editors/sed.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * sed.c - very minimalist version of sed
   4  *
   5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   7  * Copyright (C) 2002  Matt Kraai
   8  * Copyright (C) 2003 by Glenn McGrath
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * MAINTAINER: Rob Landley <rob@landley.net>
  12  *
  13  * Licensed under GPLv2, see file LICENSE in this source tree.
  14  */
  15
  16 /* Code overview.
  17  *
  18  * Files are laid out to avoid unnecessary function declarations.  So for
  19  * example, every function add_cmd calls occurs before add_cmd in this file.
  20  *
  21  * add_cmd() is called on each line of sed command text (from a file or from
  22  * the command line).  It calls get_address() and parse_cmd_args().  The
  23  * resulting sed_cmd_t structures are appended to a linked list
  24  * (G.sed_cmd_head/G.sed_cmd_tail).
  25  *
  26  * process_files() does actual sedding, reading data lines from each input FILE*
  27  * (which could be stdin) and applying the sed command list (sed_cmd_head) to
  28  * each of the resulting lines.
  29  *
  30  * sed_main() is where external code calls into this, with a command line.
  31  */
  32
  33 /* Supported features and commands in this version of sed:
  34  *
  35  * - comments ('#')
  36  * - address matching: num|/matchstr/[,num|/matchstr/|$]command
  37  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  38  * - edit commands: (a)ppend, (i)nsert, (c)hange
  39  * - file commands: (r)ead
  40  * - backreferences in substitution expressions (\0, \1, \2...\9)
  41  * - grouped commands: {cmd1;cmd2}
  42  * - transliteration (y/source-chars/dest-chars/)
  43  * - pattern space hold space storing / swapping (g, h, x)
  44  * - labels / branching (: label, b, t, T)
  45  *
  46  * (Note: Specifying an address (range) to match is *optional*; commands
  47  * default to the whole pattern space if no specific address match was
  48  * requested.)
  49  *
  50  * Todo:
  51  * - Create a wrapper around regex to make libc's regex conform with sed
  52  *
  53  * Reference
  54  * http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  55  * http://pubs.opengroup.org/onlinepubs/9699919799/utilities/sed.html
  56  * http://sed.sourceforge.net/sedfaq3.html
  57  */
  58
  59 //config:config SED
  60 //config:       bool "sed (12 kb)"
  61 //config:       default y
  62 //config:       help
  63 //config:       sed is used to perform text transformations on a file
  64 //config:       or input from a pipeline.
  65
  66 //kbuild:lib-$(CONFIG_SED) += sed.o
  67
  68 //applet:IF_SED(APPLET(sed, BB_DIR_BIN, BB_SUID_DROP))
  69
  70 //usage:#define sed_trivial_usage
  71 //usage:       "[-i[SFX]] [-nrE] [-f FILE]... [-e CMD]... [FILE]...\n"
  72 //usage:       "or: sed [-i[SFX]] [-nrE] CMD [FILE]..."
  73 //usage:#define sed_full_usage "\n\n"
  74 //usage:       "        -e CMD  Add CMD to sed commands to be executed"
  75 //usage:     "\n        -f FILE Add FILE contents to sed commands to be executed"
  76 //usage:     "\n        -i[SFX] Edit files in-place (otherwise sends to stdout)"
  77 //usage:     "\n                Optionally back files up, appending SFX"
  78 //usage:     "\n        -n      Suppress automatic printing of pattern space"
  79 //usage:     "\n        -r,-E   Use extended regex syntax"
  80 //usage:     "\n"
  81 //usage:     "\nIf no -e or -f, the first non-option argument is the sed command string."
  82 //usage:     "\nRemaining arguments are input files (stdin if none)."
  83 //usage:
  84 //usage:#define sed_example_usage
  85 //usage:       "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
  86 //usage:       "bar\n"
  87
  88 #include "libbb.h"
  89 #include "common_bufsiz.h"
  90 #include "xregex.h"
  91
  92 #if 0
  93 # define dbg(...) bb_error_msg(__VA_ARGS__)
  94 #else
  95 # define dbg(...) ((void)0)
  96 #endif
  97
  98
  99 enum {
 100         OPT_in_place = 1 << 0,
 101 };
 102
 103 /* Each sed command turns into one of these structures. */
 104 typedef struct sed_cmd_s {
 105         /* Ordered by alignment requirements: currently 36 bytes on x86 */
 106         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
 107
 108         /* address storage */
 109         regex_t *beg_match;     /* sed -e '/match/cmd' */
 110         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
 111         regex_t *sub_match;     /* For 's/sub_match/string/' */
 112         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
 113         int beg_line_orig;      /* copy of the above, needed for -i */
 114         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($). -2-N = +N */
 115         int end_line_orig;
 116
 117         FILE *sw_file;          /* File (sw) command writes to, NULL for none. */
 118         char *string;           /* Data string for (saicytb) commands. */
 119
 120         unsigned which_match;   /* (s) Which match to replace (0 for all) */
 121
 122         /* Bitfields (gcc won't group them if we don't) */
 123         unsigned invert:1;      /* the '!' after the address */
 124         unsigned in_match:1;    /* Next line also included in match? */
 125         unsigned sub_p:1;       /* (s) print option */
 126
 127         char sw_last_char;      /* Last line written by (sw) had no '\n' */
 128
 129         /* GENERAL FIELDS */
 130         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 131 } sed_cmd_t;
 132
 133 static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
 134
 135 struct globals {
 136         /* options */
 137         int be_quiet, regex_type;
 138
 139         FILE *nonstdout;
 140         char *outname, *hold_space;
 141         smallint exitcode;
 142
 143         /* list of input files */
 144         int current_input_file, last_input_file;
 145         char **input_file_list;
 146         FILE *current_fp;
 147
 148         regmatch_t regmatch[10];
 149         regex_t *previous_regex_ptr;
 150
 151         /* linked list of sed commands */
 152         sed_cmd_t *sed_cmd_head, **sed_cmd_tail;
 153
 154         /* linked list of append lines */
 155         llist_t *append_head;
 156
 157         char *add_cmd_line;
 158
 159         struct pipeline {
 160                 char *buf;  /* Space to hold string */
 161                 int idx;    /* Space used */
 162                 int len;    /* Space allocated */
 163         } pipeline;
 164 } FIX_ALIASING;
 165 #define G (*(struct globals*)bb_common_bufsiz1)
 166 #define INIT_G() do { \
 167         setup_common_bufsiz(); \
 168         BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 169         G.sed_cmd_tail = &G.sed_cmd_head; \
 170 } while (0)
 171
 172
 173 #if ENABLE_FEATURE_CLEAN_UP
 174 static void sed_free_and_close_stuff(void)
 175 {
 176         sed_cmd_t *sed_cmd = G.sed_cmd_head;
 177
 178         llist_free(G.append_head, free);
 179
 180         while (sed_cmd) {
 181                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 182
 183                 if (sed_cmd->sw_file)
 184                         fclose(sed_cmd->sw_file);
 185
 186                 if (sed_cmd->beg_match) {
 187                         regfree(sed_cmd->beg_match);
 188                         free(sed_cmd->beg_match);
 189                 }
 190                 if (sed_cmd->end_match) {
 191                         regfree(sed_cmd->end_match);
 192                         free(sed_cmd->end_match);
 193                 }
 194                 if (sed_cmd->sub_match) {
 195                         regfree(sed_cmd->sub_match);
 196                         free(sed_cmd->sub_match);
 197                 }
 198                 free(sed_cmd->string);
 199                 free(sed_cmd);
 200                 sed_cmd = sed_cmd_next;
 201         }
 202
 203         free(G.hold_space);
 204
 205         if (G.current_fp)
 206                 fclose(G.current_fp);
 207 }
 208 #else
 209 void sed_free_and_close_stuff(void);
 210 #endif
 211
 212 /* If something bad happens during -i operation, delete temp file */
 213
 214 static void cleanup_outname(void)
 215 {
 216         if (G.outname) unlink(G.outname);
 217 }
 218
 219 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
 220
 221 static unsigned parse_escapes(char *dest, const char *string, int len, char from, char to)
 222 {
 223         char *d = dest;
 224         int i = 0;
 225
 226         if (len == -1)
 227                 len = strlen(string);
 228
 229         while (i < len) {
 230                 if (string[i] == '\\') {
 231                         if (!to || string[i+1] == from) {
 232                                 if ((*d = to ? to : string[i+1]) == '\0')
 233                                         return d - dest;
 234                                 i += 2;
 235                                 d++;
 236                                 continue;
 237                         }
 238                         i++; /* skip backslash in string[] */
 239                         *d++ = '\\';
 240                         /* fall through: copy next char verbatim */
 241                 }
 242                 if ((*d = string[i++]) == '\0')
 243                         return d - dest;
 244                 d++;
 245         }
 246         *d = '\0';
 247         return d - dest;
 248 }
 249
 250 static char *copy_parsing_escapes(const char *string, int len)
 251 {
 252         const char *s;
 253         char *dest = xmalloc(len + 1);
 254
 255         /* sed recognizes \n */
 256         /* GNU sed also recognizes \t and \r */
 257         for (s = "\nn\tt\rr"; *s; s += 2) {
 258                 len = parse_escapes(dest, string, len, s[1], s[0]);
 259                 string = dest;
 260         }
 261         return dest;
 262 }
 263
 264
 265 /*
 266  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 267  * beginning at a specified index and returns the index of the next regular
 268  * expression delimiter (typically a forward slash ('/')) not preceded by
 269  * a backslash ('\').  A negative delimiter disables square bracket checking.
 270  */
 271 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
 272 {
 273         int bracket = -1;
 274         int escaped = 0;
 275         int idx = 0;
 276         char ch;
 277
 278         if (delimiter < 0) {
 279                 bracket--;
 280                 delimiter = -delimiter;
 281         }
 282
 283         for (; (ch = str[idx]) != '\0'; idx++) {
 284                 if (bracket >= 0) {
 285                         if (ch == ']'
 286                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
 287                         ) {
 288                                 bracket = -1;
 289                         }
 290                 } else if (escaped)
 291                         escaped = 0;
 292                 else if (ch == '\\')
 293                         escaped = 1;
 294                 else if (bracket == -1 && ch == '[')
 295                         bracket = idx;
 296                 else if (ch == delimiter)
 297                         return idx;
 298         }
 299
 300         /* if we make it to here, we've hit the end of the string */
 301         bb_error_msg_and_die("unmatched '%c'", delimiter);
 302 }
 303
 304 /*
 305  *  Returns the index of the third delimiter
 306  */
 307 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 308 {
 309         const char *cmdstr_ptr = cmdstr;
 310         unsigned char delimiter;
 311         int idx = 0;
 312
 313         /* verify that the 's' or 'y' is followed by something.  That something
 314          * (typically a 'slash') is now our regexp delimiter... */
 315         if (*cmdstr == '\0')
 316                 bb_error_msg_and_die("bad format in substitution expression");
 317         delimiter = *cmdstr_ptr++;
 318
 319         /* save the match string */
 320         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 321         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 322
 323         /* save the replacement string */
 324         cmdstr_ptr += idx + 1;
 325         idx = index_of_next_unescaped_regexp_delim(- (int)delimiter, cmdstr_ptr);
 326         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 327
 328         return ((cmdstr_ptr - cmdstr) + idx);
 329 }
 330
 331 /*
 332  * returns the index in the string just past where the address ends.
 333  */
 334 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
 335 {
 336         const char *pos = my_str;
 337
 338         if (isdigit(*my_str)) {
 339                 *linenum = strtol(my_str, (char**)&pos, 10);
 340                 /* endstr shouldn't ever equal NULL */
 341         } else if (*my_str == '$') {
 342                 *linenum = -1;
 343                 pos++;
 344         } else if (*my_str == '/' || *my_str == '\\') {
 345                 int next;
 346                 char delimiter;
 347                 char *temp;
 348
 349                 delimiter = '/';
 350                 if (*my_str == '\\')
 351                         delimiter = *++pos;
 352                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 353                 if (next != 0) {
 354                         temp = copy_parsing_escapes(pos, next);
 355                         G.previous_regex_ptr = *regex = xzalloc(sizeof(regex_t));
 356                         xregcomp(*regex, temp, G.regex_type);
 357                         free(temp);
 358                 } else {
 359                         *regex = G.previous_regex_ptr;
 360                         if (!G.previous_regex_ptr)
 361                                 bb_error_msg_and_die("no previous regexp");
 362                 }
 363                 /* Move position to next character after last delimiter */
 364                 pos += (next+1);
 365         }
 366         return pos - my_str;
 367 }
 368
 369 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 370 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
 371 {
 372         int start = 0, idx, hack = 0;
 373
 374         /* Skip whitespace, then grab filename to end of line */
 375         while (isspace(filecmdstr[start]))
 376                 start++;
 377         idx = start;
 378         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
 379                 idx++;
 380
 381         /* If lines glued together, put backslash back. */
 382         if (filecmdstr[idx] == '\n')
 383                 hack = 1;
 384         if (idx == start)
 385                 bb_error_msg_and_die("empty filename");
 386         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 387         if (hack)
 388                 (*retval)[idx] = '\\';
 389
 390         return idx;
 391 }
 392
 393 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
 394 {
 395         int cflags = G.regex_type;
 396         char *match;
 397         int idx;
 398
 399         /*
 400          * A substitution command should look something like this:
 401          *    s/match/replace/ #giIpw
 402          *    ||     |        |||
 403          *    mandatory       optional
 404          */
 405         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 406
 407         /* determine the number of back references in the match string */
 408         /* Note: we compute this here rather than in the do_subst_command()
 409          * function to save processor time, at the expense of a little more memory
 410          * (4 bits) per sed_cmd */
 411
 412         /* process the flags */
 413
 414         sed_cmd->which_match = 1;
 415         dbg("s flags:'%s'", substr + idx + 1);
 416         while (substr[++idx]) {
 417                 dbg("s flag:'%c'", substr[idx]);
 418                 /* Parse match number */
 419                 if (isdigit(substr[idx])) {
 420                         if (match[0] != '^') {
 421                                 /* Match 0 treated as all, multiple matches we take the last one. */
 422                                 const char *pos = substr + idx;
 423 /* FIXME: error check? */
 424                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
 425                                 idx = pos - substr - 1;
 426                         }
 427                         continue;
 428                 }
 429                 /* Skip spaces */
 430                 if (isspace(substr[idx]))
 431                         continue;
 432
 433                 switch (substr[idx]) {
 434                 /* Replace all occurrences */
 435                 case 'g':
 436                         if (match[0] != '^')
 437                                 sed_cmd->which_match = 0;
 438                         break;
 439                 /* Print pattern space */
 440                 case 'p':
 441                         sed_cmd->sub_p = 1;
 442                         break;
 443                 /* Write to file */
 444                 case 'w':
 445                 {
 446                         char *fname;
 447                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname);
 448                         sed_cmd->sw_file = xfopen_for_write(fname);
 449                         sed_cmd->sw_last_char = '\n';
 450                         free(fname);
 451                         break;
 452                 }
 453                 /* Ignore case (gnu extension) */
 454                 case 'i':
 455                 case 'I':
 456                         cflags |= REG_ICASE;
 457                         break;
 458                 /* Comment */
 459                 case '#':
 460                         // while (substr[++idx]) continue;
 461                         idx += strlen(substr + idx); // same
 462                         /* Fall through */
 463                 /* End of command */
 464                 case ';':
 465                 case '}':
 466                         goto out;
 467                 default:
 468                         dbg("s bad flags:'%s'", substr + idx);
 469                         bb_error_msg_and_die("bad option in substitution expression");
 470                 }
 471         }
 472  out:
 473         /* compile the match string into a regex */
 474         if (*match != '\0') {
 475                 /* If match is empty, we use last regex used at runtime */
 476                 sed_cmd->sub_match = xzalloc(sizeof(regex_t));
 477                 dbg("xregcomp('%s',%x)", match, cflags);
 478                 xregcomp(sed_cmd->sub_match, match, cflags);
 479                 dbg("regcomp ok");
 480         }
 481         free(match);
 482
 483         return idx;
 484 }
 485
 486 /*
 487  *  Process the commands arguments
 488  */
 489 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
 490 {
 491         static const char cmd_letters[] ALIGN1 = "saicrw:btTydDgGhHlnNpPqx={}";
 492         enum {
 493                 IDX_s = 0,
 494                 IDX_a,
 495                 IDX_i,
 496                 IDX_c,
 497                 IDX_r,
 498                 IDX_w,
 499                 IDX_colon,
 500                 IDX_b,
 501                 IDX_t,
 502                 IDX_T,
 503                 IDX_y,
 504                 IDX_d,
 505                 IDX_D,
 506                 IDX_g,
 507                 IDX_G,
 508                 IDX_h,
 509                 IDX_H,
 510                 IDX_l,
 511                 IDX_n,
 512                 IDX_N,
 513                 IDX_p,
 514                 IDX_P,
 515                 IDX_q,
 516                 IDX_x,
 517                 IDX_equal,
 518                 IDX_lbrace,
 519                 IDX_rbrace,
 520                 IDX_nul
 521         };
 522         unsigned idx;
 523
 524         BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul);
 525
 526         idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
 527
 528         /* handle (s)ubstitution command */
 529         if (idx == IDX_s) {
 530                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 531         }
 532         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 533         else if (idx <= IDX_c) { /* a,i,c */
 534                 unsigned len;
 535
 536                 if (idx < IDX_c) { /* a,i */
 537                         if (sed_cmd->end_line || sed_cmd->end_match)
 538                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 539                 }
 540                 for (;;) {
 541                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 542                                 cmdstr++;
 543                                 break;
 544                         }
 545                         if (!isspace(*cmdstr))
 546                                 break;
 547                         cmdstr++;
 548                 }
 549                 len = strlen(cmdstr);
 550                 sed_cmd->string = copy_parsing_escapes(cmdstr, len);
 551                 cmdstr += len;
 552                 /* "\anychar" -> "anychar" */
 553                 parse_escapes(sed_cmd->string, sed_cmd->string, -1, '\0', '\0');
 554         }
 555         /* handle file cmds: (r)ead */
 556         else if (idx <= IDX_w) { /* r,w */
 557                 if (idx < IDX_w) { /* r */
 558                         if (sed_cmd->end_line || sed_cmd->end_match)
 559                                 bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
 560                 }
 561                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
 562                 if (sed_cmd->cmd == 'w') {
 563                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
 564                         sed_cmd->sw_last_char = '\n';
 565                 }
 566         }
 567         /* handle branch commands */
 568         else if (idx <= IDX_T) { /* :,b,t,T */
 569                 int length;
 570
 571                 cmdstr = skip_whitespace(cmdstr);
 572                 length = strcspn(cmdstr, semicolon_whitespace);
 573                 if (length) {
 574                         sed_cmd->string = xstrndup(cmdstr, length);
 575                         cmdstr += length;
 576                 }
 577         }
 578         /* translation command */
 579         else if (idx == IDX_y) {
 580                 char *match, *replace;
 581                 int i = cmdstr[0];
 582
 583                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 584                 /* \n already parsed, but \delimiter needs unescaping. */
 585                 parse_escapes(match,   match,   -1, i, i);
 586                 parse_escapes(replace, replace, -1, i, i);
 587
 588                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 589                 for (i = 0; match[i] && replace[i]; i++) {
 590                         sed_cmd->string[i*2] = match[i];
 591                         sed_cmd->string[i*2+1] = replace[i];
 592                 }
 593                 free(match);
 594                 free(replace);
 595         }
 596         /* if it wasn't a single-letter command that takes no arguments
 597          * then it must be an invalid command.
 598          */
 599         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
 600                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 601         }
 602
 603         /* give back whatever's left over */
 604         return cmdstr;
 605 }
 606
 607
 608 /* Parse address+command sets, skipping comment lines. */
 609
 610 static void add_cmd(const char *cmdstr)
 611 {
 612         sed_cmd_t *sed_cmd;
 613         unsigned len, n;
 614
 615         /* Append this line to any unfinished line from last time. */
 616         if (G.add_cmd_line) {
 617                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
 618                 free(G.add_cmd_line);
 619                 cmdstr = G.add_cmd_line = tp;
 620         }
 621
 622         /* If this line ends with unescaped backslash, request next line. */
 623         n = len = strlen(cmdstr);
 624         while (n && cmdstr[n-1] == '\\')
 625                 n--;
 626         if ((len - n) & 1) { /* if odd number of trailing backslashes */
 627                 if (!G.add_cmd_line)
 628                         G.add_cmd_line = xstrdup(cmdstr);
 629                 G.add_cmd_line[len-1] = '\0';
 630                 return;
 631         }
 632
 633         /* Loop parsing all commands in this line. */
 634         while (*cmdstr) {
 635                 /* Skip leading whitespace and semicolons */
 636                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 637
 638                 /* If no more commands, exit. */
 639                 if (!*cmdstr) break;
 640
 641                 /* if this is a comment, jump past it and keep going */
 642                 if (*cmdstr == '#') {
 643                         /* "#n" is the same as using -n on the command line */
 644                         if (cmdstr[1] == 'n')
 645                                 G.be_quiet++;
 646                         cmdstr = strpbrk(cmdstr, "\n\r");
 647                         if (!cmdstr) break;
 648                         continue;
 649                 }
 650
 651                 /* parse the command
 652                  * format is: [addr][,addr][!]cmd
 653                  *            |----||-----||-|
 654                  *            part1 part2  part3
 655                  */
 656
 657                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 658
 659                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 660                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 661                 sed_cmd->beg_line_orig = sed_cmd->beg_line;
 662
 663                 /* second part (if present) will begin with a comma */
 664                 if (*cmdstr == ',') {
 665                         int idx;
 666
 667                         cmdstr++;
 668                         if (*cmdstr == '+' && isdigit(cmdstr[1])) {
 669                                 /* http://sed.sourceforge.net/sedfaq3.html#s3.3
 670                                  * Under GNU sed 3.02+, ssed, and sed15+, <address2>
 671                                  * may also be a notation of the form +num,
 672                                  * indicating the next num lines after <address1> is
 673                                  * matched.
 674                                  * GNU sed 4.2.1 accepts even "+" (meaning "+0").
 675                                  * We don't (we check for isdigit, see above), think
 676                                  * about the "+-3" case.
 677                                  */
 678                                 char *end;
 679                                 /* code is smaller compared to using &cmdstr here: */
 680                                 idx = strtol(cmdstr+1, &end, 10);
 681                                 sed_cmd->end_line = -2 - idx;
 682                                 cmdstr = end;
 683                         } else {
 684                                 idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 685                                 cmdstr += idx;
 686                                 idx--; /* if 0, trigger error check below */
 687                         }
 688                         if (idx < 0)
 689                                 bb_error_msg_and_die("no address after comma");
 690                         sed_cmd->end_line_orig = sed_cmd->end_line;
 691                 }
 692
 693                 /* skip whitespace before the command */
 694                 cmdstr = skip_whitespace(cmdstr);
 695
 696                 /* Check for inversion flag */
 697                 if (*cmdstr == '!') {
 698                         sed_cmd->invert = 1;
 699                         cmdstr++;
 700
 701                         /* skip whitespace before the command */
 702                         cmdstr = skip_whitespace(cmdstr);
 703                 }
 704
 705                 /* last part (mandatory) will be a command */
 706                 if (!*cmdstr)
 707                         bb_error_msg_and_die("missing command");
 708                 sed_cmd->cmd = *cmdstr++;
 709                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 710
 711                 /* cmdstr now points past args.
 712                  * GNU sed requires a separator, if there are more commands,
 713                  * else it complains "char N: extra characters after command".
 714                  * Example: "sed 'p;d'". We also allow "sed 'pd'".
 715                  */
 716
 717                 /* Add the command to the command array */
 718                 *G.sed_cmd_tail = sed_cmd;
 719                 G.sed_cmd_tail = &sed_cmd->next;
 720         }
 721
 722         /* If we glued multiple lines together, free the memory. */
 723         free(G.add_cmd_line);
 724         G.add_cmd_line = NULL;
 725 }
 726
 727 /* Append to a string, reallocating memory as necessary. */
 728
 729 #define PIPE_GROW 64
 730
 731 static void pipe_putc(char c)
 732 {
 733         if (G.pipeline.idx == G.pipeline.len) {
 734                 G.pipeline.buf = xrealloc(G.pipeline.buf,
 735                                 G.pipeline.len + PIPE_GROW);
 736                 G.pipeline.len += PIPE_GROW;
 737         }
 738         G.pipeline.buf[G.pipeline.idx++] = c;
 739 }
 740
 741 static void do_subst_w_backrefs(char *line, char *replace)
 742 {
 743         int i, j;
 744
 745         /* go through the replacement string */
 746         for (i = 0; replace[i]; i++) {
 747                 /* if we find a backreference (\1, \2, etc.) print the backref'ed text */
 748                 if (replace[i] == '\\') {
 749                         unsigned backref = replace[++i] - '0';
 750                         if (backref <= 9) {
 751                                 /* print out the text held in G.regmatch[backref] */
 752                                 if (G.regmatch[backref].rm_so != -1) {
 753                                         j = G.regmatch[backref].rm_so;
 754                                         while (j < G.regmatch[backref].rm_eo)
 755                                                 pipe_putc(line[j++]);
 756                                 }
 757                                 continue;
 758                         }
 759                         /* I _think_ it is impossible to get '\' to be
 760                          * the last char in replace string. Thus we don't check
 761                          * for replace[i] == NUL. (counterexample anyone?) */
 762                         /* if we find a backslash escaped character, print the character */
 763                         pipe_putc(replace[i]);
 764                         continue;
 765                 }
 766                 /* if we find an unescaped '&' print out the whole matched text. */
 767                 if (replace[i] == '&') {
 768                         j = G.regmatch[0].rm_so;
 769                         while (j < G.regmatch[0].rm_eo)
 770                                 pipe_putc(line[j++]);
 771                         continue;
 772                 }
 773                 /* Otherwise just output the character. */
 774                 pipe_putc(replace[i]);
 775         }
 776 }
 777
 778 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
 779 {
 780         char *line = *line_p;
 781         unsigned match_count = 0;
 782         bool altered = 0;
 783         bool prev_match_empty = 1;
 784         bool tried_at_eol = 0;
 785         regex_t *current_regex;
 786
 787         current_regex = sed_cmd->sub_match;
 788         /* Handle empty regex. */
 789         if (!current_regex) {
 790                 current_regex = G.previous_regex_ptr;
 791                 if (!current_regex)
 792                         bb_error_msg_and_die("no previous regexp");
 793         }
 794         G.previous_regex_ptr = current_regex;
 795
 796         /* Find the first match */
 797         dbg("matching '%s'", line);
 798         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) {
 799                 dbg("no match");
 800                 return 0;
 801         }
 802         dbg("match");
 803
 804         /* Initialize temporary output buffer. */
 805         G.pipeline.buf = xmalloc(PIPE_GROW);
 806         G.pipeline.len = PIPE_GROW;
 807         G.pipeline.idx = 0;
 808
 809         /* Now loop through, substituting for matches */
 810         do {
 811                 int start = G.regmatch[0].rm_so;
 812                 int end = G.regmatch[0].rm_eo;
 813                 int i;
 814
 815                 match_count++;
 816
 817                 /* If we aren't interested in this match, output old line to
 818                  * end of match and continue */
 819                 if (sed_cmd->which_match
 820                  && (sed_cmd->which_match != match_count)
 821                 ) {
 822                         for (i = 0; i < end; i++)
 823                                 pipe_putc(*line++);
 824                         /* Null match? Print one more char */
 825                         if (start == end && *line)
 826                                 pipe_putc(*line++);
 827                         goto next;
 828                 }
 829
 830                 /* Print everything before the match */
 831                 for (i = 0; i < start; i++)
 832                         pipe_putc(line[i]);
 833
 834                 /* Then print the substitution string,
 835                  * unless we just matched empty string after non-empty one.
 836                  * Example: string "cccd", pattern "c*", repl "R":
 837                  * result is "RdR", not "RRdR": first match "ccc",
 838                  * second is "" before "d", third is "" after "d".
 839                  * Second match is NOT replaced!
 840                  */
 841                 if (prev_match_empty || start != 0 || start != end) {
 842                         //dbg("%d %d %d", prev_match_empty, start, end);
 843                         dbg("inserting replacement at %d in '%s'", start, line);
 844                         do_subst_w_backrefs(line, sed_cmd->string);
 845                         /* Flag that something has changed */
 846                         altered = 1;
 847                 } else {
 848                         dbg("NOT inserting replacement at %d in '%s'", start, line);
 849                 }
 850
 851                 /* If matched string is empty (f.e. "c*" pattern),
 852                  * copy verbatim one char after it before attempting more matches
 853                  */
 854                 prev_match_empty = (start == end);
 855                 if (prev_match_empty) {
 856                         if (!line[end]) {
 857                                 tried_at_eol = 1;
 858                         } else {
 859                                 pipe_putc(line[end]);
 860                                 end++;
 861                         }
 862                 }
 863
 864                 /* Advance past the match */
 865                 dbg("line += %d", end);
 866                 line += end;
 867
 868                 /* if we're not doing this globally, get out now */
 869                 if (sed_cmd->which_match != 0)
 870                         break;
 871  next:
 872                 /* Exit if we are at EOL and already tried matching at it */
 873                 if (*line == '\0') {
 874                         if (tried_at_eol)
 875                                 break;
 876                         tried_at_eol = 1;
 877                 }
 878
 879 //maybe (end ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
 880         } while (regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
 881
 882         /* Copy rest of string into output pipeline */
 883         while (1) {
 884                 char c = *line++;
 885                 pipe_putc(c);
 886                 if (c == '\0')
 887                         break;
 888         }
 889
 890         free(*line_p);
 891         *line_p = G.pipeline.buf;
 892         return altered;
 893 }
 894
 895 /* Set command pointer to point to this label.  (Does not handle null label.) */
 896 static sed_cmd_t *branch_to(char *label)
 897 {
 898         sed_cmd_t *sed_cmd;
 899
 900         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
 901                 if (sed_cmd->cmd == ':'
 902                  && sed_cmd->string
 903                  && strcmp(sed_cmd->string, label) == 0
 904                 ) {
 905                         return sed_cmd;
 906                 }
 907         }
 908         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 909 }
 910
 911 static void append(char *s)
 912 {
 913         llist_add_to_end(&G.append_head, s);
 914 }
 915
 916 /* Output line of text. */
 917 /* Note:
 918  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 919  * Without them, we had this:
 920  * echo -n thingy >z1
 921  * echo -n again >z2
 922  * >znull
 923  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 924  * output:
 925  * gnu sed 4.1.5:
 926  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 927  * bbox:
 928  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 929  */
 930 enum {
 931         NO_EOL_CHAR = 1,
 932         LAST_IS_NUL = 2,
 933 };
 934 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 935 {
 936         char lpc = *last_puts_char;
 937
 938         /* Need to insert a '\n' between two files because first file's
 939          * last line wasn't terminated? */
 940         if (lpc != '\n' && lpc != '\0') {
 941                 fputc('\n', file);
 942                 lpc = '\n';
 943         }
 944         fputs(s, file);
 945
 946         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 947         if (s[0])
 948                 lpc = 'x';
 949
 950         /* had trailing '\0' and it was last char of file? */
 951         if (last_gets_char == LAST_IS_NUL) {
 952                 fputc('\0', file);
 953                 lpc = 'x'; /* */
 954         } else
 955         /* had trailing '\n' or '\0'? */
 956         if (last_gets_char != NO_EOL_CHAR) {
 957                 fputc(last_gets_char, file);
 958                 lpc = last_gets_char;
 959         }
 960
 961         if (ferror(file)) {
 962                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 963                 bb_error_msg_and_die(bb_msg_write_error);
 964         }
 965         *last_puts_char = lpc;
 966 }
 967
 968 static void flush_append(char *last_puts_char)
 969 {
 970         char *data;
 971
 972         /* Output appended lines. */
 973         while ((data = (char *)llist_pop(&G.append_head)) != NULL) {
 974                 /* Append command does not respect "nonterminated-ness"
 975                  * of last line. Try this:
 976                  * $ echo -n "woot" | sed -e '/woot/a woo' -
 977                  * woot
 978                  * woo
 979                  * (both lines are terminated with \n)
 980                  * Therefore we do not propagate "last_gets_char" here,
 981                  * pass '\n' instead:
 982                  */
 983                 puts_maybe_newline(data, G.nonstdout, last_puts_char, '\n');
 984                 free(data);
 985         }
 986 }
 987
 988 /* Get next line of input from G.input_file_list, flushing append buffer and
 989  * noting if we ran out of files without a newline on the last line we read.
 990  */
 991 static char *get_next_line(char *gets_char, char *last_puts_char)
 992 {
 993         char *temp = NULL;
 994         int len;
 995         char gc;
 996
 997         flush_append(last_puts_char);
 998
 999         /* will be returned if last line in the file
1000          * doesn't end with either '\n' or '\0' */
1001         gc = NO_EOL_CHAR;
1002         for (; G.current_input_file <= G.last_input_file; G.current_input_file++) {
1003                 FILE *fp = G.current_fp;
1004                 if (!fp) {
1005                         const char *path = G.input_file_list[G.current_input_file];
1006                         fp = stdin;
1007                         if (path != bb_msg_standard_input) {
1008                                 fp = fopen_or_warn(path, "r");
1009                                 if (!fp) {
1010                                         G.exitcode = EXIT_FAILURE;
1011                                         continue;
1012                                 }
1013                         }
1014                         G.current_fp = fp;
1015                 }
1016                 /* Read line up to a newline or NUL byte, inclusive,
1017                  * return malloc'ed char[]. length of the chunk read
1018                  * is stored in len. NULL if EOF/error */
1019                 temp = bb_get_chunk_from_file(fp, &len);
1020                 if (temp) {
1021                         /* len > 0 here, it's ok to do temp[len-1] */
1022                         char c = temp[len-1];
1023                         if (c == '\n' || c == '\0') {
1024                                 temp[len-1] = '\0';
1025                                 gc = c;
1026                                 if (c == '\0') {
1027                                         int ch = fgetc(fp);
1028                                         if (ch != EOF)
1029                                                 ungetc(ch, fp);
1030                                         else
1031                                                 gc = LAST_IS_NUL;
1032                                 }
1033                         }
1034                         /* else we put NO_EOL_CHAR into *gets_char */
1035                         break;
1036
1037                 /* NB: I had the idea of peeking next file(s) and returning
1038                  * NO_EOL_CHAR only if it is the *last* non-empty
1039                  * input file. But there is a case where this won't work:
1040                  * file1: "a woo\nb woo"
1041                  * file2: "c no\nd no"
1042                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
1043                  * (note: *no* newline after "b bang"!) */
1044                 }
1045                 /* Close this file and advance to next one */
1046                 fclose_if_not_stdin(fp);
1047                 G.current_fp = NULL;
1048         }
1049         *gets_char = gc;
1050         return temp;
1051 }
1052
1053 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
1054
1055 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
1056 {
1057         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
1058         if (retval)
1059                 G.previous_regex_ptr = sed_cmd->beg_match;
1060         return retval;
1061 }
1062
1063 /* Process all the lines in all the files */
1064
1065 static void process_files(void)
1066 {
1067         char *pattern_space, *next_line;
1068         int linenum = 0;
1069         char last_puts_char = '\n';
1070         char last_gets_char, next_gets_char;
1071         sed_cmd_t *sed_cmd;
1072         int substituted;
1073
1074         /* Prime the pump */
1075         next_line = get_next_line(&next_gets_char, &last_puts_char);
1076
1077         /* Go through every line in each file */
1078  again:
1079         substituted = 0;
1080
1081         /* Advance to next line.  Stop if out of lines. */
1082         pattern_space = next_line;
1083         if (!pattern_space)
1084                 return;
1085         last_gets_char = next_gets_char;
1086
1087         /* Read one line in advance so we can act on the last line,
1088          * the '$' address */
1089         next_line = get_next_line(&next_gets_char, &last_puts_char);
1090         linenum++;
1091
1092         /* For every line, go through all the commands */
1093  restart:
1094         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1095                 int old_matched, matched;
1096
1097                 old_matched = sed_cmd->in_match;
1098
1099                 /* Determine if this command matches this line: */
1100
1101                 dbg("match1:%d", sed_cmd->in_match);
1102                 dbg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
1103                                 && !sed_cmd->beg_match && !sed_cmd->end_match));
1104                 dbg("match3:%d", (sed_cmd->beg_line > 0
1105                         && (sed_cmd->end_line || sed_cmd->end_match
1106                             ? (sed_cmd->beg_line <= linenum)
1107                             : (sed_cmd->beg_line == linenum)
1108                             )
1109                         ));
1110                 dbg("match4:%d", (beg_match(sed_cmd, pattern_space)));
1111                 dbg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
1112
1113                 /* Are we continuing a previous multi-line match? */
1114                 sed_cmd->in_match = sed_cmd->in_match
1115                         /* Or is no range necessary? */
1116                         || (!sed_cmd->beg_line && !sed_cmd->end_line
1117                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
1118                         /* Or did we match the start of a numerical range? */
1119                         || (sed_cmd->beg_line > 0
1120                             && (sed_cmd->end_line || sed_cmd->end_match
1121                                   /* note: even if end is numeric and is < linenum too,
1122                                    * GNU sed matches! We match too, therefore we don't
1123                                    * check here that linenum <= end.
1124                                    * Example:
1125                                    * printf '1\n2\n3\n4\n' | sed -n '1{N;N;d};1p;2,3p;3p;4p'
1126                                    * first three input lines are deleted;
1127                                    * 4th line is matched and printed
1128                                    * by "2,3" (!) and by "4" ranges
1129                                    */
1130                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
1131                                 : (sed_cmd->beg_line == linenum)    /* N */
1132                                 )
1133                             )
1134                         /* Or does this line match our begin address regex? */
1135                         || (beg_match(sed_cmd, pattern_space))
1136                         /* Or did we match last line of input? */
1137                         || (sed_cmd->beg_line == -1 && next_line == NULL);
1138
1139                 /* Snapshot the value */
1140                 matched = sed_cmd->in_match;
1141
1142                 dbg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
1143                         sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
1144
1145                 /* Is this line the end of the current match? */
1146
1147                 if (matched) {
1148                         if (sed_cmd->end_line <= -2) {
1149                                 /* address2 is +N, i.e. N lines from beg_line */
1150                                 sed_cmd->end_line = linenum + (-sed_cmd->end_line - 2);
1151                         }
1152                         /* once matched, "n,xxx" range is dead, disabling it */
1153                         if (sed_cmd->beg_line > 0) {
1154                                 sed_cmd->beg_line = -2;
1155                         }
1156                         dbg("end1:%d", sed_cmd->end_line ? sed_cmd->end_line == -1
1157                                                 ? !next_line : (sed_cmd->end_line <= linenum)
1158                                         : !sed_cmd->end_match);
1159                         dbg("end2:%d", sed_cmd->end_match && old_matched
1160                                         && !regexec(sed_cmd->end_match,pattern_space, 0, NULL, 0));
1161                         sed_cmd->in_match = !(
1162                                 /* has the ending line come, or is this a single address command? */
1163                                 (sed_cmd->end_line
1164                                         ? sed_cmd->end_line == -1
1165                                                 ? !next_line
1166                                                 : (sed_cmd->end_line <= linenum)
1167                                         : !sed_cmd->end_match
1168                                 )
1169                                 /* or does this line matches our last address regex */
1170                                 || (sed_cmd->end_match && old_matched
1171                                      && (regexec(sed_cmd->end_match,
1172                                                 pattern_space, 0, NULL, 0) == 0)
1173                                 )
1174                         );
1175                 }
1176
1177                 /* Skip blocks of commands we didn't match */
1178                 if (sed_cmd->cmd == '{') {
1179                         if (sed_cmd->invert ? matched : !matched) {
1180                                 unsigned nest_cnt = 0;
1181                                 while (1) {
1182                                         if (sed_cmd->cmd == '{')
1183                                                 nest_cnt++;
1184                                         if (sed_cmd->cmd == '}') {
1185                                                 nest_cnt--;
1186                                                 if (nest_cnt == 0)
1187                                                         break;
1188                                         }
1189                                         sed_cmd = sed_cmd->next;
1190                                         if (!sed_cmd)
1191                                                 bb_error_msg_and_die("unterminated {");
1192                                 }
1193                         }
1194                         continue;
1195                 }
1196
1197                 /* Okay, so did this line match? */
1198                 if (sed_cmd->invert ? matched : !matched)
1199                         continue; /* no */
1200
1201                 /* Update last used regex in case a blank substitute BRE is found */
1202                 if (sed_cmd->beg_match) {
1203                         G.previous_regex_ptr = sed_cmd->beg_match;
1204                 }
1205
1206                 /* actual sedding */
1207                 dbg("pattern_space:'%s' next_line:'%s' cmd:%c",
1208                                 pattern_space, next_line, sed_cmd->cmd);
1209                 switch (sed_cmd->cmd) {
1210
1211                 /* Print line number */
1212                 case '=':
1213                         fprintf(G.nonstdout, "%d\n", linenum);
1214                         break;
1215
1216                 /* Write the current pattern space up to the first newline */
1217                 case 'P':
1218                 {
1219                         char *tmp = strchr(pattern_space, '\n');
1220                         if (tmp) {
1221                                 *tmp = '\0';
1222                                 /* TODO: explain why '\n' below */
1223                                 sed_puts(pattern_space, '\n');
1224                                 *tmp = '\n';
1225                                 break;
1226                         }
1227                         /* Fall Through */
1228                 }
1229
1230                 /* Write the current pattern space to output */
1231                 case 'p':
1232                         /* NB: we print this _before_ the last line
1233                          * (of current file) is printed. Even if
1234                          * that line is nonterminated, we print
1235                          * '\n' here (gnu sed does the same) */
1236                         sed_puts(pattern_space, '\n');
1237                         break;
1238                 /* Delete up through first newline */
1239                 case 'D':
1240                 {
1241                         char *tmp = strchr(pattern_space, '\n');
1242                         if (tmp) {
1243                                 overlapping_strcpy(pattern_space, tmp + 1);
1244                                 goto restart;
1245                         }
1246                 }
1247                 /* discard this line. */
1248                 case 'd':
1249                         goto discard_line;
1250
1251                 /* Substitute with regex */
1252                 case 's':
1253                         if (!do_subst_command(sed_cmd, &pattern_space))
1254                                 break;
1255                         dbg("do_subst_command succeeded:'%s'", pattern_space);
1256                         substituted |= 1;
1257
1258                         /* handle p option */
1259                         if (sed_cmd->sub_p)
1260                                 sed_puts(pattern_space, last_gets_char);
1261                         /* handle w option */
1262                         if (sed_cmd->sw_file)
1263                                 puts_maybe_newline(
1264                                         pattern_space, sed_cmd->sw_file,
1265                                         &sed_cmd->sw_last_char, last_gets_char);
1266                         break;
1267
1268                 /* Append line to linked list to be printed later */
1269                 case 'a':
1270                         append(xstrdup(sed_cmd->string));
1271                         break;
1272
1273                 /* Insert text before this line */
1274                 case 'i':
1275                         sed_puts(sed_cmd->string, '\n');
1276                         break;
1277
1278                 /* Cut and paste text (replace) */
1279                 case 'c':
1280                         /* Only triggers on last line of a matching range. */
1281                         if (!sed_cmd->in_match)
1282                                 sed_puts(sed_cmd->string, '\n');
1283                         goto discard_line;
1284
1285                 /* Read file, append contents to output */
1286                 case 'r':
1287                 {
1288                         FILE *rfile;
1289                         rfile = fopen_for_read(sed_cmd->string);
1290                         if (rfile) {
1291                                 char *line;
1292                                 while ((line = xmalloc_fgetline(rfile))
1293                                                 != NULL)
1294                                         append(line);
1295                                 fclose(rfile);
1296                         }
1297
1298                         break;
1299                 }
1300
1301                 /* Write pattern space to file. */
1302                 case 'w':
1303                         puts_maybe_newline(
1304                                 pattern_space, sed_cmd->sw_file,
1305                                 &sed_cmd->sw_last_char, last_gets_char);
1306                         break;
1307
1308                 /* Read next line from input */
1309                 case 'n':
1310                         if (!G.be_quiet)
1311                                 sed_puts(pattern_space, last_gets_char);
1312                         if (next_line == NULL) {
1313                                 /* If no next line, jump to end of script and exit. */
1314                                 goto discard_line;
1315                         }
1316                         free(pattern_space);
1317                         pattern_space = next_line;
1318                         last_gets_char = next_gets_char;
1319                         next_line = get_next_line(&next_gets_char, &last_puts_char);
1320                         substituted = 0;
1321                         linenum++;
1322                         break;
1323
1324                 /* Quit.  End of script, end of input. */
1325                 case 'q':
1326                         /* Exit the outer while loop */
1327                         free(next_line);
1328                         next_line = NULL;
1329                         goto discard_commands;
1330
1331                 /* Append the next line to the current line */
1332                 case 'N':
1333                 {
1334                         int len;
1335                         /* If no next line, jump to end of script and exit. */
1336                         /* http://www.gnu.org/software/sed/manual/sed.html:
1337                          * "Most versions of sed exit without printing anything
1338                          * when the N command is issued on the last line of
1339                          * a file. GNU sed prints pattern space before exiting
1340                          * unless of course the -n command switch has been
1341                          * specified. This choice is by design."
1342                          */
1343                         if (next_line == NULL) {
1344                                 //goto discard_line;
1345                                 goto discard_commands; /* GNU behavior */
1346                         }
1347                         /* Append next_line, read new next_line. */
1348                         len = strlen(pattern_space);
1349                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1350                         pattern_space[len] = '\n';
1351                         strcpy(pattern_space + len+1, next_line);
1352                         last_gets_char = next_gets_char;
1353                         next_line = get_next_line(&next_gets_char, &last_puts_char);
1354                         linenum++;
1355                         break;
1356                 }
1357
1358                 /* Test/branch if substitution occurred */
1359                 case 't':
1360                         if (!substituted) break;
1361                         substituted = 0;
1362                         /* Fall through */
1363                 /* Test/branch if substitution didn't occur */
1364                 case 'T':
1365                         if (substituted) break;
1366                         /* Fall through */
1367                 /* Branch to label */
1368                 case 'b':
1369                         if (!sed_cmd->string) goto discard_commands;
1370                         else sed_cmd = branch_to(sed_cmd->string);
1371                         break;
1372                 /* Transliterate characters */
1373                 case 'y':
1374                 {
1375                         int i, j;
1376                         for (i = 0; pattern_space[i]; i++) {
1377                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1378                                         if (pattern_space[i] == sed_cmd->string[j]) {
1379                                                 pattern_space[i] = sed_cmd->string[j + 1];
1380                                                 break;
1381                                         }
1382                                 }
1383                         }
1384
1385                         break;
1386                 }
1387                 case 'g':       /* Replace pattern space with hold space */
1388                         free(pattern_space);
1389                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1390                         break;
1391                 case 'G':       /* Append newline and hold space to pattern space */
1392                 {
1393                         int pattern_space_size = 2;
1394                         int hold_space_size = 0;
1395
1396                         if (pattern_space)
1397                                 pattern_space_size += strlen(pattern_space);
1398                         if (G.hold_space)
1399                                 hold_space_size = strlen(G.hold_space);
1400                         pattern_space = xrealloc(pattern_space,
1401                                         pattern_space_size + hold_space_size);
1402                         if (pattern_space_size == 2)
1403                                 pattern_space[0] = 0;
1404                         strcat(pattern_space, "\n");
1405                         if (G.hold_space)
1406                                 strcat(pattern_space, G.hold_space);
1407                         last_gets_char = '\n';
1408
1409                         break;
1410                 }
1411                 case 'h':       /* Replace hold space with pattern space */
1412                         free(G.hold_space);
1413                         G.hold_space = xstrdup(pattern_space);
1414                         break;
1415                 case 'H':       /* Append newline and pattern space to hold space */
1416                 {
1417                         int hold_space_size = 2;
1418                         int pattern_space_size = 0;
1419
1420                         if (G.hold_space)
1421                                 hold_space_size += strlen(G.hold_space);
1422                         if (pattern_space)
1423                                 pattern_space_size = strlen(pattern_space);
1424                         G.hold_space = xrealloc(G.hold_space,
1425                                         hold_space_size + pattern_space_size);
1426
1427                         if (hold_space_size == 2)
1428                                 *G.hold_space = 0;
1429                         strcat(G.hold_space, "\n");
1430                         if (pattern_space)
1431                                 strcat(G.hold_space, pattern_space);
1432
1433                         break;
1434                 }
1435                 case 'x': /* Exchange hold and pattern space */
1436                 {
1437                         char *tmp = pattern_space;
1438                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1439                         last_gets_char = '\n';
1440                         G.hold_space = tmp;
1441                         break;
1442                 }
1443                 } /* switch */
1444         } /* for each cmd */
1445
1446         /*
1447          * Exit point from sedding...
1448          */
1449  discard_commands:
1450         /* we will print the line unless we were told to be quiet ('-n')
1451            or if the line was suppressed (ala 'd'elete) */
1452         if (!G.be_quiet)
1453                 sed_puts(pattern_space, last_gets_char);
1454
1455         /* Delete and such jump here. */
1456  discard_line:
1457         flush_append(&last_puts_char /*,last_gets_char*/);
1458         free(pattern_space);
1459
1460         goto again;
1461 }
1462
1463 /* It is possible to have a command line argument with embedded
1464  * newlines.  This counts as multiple command lines.
1465  * However, newline can be escaped: 's/e/z\<newline>z/'
1466  * add_cmd() handles this.
1467  */
1468
1469 static void add_cmd_block(char *cmdstr)
1470 {
1471         char *sv, *eol;
1472
1473         cmdstr = sv = xstrdup(cmdstr);
1474         do {
1475                 eol = strchr(cmdstr, '\n');
1476                 if (eol)
1477                         *eol = '\0';
1478                 add_cmd(cmdstr);
1479                 cmdstr = eol + 1;
1480         } while (eol);
1481         free(sv);
1482 }
1483
1484 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1485 int sed_main(int argc UNUSED_PARAM, char **argv)
1486 {
1487         unsigned opt;
1488         llist_t *opt_e, *opt_f;
1489         char *opt_i;
1490
1491 #if ENABLE_LONG_OPTS
1492         static const char sed_longopts[] ALIGN1 =
1493                 /* name             has_arg             short */
1494                 "in-place\0"        Optional_argument   "i"
1495                 "regexp-extended\0" No_argument         "r"
1496                 "quiet\0"           No_argument         "n"
1497                 "silent\0"          No_argument         "n"
1498                 "expression\0"      Required_argument   "e"
1499                 "file\0"            Required_argument   "f";
1500 #endif
1501
1502         INIT_G();
1503
1504         /* destroy command strings on exit */
1505         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1506
1507         /* Lie to autoconf when it starts asking stupid questions. */
1508         if (argv[1] && strcmp(argv[1], "--version") == 0) {
1509                 puts("This is not GNU sed version 4.0");
1510                 return 0;
1511         }
1512
1513         /* do normal option parsing */
1514         opt_e = opt_f = NULL;
1515         opt_i = NULL;
1516         /* -i must be first, to match OPT_in_place definition */
1517         /* -E is a synonym of -r:
1518          * GNU sed 4.2.1 mentions it in neither --help
1519          * nor manpage, but does recognize it.
1520          */
1521         opt = getopt32long(argv, "^"
1522                         "i::rEne:*f:*"
1523                         "\0" "nn"/*count -n*/,
1524                         sed_longopts,
1525                         &opt_i, &opt_e, &opt_f,
1526                         &G.be_quiet); /* counter for -n */
1527         //argc -= optind;
1528         argv += optind;
1529         if (opt & OPT_in_place) { // -i
1530                 die_func = cleanup_outname;
1531         }
1532         if (opt & (2|4))
1533                 G.regex_type |= REG_EXTENDED; // -r or -E
1534         //if (opt & 8)
1535         //      G.be_quiet++; // -n (implemented with a counter instead)
1536         while (opt_e) { // -e
1537                 add_cmd_block(llist_pop(&opt_e));
1538         }
1539         while (opt_f) { // -f
1540                 char *line;
1541                 FILE *cmdfile;
1542                 cmdfile = xfopen_stdin(llist_pop(&opt_f));
1543                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1544                         add_cmd(line);
1545                         free(line);
1546                 }
1547                 fclose_if_not_stdin(cmdfile);
1548         }
1549         /* if we didn't get a pattern from -e or -f, use argv[0] */
1550         if (!(opt & 0x30)) {
1551                 if (!*argv)
1552                         bb_show_usage();
1553                 add_cmd_block(*argv++);
1554         }
1555         /* Flush any unfinished commands. */
1556         add_cmd("");
1557
1558         /* By default, we write to stdout */
1559         G.nonstdout = stdout;
1560
1561         /* argv[0..(argc-1)] should be names of file to process. If no
1562          * files were specified or '-' was specified, take input from stdin.
1563          * Otherwise, we process all the files specified. */
1564         G.input_file_list = argv;
1565         if (!argv[0]) {
1566                 if (opt & OPT_in_place)
1567                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1568                 argv[0] = (char*)bb_msg_standard_input;
1569                 /* G.last_input_file = 0; - already is */
1570         } else {
1571                 goto start;
1572
1573                 for (; *argv; argv++) {
1574                         struct stat statbuf;
1575                         int nonstdoutfd;
1576                         sed_cmd_t *sed_cmd;
1577
1578                         G.last_input_file++;
1579  start:
1580                         if (!(opt & OPT_in_place)) {
1581                                 if (LONE_DASH(*argv)) {
1582                                         *argv = (char*)bb_msg_standard_input;
1583                                         process_files();
1584                                 }
1585                                 continue;
1586                         }
1587
1588                         /* -i: process each FILE separately: */
1589
1590                         if (stat(*argv, &statbuf) != 0) {
1591                                 bb_simple_perror_msg(*argv);
1592                                 G.exitcode = EXIT_FAILURE;
1593                                 G.current_input_file++;
1594                                 continue;
1595                         }
1596                         G.outname = xasprintf("%sXXXXXX", *argv);
1597                         nonstdoutfd = xmkstemp(G.outname);
1598                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1599                         /* Set permissions/owner of output file */
1600                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1601                          * but GNU sed 4.2.1 does not preserve them either */
1602                         fchmod(nonstdoutfd, statbuf.st_mode);
1603                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1604
1605                         process_files();
1606                         fclose(G.nonstdout);
1607                         G.nonstdout = stdout;
1608
1609                         if (opt_i) {
1610                                 char *backupname = xasprintf("%s%s", *argv, opt_i);
1611                                 xrename(*argv, backupname);
1612                                 free(backupname);
1613                         }
1614                         /* else unlink(*argv); - rename below does this */
1615                         xrename(G.outname, *argv); //TODO: rollback backup on error?
1616                         free(G.outname);
1617                         G.outname = NULL;
1618
1619                         /* Fix disabled range matches and mangled ",+N" ranges */
1620                         for (sed_cmd = G.sed_cmd_head; sed_cmd; sed_cmd = sed_cmd->next) {
1621                                 sed_cmd->beg_line = sed_cmd->beg_line_orig;
1622                                 sed_cmd->end_line = sed_cmd->end_line_orig;
1623                         }
1624                 }
1625                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1626                  * if (G.current_input_file[G.current_input_file] == NULL)
1627                  *      return G.exitcode;
1628                  * but it's not needed since process_files() works correctly
1629                  * in this case too. */
1630         }
1631
1632         process_files();
1633
1634         return G.exitcode;
1635 }