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