot the value */
1068                 matched = sed_cmd->in_match;
1069
1070                 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1071                         sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
1072
1073                 /* Is this line the end of the current match? */
1074
1075                 if (matched) {
1076                         /* once matched, "n,xxx" range is dead, disabling it */
1077                         if (sed_cmd->beg_line > 0) {
1078                                 sed_cmd->beg_line = -2;
1079                         }
1080                         sed_cmd->in_match = !(
1081                                 /* has the ending line come, or is this a single address command? */
1082                                 (sed_cmd->end_line
1083                                         ? sed_cmd->end_line == -1
1084                                                 ? !next_line
1085                                                 : (sed_cmd->end_line <= linenum)
1086                                         : !sed_cmd->end_match
1087                                 )
1088                                 /* or does this line matches our last address regex */
1089                                 || (sed_cmd->end_match && old_matched
1090                                      && (regexec(sed_cmd->end_match,
1091                                                 pattern_space, 0, NULL, 0) == 0)
1092                                 )
1093                         );
1094                 }
1095
1096                 /* Skip blocks of commands we didn't match */
1097                 if (sed_cmd->cmd == '{') {
1098                         if (sed_cmd->invert ? matched : !matched) {
1099                                 unsigned nest_cnt = 0;
1100                                 while (1) {
1101                                         if (sed_cmd->cmd == '{')
1102                                                 nest_cnt++;
1103                                         if (sed_cmd->cmd == '}') {
1104                                                 nest_cnt--;
1105                                                 if (nest_cnt == 0)
1106                                                         break;
1107                                         }
1108                                         sed_cmd = sed_cmd->next;
1109                                         if (!sed_cmd)
1110                                                 bb_error_msg_and_die("unterminated {");
1111                                 }
1112                         }
1113                         continue;
1114                 }
1115
1116                 /* Okay, so did this line match? */
1117                 if (sed_cmd->invert ? matched : !matched)
1118                         continue; /* no */
1119
1120                 /* Update last used regex in case a blank substitute BRE is found */
1121                 if (sed_cmd->beg_match) {
1122                         G.previous_regex_ptr = sed_cmd->beg_match;
1123                 }
1124
1125                 /* actual sedding */
1126                 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1127                                 pattern_space, next_line, sed_cmd->cmd);
1128                 switch (sed_cmd->cmd) {
1129
1130                 /* Print line number */
1131                 case '=':
1132                         fprintf(G.nonstdout, "%d\n", linenum);
1133                         break;
1134
1135                 /* Write the current pattern space up to the first newline */
1136                 case 'P':
1137                 {
1138                         char *tmp = strchr(pattern_space, '\n');
1139                         if (tmp) {
1140                                 *tmp = '\0';
1141                                 /* TODO: explain why '\n' below */
1142                                 sed_puts(pattern_space, '\n');
1143                                 *tmp = '\n';
1144                                 break;
1145                         }
1146                         /* Fall Through */
1147                 }
1148
1149                 /* Write the current pattern space to output */
1150                 case 'p':
1151                         /* NB: we print this _before_ the last line
1152                          * (of current file) is printed. Even if
1153                          * that line is nonterminated, we print
1154                          * '\n' here (gnu sed does the same) */
1155                         sed_puts(pattern_space, '\n');
1156                         break;
1157                 /* Delete up through first newline */
1158                 case 'D':
1159                 {
1160                         char *tmp = strchr(pattern_space, '\n');
1161                         if (tmp) {
1162                                 overlapping_strcpy(pattern_space, tmp + 1);
1163                                 goto restart;
1164                         }
1165                 }
1166                 /* discard this line. */
1167                 case 'd':
1168                         goto discard_line;
1169
1170                 /* Substitute with regex */
1171                 case 's':
1172                         if (!do_subst_command(sed_cmd, &pattern_space))
1173                                 break;
1174                         dbg("do_subst_command succeeded:'%s'", pattern_space);
1175                         substituted |= 1;
1176
1177                         /* handle p option */
1178                         if (sed_cmd->sub_p)
1179                                 sed_puts(pattern_space, last_gets_char);
1180                         /* handle w option */
1181                         if (sed_cmd->sw_file)
1182                                 puts_maybe_newline(
1183                                         pattern_space, sed_cmd->sw_file,
1184                                         &sed_cmd->sw_last_char, last_gets_char);
1185                         break;
1186
1187                 /* Append line to linked list to be printed later */
1188                 case 'a':
1189                         append(xstrdup(sed_cmd->string));
1190                         break;
1191
1192                 /* Insert text before this line */
1193                 case 'i':
1194                         sed_puts(sed_cmd->string, '\n');
1195                         break;
1196
1197                 /* Cut and paste text (replace) */
1198                 case 'c':
1199                         /* Only triggers on last line of a matching range. */
1200                         if (!sed_cmd->in_match)
1201                                 sed_puts(sed_cmd->string, '\n');
1202                         goto discard_line;
1203
1204                 /* Read file, append contents to output */
1205                 case 'r':
1206                 {
1207                         FILE *rfile;
1208                         rfile = fopen_for_read(sed_cmd->string);
1209                         if (rfile) {
1210                                 char *line;
1211                                 while ((line = xmalloc_fgetline(rfile))
1212                                                 != NULL)
1213                                         append(line);
1214                                 fclose(rfile);
1215                         }
1216
1217                         break;
1218                 }
1219
1220                 /* Write pattern space to file. */
1221                 case 'w':
1222                         puts_maybe_newline(
1223                                 pattern_space, sed_cmd->sw_file,
1224                                 &sed_cmd->sw_last_char, last_gets_char);
1225                         break;
1226
1227                 /* Read next line from input */
1228                 case 'n':
1229                         if (!G.be_quiet)
1230                                 sed_puts(pattern_space, last_gets_char);
1231                         if (next_line) {
1232                                 free(pattern_space);
1233                                 pattern_space = next_line;
1234                                 last_gets_char = next_gets_char;
1235                                 next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1236                                 substituted = 0;
1237                                 linenum++;
1238                                 break;
1239                         }
1240                         /* fall through */
1241
1242                 /* Quit.  End of script, end of input. */
1243                 case 'q':
1244                         /* Exit the outer while loop */
1245                         free(next_line);
1246                         next_line = NULL;
1247                         goto discard_commands;
1248
1249                 /* Append the next line to the current line */
1250                 case 'N':
1251                 {
1252                         int len;
1253                         /* If no next line, jump to end of script and exit. */
1254                         /* http://www.gnu.org/software/sed/manual/sed.html:
1255                          * "Most versions of sed exit without printing anything
1256                          * when the N command is issued on the last line of
1257                          * a file. GNU sed prints pattern space before exiting
1258                          * unless of course the -n command switch has been
1259                          * specified. This choice is by design."
1260                          */
1261                         if (next_line == NULL) {
1262                                 //goto discard_line;
1263                                 goto discard_commands; /* GNU behavior */
1264                         }
1265                         /* Append next_line, read new next_line. */
1266                         len = strlen(pattern_space);
1267                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1268                         pattern_space[len] = '\n';
1269                         strcpy(pattern_space + len+1, next_line);
1270                         last_gets_char = next_gets_char;
1271                         next_line = get_next_line(&next_gets_char, &last_puts_char, last_gets_char);
1272                         linenum++;
1273                         break;
1274                 }
1275
1276                 /* Test/branch if substitution occurred */
1277                 case 't':
1278                         if (!substituted) break;
1279                         substituted = 0;
1280                         /* Fall through */
1281                 /* Test/branch if substitution didn't occur */
1282                 case 'T':
1283                         if (substituted) break;
1284                         /* Fall through */
1285                 /* Branch to label */
1286                 case 'b':
1287                         if (!sed_cmd->string) goto discard_commands;
1288                         else sed_cmd = branch_to(sed_cmd->string);
1289                         break;
1290                 /* Transliterate characters */
1291                 case 'y':
1292                 {
1293                         int i, j;
1294                         for (i = 0; pattern_space[i]; i++) {
1295                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1296                                         if (pattern_space[i] == sed_cmd->string[j]) {
1297                                                 pattern_space[i] = sed_cmd->string[j + 1];
1298                                                 break;
1299                                         }
1300                                 }
1301                         }
1302
1303                         break;
1304                 }
1305                 case 'g':       /* Replace pattern space with hold space */
1306                         free(pattern_space);
1307                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1308                         break;
1309                 case 'G':       /* Append newline and hold space to pattern space */
1310                 {
1311                         int pattern_space_size = 2;
1312                         int hold_space_size = 0;
1313
1314                         if (pattern_space)
1315                                 pattern_space_size += strlen(pattern_space);
1316                         if (G.hold_space)
1317                                 hold_space_size = strlen(G.hold_space);
1318                         pattern_space = xrealloc(pattern_space,
1319                                         pattern_space_size + hold_space_size);
1320                         if (pattern_space_size == 2)
1321                                 pattern_space[0] = 0;
1322                         strcat(pattern_space, "\n");
1323                         if (G.hold_space)
1324                                 strcat(pattern_space, G.hold_space);
1325                         last_gets_char = '\n';
1326
1327                         break;
1328                 }
1329                 case 'h':       /* Replace hold space with pattern space */
1330                         free(G.hold_space);
1331                         G.hold_space = xstrdup(pattern_space);
1332                         break;
1333                 case 'H':       /* Append newline and pattern space to hold space */
1334                 {
1335                         int hold_space_size = 2;
1336                         int pattern_space_size = 0;
1337
1338                         if (G.hold_space)
1339                                 hold_space_size += strlen(G.hold_space);
1340                         if (pattern_space)
1341                                 pattern_space_size = strlen(pattern_space);
1342                         G.hold_space = xrealloc(G.hold_space,
1343                                         hold_space_size + pattern_space_size);
1344
1345                         if (hold_space_size == 2)
1346                                 *G.hold_space = 0;
1347                         strcat(G.hold_space, "\n");
1348                         if (pattern_space)
1349                                 strcat(G.hold_space, pattern_space);
1350
1351                         break;
1352                 }
1353                 case 'x': /* Exchange hold and pattern space */
1354                 {
1355                         char *tmp = pattern_space;
1356                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1357                         last_gets_char = '\n';
1358                         G.hold_space = tmp;
1359                         break;
1360                 }
1361                 } /* switch */
1362         } /* for each cmd */
1363
1364         /*
1365          * Exit point from sedding...
1366          */
1367  discard_commands:
1368         /* we will print the line unless we were told to be quiet ('-n')
1369            or if the line was suppressed (ala 'd'elete) */
1370         if (!G.be_quiet)
1371                 sed_puts(pattern_space, last_gets_char);
1372
1373         /* Delete and such jump here. */
1374  discard_line:
1375         flush_append(&last_puts_char, last_gets_char);
1376         free(pattern_space);
1377
1378         goto again;
1379 }
1380
1381 /* It is possible to have a command line argument with embedded
1382  * newlines.  This counts as multiple command lines.
1383  * However, newline can be escaped: 's/e/z\<newline>z/'
1384  * add_cmd() handles this.
1385  */
1386
1387 static void add_cmd_block(char *cmdstr)
1388 {
1389         char *sv, *eol;
1390
1391         cmdstr = sv = xstrdup(cmdstr);
1392         do {
1393                 eol = strchr(cmdstr, '\n');
1394                 if (eol)
1395                         *eol = '\0';
1396                 add_cmd(cmdstr);
1397                 cmdstr = eol + 1;
1398         } while (eol);
1399         free(sv);
1400 }
1401
1402 static void add_input_file(FILE *file)
1403 {
1404         G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count);
1405         G.input_file_list[G.input_file_count++] = file;
1406 }
1407
1408 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1409 int sed_main(int argc UNUSED_PARAM, char **argv)
1410 {
1411         unsigned opt;
1412         llist_t *opt_e, *opt_f;
1413         char *opt_i;
1414
1415 #if ENABLE_LONG_OPTS
1416         static const char sed_longopts[] ALIGN1 =
1417                 /* name             has_arg             short */
1418                 "in-place\0"        Optional_argument   "i"
1419                 "regexp-extended\0" No_argument         "r"
1420                 "quiet\0"           No_argument         "n"
1421                 "silent\0"          No_argument         "n"
1422                 "expression\0"      Required_argument   "e"
1423                 "file\0"            Required_argument   "f";
1424 #endif
1425
1426         int status = EXIT_SUCCESS;
1427
1428         INIT_G();
1429
1430         /* destroy command strings on exit */
1431         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1432
1433         /* Lie to autoconf when it starts asking stupid questions. */
1434         if (argv[1] && strcmp(argv[1], "--version") == 0) {
1435                 puts("This is not GNU sed version 4.0");
1436                 return 0;
1437         }
1438
1439         /* do normal option parsing */
1440         opt_e = opt_f = NULL;
1441         opt_i = NULL;
1442         opt_complementary = "e::f::" /* can occur multiple times */
1443                             "nn"; /* count -n */
1444
1445         IF_LONG_OPTS(applet_long_options = sed_longopts);
1446
1447         /* -i must be first, to match OPT_in_place definition */
1448         /* -E is a synonym of -r:
1449          * GNU sed 4.2.1 mentions it in neither --help
1450          * nor manpage, but does recognize it.
1451          */
1452         opt = getopt32(argv, "i::rEne:f:", &opt_i, &opt_e, &opt_f,
1453                             &G.be_quiet); /* counter for -n */
1454         //argc -= optind;
1455         argv += optind;
1456         if (opt & OPT_in_place) { // -i
1457                 atexit(cleanup_outname);
1458         }
1459         if (opt & (2|4))
1460                 G.regex_type |= REG_EXTENDED; // -r or -E
1461         //if (opt & 8)
1462         //      G.be_quiet++; // -n (implemented with a counter instead)
1463         while (opt_e) { // -e
1464                 add_cmd_block(llist_pop(&opt_e));
1465         }
1466         while (opt_f) { // -f
1467                 char *line;
1468                 FILE *cmdfile;
1469                 cmdfile = xfopen_for_read(llist_pop(&opt_f));
1470                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1471                         add_cmd(line);
1472                         free(line);
1473                 }
1474                 fclose(cmdfile);
1475         }
1476         /* if we didn't get a pattern from -e or -f, use argv[0] */
1477         if (!(opt & 0x30)) {
1478                 if (!*argv)
1479                         bb_show_usage();
1480                 add_cmd_block(*argv++);
1481         }
1482         /* Flush any unfinished commands. */
1483         add_cmd("");
1484
1485         /* By default, we write to stdout */
1486         G.nonstdout = stdout;
1487
1488         /* argv[0..(argc-1)] should be names of file to process. If no
1489          * files were specified or '-' was specified, take input from stdin.
1490          * Otherwise, we process all the files specified. */
1491         if (argv[0] == NULL) {
1492                 if (opt & OPT_in_place)
1493                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1494                 add_input_file(stdin);
1495         } else {
1496                 int i;
1497
1498                 for (i = 0; argv[i]; i++) {
1499                         struct stat statbuf;
1500                         int nonstdoutfd;
1501                         FILE *file;
1502                         sed_cmd_t *sed_cmd;
1503
1504                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1505                                 add_input_file(stdin);
1506                                 process_files();
1507                                 continue;
1508                         }
1509                         file = fopen_or_warn(argv[i], "r");
1510                         if (!file) {
1511                                 status = EXIT_FAILURE;
1512                                 continue;
1513                         }
1514                         add_input_file(file);
1515                         if (!(opt & OPT_in_place)) {
1516                                 continue;
1517                         }
1518
1519                         /* -i: process each FILE separately: */
1520
1521                         G.outname = xasprintf("%sXXXXXX", argv[i]);
1522                         nonstdoutfd = xmkstemp(G.outname);
1523                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1524
1525                         /* Set permissions/owner of output file */
1526                         fstat(fileno(file), &statbuf);
1527                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1528                          * but GNU sed 4.2.1 does not preserve them either */
1529                         fchmod(nonstdoutfd, statbuf.st_mode);
1530                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1531
1532                         process_files();
1533                         fclose(G.nonstdout);
1534                         G.nonstdout = stdout;
1535
1536                         if (opt_i) {
1537                                 char *backupname = xasprintf("%s%s", argv[i], opt_i);
1538                                 xrename(argv[i], backupname);
1539                                 free(backupname);
1540                         }
1541                         /* else unlink(argv[i]); - rename below does this */
1542                         xrename(G.outname, argv[i]); //TODO: rollback backup on error?
1543                         free(G.outname);
1544                         G.outname = NULL;
1545
1546                         /* Re-enable disabled range matches */
1547                         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1548                                 sed_cmd->beg_line = sed_cmd->beg_line_orig;
1549                         }
1550                 }
1551                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1552                  * if (G.current_input_file >= G.input_file_count)
1553                  *      return status;
1554                  * but it's not needed since process_files() works correctly
1555                  * in this case too. */
1556         }
1557         process_files();
1558
1559         return status;
1560 }