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/|$]command
  40  * - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  41  * - edit commands: (a)ppend, (i)nsert, (c)hange
  42  * - file commands: (r)ead
  43  * - backreferences in substitution expressions (\0, \1, \2...\9)
  44  * - grouped commands: {cmd1;cmd2}
  45  * - transliteration (y/source-chars/dest-chars/)
  46  * - pattern space hold space storing / swapping (g, h, x)
  47  * - labels / branching (: label, b, t, T)
  48  *
  49  * (Note: Specifying an address (range) to match is *optional*; commands
  50  * default to the whole pattern space if no specific address match was
  51  * requested.)
  52  *
  53  * Todo:
  54  * - Create a wrapper around regex to make libc's regex conform with sed
  55  *
  56  * Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  57  */
  58
  59 //usage:#define sed_trivial_usage
  60 //usage:       "[-efinr] SED_CMD [FILE]..."
  61 //usage:#define sed_full_usage "\n\n"
  62 //usage:       "Options:"
  63 //usage:     "\n        -e CMD  Add CMD to sed commands to be executed"
  64 //usage:     "\n        -f FILE Add FILE contents to sed commands to be executed"
  65 //usage:     "\n        -i      Edit files in-place (else sends result to stdout)"
  66 //usage:     "\n        -n      Suppress automatic printing of pattern space"
  67 //usage:     "\n        -r      Use extended regex syntax"
  68 //usage:     "\n"
  69 //usage:     "\nIf no -e or -f, the first non-option argument is the sed command string."
  70 //usage:     "\nRemaining arguments are input files (stdin if none)."
  71 //usage:
  72 //usage:#define sed_example_usage
  73 //usage:       "$ echo \"foo\" | sed -e 's/f[a-zA-Z]o/bar/g'\n"
  74 //usage:       "bar\n"
  75
  76 #include "libbb.h"
  77 #include "xregex.h"
  78
  79 enum {
  80         OPT_in_place = 1 << 0,
  81 };
  82
  83 /* Each sed command turns into one of these structures. */
  84 typedef struct sed_cmd_s {
  85         /* Ordered by alignment requirements: currently 36 bytes on x86 */
  86         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
  87
  88         /* address storage */
  89         regex_t *beg_match;     /* sed -e '/match/cmd' */
  90         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
  91         regex_t *sub_match;     /* For 's/sub_match/string/' */
  92         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
  93         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  94
  95         FILE *sw_file;          /* File (sw) command writes to, -1 for none. */
  96         char *string;           /* Data string for (saicytb) commands. */
  97
  98         unsigned which_match;   /* (s) Which match to replace (0 for all) */
  99
 100         /* Bitfields (gcc won't group them if we don't) */
 101         unsigned invert:1;      /* the '!' after the address */
 102         unsigned in_match:1;    /* Next line also included in match? */
 103         unsigned sub_p:1;       /* (s) print option */
 104
 105         char sw_last_char;      /* Last line written by (sw) had no '\n' */
 106
 107         /* GENERAL FIELDS */
 108         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
 109 } sed_cmd_t;
 110
 111 static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v";
 112
 113 struct globals {
 114         /* options */
 115         int be_quiet, regex_type;
 116         FILE *nonstdout;
 117         char *outname, *hold_space;
 118
 119         /* List of input files */
 120         int input_file_count, current_input_file;
 121         FILE **input_file_list;
 122
 123         regmatch_t regmatch[10];
 124         regex_t *previous_regex_ptr;
 125
 126         /* linked list of sed commands */
 127         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 128
 129         /* Linked list of append lines */
 130         llist_t *append_head;
 131
 132         char *add_cmd_line;
 133
 134         struct pipeline {
 135                 char *buf;  /* Space to hold string */
 136                 int idx;    /* Space used */
 137                 int len;    /* Space allocated */
 138         } pipeline;
 139 } FIX_ALIASING;
 140 #define G (*(struct globals*)&bb_common_bufsiz1)
 141 struct BUG_G_too_big {
 142         char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
 143 };
 144 #define INIT_G() do { \
 145         G.sed_cmd_tail = &G.sed_cmd_head; \
 146 } while (0)
 147
 148
 149 #if ENABLE_FEATURE_CLEAN_UP
 150 static void sed_free_and_close_stuff(void)
 151 {
 152         sed_cmd_t *sed_cmd = G.sed_cmd_head.next;
 153
 154         llist_free(G.append_head, free);
 155
 156         while (sed_cmd) {
 157                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 158
 159                 if (sed_cmd->sw_file)
 160                         xprint_and_close_file(sed_cmd->sw_file);
 161
 162                 if (sed_cmd->beg_match) {
 163                         regfree(sed_cmd->beg_match);
 164                         free(sed_cmd->beg_match);
 165                 }
 166                 if (sed_cmd->end_match) {
 167                         regfree(sed_cmd->end_match);
 168                         free(sed_cmd->end_match);
 169                 }
 170                 if (sed_cmd->sub_match) {
 171                         regfree(sed_cmd->sub_match);
 172                         free(sed_cmd->sub_match);
 173                 }
 174                 free(sed_cmd->string);
 175                 free(sed_cmd);
 176                 sed_cmd = sed_cmd_next;
 177         }
 178
 179         free(G.hold_space);
 180
 181         while (G.current_input_file < G.input_file_count)
 182                 fclose(G.input_file_list[G.current_input_file++]);
 183 }
 184 #else
 185 void sed_free_and_close_stuff(void);
 186 #endif
 187
 188 /* If something bad happens during -i operation, delete temp file */
 189
 190 static void cleanup_outname(void)
 191 {
 192         if (G.outname) unlink(G.outname);
 193 }
 194
 195 /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */
 196
 197 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 198 {
 199         int i = 0;
 200
 201         while (i < len) {
 202                 if (string[i] == '\\') {
 203                         if (!to || string[i+1] == from) {
 204                                 *dest++ = to ? to : string[i+1];
 205                                 i += 2;
 206                                 continue;
 207                         }
 208                         *dest++ = string[i++];
 209                 }
 210                 /* TODO: is it safe wrt a string with trailing '\\' ? */
 211                 *dest++ = string[i++];
 212         }
 213         *dest = '\0';
 214 }
 215
 216 static char *copy_parsing_escapes(const char *string, int len)
 217 {
 218         char *dest = xmalloc(len + 1);
 219
 220         parse_escapes(dest, string, len, 'n', '\n');
 221         /* GNU sed also recognizes \t and \r */
 222         parse_escapes(dest, dest, strlen(dest), 't', '\t');
 223         parse_escapes(dest, dest, strlen(dest), 'r', '\r');
 224         return dest;
 225 }
 226
 227
 228 /*
 229  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 230  * beginning at a specified index and returns the index of the next regular
 231  * expression delimiter (typically a forward slash ('/')) not preceded by
 232  * a backslash ('\').  A negative delimiter disables square bracket checking.
 233  */
 234 static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str)
 235 {
 236         int bracket = -1;
 237         int escaped = 0;
 238         int idx = 0;
 239         char ch;
 240
 241         if (delimiter < 0) {
 242                 bracket--;
 243                 delimiter = -delimiter;
 244         }
 245
 246         for (; (ch = str[idx]) != '\0'; idx++) {
 247                 if (bracket >= 0) {
 248                         if (ch == ']'
 249                          && !(bracket == idx - 1 || (bracket == idx - 2 && str[idx - 1] == '^'))
 250                         ) {
 251                                 bracket = -1;
 252                         }
 253                 } else if (escaped)
 254                         escaped = 0;
 255                 else if (ch == '\\')
 256                         escaped = 1;
 257                 else if (bracket == -1 && ch == '[')
 258                         bracket = idx;
 259                 else if (ch == delimiter)
 260                         return idx;
 261         }
 262
 263         /* if we make it to here, we've hit the end of the string */
 264         bb_error_msg_and_die("unmatched '%c'", delimiter);
 265 }
 266
 267 /*
 268  *  Returns the index of the third delimiter
 269  */
 270 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 271 {
 272         const char *cmdstr_ptr = cmdstr;
 273         char delimiter;
 274         int idx = 0;
 275
 276         /* verify that the 's' or 'y' is followed by something.  That something
 277          * (typically a 'slash') is now our regexp delimiter... */
 278         if (*cmdstr == '\0')
 279                 bb_error_msg_and_die("bad format in substitution expression");
 280         delimiter = *cmdstr_ptr++;
 281
 282         /* save the match string */
 283         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 284         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 285
 286         /* save the replacement string */
 287         cmdstr_ptr += idx + 1;
 288         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 289         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 290
 291         return ((cmdstr_ptr - cmdstr) + idx);
 292 }
 293
 294 /*
 295  * returns the index in the string just past where the address ends.
 296  */
 297 static int get_address(const char *my_str, int *linenum, regex_t ** regex)
 298 {
 299         const char *pos = my_str;
 300
 301         if (isdigit(*my_str)) {
 302                 *linenum = strtol(my_str, (char**)&pos, 10);
 303                 /* endstr shouldnt ever equal NULL */
 304         } else if (*my_str == '$') {
 305                 *linenum = -1;
 306                 pos++;
 307         } else if (*my_str == '/' || *my_str == '\\') {
 308                 int next;
 309                 char delimiter;
 310                 char *temp;
 311
 312                 delimiter = '/';
 313                 if (*my_str == '\\') delimiter = *++pos;
 314                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 315                 temp = copy_parsing_escapes(pos, next);
 316                 *regex = xmalloc(sizeof(regex_t));
 317                 xregcomp(*regex, temp, G.regex_type|REG_NEWLINE);
 318                 free(temp);
 319                 /* Move position to next character after last delimiter */
 320                 pos += (next+1);
 321         }
 322         return pos - my_str;
 323 }
 324
 325 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 326 static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval)
 327 {
 328         int start = 0, idx, hack = 0;
 329
 330         /* Skip whitespace, then grab filename to end of line */
 331         while (isspace(filecmdstr[start]))
 332                 start++;
 333         idx = start;
 334         while (filecmdstr[idx] && filecmdstr[idx] != '\n')
 335                 idx++;
 336
 337         /* If lines glued together, put backslash back. */
 338         if (filecmdstr[idx] == '\n')
 339                 hack = 1;
 340         if (idx == start)
 341                 bb_error_msg_and_die("empty filename");
 342         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 343         if (hack)
 344                 (*retval)[idx] = '\\';
 345
 346         return idx;
 347 }
 348
 349 static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
 350 {
 351         int cflags = G.regex_type;
 352         char *match;
 353         int idx;
 354
 355         /*
 356          * A substitution command should look something like this:
 357          *    s/match/replace/ #gIpw
 358          *    ||     |        |||
 359          *    mandatory       optional
 360          */
 361         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 362
 363         /* determine the number of back references in the match string */
 364         /* Note: we compute this here rather than in the do_subst_command()
 365          * function to save processor time, at the expense of a little more memory
 366          * (4 bits) per sed_cmd */
 367
 368         /* process the flags */
 369
 370         sed_cmd->which_match = 1;
 371         while (substr[++idx]) {
 372                 /* Parse match number */
 373                 if (isdigit(substr[idx])) {
 374                         if (match[0] != '^') {
 375                                 /* Match 0 treated as all, multiple matches we take the last one. */
 376                                 const char *pos = substr + idx;
 377 /* FIXME: error check? */
 378                                 sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
 379                                 idx = pos - substr;
 380                         }
 381                         continue;
 382                 }
 383                 /* Skip spaces */
 384                 if (isspace(substr[idx]))
 385                         continue;
 386
 387                 switch (substr[idx]) {
 388                 /* Replace all occurrences */
 389                 case 'g':
 390                         if (match[0] != '^')
 391                                 sed_cmd->which_match = 0;
 392                         break;
 393                 /* Print pattern space */
 394                 case 'p':
 395                         sed_cmd->sub_p = 1;
 396                         break;
 397                 /* Write to file */
 398                 case 'w':
 399                 {
 400                         char *temp;
 401                         idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp);
 402                         break;
 403                 }
 404                 /* Ignore case (gnu exension) */
 405                 case 'I':
 406                         cflags |= REG_ICASE;
 407                         break;
 408                 /* Comment */
 409                 case '#':
 410                         // while (substr[++idx]) continue;
 411                         idx += strlen(substr + idx); // same
 412                         /* Fall through */
 413                 /* End of command */
 414                 case ';':
 415                 case '}':
 416                         goto out;
 417                 default:
 418                         bb_error_msg_and_die("bad option in substitution expression");
 419                 }
 420         }
 421  out:
 422         /* compile the match string into a regex */
 423         if (*match != '\0') {
 424                 /* If match is empty, we use last regex used at runtime */
 425                 sed_cmd->sub_match = xmalloc(sizeof(regex_t));
 426                 xregcomp(sed_cmd->sub_match, match, cflags);
 427         }
 428         free(match);
 429
 430         return idx;
 431 }
 432
 433 /*
 434  *  Process the commands arguments
 435  */
 436 static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
 437 {
 438         static const char cmd_letters[] = "saicrw:btTydDgGhHlnNpPqx={}";
 439         enum {
 440                 IDX_s = 0,
 441                 IDX_a,
 442                 IDX_i,
 443                 IDX_c,
 444                 IDX_r,
 445                 IDX_w,
 446                 IDX_colon,
 447                 IDX_b,
 448                 IDX_t,
 449                 IDX_T,
 450                 IDX_y,
 451                 IDX_d,
 452                 IDX_D,
 453                 IDX_g,
 454                 IDX_G,
 455                 IDX_h,
 456                 IDX_H,
 457                 IDX_l,
 458                 IDX_n,
 459                 IDX_N,
 460                 IDX_p,
 461                 IDX_P,
 462                 IDX_q,
 463                 IDX_x,
 464                 IDX_equal,
 465                 IDX_lbrace,
 466                 IDX_rbrace,
 467                 IDX_nul
 468         };
 469         struct chk { char chk[sizeof(cmd_letters)-1 == IDX_nul ? 1 : -1]; };
 470
 471         unsigned idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
 472
 473         /* handle (s)ubstitution command */
 474         if (idx == IDX_s) {
 475                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 476         }
 477         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 478         else if (idx <= IDX_c) { /* a,i,c */
 479                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 480                         bb_error_msg_and_die("only a beginning address can be specified for edit commands");
 481                 for (;;) {
 482                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 483                                 cmdstr++;
 484                                 break;
 485                         }
 486                         if (!isspace(*cmdstr))
 487                                 break;
 488                         cmdstr++;
 489                 }
 490                 sed_cmd->string = xstrdup(cmdstr);
 491                 /* "\anychar" -> "anychar" */
 492                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0');
 493                 cmdstr += strlen(cmdstr);
 494         }
 495         /* handle file cmds: (r)ead */
 496         else if (idx <= IDX_w) { /* r,w */
 497                 if (sed_cmd->end_line || sed_cmd->end_match)
 498                         bb_error_msg_and_die("command only uses one address");
 499                 cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
 500                 if (sed_cmd->cmd == 'w') {
 501                         sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
 502                         sed_cmd->sw_last_char = '\n';
 503                 }
 504         }
 505         /* handle branch commands */
 506         else if (idx <= IDX_T) { /* :,b,t,T */
 507                 int length;
 508
 509                 cmdstr = skip_whitespace(cmdstr);
 510                 length = strcspn(cmdstr, semicolon_whitespace);
 511                 if (length) {
 512                         sed_cmd->string = xstrndup(cmdstr, length);
 513                         cmdstr += length;
 514                 }
 515         }
 516         /* translation command */
 517         else if (idx == IDX_y) {
 518                 char *match, *replace;
 519                 int i = cmdstr[0];
 520
 521                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 522                 /* \n already parsed, but \delimiter needs unescaping. */
 523                 parse_escapes(match, match, strlen(match), i, i);
 524                 parse_escapes(replace, replace, strlen(replace), i, i);
 525
 526                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 527                 for (i = 0; match[i] && replace[i]; i++) {
 528                         sed_cmd->string[i*2] = match[i];
 529                         sed_cmd->string[i*2+1] = replace[i];
 530                 }
 531                 free(match);
 532                 free(replace);
 533         }
 534         /* if it wasnt a single-letter command that takes no arguments
 535          * then it must be an invalid command.
 536          */
 537         else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
 538                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 539         }
 540
 541         /* give back whatever's left over */
 542         return cmdstr;
 543 }
 544
 545
 546 /* Parse address+command sets, skipping comment lines. */
 547
 548 static void add_cmd(const char *cmdstr)
 549 {
 550         sed_cmd_t *sed_cmd;
 551         unsigned len, n;
 552
 553         /* Append this line to any unfinished line from last time. */
 554         if (G.add_cmd_line) {
 555                 char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr);
 556                 free(G.add_cmd_line);
 557                 cmdstr = G.add_cmd_line = tp;
 558         }
 559
 560         /* If this line ends with unescaped backslash, request next line. */
 561         n = len = strlen(cmdstr);
 562         while (n && cmdstr[n-1] == '\\')
 563                 n--;
 564         if ((len - n) & 1) { /* if odd number of trailing backslashes */
 565                 if (!G.add_cmd_line)
 566                         G.add_cmd_line = xstrdup(cmdstr);
 567                 G.add_cmd_line[len-1] = '\0';
 568                 return;
 569         }
 570
 571         /* Loop parsing all commands in this line. */
 572         while (*cmdstr) {
 573                 /* Skip leading whitespace and semicolons */
 574                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 575
 576                 /* If no more commands, exit. */
 577                 if (!*cmdstr) break;
 578
 579                 /* if this is a comment, jump past it and keep going */
 580                 if (*cmdstr == '#') {
 581                         /* "#n" is the same as using -n on the command line */
 582                         if (cmdstr[1] == 'n')
 583                                 G.be_quiet++;
 584                         cmdstr = strpbrk(cmdstr, "\n\r");
 585                         if (!cmdstr) break;
 586                         continue;
 587                 }
 588
 589                 /* parse the command
 590                  * format is: [addr][,addr][!]cmd
 591                  *            |----||-----||-|
 592                  *            part1 part2  part3
 593                  */
 594
 595                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 596
 597                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 598                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 599
 600                 /* second part (if present) will begin with a comma */
 601                 if (*cmdstr == ',') {
 602                         int idx;
 603
 604                         cmdstr++;
 605                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 606                         if (!idx)
 607                                 bb_error_msg_and_die("no address after comma");
 608                         cmdstr += idx;
 609                 }
 610
 611                 /* skip whitespace before the command */
 612                 cmdstr = skip_whitespace(cmdstr);
 613
 614                 /* Check for inversion flag */
 615                 if (*cmdstr == '!') {
 616                         sed_cmd->invert = 1;
 617                         cmdstr++;
 618
 619                         /* skip whitespace before the command */
 620                         cmdstr = skip_whitespace(cmdstr);
 621                 }
 622
 623                 /* last part (mandatory) will be a command */
 624                 if (!*cmdstr)
 625                         bb_error_msg_and_die("missing command");
 626                 sed_cmd->cmd = *cmdstr++;
 627                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 628
 629                 /* Add the command to the command array */
 630                 G.sed_cmd_tail->next = sed_cmd;
 631                 G.sed_cmd_tail = G.sed_cmd_tail->next;
 632         }
 633
 634         /* If we glued multiple lines together, free the memory. */
 635         free(G.add_cmd_line);
 636         G.add_cmd_line = NULL;
 637 }
 638
 639 /* Append to a string, reallocating memory as necessary. */
 640
 641 #define PIPE_GROW 64
 642
 643 static void pipe_putc(char c)
 644 {
 645         if (G.pipeline.idx == G.pipeline.len) {
 646                 G.pipeline.buf = xrealloc(G.pipeline.buf,
 647                                 G.pipeline.len + PIPE_GROW);
 648                 G.pipeline.len += PIPE_GROW;
 649         }
 650         G.pipeline.buf[G.pipeline.idx++] = c;
 651 }
 652
 653 static void do_subst_w_backrefs(char *line, char *replace)
 654 {
 655         int i, j;
 656
 657         /* go through the replacement string */
 658         for (i = 0; replace[i]; i++) {
 659                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 660                 if (replace[i] == '\\') {
 661                         unsigned backref = replace[++i] - '0';
 662                         if (backref <= 9) {
 663                                 /* print out the text held in G.regmatch[backref] */
 664                                 if (G.regmatch[backref].rm_so != -1) {
 665                                         j = G.regmatch[backref].rm_so;
 666                                         while (j < G.regmatch[backref].rm_eo)
 667                                                 pipe_putc(line[j++]);
 668                                 }
 669                                 continue;
 670                         }
 671                         /* I _think_ it is impossible to get '\' to be
 672                          * the last char in replace string. Thus we dont check
 673                          * for replace[i] == NUL. (counterexample anyone?) */
 674                         /* if we find a backslash escaped character, print the character */
 675                         pipe_putc(replace[i]);
 676                         continue;
 677                 }
 678                 /* if we find an unescaped '&' print out the whole matched text. */
 679                 if (replace[i] == '&') {
 680                         j = G.regmatch[0].rm_so;
 681                         while (j < G.regmatch[0].rm_eo)
 682                                 pipe_putc(line[j++]);
 683                         continue;
 684                 }
 685                 /* Otherwise just output the character. */
 686                 pipe_putc(replace[i]);
 687         }
 688 }
 689
 690 static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p)
 691 {
 692         char *line = *line_p;
 693         int altered = 0;
 694         unsigned match_count = 0;
 695         regex_t *current_regex;
 696
 697         current_regex = sed_cmd->sub_match;
 698         /* Handle empty regex. */
 699         if (!current_regex) {
 700                 current_regex = G.previous_regex_ptr;
 701                 if (!current_regex)
 702                         bb_error_msg_and_die("no previous regexp");
 703         }
 704         G.previous_regex_ptr = current_regex;
 705
 706         /* Find the first match */
 707         if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0))
 708                 return 0;
 709
 710         /* Initialize temporary output buffer. */
 711         G.pipeline.buf = xmalloc(PIPE_GROW);
 712         G.pipeline.len = PIPE_GROW;
 713         G.pipeline.idx = 0;
 714
 715         /* Now loop through, substituting for matches */
 716         do {
 717                 int i;
 718
 719                 /* Work around bug in glibc regexec, demonstrated by:
 720                    echo " a.b" | busybox sed 's [^ .]* x g'
 721                    The match_count check is so not to break
 722                    echo "hi" | busybox sed 's/^/!/g' */
 723                 if (!G.regmatch[0].rm_so && !G.regmatch[0].rm_eo && match_count) {
 724                         pipe_putc(*line++);
 725                         continue;
 726                 }
 727
 728                 match_count++;
 729
 730                 /* If we aren't interested in this match, output old line to
 731                    end of match and continue */
 732                 if (sed_cmd->which_match
 733                  && (sed_cmd->which_match != match_count)
 734                 ) {
 735                         for (i = 0; i < G.regmatch[0].rm_eo; i++)
 736                                 pipe_putc(*line++);
 737                         continue;
 738                 }
 739
 740                 /* print everything before the match */
 741                 for (i = 0; i < G.regmatch[0].rm_so; i++)
 742                         pipe_putc(line[i]);
 743
 744                 /* then print the substitution string */
 745                 do_subst_w_backrefs(line, sed_cmd->string);
 746
 747                 /* advance past the match */
 748                 line += G.regmatch[0].rm_eo;
 749                 /* flag that something has changed */
 750                 altered++;
 751
 752                 /* if we're not doing this globally, get out now */
 753                 if (sed_cmd->which_match)
 754                         break;
 755
 756 //maybe (G.regmatch[0].rm_eo ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL?
 757         } while (*line && regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH);
 758
 759         /* Copy rest of string into output pipeline */
 760         while (1) {
 761                 char c = *line++;
 762                 pipe_putc(c);
 763                 if (c == '\0')
 764                         break;
 765         }
 766
 767         free(*line_p);
 768         *line_p = G.pipeline.buf;
 769         return altered;
 770 }
 771
 772 /* Set command pointer to point to this label.  (Does not handle null label.) */
 773 static sed_cmd_t *branch_to(char *label)
 774 {
 775         sed_cmd_t *sed_cmd;
 776
 777         for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 778                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 779                         return sed_cmd;
 780                 }
 781         }
 782         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 783 }
 784
 785 static void append(char *s)
 786 {
 787         llist_add_to_end(&G.append_head, xstrdup(s));
 788 }
 789
 790 static void flush_append(void)
 791 {
 792         char *data;
 793
 794         /* Output appended lines. */
 795         while ((data = (char *)llist_pop(&G.append_head))) {
 796                 fprintf(G.nonstdout, "%s\n", data);
 797                 free(data);
 798         }
 799 }
 800
 801 static void add_input_file(FILE *file)
 802 {
 803         G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count);
 804         G.input_file_list[G.input_file_count++] = file;
 805 }
 806
 807 /* Get next line of input from G.input_file_list, flushing append buffer and
 808  * noting if we ran out of files without a newline on the last line we read.
 809  */
 810 enum {
 811         NO_EOL_CHAR = 1,
 812         LAST_IS_NUL = 2,
 813 };
 814 static char *get_next_line(char *gets_char)
 815 {
 816         char *temp = NULL;
 817         int len;
 818         char gc;
 819
 820         flush_append();
 821
 822         /* will be returned if last line in the file
 823          * doesn't end with either '\n' or '\0' */
 824         gc = NO_EOL_CHAR;
 825         while (G.current_input_file < G.input_file_count) {
 826                 FILE *fp = G.input_file_list[G.current_input_file];
 827                 /* Read line up to a newline or NUL byte, inclusive,
 828                  * return malloc'ed char[]. length of the chunk read
 829                  * is stored in len. NULL if EOF/error */
 830                 temp = bb_get_chunk_from_file(fp, &len);
 831                 if (temp) {
 832                         /* len > 0 here, it's ok to do temp[len-1] */
 833                         char c = temp[len-1];
 834                         if (c == '\n' || c == '\0') {
 835                                 temp[len-1] = '\0';
 836                                 gc = c;
 837                                 if (c == '\0') {
 838                                         int ch = fgetc(fp);
 839                                         if (ch != EOF)
 840                                                 ungetc(ch, fp);
 841                                         else
 842                                                 gc = LAST_IS_NUL;
 843                                 }
 844                         }
 845                         /* else we put NO_EOL_CHAR into *gets_char */
 846                         break;
 847
 848                 /* NB: I had the idea of peeking next file(s) and returning
 849                  * NO_EOL_CHAR only if it is the *last* non-empty
 850                  * input file. But there is a case where this won't work:
 851                  * file1: "a woo\nb woo"
 852                  * file2: "c no\nd no"
 853                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
 854                  * (note: *no* newline after "b bang"!) */
 855                 }
 856                 /* Close this file and advance to next one */
 857                 fclose(fp);
 858                 G.current_input_file++;
 859         }
 860         *gets_char = gc;
 861         return temp;
 862 }
 863
 864 /* Output line of text. */
 865 /* Note:
 866  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 867  * Without them, we had this:
 868  * echo -n thingy >z1
 869  * echo -n again >z2
 870  * >znull
 871  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 872  * output:
 873  * gnu sed 4.1.5:
 874  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 875  * bbox:
 876  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 877  */
 878 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 879 {
 880         char lpc = *last_puts_char;
 881
 882         /* Need to insert a '\n' between two files because first file's
 883          * last line wasn't terminated? */
 884         if (lpc != '\n' && lpc != '\0') {
 885                 fputc('\n', file);
 886                 lpc = '\n';
 887         }
 888         fputs(s, file);
 889
 890         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 891         if (s[0])
 892                 lpc = 'x';
 893
 894         /* had trailing '\0' and it was last char of file? */
 895         if (last_gets_char == LAST_IS_NUL) {
 896                 fputc('\0', file);
 897                 lpc = 'x'; /* */
 898         } else
 899         /* had trailing '\n' or '\0'? */
 900         if (last_gets_char != NO_EOL_CHAR) {
 901                 fputc(last_gets_char, file);
 902                 lpc = last_gets_char;
 903         }
 904
 905         if (ferror(file)) {
 906                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 907                 bb_error_msg_and_die(bb_msg_write_error);
 908         }
 909         *last_puts_char = lpc;
 910 }
 911
 912 #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n))
 913
 914 static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space)
 915 {
 916         int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0);
 917         if (retval)
 918                 G.previous_regex_ptr = sed_cmd->beg_match;
 919         return retval;
 920 }
 921
 922 /* Process all the lines in all the files */
 923
 924 static void process_files(void)
 925 {
 926         char *pattern_space, *next_line;
 927         int linenum = 0;
 928         char last_puts_char = '\n';
 929         char last_gets_char, next_gets_char;
 930         sed_cmd_t *sed_cmd;
 931         int substituted;
 932
 933         /* Prime the pump */
 934         next_line = get_next_line(&next_gets_char);
 935
 936         /* Go through every line in each file */
 937  again:
 938         substituted = 0;
 939
 940         /* Advance to next line.  Stop if out of lines. */
 941         pattern_space = next_line;
 942         if (!pattern_space)
 943                 return;
 944         last_gets_char = next_gets_char;
 945
 946         /* Read one line in advance so we can act on the last line,
 947          * the '$' address */
 948         next_line = get_next_line(&next_gets_char);
 949         linenum++;
 950
 951         /* For every line, go through all the commands */
 952  restart:
 953         for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 954                 int old_matched, matched;
 955
 956                 old_matched = sed_cmd->in_match;
 957
 958                 /* Determine if this command matches this line: */
 959
 960                 //bb_error_msg("match1:%d", sed_cmd->in_match);
 961                 //bb_error_msg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line
 962                 //              && !sed_cmd->beg_match && !sed_cmd->end_match));
 963                 //bb_error_msg("match3:%d", (sed_cmd->beg_line > 0
 964                 //      && (sed_cmd->end_line || sed_cmd->end_match
 965                 //          ? (sed_cmd->beg_line <= linenum)
 966                 //          : (sed_cmd->beg_line == linenum)
 967                 //          )
 968                 //      )
 969                 //bb_error_msg("match4:%d", (beg_match(sed_cmd, pattern_space)));
 970                 //bb_error_msg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL));
 971
 972                 /* Are we continuing a previous multi-line match? */
 973                 sed_cmd->in_match = sed_cmd->in_match
 974                         /* Or is no range necessary? */
 975                         || (!sed_cmd->beg_line && !sed_cmd->end_line
 976                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
 977                         /* Or did we match the start of a numerical range? */
 978                         || (sed_cmd->beg_line > 0
 979                             && (sed_cmd->end_line || sed_cmd->end_match
 980                                   /* note: even if end is numeric and is < linenum too,
 981                                    * GNU sed matches! We match too */
 982                                 ? (sed_cmd->beg_line <= linenum)    /* N,end */
 983                                 : (sed_cmd->beg_line == linenum)    /* N */
 984                                 )
 985                             )
 986                         /* Or does this line match our begin address regex? */
 987                         || (beg_match(sed_cmd, pattern_space))
 988                         /* Or did we match last line of input? */
 989                         || (sed_cmd->beg_line == -1 && next_line == NULL);
 990
 991                 /* Snapshot the value */
 992                 matched = sed_cmd->in_match;
 993
 994                 //bb_error_msg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d",
 995                 //sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum);
 996
 997                 /* Is this line the end of the current match? */
 998
 999                 if (matched) {
1000                         /* once matched, "n,xxx" range is dead, disabling it */
1001                         if (sed_cmd->beg_line > 0
1002                          && !(option_mask32 & OPT_in_place) /* but not for -i */
1003                         ) {
1004                                 sed_cmd->beg_line = -2;
1005                         }
1006                         sed_cmd->in_match = !(
1007                                 /* has the ending line come, or is this a single address command? */
1008                                 (sed_cmd->end_line
1009                                         ? sed_cmd->end_line == -1
1010                                                 ? !next_line
1011                                                 : (sed_cmd->end_line <= linenum)
1012                                         : !sed_cmd->end_match
1013                                 )
1014                                 /* or does this line matches our last address regex */
1015                                 || (sed_cmd->end_match && old_matched
1016                                      && (regexec(sed_cmd->end_match,
1017                                                  pattern_space, 0, NULL, 0) == 0))
1018                         );
1019                 }
1020
1021                 /* Skip blocks of commands we didn't match */
1022                 if (sed_cmd->cmd == '{') {
1023                         if (sed_cmd->invert ? matched : !matched) {
1024                                 unsigned nest_cnt = 0;
1025                                 while (1) {
1026                                         if (sed_cmd->cmd == '{')
1027                                                 nest_cnt++;
1028                                         if (sed_cmd->cmd == '}') {
1029                                                 nest_cnt--;
1030                                                 if (nest_cnt == 0)
1031                                                         break;
1032                                         }
1033                                         sed_cmd = sed_cmd->next;
1034                                         if (!sed_cmd)
1035                                                 bb_error_msg_and_die("unterminated {");
1036                                 }
1037                         }
1038                         continue;
1039                 }
1040
1041                 /* Okay, so did this line match? */
1042                 if (sed_cmd->invert ? matched : !matched)
1043                         continue; /* no */
1044
1045                 /* Update last used regex in case a blank substitute BRE is found */
1046                 if (sed_cmd->beg_match) {
1047                         G.previous_regex_ptr = sed_cmd->beg_match;
1048                 }
1049
1050                 /* actual sedding */
1051                 //bb_error_msg("pattern_space:'%s' next_line:'%s' cmd:%c",
1052                 //pattern_space, next_line, sed_cmd->cmd);
1053                 switch (sed_cmd->cmd) {
1054
1055                 /* Print line number */
1056                 case '=':
1057                         fprintf(G.nonstdout, "%d\n", linenum);
1058                         break;
1059
1060                 /* Write the current pattern space up to the first newline */
1061                 case 'P':
1062                 {
1063                         char *tmp = strchr(pattern_space, '\n');
1064                         if (tmp) {
1065                                 *tmp = '\0';
1066                                 /* TODO: explain why '\n' below */
1067                                 sed_puts(pattern_space, '\n');
1068                                 *tmp = '\n';
1069                                 break;
1070                         }
1071                         /* Fall Through */
1072                 }
1073
1074                 /* Write the current pattern space to output */
1075                 case 'p':
1076                         /* NB: we print this _before_ the last line
1077                          * (of current file) is printed. Even if
1078                          * that line is nonterminated, we print
1079                          * '\n' here (gnu sed does the same) */
1080                         sed_puts(pattern_space, '\n');
1081                         break;
1082                 /* Delete up through first newline */
1083                 case 'D':
1084                 {
1085                         char *tmp = strchr(pattern_space, '\n');
1086                         if (tmp) {
1087                                 overlapping_strcpy(pattern_space, tmp + 1);
1088                                 goto restart;
1089                         }
1090                 }
1091                 /* discard this line. */
1092                 case 'd':
1093                         goto discard_line;
1094
1095                 /* Substitute with regex */
1096                 case 's':
1097                         if (!do_subst_command(sed_cmd, &pattern_space))
1098                                 break;
1099                         substituted |= 1;
1100
1101                         /* handle p option */
1102                         if (sed_cmd->sub_p)
1103                                 sed_puts(pattern_space, last_gets_char);
1104                         /* handle w option */
1105                         if (sed_cmd->sw_file)
1106                                 puts_maybe_newline(
1107                                         pattern_space, sed_cmd->sw_file,
1108                                         &sed_cmd->sw_last_char, last_gets_char);
1109                         break;
1110
1111                 /* Append line to linked list to be printed later */
1112                 case 'a':
1113                         append(sed_cmd->string);
1114                         break;
1115
1116                 /* Insert text before this line */
1117                 case 'i':
1118                         sed_puts(sed_cmd->string, '\n');
1119                         break;
1120
1121                 /* Cut and paste text (replace) */
1122                 case 'c':
1123                         /* Only triggers on last line of a matching range. */
1124                         if (!sed_cmd->in_match)
1125                                 sed_puts(sed_cmd->string, '\n');
1126                         goto discard_line;
1127
1128                 /* Read file, append contents to output */
1129                 case 'r':
1130                 {
1131                         FILE *rfile;
1132                         rfile = fopen_for_read(sed_cmd->string);
1133                         if (rfile) {
1134                                 char *line;
1135
1136                                 while ((line = xmalloc_fgetline(rfile))
1137                                                 != NULL)
1138                                         append(line);
1139                                 xprint_and_close_file(rfile);
1140                         }
1141
1142                         break;
1143                 }
1144
1145                 /* Write pattern space to file. */
1146                 case 'w':
1147                         puts_maybe_newline(
1148                                 pattern_space, sed_cmd->sw_file,
1149                                 &sed_cmd->sw_last_char, last_gets_char);
1150                         break;
1151
1152                 /* Read next line from input */
1153                 case 'n':
1154                         if (!G.be_quiet)
1155                                 sed_puts(pattern_space, last_gets_char);
1156                         if (next_line) {
1157                                 free(pattern_space);
1158                                 pattern_space = next_line;
1159                                 last_gets_char = next_gets_char;
1160                                 next_line = get_next_line(&next_gets_char);
1161                                 substituted = 0;
1162                                 linenum++;
1163                                 break;
1164                         }
1165                         /* fall through */
1166
1167                 /* Quit.  End of script, end of input. */
1168                 case 'q':
1169                         /* Exit the outer while loop */
1170                         free(next_line);
1171                         next_line = NULL;
1172                         goto discard_commands;
1173
1174                 /* Append the next line to the current line */
1175                 case 'N':
1176                 {
1177                         int len;
1178                         /* If no next line, jump to end of script and exit. */
1179                         /* http://www.gnu.org/software/sed/manual/sed.html:
1180                          * "Most versions of sed exit without printing anything
1181                          * when the N command is issued on the last line of
1182                          * a file. GNU sed prints pattern space before exiting
1183                          * unless of course the -n command switch has been
1184                          * specified. This choice is by design."
1185                          */
1186                         if (next_line == NULL) {
1187                                 //goto discard_line;
1188                                 goto discard_commands; /* GNU behavior */
1189                         }
1190                         /* Append next_line, read new next_line. */
1191                         len = strlen(pattern_space);
1192                         pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2);
1193                         pattern_space[len] = '\n';
1194                         strcpy(pattern_space + len+1, next_line);
1195                         last_gets_char = next_gets_char;
1196                         next_line = get_next_line(&next_gets_char);
1197                         linenum++;
1198                         break;
1199                 }
1200
1201                 /* Test/branch if substitution occurred */
1202                 case 't':
1203                         if (!substituted) break;
1204                         substituted = 0;
1205                         /* Fall through */
1206                 /* Test/branch if substitution didn't occur */
1207                 case 'T':
1208                         if (substituted) break;
1209                         /* Fall through */
1210                 /* Branch to label */
1211                 case 'b':
1212                         if (!sed_cmd->string) goto discard_commands;
1213                         else sed_cmd = branch_to(sed_cmd->string);
1214                         break;
1215                 /* Transliterate characters */
1216                 case 'y':
1217                 {
1218                         int i, j;
1219                         for (i = 0; pattern_space[i]; i++) {
1220                                 for (j = 0; sed_cmd->string[j]; j += 2) {
1221                                         if (pattern_space[i] == sed_cmd->string[j]) {
1222                                                 pattern_space[i] = sed_cmd->string[j + 1];
1223                                                 break;
1224                                         }
1225                                 }
1226                         }
1227
1228                         break;
1229                 }
1230                 case 'g':       /* Replace pattern space with hold space */
1231                         free(pattern_space);
1232                         pattern_space = xstrdup(G.hold_space ? G.hold_space : "");
1233                         break;
1234                 case 'G':       /* Append newline and hold space to pattern space */
1235                 {
1236                         int pattern_space_size = 2;
1237                         int hold_space_size = 0;
1238
1239                         if (pattern_space)
1240                                 pattern_space_size += strlen(pattern_space);
1241                         if (G.hold_space)
1242                                 hold_space_size = strlen(G.hold_space);
1243                         pattern_space = xrealloc(pattern_space,
1244                                         pattern_space_size + hold_space_size);
1245                         if (pattern_space_size == 2)
1246                                 pattern_space[0] = 0;
1247                         strcat(pattern_space, "\n");
1248                         if (G.hold_space)
1249                                 strcat(pattern_space, G.hold_space);
1250                         last_gets_char = '\n';
1251
1252                         break;
1253                 }
1254                 case 'h':       /* Replace hold space with pattern space */
1255                         free(G.hold_space);
1256                         G.hold_space = xstrdup(pattern_space);
1257                         break;
1258                 case 'H':       /* Append newline and pattern space to hold space */
1259                 {
1260                         int hold_space_size = 2;
1261                         int pattern_space_size = 0;
1262
1263                         if (G.hold_space)
1264                                 hold_space_size += strlen(G.hold_space);
1265                         if (pattern_space)
1266                                 pattern_space_size = strlen(pattern_space);
1267                         G.hold_space = xrealloc(G.hold_space,
1268                                         hold_space_size + pattern_space_size);
1269
1270                         if (hold_space_size == 2)
1271                                 *G.hold_space = 0;
1272                         strcat(G.hold_space, "\n");
1273                         if (pattern_space)
1274                                 strcat(G.hold_space, pattern_space);
1275
1276                         break;
1277                 }
1278                 case 'x': /* Exchange hold and pattern space */
1279                 {
1280                         char *tmp = pattern_space;
1281                         pattern_space = G.hold_space ? G.hold_space : xzalloc(1);
1282                         last_gets_char = '\n';
1283                         G.hold_space = tmp;
1284                         break;
1285                 }
1286                 } /* switch */
1287         } /* for each cmd */
1288
1289         /*
1290          * Exit point from sedding...
1291          */
1292  discard_commands:
1293         /* we will print the line unless we were told to be quiet ('-n')
1294            or if the line was suppressed (ala 'd'elete) */
1295         if (!G.be_quiet)
1296                 sed_puts(pattern_space, last_gets_char);
1297
1298         /* Delete and such jump here. */
1299  discard_line:
1300         flush_append();
1301         free(pattern_space);
1302
1303         goto again;
1304 }
1305
1306 /* It is possible to have a command line argument with embedded
1307  * newlines.  This counts as multiple command lines.
1308  * However, newline can be escaped: 's/e/z\<newline>z/'
1309  * We check for this.
1310  */
1311
1312 static void add_cmd_block(char *cmdstr)
1313 {
1314         char *sv, *eol;
1315
1316         cmdstr = sv = xstrdup(cmdstr);
1317         do {
1318                 eol = strchr(cmdstr, '\n');
1319  next:
1320                 if (eol) {
1321                         /* Count preceding slashes */
1322                         int slashes = 0;
1323                         char *sl = eol;
1324
1325                         while (sl != cmdstr && *--sl == '\\')
1326                                 slashes++;
1327                         /* Odd number of preceding slashes - newline is escaped */
1328                         if (slashes & 1) {
1329                                 overlapping_strcpy(eol - 1, eol);
1330                                 eol = strchr(eol, '\n');
1331                                 goto next;
1332                         }
1333                         *eol = '\0';
1334                 }
1335                 add_cmd(cmdstr);
1336                 cmdstr = eol + 1;
1337         } while (eol);
1338         free(sv);
1339 }
1340
1341 int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
1342 int sed_main(int argc UNUSED_PARAM, char **argv)
1343 {
1344         unsigned opt;
1345         llist_t *opt_e, *opt_f;
1346         int status = EXIT_SUCCESS;
1347
1348         INIT_G();
1349
1350         /* destroy command strings on exit */
1351         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1352
1353         /* Lie to autoconf when it starts asking stupid questions. */
1354         if (argv[1] && !strcmp(argv[1], "--version")) {
1355                 puts("This is not GNU sed version 4.0");
1356                 return 0;
1357         }
1358
1359         /* do normal option parsing */
1360         opt_e = opt_f = NULL;
1361         opt_complementary = "e::f::" /* can occur multiple times */
1362                             "nn"; /* count -n */
1363         /* -i must be first, to match OPT_in_place definition */
1364         opt = getopt32(argv, "irne:f:", &opt_e, &opt_f,
1365                             &G.be_quiet); /* counter for -n */
1366         //argc -= optind;
1367         argv += optind;
1368         if (opt & OPT_in_place) { // -i
1369                 atexit(cleanup_outname);
1370         }
1371         if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r
1372         //if (opt & 0x4) G.be_quiet++; // -n
1373         while (opt_e) { // -e
1374                 add_cmd_block(llist_pop(&opt_e));
1375         }
1376         while (opt_f) { // -f
1377                 char *line;
1378                 FILE *cmdfile;
1379                 cmdfile = xfopen_for_read(llist_pop(&opt_f));
1380                 while ((line = xmalloc_fgetline(cmdfile)) != NULL) {
1381                         add_cmd(line);
1382                         free(line);
1383                 }
1384                 fclose(cmdfile);
1385         }
1386         /* if we didn't get a pattern from -e or -f, use argv[0] */
1387         if (!(opt & 0x18)) {
1388                 if (!*argv)
1389                         bb_show_usage();
1390                 add_cmd_block(*argv++);
1391         }
1392         /* Flush any unfinished commands. */
1393         add_cmd("");
1394
1395         /* By default, we write to stdout */
1396         G.nonstdout = stdout;
1397
1398         /* argv[0..(argc-1)] should be names of file to process. If no
1399          * files were specified or '-' was specified, take input from stdin.
1400          * Otherwise, we process all the files specified. */
1401         if (argv[0] == NULL) {
1402                 if (opt & OPT_in_place)
1403                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1404                 add_input_file(stdin);
1405         } else {
1406                 int i;
1407                 FILE *file;
1408
1409                 for (i = 0; argv[i]; i++) {
1410                         struct stat statbuf;
1411                         int nonstdoutfd;
1412
1413                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1414                                 add_input_file(stdin);
1415                                 process_files();
1416                                 continue;
1417                         }
1418                         file = fopen_or_warn(argv[i], "r");
1419                         if (!file) {
1420                                 status = EXIT_FAILURE;
1421                                 continue;
1422                         }
1423                         if (!(opt & OPT_in_place)) {
1424                                 add_input_file(file);
1425                                 continue;
1426                         }
1427
1428                         G.outname = xasprintf("%sXXXXXX", argv[i]);
1429                         nonstdoutfd = xmkstemp(G.outname);
1430                         G.nonstdout = xfdopen_for_write(nonstdoutfd);
1431
1432                         /* Set permissions/owner of output file */
1433                         fstat(fileno(file), &statbuf);
1434                         /* chmod'ing AFTER chown would preserve suid/sgid bits,
1435                          * but GNU sed 4.2.1 does not preserve them either */
1436                         fchmod(nonstdoutfd, statbuf.st_mode);
1437                         fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid);
1438                         add_input_file(file);
1439                         process_files();
1440                         fclose(G.nonstdout);
1441
1442                         G.nonstdout = stdout;
1443                         /* unlink(argv[i]); */
1444                         xrename(G.outname, argv[i]);
1445                         free(G.outname);
1446                         G.outname = NULL;
1447                 }
1448                 /* Here, to handle "sed 'cmds' nonexistent_file" case we did:
1449                  * if (G.current_input_file >= G.input_file_count)
1450                  *      return status;
1451                  * but it's not needed since process_files() works correctly
1452                  * in this case too. */
1453         }
1454         process_files();
1455
1456         return status;
1457 }