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 <bug1@iinet.net.au>
   9  * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net>
  10  *
  11  * MAINTAINER: Rob Landley <rob@landley.net>
  12  *
  13  * Licensed under GPL version 2, see file LICENSE in this tarball for details.
  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   (bbg.sed_cmd_head/bbg.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
  37 /*
  38         Supported features and commands in this version of sed:
  39
  40          - comments ('#')
  41          - address matching: num|/matchstr/[,num|/matchstr/|$]command
  42          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  43          - edit commands: (a)ppend, (i)nsert, (c)hange
  44          - file commands: (r)ead
  45          - backreferences in substitution expressions (\0, \1, \2...\9)
  46          - grouped commands: {cmd1;cmd2}
  47          - transliteration (y/source-chars/dest-chars/)
  48          - pattern space hold space storing / swapping (g, h, x)
  49          - labels / branching (: label, b, t, T)
  50
  51          (Note: Specifying an address (range) to match is *optional*; commands
  52          default to the whole pattern space if no specific address match was
  53          requested.)
  54
  55         Todo:
  56          - Create a wrapper around regex to make libc's regex conform with sed
  57
  58         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  59 */
  60
  61 #include "busybox.h"
  62 #include "xregex.h"
  63
  64 /* Each sed command turns into one of these structures. */
  65 typedef struct sed_cmd_s {
  66         /* Ordered by alignment requirements: currently 36 bytes on x86 */
  67
  68         /* address storage */
  69         regex_t *beg_match;     /* sed -e '/match/cmd' */
  70         regex_t *end_match;     /* sed -e '/match/,/end_match/cmd' */
  71         regex_t *sub_match;     /* For 's/sub_match/string/' */
  72         int beg_line;           /* 'sed 1p'   0 == apply commands to all lines */
  73         int end_line;           /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  74
  75         FILE *sw_file;          /* File (sw) command writes to, -1 for none. */
  76         char *string;           /* Data string for (saicytb) commands. */
  77
  78         unsigned short which_match;     /* (s) Which match to replace (0 for all) */
  79
  80         /* Bitfields (gcc won't group them if we don't) */
  81         unsigned int invert:1;          /* the '!' after the address */
  82         unsigned int in_match:1;        /* Next line also included in match? */
  83         unsigned int sub_p:1;           /* (s) print option */
  84
  85         char sw_last_char;                  /* Last line written by (sw) had no '\n' */
  86
  87         /* GENERAL FIELDS */
  88         char cmd;               /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
  89         struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */
  90 } sed_cmd_t;
  91
  92 static const char *const semicolon_whitespace = "; \n\r\t\v";
  93
  94 struct sed_globals {
  95         /* options */
  96         int be_quiet, regex_type;
  97         FILE *nonstdout;
  98         char *outname, *hold_space;
  99
 100         /* List of input files */
 101         int input_file_count, current_input_file;
 102         FILE **input_file_list;
 103
 104         regmatch_t regmatch[10];
 105         regex_t *previous_regex_ptr;
 106
 107         /* linked list of sed commands */
 108         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 109
 110         /* Linked list of append lines */
 111         llist_t *append_head;
 112
 113         char *add_cmd_line;
 114
 115         struct pipeline {
 116                 char *buf;      /* Space to hold string */
 117                 int idx;        /* Space used */
 118                 int len;        /* Space allocated */
 119         } pipeline;
 120 } bbg;
 121
 122
 123 #if ENABLE_FEATURE_CLEAN_UP
 124 static void sed_free_and_close_stuff(void)
 125 {
 126         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 127
 128         llist_free(bbg.append_head, free);
 129
 130         while (sed_cmd) {
 131                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 132
 133                 if (sed_cmd->sw_file)
 134                         xprint_and_close_file(sed_cmd->sw_file);
 135
 136                 if (sed_cmd->beg_match) {
 137                         regfree(sed_cmd->beg_match);
 138                         free(sed_cmd->beg_match);
 139                 }
 140                 if (sed_cmd->end_match) {
 141                         regfree(sed_cmd->end_match);
 142                         free(sed_cmd->end_match);
 143                 }
 144                 if (sed_cmd->sub_match) {
 145                         regfree(sed_cmd->sub_match);
 146                         free(sed_cmd->sub_match);
 147                 }
 148                 free(sed_cmd->string);
 149                 free(sed_cmd);
 150                 sed_cmd = sed_cmd_next;
 151         }
 152
 153         if (bbg.hold_space) free(bbg.hold_space);
 154
 155         while (bbg.current_input_file < bbg.input_file_count)
 156                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 157 }
 158 #else
 159 void sed_free_and_close_stuff(void);
 160 #endif
 161
 162 /* If something bad happens during -i operation, delete temp file */
 163
 164 static void cleanup_outname(void)
 165 {
 166         if (bbg.outname) unlink(bbg.outname);
 167 }
 168
 169 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 170
 171 static void parse_escapes(char *dest, char *string, int len, char from, char to)
 172 {
 173         int i = 0;
 174
 175         while (i < len) {
 176                 if (string[i] == '\\') {
 177                         if (!to || string[i+1] == from) {
 178                                 *dest++ = to ? to : string[i+1];
 179                                 i += 2;
 180                                 continue;
 181                         }
 182                         *dest++ = string[i++];
 183                 }
 184                 *dest++ = string[i++];
 185         }
 186         *dest = 0;
 187 }
 188
 189 static char *copy_parsing_escapes(char *string, int len)
 190 {
 191         char *dest = xmalloc(len + 1);
 192
 193         parse_escapes(dest, string, len, 'n', '\n');
 194         return dest;
 195 }
 196
 197
 198 /*
 199  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 200  * beginning at a specified index and returns the index of the next regular
 201  * expression delimiter (typically a forward * slash ('/')) not preceded by
 202  * a backslash ('\').  A negative delimiter disables square bracket checking.
 203  */
 204 static int index_of_next_unescaped_regexp_delim(int delimiter, char *str)
 205 {
 206         int bracket = -1;
 207         int escaped = 0;
 208         int idx = 0;
 209         char ch;
 210
 211         if (delimiter < 0) {
 212                 bracket--;
 213                 delimiter = -delimiter;
 214         }
 215
 216         for (; (ch = str[idx]); idx++) {
 217                 if (bracket >= 0) {
 218                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 219                                         && str[idx - 1] == '^')))
 220                                 bracket = -1;
 221                 } else if (escaped)
 222                         escaped = 0;
 223                 else if (ch == '\\')
 224                         escaped = 1;
 225                 else if (bracket == -1 && ch == '[')
 226                         bracket = idx;
 227                 else if (ch == delimiter)
 228                         return idx;
 229         }
 230
 231         /* if we make it to here, we've hit the end of the string */
 232         bb_error_msg_and_die("unmatched '%c'", delimiter);
 233 }
 234
 235 /*
 236  *  Returns the index of the third delimiter
 237  */
 238 static int parse_regex_delim(char *cmdstr, char **match, char **replace)
 239 {
 240         char *cmdstr_ptr = cmdstr;
 241         char delimiter;
 242         int idx = 0;
 243
 244         /* verify that the 's' or 'y' is followed by something.  That something
 245          * (typically a 'slash') is now our regexp delimiter... */
 246         if (*cmdstr == '\0')
 247                 bb_error_msg_and_die("bad format in substitution expression");
 248         delimiter = *cmdstr_ptr++;
 249
 250         /* save the match string */
 251         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 252         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 253
 254         /* save the replacement string */
 255         cmdstr_ptr += idx + 1;
 256         idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr);
 257         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 258
 259         return ((cmdstr_ptr - cmdstr) + idx);
 260 }
 261
 262 /*
 263  * returns the index in the string just past where the address ends.
 264  */
 265 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 266 {
 267         char *pos = my_str;
 268
 269         if (isdigit(*my_str)) {
 270                 *linenum = strtol(my_str, &pos, 10);
 271                 /* endstr shouldnt ever equal NULL */
 272         } else if (*my_str == '$') {
 273                 *linenum = -1;
 274                 pos++;
 275         } else if (*my_str == '/' || *my_str == '\\') {
 276                 int next;
 277                 char delimiter;
 278                 char *temp;
 279
 280                 delimiter = '/';
 281                 if (*my_str == '\\') delimiter = *++pos;
 282                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 283                 temp = copy_parsing_escapes(pos, next);
 284                 *regex = xmalloc(sizeof(regex_t));
 285                 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
 286                 free(temp);
 287                 /* Move position to next character after last delimiter */
 288                 pos += (next+1);
 289         }
 290         return pos - my_str;
 291 }
 292
 293 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 294 static int parse_file_cmd(sed_cmd_t *sed_cmd, char *filecmdstr, char **retval)
 295 {
 296         int start = 0, idx, hack = 0;
 297
 298         /* Skip whitespace, then grab filename to end of line */
 299         while (isspace(filecmdstr[start])) start++;
 300         idx = start;
 301         while (filecmdstr[idx] && filecmdstr[idx] != '\n') idx++;
 302
 303         /* If lines glued together, put backslash back. */
 304         if (filecmdstr[idx] == '\n') hack = 1;
 305         if (idx == start)
 306                 bb_error_msg_and_die("empty filename");
 307         *retval = xstrndup(filecmdstr+start, idx-start+hack+1);
 308         if (hack) (*retval)[idx] = '\\';
 309
 310         return idx;
 311 }
 312
 313 static int parse_subst_cmd(sed_cmd_t *sed_cmd, char *substr)
 314 {
 315         int cflags = bbg.regex_type;
 316         char *match;
 317         int idx = 0;
 318
 319         /*
 320          * A substitution command should look something like this:
 321          *    s/match/replace/ #gIpw
 322          *    ||     |        |||
 323          *    mandatory       optional
 324          */
 325         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 326
 327         /* determine the number of back references in the match string */
 328         /* Note: we compute this here rather than in the do_subst_command()
 329          * function to save processor time, at the expense of a little more memory
 330          * (4 bits) per sed_cmd */
 331
 332         /* process the flags */
 333
 334         sed_cmd->which_match = 1;
 335         while (substr[++idx]) {
 336                 /* Parse match number */
 337                 if (isdigit(substr[idx])) {
 338                         if (match[0] != '^') {
 339                                 /* Match 0 treated as all, multiple matches we take the last one. */
 340                                 char *pos = substr + idx;
 341                                 /* FIXME: error check? */
 342                                 sed_cmd->which_match = (unsigned short)strtol(substr+idx, &pos, 10);
 343                                 idx = pos - substr;
 344                         }
 345                         continue;
 346                 }
 347                 /* Skip spaces */
 348                 if (isspace(substr[idx])) continue;
 349
 350                 switch (substr[idx]) {
 351                 /* Replace all occurrences */
 352                 case 'g':
 353                         if (match[0] != '^') sed_cmd->which_match = 0;
 354                         break;
 355                 /* Print pattern space */
 356                 case 'p':
 357                         sed_cmd->sub_p = 1;
 358                         break;
 359                 /* Write to file */
 360                 case 'w':
 361                 {
 362                         char *temp;
 363                         idx += parse_file_cmd(sed_cmd, substr+idx, &temp);
 364
 365                         break;
 366                 }
 367                 /* Ignore case (gnu exension) */
 368                 case 'I':
 369                         cflags |= REG_ICASE;
 370                         break;
 371                 /* Comment */
 372                 case '#':
 373                         while (substr[++idx]) /*skip all*/;
 374                         /* Fall through */
 375                 /* End of command */
 376                 case ';':
 377                 case '}':
 378                         goto out;
 379                 default:
 380                         bb_error_msg_and_die("bad option in substitution expression");
 381                 }
 382         }
 383 out:
 384         /* compile the match string into a regex */
 385         if (*match != '\0') {
 386                 /* If match is empty, we use last regex used at runtime */
 387                 sed_cmd->sub_match = xmalloc(sizeof(regex_t));
 388                 xregcomp(sed_cmd->sub_match, match, cflags);
 389         }
 390         free(match);
 391
 392         return idx;
 393 }
 394
 395 /*
 396  *  Process the commands arguments
 397  */
 398 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 399 {
 400         /* handle (s)ubstitution command */
 401         if (sed_cmd->cmd == 's')
 402                 cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 403         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 404         else if (strchr("aic", sed_cmd->cmd)) {
 405                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 406                         bb_error_msg_and_die
 407                                 ("only a beginning address can be specified for edit commands");
 408                 for (;;) {
 409                         if (*cmdstr == '\n' || *cmdstr == '\\') {
 410                                 cmdstr++;
 411                                 break;
 412                         } else if (isspace(*cmdstr))
 413                                 cmdstr++;
 414                         else
 415                                 break;
 416                 }
 417                 sed_cmd->string = xstrdup(cmdstr);
 418                 parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), 0, 0);
 419                 cmdstr += strlen(cmdstr);
 420         /* handle file cmds: (r)ead */
 421         } else if (strchr("rw", sed_cmd->cmd)) {
 422                 if (sed_cmd->end_line || sed_cmd->end_match)
 423                         bb_error_msg_and_die("command only uses one address");
 424                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 425                 if (sed_cmd->cmd == 'w') {
 426                         sed_cmd->sw_file = xfopen(sed_cmd->string, "w");
 427                         sed_cmd->sw_last_char = '\n';
 428                 }
 429         /* handle branch commands */
 430         } else if (strchr(":btT", sed_cmd->cmd)) {
 431                 int length;
 432
 433                 cmdstr = skip_whitespace(cmdstr);
 434                 length = strcspn(cmdstr, semicolon_whitespace);
 435                 if (length) {
 436                         sed_cmd->string = xstrndup(cmdstr, length);
 437                         cmdstr += length;
 438                 }
 439         }
 440         /* translation command */
 441         else if (sed_cmd->cmd == 'y') {
 442                 char *match, *replace;
 443                 int i = cmdstr[0];
 444
 445                 cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
 446                 /* \n already parsed, but \delimiter needs unescaping. */
 447                 parse_escapes(match, match, strlen(match), i, i);
 448                 parse_escapes(replace, replace, strlen(replace), i, i);
 449
 450                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 451                 for (i = 0; match[i] && replace[i]; i++) {
 452                         sed_cmd->string[i*2] = match[i];
 453                         sed_cmd->string[i*2+1] = replace[i];
 454                 }
 455                 free(match);
 456                 free(replace);
 457         }
 458         /* if it wasnt a single-letter command that takes no arguments
 459          * then it must be an invalid command.
 460          */
 461         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 462                 bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
 463         }
 464
 465         /* give back whatever's left over */
 466         return cmdstr;
 467 }
 468
 469
 470 /* Parse address+command sets, skipping comment lines. */
 471
 472 static void add_cmd(char *cmdstr)
 473 {
 474         sed_cmd_t *sed_cmd;
 475         int temp;
 476
 477         /* Append this line to any unfinished line from last time. */
 478         if (bbg.add_cmd_line) {
 479                 cmdstr = xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 480                 free(bbg.add_cmd_line);
 481                 bbg.add_cmd_line = cmdstr;
 482         }
 483
 484         /* If this line ends with backslash, request next line. */
 485         temp = strlen(cmdstr);
 486         if (temp && cmdstr[temp-1] == '\\') {
 487                 if (!bbg.add_cmd_line)
 488                         bbg.add_cmd_line = xstrdup(cmdstr);
 489                 bbg.add_cmd_line[temp-1] = 0;
 490                 return;
 491         }
 492
 493         /* Loop parsing all commands in this line. */
 494         while (*cmdstr) {
 495                 /* Skip leading whitespace and semicolons */
 496                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 497
 498                 /* If no more commands, exit. */
 499                 if (!*cmdstr) break;
 500
 501                 /* if this is a comment, jump past it and keep going */
 502                 if (*cmdstr == '#') {
 503                         /* "#n" is the same as using -n on the command line */
 504                         if (cmdstr[1] == 'n')
 505                                 bbg.be_quiet++;
 506                         cmdstr = strpbrk(cmdstr, "\n\r");
 507                         if (!cmdstr) break;
 508                         continue;
 509                 }
 510
 511                 /* parse the command
 512                  * format is: [addr][,addr][!]cmd
 513                  *            |----||-----||-|
 514                  *            part1 part2  part3
 515                  */
 516
 517                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 518
 519                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 520                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 521
 522                 /* second part (if present) will begin with a comma */
 523                 if (*cmdstr == ',') {
 524                         int idx;
 525
 526                         cmdstr++;
 527                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 528                         if (!idx)
 529                                 bb_error_msg_and_die("no address after comma");
 530                         cmdstr += idx;
 531                 }
 532
 533                 /* skip whitespace before the command */
 534                 cmdstr = skip_whitespace(cmdstr);
 535
 536                 /* Check for inversion flag */
 537                 if (*cmdstr == '!') {
 538                         sed_cmd->invert = 1;
 539                         cmdstr++;
 540
 541                         /* skip whitespace before the command */
 542                         cmdstr = skip_whitespace(cmdstr);
 543                 }
 544
 545                 /* last part (mandatory) will be a command */
 546                 if (!*cmdstr)
 547                         bb_error_msg_and_die("missing command");
 548                 sed_cmd->cmd = *(cmdstr++);
 549                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 550
 551                 /* Add the command to the command array */
 552                 bbg.sed_cmd_tail->next = sed_cmd;
 553                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 554         }
 555
 556         /* If we glued multiple lines together, free the memory. */
 557         free(bbg.add_cmd_line);
 558         bbg.add_cmd_line = NULL;
 559 }
 560
 561 /* Append to a string, reallocating memory as necessary. */
 562
 563 #define PIPE_GROW 64
 564
 565 static void pipe_putc(char c)
 566 {
 567         if (bbg.pipeline.idx == bbg.pipeline.len) {
 568                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 569                                 bbg.pipeline.len + PIPE_GROW);
 570                 bbg.pipeline.len += PIPE_GROW;
 571         }
 572         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 573 }
 574
 575 static void do_subst_w_backrefs(char *line, char *replace)
 576 {
 577         int i,j;
 578
 579         /* go through the replacement string */
 580         for (i = 0; replace[i]; i++) {
 581                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 582                 if (replace[i] == '\\') {
 583                         unsigned backref = replace[++i] - '0';
 584                         if (backref <= 9) {
 585                                 /* print out the text held in bbg.regmatch[backref] */
 586                                 if (bbg.regmatch[backref].rm_so != -1) {
 587                                         j = bbg.regmatch[backref].rm_so;
 588                                         while (j < bbg.regmatch[backref].rm_eo)
 589                                                 pipe_putc(line[j++]);
 590                                 }
 591                                 continue;
 592                         }
 593                         /* I _think_ it is impossible to get '\' to be
 594                          * the last char in replace string. Thus we dont check
 595                          * for replace[i] == NUL. (counterexample anyone?) */
 596                         /* if we find a backslash escaped character, print the character */
 597                         pipe_putc(replace[i]);
 598                         continue;
 599                 }
 600                 /* if we find an unescaped '&' print out the whole matched text. */
 601                 if (replace[i] == '&') {
 602                         j = bbg.regmatch[0].rm_so;
 603                         while (j < bbg.regmatch[0].rm_eo)
 604                                 pipe_putc(line[j++]);
 605                         continue;
 606                 }
 607                 /* Otherwise just output the character. */
 608                 pipe_putc(replace[i]);
 609         }
 610 }
 611
 612 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 613 {
 614         char *oldline = *line;
 615         int altered = 0;
 616         int match_count = 0;
 617         regex_t *current_regex;
 618
 619         /* Handle empty regex. */
 620         if (sed_cmd->sub_match == NULL) {
 621                 current_regex = bbg.previous_regex_ptr;
 622                 if (!current_regex)
 623                         bb_error_msg_and_die("no previous regexp");
 624         } else
 625                 bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 626
 627         /* Find the first match */
 628         if (REG_NOMATCH == regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 629                 return 0;
 630
 631         /* Initialize temporary output buffer. */
 632         bbg.pipeline.buf = xmalloc(PIPE_GROW);
 633         bbg.pipeline.len = PIPE_GROW;
 634         bbg.pipeline.idx = 0;
 635
 636         /* Now loop through, substituting for matches */
 637         do {
 638                 int i;
 639
 640                 /* Work around bug in glibc regexec, demonstrated by:
 641                    echo " a.b" | busybox sed 's [^ .]* x g'
 642                    The match_count check is so not to break
 643                    echo "hi" | busybox sed 's/^/!/g' */
 644                 if (!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
 645                         pipe_putc(*oldline++);
 646                         continue;
 647                 }
 648
 649                 match_count++;
 650
 651                 /* If we aren't interested in this match, output old line to
 652                    end of match and continue */
 653                 if (sed_cmd->which_match && sed_cmd->which_match != match_count) {
 654                         for (i = 0; i < bbg.regmatch[0].rm_eo; i++)
 655                                 pipe_putc(*oldline++);
 656                         continue;
 657                 }
 658
 659                 /* print everything before the match */
 660                 for (i = 0; i < bbg.regmatch[0].rm_so; i++)
 661                         pipe_putc(oldline[i]);
 662
 663                 /* then print the substitution string */
 664                 do_subst_w_backrefs(oldline, sed_cmd->string);
 665
 666                 /* advance past the match */
 667                 oldline += bbg.regmatch[0].rm_eo;
 668                 /* flag that something has changed */
 669                 altered++;
 670
 671                 /* if we're not doing this globally, get out now */
 672                 if (sed_cmd->which_match) break;
 673         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 674
 675         /* Copy rest of string into output pipeline */
 676
 677         while (*oldline)
 678                 pipe_putc(*oldline++);
 679         pipe_putc(0);
 680
 681         free(*line);
 682         *line = bbg.pipeline.buf;
 683         return altered;
 684 }
 685
 686 /* Set command pointer to point to this label.  (Does not handle null label.) */
 687 static sed_cmd_t *branch_to(char *label)
 688 {
 689         sed_cmd_t *sed_cmd;
 690
 691         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 692                 if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) {
 693                         return sed_cmd;
 694                 }
 695         }
 696         bb_error_msg_and_die("can't find label for jump to '%s'", label);
 697 }
 698
 699 static void append(char *s)
 700 {
 701         llist_add_to_end(&bbg.append_head, xstrdup(s));
 702 }
 703
 704 static void flush_append(void)
 705 {
 706         char *data;
 707
 708         /* Output appended lines. */
 709         while ((data = (char *)llist_pop(&bbg.append_head))) {
 710                 fprintf(bbg.nonstdout, "%s\n", data);
 711                 free(data);
 712         }
 713 }
 714
 715 static void add_input_file(FILE *file)
 716 {
 717         bbg.input_file_list = xrealloc(bbg.input_file_list,
 718                         (bbg.input_file_count + 1) * sizeof(FILE *));
 719         bbg.input_file_list[bbg.input_file_count++] = file;
 720 }
 721
 722 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 723  * noting if we ran out of files without a newline on the last line we read.
 724  */
 725 enum {
 726         NO_EOL_CHAR = 1,
 727 };
 728 static char *get_next_line(char *gets_char)
 729 {
 730         char *temp = NULL;
 731         int len;
 732         char gc;
 733
 734         flush_append();
 735
 736         /* will be returned if last line in the file
 737          * doesn't end with either '\n' or '\0' */
 738         gc = NO_EOL_CHAR;
 739         while (bbg.current_input_file < bbg.input_file_count) {
 740                 /* Read line up to a newline or NUL byte, inclusive,
 741                  * return malloc'ed char[]. length of the chunk read
 742                  * is stored in len. NULL if EOF/error */
 743                 temp = bb_get_chunk_from_file(
 744                         bbg.input_file_list[bbg.current_input_file], &len);
 745                 if (temp) {
 746                         /* len > 0 here, it's ok to do temp[len-1] */
 747                         char c = temp[len-1];
 748                         if (c == '\n' || c == '\0') {
 749                                 temp[len-1] = '\0';
 750                                 gc = c;
 751                         }
 752                         /* else we put NO_EOL_CHAR into *gets_char */
 753                         break;
 754
 755                 /* NB: I had the idea of peeking next file(s) and returning
 756                  * NO_EOL_CHAR only if it is the *last* non-empty
 757                  * input file. But there is a case where this won't work:
 758                  * file1: "a woo\nb woo"
 759                  * file2: "c no\nd no"
 760                  * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang"
 761                  * (note: *no* newline after "b bang"!) */
 762                 }
 763                 /* Close this file and advance to next one */
 764                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 765         }
 766         *gets_char = gc;
 767         return temp;
 768 }
 769
 770 /* Output line of text. */
 771 /* Note:
 772  * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed.
 773  * Without them, we had this:
 774  * echo -n thingy >z1
 775  * echo -n again >z2
 776  * >znull
 777  * sed "s/i/z/" z1 z2 znull | hexdump -vC
 778  * output:
 779  * gnu sed 4.1.5:
 780  * 00000000  74 68 7a 6e 67 79 0a 61  67 61 7a 6e              |thzngy.agazn|
 781  * bbox:
 782  * 00000000  74 68 7a 6e 67 79 61 67  61 7a 6e                 |thzngyagazn|
 783  */
 784 static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char)
 785 {
 786         char lpc = *last_puts_char;
 787
 788         /* Is this a first line from new file
 789          * and old file didn't end with '\n' or '\0'? */
 790         if (lpc != '\n' && lpc != '\0') {
 791                 fputc('\n', file);
 792                 lpc = '\n';
 793         }
 794         fputs(s, file);
 795         /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */
 796         if (s[0])
 797                 lpc = 'x';
 798         if (last_gets_char != NO_EOL_CHAR) { /* had trailing '\n' or '\0'? */
 799                 fputc(last_gets_char, file);
 800                 lpc = last_gets_char;
 801         }
 802         if (ferror(file)) {
 803                 xfunc_error_retval = 4;  /* It's what gnu sed exits with... */
 804                 bb_error_msg_and_die(bb_msg_write_error);
 805         }
 806         *last_puts_char = lpc;
 807 }
 808
 809 #define sed_puts(s, n) (puts_maybe_newline(s, bbg.nonstdout, &last_puts_char, n))
 810
 811 /* Process all the lines in all the files */
 812
 813 static void process_files(void)
 814 {
 815         char *pattern_space, *next_line;
 816         int linenum = 0;
 817         char last_puts_char = '\n';
 818         char last_gets_char, next_gets_char;
 819         sed_cmd_t *sed_cmd;
 820         int substituted;
 821
 822         /* Prime the pump */
 823         next_line = get_next_line(&next_gets_char);
 824
 825         /* go through every line in each file */
 826 again:
 827         substituted = 0;
 828
 829         /* Advance to next line.  Stop if out of lines. */
 830         pattern_space = next_line;
 831         if (!pattern_space) return;
 832         last_gets_char = next_gets_char;
 833
 834         /* Read one line in advance so we can act on the last line,
 835          * the '$' address */
 836         next_line = get_next_line(&next_gets_char);
 837         linenum++;
 838 restart:
 839         /* for every line, go through all the commands */
 840         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 841                 int old_matched, matched;
 842
 843                 old_matched = sed_cmd->in_match;
 844
 845                 /* Determine if this command matches this line: */
 846
 847                 /* Are we continuing a previous multi-line match? */
 848                 sed_cmd->in_match = sed_cmd->in_match
 849                         /* Or is no range necessary? */
 850                         || (!sed_cmd->beg_line && !sed_cmd->end_line
 851                                 && !sed_cmd->beg_match && !sed_cmd->end_match)
 852                         /* Or did we match the start of a numerical range? */
 853                         || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 854                         /* Or does this line match our begin address regex? */
 855                         || (sed_cmd->beg_match &&
 856                             !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 857                         /* Or did we match last line of input? */
 858                         || (sed_cmd->beg_line == -1 && next_line == NULL);
 859
 860                 /* Snapshot the value */
 861
 862                 matched = sed_cmd->in_match;
 863
 864                 /* Is this line the end of the current match? */
 865
 866                 if (matched) {
 867                         sed_cmd->in_match = !(
 868                                 /* has the ending line come, or is this a single address command? */
 869                                 (sed_cmd->end_line ?
 870                                         sed_cmd->end_line == -1 ?
 871                                                 !next_line
 872                                                 : (sed_cmd->end_line <= linenum)
 873                                         : !sed_cmd->end_match
 874                                 )
 875                                 /* or does this line matches our last address regex */
 876                                 || (sed_cmd->end_match && old_matched
 877                                      && (regexec(sed_cmd->end_match,
 878                                                  pattern_space, 0, NULL, 0) == 0))
 879                         );
 880                 }
 881
 882                 /* Skip blocks of commands we didn't match. */
 883                 if (sed_cmd->cmd == '{') {
 884                         if (sed_cmd->invert ? matched : !matched) {
 885                                 while (sed_cmd->cmd != '}') {
 886                                         sed_cmd = sed_cmd->next;
 887                                         if (!sed_cmd)
 888                                                 bb_error_msg_and_die("unterminated {");
 889                                 }
 890                         }
 891                         continue;
 892                 }
 893
 894                 /* Okay, so did this line match? */
 895                 if (sed_cmd->invert ? !matched : matched) {
 896                         /* Update last used regex in case a blank substitute BRE is found */
 897                         if (sed_cmd->beg_match) {
 898                                 bbg.previous_regex_ptr = sed_cmd->beg_match;
 899                         }
 900
 901                         /* actual sedding */
 902                         switch (sed_cmd->cmd) {
 903
 904                         /* Print line number */
 905                         case '=':
 906                                 fprintf(bbg.nonstdout, "%d\n", linenum);
 907                                 break;
 908
 909                         /* Write the current pattern space up to the first newline */
 910                         case 'P':
 911                         {
 912                                 char *tmp = strchr(pattern_space, '\n');
 913
 914                                 if (tmp) {
 915                                         *tmp = '\0';
 916                                         /* TODO: explain why '\n' below */
 917                                         sed_puts(pattern_space, '\n');
 918                                         *tmp = '\n';
 919                                         break;
 920                                 }
 921                                 /* Fall Through */
 922                         }
 923
 924                         /* Write the current pattern space to output */
 925                         case 'p':
 926                                 /* NB: we print this _before_ the last line
 927                                  * (of current file) is printed. Even if
 928                                  * that line is nonterminated, we print
 929                                  * '\n' here (gnu sed does the same) */
 930                                 sed_puts(pattern_space, '\n');
 931                                 break;
 932                         /* Delete up through first newline */
 933                         case 'D':
 934                         {
 935                                 char *tmp = strchr(pattern_space, '\n');
 936
 937                                 if (tmp) {
 938                                         tmp = xstrdup(tmp+1);
 939                                         free(pattern_space);
 940                                         pattern_space = tmp;
 941                                         goto restart;
 942                                 }
 943                         }
 944                         /* discard this line. */
 945                         case 'd':
 946                                 goto discard_line;
 947
 948                         /* Substitute with regex */
 949                         case 's':
 950                                 if (!do_subst_command(sed_cmd, &pattern_space))
 951                                         break;
 952                                 substituted |= 1;
 953
 954                                 /* handle p option */
 955                                 if (sed_cmd->sub_p)
 956                                         sed_puts(pattern_space, last_gets_char);
 957                                 /* handle w option */
 958                                 if (sed_cmd->sw_file)
 959                                         puts_maybe_newline(
 960                                                 pattern_space, sed_cmd->sw_file,
 961                                                 &sed_cmd->sw_last_char, last_gets_char);
 962                                 break;
 963
 964                         /* Append line to linked list to be printed later */
 965                         case 'a':
 966                                 append(sed_cmd->string);
 967                                 break;
 968
 969                         /* Insert text before this line */
 970                         case 'i':
 971                                 sed_puts(sed_cmd->string, '\n');
 972                                 break;
 973
 974                         /* Cut and paste text (replace) */
 975                         case 'c':
 976                                 /* Only triggers on last line of a matching range. */
 977                                 if (!sed_cmd->in_match)
 978                                         sed_puts(sed_cmd->string, NO_EOL_CHAR);
 979                                 goto discard_line;
 980
 981                         /* Read file, append contents to output */
 982                         case 'r':
 983                         {
 984                                 FILE *rfile;
 985
 986                                 rfile = fopen(sed_cmd->string, "r");
 987                                 if (rfile) {
 988                                         char *line;
 989
 990                                         while ((line = xmalloc_getline(rfile))
 991                                                         != NULL)
 992                                                 append(line);
 993                                         xprint_and_close_file(rfile);
 994                                 }
 995
 996                                 break;
 997                         }
 998
 999                         /* Write pattern space to file. */
1000                         case 'w':
1001                                 puts_maybe_newline(
1002                                         pattern_space, sed_cmd->sw_file,
1003                                         &sed_cmd->sw_last_char, last_gets_char);
1004                                 break;
1005
1006                         /* Read next line from input */
1007                         case 'n':
1008                                 if (!bbg.be_quiet)
1009                                         sed_puts(pattern_space, last_gets_char);
1010                                 if (next_line) {
1011                                         free(pattern_space);
1012                                         pattern_space = next_line;
1013                                         last_gets_char = next_gets_char;
1014                                         next_line = get_next_line(&next_gets_char);
1015                                         linenum++;
1016                                         break;
1017                                 }
1018                                 /* fall through */
1019
1020                         /* Quit.  End of script, end of input. */
1021                         case 'q':
1022                                 /* Exit the outer while loop */
1023                                 free(next_line);
1024                                 next_line = NULL;
1025                                 goto discard_commands;
1026
1027                         /* Append the next line to the current line */
1028                         case 'N':
1029                         {
1030                                 int len;
1031                                 /* If no next line, jump to end of script and exit. */
1032                                 if (next_line == NULL) {
1033                                         /* Jump to end of script and exit */
1034                                         free(next_line);
1035                                         next_line = NULL;
1036                                         goto discard_line;
1037                                 /* append next_line, read new next_line. */
1038                                 }
1039                                 len = strlen(pattern_space);
1040                                 pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
1041                                 pattern_space[len] = '\n';
1042                                 strcpy(pattern_space + len+1, next_line);
1043                                 last_gets_char = next_gets_char;
1044                                 next_line = get_next_line(&next_gets_char);
1045                                 linenum++;
1046                                 break;
1047                         }
1048
1049                         /* Test/branch if substitution occurred */
1050                         case 't':
1051                                 if (!substituted) break;
1052                                 substituted = 0;
1053                                 /* Fall through */
1054                         /* Test/branch if substitution didn't occur */
1055                         case 'T':
1056                                 if (substituted) break;
1057                                 /* Fall through */
1058                         /* Branch to label */
1059                         case 'b':
1060                                 if (!sed_cmd->string) goto discard_commands;
1061                                 else sed_cmd = branch_to(sed_cmd->string);
1062                                 break;
1063                         /* Transliterate characters */
1064                         case 'y':
1065                         {
1066                                 int i, j;
1067
1068                                 for (i = 0; pattern_space[i]; i++) {
1069                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1070                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1071                                                         pattern_space[i] = sed_cmd->string[j + 1];
1072                                                         break;
1073                                                 }
1074                                         }
1075                                 }
1076
1077                                 break;
1078                         }
1079                         case 'g':       /* Replace pattern space with hold space */
1080                                 free(pattern_space);
1081                                 pattern_space = xstrdup(bbg.hold_space ? bbg.hold_space : "");
1082                                 break;
1083                         case 'G':       /* Append newline and hold space to pattern space */
1084                         {
1085                                 int pattern_space_size = 2;
1086                                 int hold_space_size = 0;
1087
1088                                 if (pattern_space)
1089                                         pattern_space_size += strlen(pattern_space);
1090                                 if (bbg.hold_space)
1091                                         hold_space_size = strlen(bbg.hold_space);
1092                                 pattern_space = xrealloc(pattern_space,
1093                                                 pattern_space_size + hold_space_size);
1094                                 if (pattern_space_size == 2)
1095                                         pattern_space[0] = 0;
1096                                 strcat(pattern_space, "\n");
1097                                 if (bbg.hold_space)
1098                                         strcat(pattern_space, bbg.hold_space);
1099                                 last_gets_char = '\n';
1100
1101                                 break;
1102                         }
1103                         case 'h':       /* Replace hold space with pattern space */
1104                                 free(bbg.hold_space);
1105                                 bbg.hold_space = xstrdup(pattern_space);
1106                                 break;
1107                         case 'H':       /* Append newline and pattern space to hold space */
1108                         {
1109                                 int hold_space_size = 2;
1110                                 int pattern_space_size = 0;
1111
1112                                 if (bbg.hold_space)
1113                                         hold_space_size += strlen(bbg.hold_space);
1114                                 if (pattern_space)
1115                                         pattern_space_size = strlen(pattern_space);
1116                                 bbg.hold_space = xrealloc(bbg.hold_space,
1117                                                 hold_space_size + pattern_space_size);
1118
1119                                 if (hold_space_size == 2)
1120                                         *bbg.hold_space = 0;
1121                                 strcat(bbg.hold_space, "\n");
1122                                 if (pattern_space)
1123                                         strcat(bbg.hold_space, pattern_space);
1124
1125                                 break;
1126                         }
1127                         case 'x': /* Exchange hold and pattern space */
1128                         {
1129                                 char *tmp = pattern_space;
1130                                 pattern_space = bbg.hold_space ? : xzalloc(1);
1131                                 last_gets_char = '\n';
1132                                 bbg.hold_space = tmp;
1133                                 break;
1134                         }
1135                         }
1136                 }
1137         }
1138
1139         /*
1140          * exit point from sedding...
1141          */
1142  discard_commands:
1143         /* we will print the line unless we were told to be quiet ('-n')
1144            or if the line was suppressed (ala 'd'elete) */
1145         if (!bbg.be_quiet)
1146                 sed_puts(pattern_space, last_gets_char);
1147
1148         /* Delete and such jump here. */
1149  discard_line:
1150         flush_append();
1151         free(pattern_space);
1152
1153         goto again;
1154 }
1155
1156 /* It is possible to have a command line argument with embedded
1157    newlines.  This counts as multiple command lines. */
1158
1159 static void add_cmd_block(char *cmdstr)
1160 {
1161         int go = 1;
1162         char *temp = xstrdup(cmdstr), *temp2 = temp;
1163
1164         while (go) {
1165                 int len = strcspn(temp2, "\n");
1166                 if (!temp2[len]) go = 0;
1167                 else temp2[len] = 0;
1168                 add_cmd(temp2);
1169                 temp2 += len+1;
1170         }
1171         free(temp);
1172 }
1173
1174 static void add_cmds_link(llist_t *opt_e)
1175 {
1176         if (!opt_e) return;
1177         add_cmds_link(opt_e->link);
1178         add_cmd_block(opt_e->data);
1179         free(opt_e);
1180 }
1181
1182 static void add_files_link(llist_t *opt_f)
1183 {
1184         char *line;
1185         FILE *cmdfile;
1186         if (!opt_f) return;
1187         add_files_link(opt_f->link);
1188         cmdfile = xfopen(opt_f->data, "r");
1189         while ((line = xmalloc_getline(cmdfile)) != NULL) {
1190                 add_cmd(line);
1191                 free(line);
1192         }
1193         xprint_and_close_file(cmdfile);
1194         free(opt_f);
1195 }
1196
1197 int sed_main(int argc, char **argv)
1198 {
1199         enum {
1200                 OPT_in_place = 1 << 0,
1201         };
1202         unsigned opt;
1203         llist_t *opt_e, *opt_f;
1204         int status = EXIT_SUCCESS;
1205
1206         bbg.sed_cmd_tail = &bbg.sed_cmd_head;
1207
1208         /* destroy command strings on exit */
1209         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1210
1211         /* Lie to autoconf when it starts asking stupid questions. */
1212         if (argc == 2 && !strcmp(argv[1], "--version")) {
1213                 puts("This is not GNU sed version 4.0");
1214                 return 0;
1215         }
1216
1217         /* do normal option parsing */
1218         opt_e = opt_f = NULL;
1219         opt_complementary = "e::f::" /* can occur multiple times */
1220                             "nn"; /* count -n */
1221         opt = getopt32(argc, argv, "irne:f:", &opt_e, &opt_f,
1222                             &bbg.be_quiet); /* counter for -n */
1223         argc -= optind;
1224         argv += optind;
1225         if (opt & OPT_in_place) { // -i
1226                 atexit(cleanup_outname);
1227         }
1228         if (opt & 0x2) bbg.regex_type |= REG_EXTENDED; // -r
1229         //if (opt & 0x4) bbg.be_quiet++; // -n
1230         if (opt & 0x8) { // -e
1231                 /* getopt32 reverses order of arguments, handle it */
1232                 add_cmds_link(opt_e);
1233         }
1234         if (opt & 0x10) { // -f
1235                 /* getopt32 reverses order of arguments, handle it */
1236                 add_files_link(opt_f);
1237         }
1238         /* if we didn't get a pattern from -e or -f, use argv[0] */
1239         if (!(opt & 0x18)) {
1240                 if (!argc)
1241                         bb_show_usage();
1242                 add_cmd_block(*argv++);
1243                 argc--;
1244         }
1245         /* Flush any unfinished commands. */
1246         add_cmd("");
1247
1248         /* By default, we write to stdout */
1249         bbg.nonstdout = stdout;
1250
1251         /* argv[0..(argc-1)] should be names of file to process. If no
1252          * files were specified or '-' was specified, take input from stdin.
1253          * Otherwise, we process all the files specified. */
1254         if (argv[0] == NULL) {
1255                 if (opt & OPT_in_place)
1256                         bb_error_msg_and_die(bb_msg_requires_arg, "-i");
1257                 add_input_file(stdin);
1258                 process_files();
1259         } else {
1260                 int i;
1261                 FILE *file;
1262
1263                 for (i = 0; i < argc; i++) {
1264                         struct stat statbuf;
1265                         int nonstdoutfd;
1266
1267                         if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) {
1268                                 add_input_file(stdin);
1269                                 process_files();
1270                                 continue;
1271                         }
1272                         file = fopen_or_warn(argv[i], "r");
1273                         if (!file) {
1274                                 status = EXIT_FAILURE;
1275                                 continue;
1276                         }
1277                         if (!(opt & OPT_in_place)) {
1278                                 add_input_file(file);
1279                                 continue;
1280                         }
1281
1282                         bbg.outname = xasprintf("%sXXXXXX", argv[i]);
1283                         nonstdoutfd = mkstemp(bbg.outname);
1284                         if (-1 == nonstdoutfd)
1285                                 bb_error_msg_and_die("no temp file");
1286                         bbg.nonstdout = fdopen(nonstdoutfd, "w");
1287
1288                         /* Set permissions of output file */
1289
1290                         fstat(fileno(file), &statbuf);
1291                         fchmod(nonstdoutfd, statbuf.st_mode);
1292                         add_input_file(file);
1293                         process_files();
1294                         fclose(bbg.nonstdout);
1295
1296                         bbg.nonstdout = stdout;
1297                         /* unlink(argv[i]); */
1298                         // FIXME: error check / message?
1299                         rename(bbg.outname, argv[i]);
1300                         free(bbg.outname);
1301                         bbg.outname = 0;
1302                 }
1303                 if (bbg.input_file_count > bbg.current_input_file)
1304                         process_files();
1305         }
1306
1307         return status;
1308 }