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 GPLv2 or later, 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/|$]command
  42          - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
  43          - edit commands: (a)ppend, (i)nsert, (c)hange
  44          - file commands: (r)ead
  45          - backreferences in substitution expressions (\0, \1, \2...\9)
  46          - grouped commands: {cmd1;cmd2}
  47          - transliteration (y/source-chars/dest-chars/)
  48          - pattern space hold space storing / swapping (g, h, x)
  49          - labels / branching (: label, b, t, T)
  50
  51          (Note: Specifying an address (range) to match is *optional*; commands
  52          default to the whole pattern space if no specific address match was
  53          requested.)
  54
  55         Todo:
  56          - Create a wrapper around regex to make libc's regex conform with sed
  57
  58         Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html
  59 */
  60
  61 #include <stdio.h>
  62 #include <unistd.h>             /* for getopt() */
  63 #include <errno.h>
  64 #include <ctype.h>              /* for isspace() */
  65 #include <stdlib.h>
  66 #include <string.h>
  67 #include "busybox.h"
  68 #include "xregex.h"
  69
  70 /* Each sed command turns into one of these structures. */
  71 typedef struct sed_cmd_s {
  72     /* Ordered by alignment requirements: currently 36 bytes on x86 */
  73
  74     /* address storage */
  75     regex_t *beg_match; /* sed -e '/match/cmd' */
  76     regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
  77     regex_t *sub_match; /* For 's/sub_match/string/' */
  78     int beg_line;               /* 'sed 1p'   0 == apply commands to all lines */
  79     int end_line;               /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */
  80
  81     FILE *file;                 /* File (sw) command writes to, -1 for none. */
  82     char *string;               /* Data string for (saicytb) commands. */
  83
  84     unsigned short which_match;         /* (s) Which match to replace (0 for all) */
  85
  86     /* Bitfields (gcc won't group them if we don't) */
  87     unsigned int invert:1;                      /* the '!' after the address */
  88     unsigned int in_match:1;            /* Next line also included in match? */
  89     unsigned int no_newline:1;          /* Last line written by (sw) had no '\n' */
  90     unsigned int sub_p:1;                       /* (s) print option */
  91
  92     /* GENERAL FIELDS */
  93     char cmd;                           /* The command char: abcdDgGhHilnNpPqrstwxy:={} */
  94     struct sed_cmd_s *next;     /* Next command (linked list, NULL terminated) */
  95 } sed_cmd_t;
  96
  97 static const char bad_format_in_subst[] =
  98         "bad format in substitution expression";
  99 static const char *const semicolon_whitespace = "; \n\r\t\v";
 100
 101 struct sed_globals
 102 {
 103         /* options */
 104         int be_quiet, in_place, regex_type;
 105         FILE *nonstdout;
 106         char *outname, *hold_space;
 107
 108         /* List of input files */
 109         int input_file_count,current_input_file;
 110         FILE **input_file_list;
 111
 112         regmatch_t regmatch[10];
 113         regex_t *previous_regex_ptr;
 114
 115         /* linked list of sed commands */
 116         sed_cmd_t sed_cmd_head, *sed_cmd_tail;
 117
 118         /* Linked list of append lines */
 119         llist_t *append_head;
 120
 121         char *add_cmd_line;
 122
 123         struct pipeline {
 124                 char *buf;      /* Space to hold string */
 125                 int idx;        /* Space used */
 126                 int len;        /* Space allocated */
 127         } pipeline;
 128 } bbg;
 129
 130
 131 void sed_free_and_close_stuff(void);
 132 #if ENABLE_FEATURE_CLEAN_UP
 133 void sed_free_and_close_stuff(void)
 134 {
 135         sed_cmd_t *sed_cmd = bbg.sed_cmd_head.next;
 136
 137         llist_free(bbg.append_head, free);
 138
 139         while (sed_cmd) {
 140                 sed_cmd_t *sed_cmd_next = sed_cmd->next;
 141
 142                 if(sed_cmd->file)
 143                         bb_xprint_and_close_file(sed_cmd->file);
 144
 145                 if (sed_cmd->beg_match) {
 146                         regfree(sed_cmd->beg_match);
 147                         free(sed_cmd->beg_match);
 148                 }
 149                 if (sed_cmd->end_match) {
 150                         regfree(sed_cmd->end_match);
 151                         free(sed_cmd->end_match);
 152                 }
 153                 if (sed_cmd->sub_match) {
 154                         regfree(sed_cmd->sub_match);
 155                         free(sed_cmd->sub_match);
 156                 }
 157                 free(sed_cmd->string);
 158                 free(sed_cmd);
 159                 sed_cmd = sed_cmd_next;
 160         }
 161
 162         if(bbg.hold_space) free(bbg.hold_space);
 163
 164     while(bbg.current_input_file<bbg.input_file_count)
 165                 fclose(bbg.input_file_list[bbg.current_input_file++]);
 166 }
 167 #endif
 168
 169 /* If something bad happens during -i operation, delete temp file */
 170
 171 static void cleanup_outname(void)
 172 {
 173   if(bbg.outname) unlink(bbg.outname);
 174 }
 175
 176 /* strdup, replacing "\n" with '\n', and "\delimiter" with 'delimiter' */
 177
 178 static void parse_escapes(char *dest, const char *string, int len, char from, char to)
 179 {
 180         int i=0;
 181
 182         while(i<len) {
 183                 if(string[i] == '\\') {
 184                         if(!to || string[i+1] == from) {
 185                                 *(dest++) = to ? to : string[i+1];
 186                                 i+=2;
 187                                 continue;
 188                         } else *(dest++)=string[i++];
 189                 }
 190                 *(dest++) = string[i++];
 191         }
 192         *dest=0;
 193 }
 194
 195 static char *copy_parsing_escapes(const char *string, int len)
 196 {
 197         char *dest=xmalloc(len+1);
 198
 199         parse_escapes(dest,string,len,'n','\n');
 200         return dest;
 201 }
 202
 203
 204 /*
 205  * index_of_next_unescaped_regexp_delim - walks left to right through a string
 206  * beginning at a specified index and returns the index of the next regular
 207  * expression delimiter (typically a forward * slash ('/')) not preceded by
 208  * a backslash ('\').
 209  */
 210 static int index_of_next_unescaped_regexp_delim(const char delimiter,
 211         const char *str)
 212 {
 213         int bracket = -1;
 214         int escaped = 0;
 215         int idx = 0;
 216         char ch;
 217
 218         for (; (ch = str[idx]); idx++) {
 219                 if (bracket != -1) {
 220                         if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2
 221                                         && str[idx - 1] == '^')))
 222                                 bracket = -1;
 223                 } else if (escaped)
 224                         escaped = 0;
 225                 else if (ch == '\\')
 226                         escaped = 1;
 227                 else if (ch == '[')
 228                         bracket = idx;
 229                 else if (ch == delimiter)
 230                         return idx;
 231         }
 232
 233         /* if we make it to here, we've hit the end of the string */
 234         return -1;
 235 }
 236
 237 /*
 238  *  Returns the index of the third delimiter
 239  */
 240 static int parse_regex_delim(const char *cmdstr, char **match, char **replace)
 241 {
 242         const char *cmdstr_ptr = cmdstr;
 243         char delimiter;
 244         int idx = 0;
 245
 246         /* verify that the 's' or 'y' is followed by something.  That something
 247          * (typically a 'slash') is now our regexp delimiter... */
 248         if (*cmdstr == '\0') bb_error_msg_and_die(bad_format_in_subst);
 249         delimiter = *(cmdstr_ptr++);
 250
 251         /* save the match string */
 252         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 253         if (idx == -1) {
 254                 bb_error_msg_and_die(bad_format_in_subst);
 255         }
 256         *match = copy_parsing_escapes(cmdstr_ptr, idx);
 257
 258         /* save the replacement string */
 259         cmdstr_ptr += idx + 1;
 260         idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr);
 261         if (idx == -1) {
 262                 bb_error_msg_and_die(bad_format_in_subst);
 263         }
 264         *replace = copy_parsing_escapes(cmdstr_ptr, idx);
 265
 266         return ((cmdstr_ptr - cmdstr) + idx);
 267 }
 268
 269 /*
 270  * returns the index in the string just past where the address ends.
 271  */
 272 static int get_address(char *my_str, int *linenum, regex_t ** regex)
 273 {
 274         char *pos = my_str;
 275
 276         if (isdigit(*my_str)) {
 277                 *linenum = strtol(my_str, &pos, 10);
 278                 /* endstr shouldnt ever equal NULL */
 279         } else if (*my_str == '$') {
 280                 *linenum = -1;
 281                 pos++;
 282         } else if (*my_str == '/' || *my_str == '\\') {
 283                 int next;
 284                 char delimiter;
 285                 char *temp;
 286
 287                 if (*my_str == '\\') delimiter = *(++pos);
 288                 else delimiter = '/';
 289                 next = index_of_next_unescaped_regexp_delim(delimiter, ++pos);
 290                 if (next == -1)
 291                         bb_error_msg_and_die("unterminated match expression");
 292
 293                 temp=copy_parsing_escapes(pos,next);
 294                 *regex = (regex_t *) xmalloc(sizeof(regex_t));
 295                 xregcomp(*regex, temp, bbg.regex_type|REG_NEWLINE);
 296                 free(temp);
 297                 /* Move position to next character after last delimiter */
 298                 pos+=(next+1);
 299         }
 300         return pos - my_str;
 301 }
 302
 303 /* Grab a filename.  Whitespace at start is skipped, then goes to EOL. */
 304 static int parse_file_cmd(sed_cmd_t *sed_cmd, const char *filecmdstr, char **retval)
 305 {
 306         int start = 0, idx, hack=0;
 307
 308         /* Skip whitespace, then grab filename to end of line */
 309         while (isspace(filecmdstr[start])) start++;
 310         idx=start;
 311         while(filecmdstr[idx] && filecmdstr[idx]!='\n') idx++;
 312         /* If lines glued together, put backslash back. */
 313         if(filecmdstr[idx]=='\n') hack=1;
 314         if(idx==start) bb_error_msg_and_die("Empty filename");
 315         *retval = bb_xstrndup(filecmdstr+start, idx-start+hack+1);
 316         if(hack) *(idx+*retval)='\\';
 317
 318         return idx;
 319 }
 320
 321 static int parse_subst_cmd(sed_cmd_t *const sed_cmd, char *substr)
 322 {
 323         int cflags = bbg.regex_type;
 324         char *match;
 325         int idx = 0;
 326
 327         /*
 328          * A substitution command should look something like this:
 329          *    s/match/replace/ #gIpw
 330          *    ||     |        |||
 331          *    mandatory       optional
 332          */
 333         idx = parse_regex_delim(substr, &match, &sed_cmd->string);
 334
 335         /* determine the number of back references in the match string */
 336         /* Note: we compute this here rather than in the do_subst_command()
 337          * function to save processor time, at the expense of a little more memory
 338          * (4 bits) per sed_cmd */
 339
 340         /* process the flags */
 341
 342         sed_cmd->which_match=1;
 343         while (substr[++idx]) {
 344                 /* Parse match number */
 345                 if(isdigit(substr[idx])) {
 346                         if(match[0]!='^') {
 347                                 /* Match 0 treated as all, multiple matches we take the last one. */
 348                                 char *pos=substr+idx;
 349                                 sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
 350                                 idx=pos-substr;
 351                         }
 352                         continue;
 353                 }
 354                 /* Skip spaces */
 355                 if(isspace(substr[idx])) continue;
 356
 357                 switch (substr[idx]) {
 358                         /* Replace all occurrences */
 359                         case 'g':
 360                                 if (match[0] != '^') sed_cmd->which_match = 0;
 361                                 break;
 362                         /* Print pattern space */
 363                         case 'p':
 364                                 sed_cmd->sub_p = 1;
 365                                 break;
 366                         /* Write to file */
 367                         case 'w':
 368                         {
 369                                 char *temp;
 370                                 idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
 371
 372                                 break;
 373                         }
 374                         /* Ignore case (gnu exension) */
 375                         case 'I':
 376                                 cflags |= REG_ICASE;
 377                                 break;
 378                         /* Comment */
 379                         case '#':
 380                                 while(substr[++idx]);
 381                                 /* Fall through */
 382                         /* End of command */
 383                         case ';':
 384                         case '}':
 385                                 goto out;
 386                         default:
 387                                 bb_error_msg_and_die("bad option in substitution expression");
 388                 }
 389         }
 390 out:
 391         /* compile the match string into a regex */
 392         if (*match != '\0') {
 393                 /* If match is empty, we use last regex used at runtime */
 394                 sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
 395                 xregcomp(sed_cmd->sub_match, match, cflags);
 396         }
 397         free(match);
 398
 399         return idx;
 400 }
 401
 402 /*
 403  *  Process the commands arguments
 404  */
 405 static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
 406 {
 407         /* handle (s)ubstitution command */
 408         if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
 409         /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
 410         else if (strchr("aic", sed_cmd->cmd)) {
 411                 if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
 412                         bb_error_msg_and_die
 413                                 ("only a beginning address can be specified for edit commands");
 414                 for(;;) {
 415                         if(*cmdstr=='\n' || *cmdstr=='\\') {
 416                                 cmdstr++;
 417                                 break;
 418                         } else if(isspace(*cmdstr)) cmdstr++;
 419                         else break;
 420                 }
 421                 sed_cmd->string = bb_xstrdup(cmdstr);
 422                 parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
 423                 cmdstr += strlen(cmdstr);
 424         /* handle file cmds: (r)ead */
 425         } else if(strchr("rw", sed_cmd->cmd)) {
 426                 if (sed_cmd->end_line || sed_cmd->end_match)
 427                         bb_error_msg_and_die("Command only uses one address");
 428                 cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
 429                 if(sed_cmd->cmd=='w')
 430                         sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
 431         /* handle branch commands */
 432         } else if (strchr(":btT", sed_cmd->cmd)) {
 433                 int length;
 434
 435                 while(isspace(*cmdstr)) cmdstr++;
 436                 length = strcspn(cmdstr, semicolon_whitespace);
 437                 if (length) {
 438                         sed_cmd->string = bb_xstrndup(cmdstr, length);
 439                         cmdstr += length;
 440                 }
 441         }
 442         /* translation command */
 443         else if (sed_cmd->cmd == 'y') {
 444                 char *match, *replace;
 445                 int i=cmdstr[0];
 446
 447                 cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
 448                 /* \n already parsed, but \delimiter needs unescaping. */
 449                 parse_escapes(match,match,strlen(match),i,i);
 450                 parse_escapes(replace,replace,strlen(replace),i,i);
 451
 452                 sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
 453                 for (i = 0; match[i] && replace[i]; i++) {
 454                         sed_cmd->string[i * 2] = match[i];
 455                         sed_cmd->string[(i * 2) + 1] = replace[i];
 456                 }
 457                 free(match);
 458                 free(replace);
 459         }
 460         /* if it wasnt a single-letter command that takes no arguments
 461          * then it must be an invalid command.
 462          */
 463         else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
 464                 bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
 465         }
 466
 467         /* give back whatever's left over */
 468         return (cmdstr);
 469 }
 470
 471
 472 /* Parse address+command sets, skipping comment lines. */
 473
 474 static void add_cmd(char *cmdstr)
 475 {
 476         sed_cmd_t *sed_cmd;
 477         int temp;
 478
 479         /* Append this line to any unfinished line from last time. */
 480         if (bbg.add_cmd_line) {
 481                 cmdstr = bb_xasprintf("%s\n%s", bbg.add_cmd_line, cmdstr);
 482                 free(bbg.add_cmd_line);
 483                 bbg.add_cmd_line = cmdstr;
 484         } else bbg.add_cmd_line=NULL;
 485
 486         /* If this line ends with backslash, request next line. */
 487         temp=strlen(cmdstr);
 488         if(temp && cmdstr[temp-1]=='\\') {
 489                 if (!bbg.add_cmd_line) bbg.add_cmd_line = bb_xstrdup(cmdstr);
 490                 bbg.add_cmd_line[temp-1] = 0;
 491                 return;
 492         }
 493
 494         /* Loop parsing all commands in this line. */
 495         while(*cmdstr) {
 496                 /* Skip leading whitespace and semicolons */
 497                 cmdstr += strspn(cmdstr, semicolon_whitespace);
 498
 499                 /* If no more commands, exit. */
 500                 if(!*cmdstr) break;
 501
 502                 /* if this is a comment, jump past it and keep going */
 503                 if (*cmdstr == '#') {
 504                         /* "#n" is the same as using -n on the command line */
 505                         if (cmdstr[1] == 'n') bbg.be_quiet++;
 506                         if(!(cmdstr=strpbrk(cmdstr, "\n\r"))) break;
 507                         continue;
 508                 }
 509
 510                 /* parse the command
 511                  * format is: [addr][,addr][!]cmd
 512                  *            |----||-----||-|
 513                  *            part1 part2  part3
 514                  */
 515
 516                 sed_cmd = xzalloc(sizeof(sed_cmd_t));
 517
 518                 /* first part (if present) is an address: either a '$', a number or a /regex/ */
 519                 cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
 520
 521                 /* second part (if present) will begin with a comma */
 522                 if (*cmdstr == ',') {
 523                         int idx;
 524
 525                         cmdstr++;
 526                         idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match);
 527                         if (!idx) bb_error_msg_and_die("no address after comma\n");
 528                         cmdstr += idx;
 529                 }
 530
 531                 /* skip whitespace before the command */
 532                 while (isspace(*cmdstr)) cmdstr++;
 533
 534                 /* Check for inversion flag */
 535                 if (*cmdstr == '!') {
 536                         sed_cmd->invert = 1;
 537                         cmdstr++;
 538
 539                         /* skip whitespace before the command */
 540                         while (isspace(*cmdstr)) cmdstr++;
 541                 }
 542
 543                 /* last part (mandatory) will be a command */
 544                 if (!*cmdstr) bb_error_msg_and_die("missing command");
 545                 sed_cmd->cmd = *(cmdstr++);
 546                 cmdstr = parse_cmd_args(sed_cmd, cmdstr);
 547
 548                 /* Add the command to the command array */
 549                 bbg.sed_cmd_tail->next = sed_cmd;
 550                 bbg.sed_cmd_tail = bbg.sed_cmd_tail->next;
 551         }
 552
 553         /* If we glued multiple lines together, free the memory. */
 554         free(bbg.add_cmd_line);
 555         bbg.add_cmd_line = NULL;
 556 }
 557
 558 /* Append to a string, reallocating memory as necessary. */
 559
 560 #define PIPE_GROW 64
 561
 562 static void pipe_putc(char c)
 563 {
 564         if(bbg.pipeline.idx==bbg.pipeline.len) {
 565                 bbg.pipeline.buf = xrealloc(bbg.pipeline.buf,
 566                                                                 bbg.pipeline.len + PIPE_GROW);
 567                 bbg.pipeline.len+=PIPE_GROW;
 568         }
 569         bbg.pipeline.buf[bbg.pipeline.idx++] = c;
 570 }
 571
 572 static void do_subst_w_backrefs(const char *line, const char *replace)
 573 {
 574         int i,j;
 575
 576         /* go through the replacement string */
 577         for (i = 0; replace[i]; i++) {
 578                 /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
 579                 if (replace[i] == '\\' && replace[i+1]>='0' && replace[i+1]<='9') {
 580                         int backref=replace[++i]-'0';
 581
 582                         /* print out the text held in bbg.regmatch[backref] */
 583                         if(bbg.regmatch[backref].rm_so != -1)
 584                                 for (j = bbg.regmatch[backref].rm_so;
 585                                         j < bbg.regmatch[backref].rm_eo; j++) pipe_putc(line[j]);
 586                 }
 587
 588                 /* if we find a backslash escaped character, print the character */
 589                 else if (replace[i] == '\\') pipe_putc(replace[++i]);
 590
 591                 /* if we find an unescaped '&' print out the whole matched text. */
 592                 else if (replace[i] == '&')
 593                         for (j = bbg.regmatch[0].rm_so; j < bbg.regmatch[0].rm_eo; j++)
 594                                 pipe_putc(line[j]);
 595                 /* Otherwise just output the character. */
 596                 else pipe_putc(replace[i]);
 597         }
 598 }
 599
 600 static int do_subst_command(sed_cmd_t *sed_cmd, char **line)
 601 {
 602         char *oldline = *line;
 603         int altered = 0;
 604         int match_count=0;
 605         regex_t *current_regex;
 606
 607         /* Handle empty regex. */
 608         if (sed_cmd->sub_match == NULL) {
 609                 current_regex = bbg.previous_regex_ptr;
 610                 if(!current_regex)
 611                         bb_error_msg_and_die("No previous regexp.");
 612         } else bbg.previous_regex_ptr = current_regex = sed_cmd->sub_match;
 613
 614         /* Find the first match */
 615         if(REG_NOMATCH==regexec(current_regex, oldline, 10, bbg.regmatch, 0))
 616                 return 0;
 617
 618         /* Initialize temporary output buffer. */
 619         bbg.pipeline.buf=xmalloc(PIPE_GROW);
 620         bbg.pipeline.len=PIPE_GROW;
 621         bbg.pipeline.idx=0;
 622
 623         /* Now loop through, substituting for matches */
 624         do {
 625                 int i;
 626
 627                 /* Work around bug in glibc regexec, demonstrated by:
 628                    echo " a.b" | busybox sed 's [^ .]* x g'
 629                    The match_count check is so not to break
 630                    echo "hi" | busybox sed 's/^/!/g' */
 631                 if(!bbg.regmatch[0].rm_so && !bbg.regmatch[0].rm_eo && match_count) {
 632                         pipe_putc(*(oldline++));
 633                         continue;
 634                 }
 635
 636                 match_count++;
 637
 638                 /* If we aren't interested in this match, output old line to
 639                    end of match and continue */
 640                 if(sed_cmd->which_match && sed_cmd->which_match!=match_count) {
 641                         for(i=0;i<bbg.regmatch[0].rm_eo;i++)
 642                                 pipe_putc(oldline[i]);
 643                         continue;
 644                 }
 645
 646                 /* print everything before the match */
 647                 for (i = 0; i < bbg.regmatch[0].rm_so; i++) pipe_putc(oldline[i]);
 648
 649                 /* then print the substitution string */
 650                 do_subst_w_backrefs(oldline, sed_cmd->string);
 651
 652                 /* advance past the match */
 653                 oldline += bbg.regmatch[0].rm_eo;
 654                 /* flag that something has changed */
 655                 altered++;
 656
 657                 /* if we're not doing this globally, get out now */
 658                 if (sed_cmd->which_match) break;
 659         } while (*oldline && (regexec(current_regex, oldline, 10, bbg.regmatch, 0) != REG_NOMATCH));
 660
 661         /* Copy rest of string into output pipeline */
 662
 663         while(*oldline) pipe_putc(*(oldline++));
 664         pipe_putc(0);
 665
 666         free(*line);
 667         *line = bbg.pipeline.buf;
 668         return altered;
 669 }
 670
 671 /* Set command pointer to point to this label.  (Does not handle null label.) */
 672 static sed_cmd_t *branch_to(const char *label)
 673 {
 674         sed_cmd_t *sed_cmd;
 675
 676         for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) {
 677                 if ((sed_cmd->cmd == ':') && (sed_cmd->string) && (strcmp(sed_cmd->string, label) == 0)) {
 678                         return (sed_cmd);
 679                 }
 680         }
 681         bb_error_msg_and_die("Can't find label for jump to `%s'", label);
 682 }
 683
 684 static void append(char *s)
 685 {
 686         bbg.append_head = llist_add_to_end(bbg.append_head, bb_xstrdup(s));
 687 }
 688
 689 static void flush_append(void)
 690 {
 691         char *data;
 692
 693         /* Output appended lines. */
 694         while((data = (char *)llist_pop(&bbg.append_head))) {
 695                 fprintf(bbg.nonstdout,"%s\n",data);
 696                 free(data);
 697         }
 698 }
 699
 700 static void add_input_file(FILE *file)
 701 {
 702         bbg.input_file_list=xrealloc(bbg.input_file_list,
 703                         (bbg.input_file_count + 1) * sizeof(FILE *));
 704         bbg.input_file_list[bbg.input_file_count++] = file;
 705 }
 706
 707 /* Get next line of input from bbg.input_file_list, flushing append buffer and
 708  * noting if we ran out of files without a newline on the last line we read.
 709  */
 710 static char *get_next_line(int *no_newline)
 711 {
 712         char *temp=NULL;
 713         int len;
 714
 715         flush_append();
 716         while (bbg.current_input_file<bbg.input_file_count) {
 717                 temp = bb_get_chunk_from_file(bbg.input_file_list[bbg.current_input_file],&len);
 718                 if (temp) {
 719                         *no_newline = !(len && temp[len-1]=='\n');
 720                         if (!*no_newline) temp[len-1] = 0;
 721                         break;
 722                 // Close this file and advance to next one
 723                 } else fclose(bbg.input_file_list[bbg.current_input_file++]);
 724         }
 725
 726         return temp;
 727 }
 728
 729 /* Output line of text.  missing_newline means the last line output did not
 730    end with a newline.  no_newline means this line does not end with a
 731    newline. */
 732
 733 static int puts_maybe_newline(char *s, FILE *file, int missing_newline, int no_newline)
 734 {
 735         if(missing_newline) fputc('\n',file);
 736         fputs(s,file);
 737         if(!no_newline) fputc('\n',file);
 738
 739     if(ferror(file)) {
 740                 fprintf(stderr,"Write failed.\n");
 741                 exit(4);  /* It's what gnu sed exits with... */
 742         }
 743
 744         return no_newline;
 745 }
 746
 747 #define sed_puts(s,n) missing_newline=puts_maybe_newline(s,bbg.nonstdout,missing_newline,n)
 748
 749 /* Process all the lines in all the files */
 750
 751 static void process_files(void)
 752 {
 753         char *pattern_space, *next_line;
 754         int linenum = 0, missing_newline=0;
 755         int no_newline,next_no_newline=0;
 756
 757         /* Prime the pump */
 758         next_line = get_next_line(&next_no_newline);
 759
 760         /* go through every line in each file */
 761         for(;;) {
 762                 sed_cmd_t *sed_cmd;
 763                 int substituted=0;
 764
 765                 /* Advance to next line.  Stop if out of lines. */
 766                 if(!(pattern_space=next_line)) break;
 767                 no_newline=next_no_newline;
 768
 769                 /* Read one line in advance so we can act on the last line,
 770                  * the '$' address */
 771                 next_line = get_next_line(&next_no_newline);
 772                 linenum++;
 773 restart:
 774                 /* for every line, go through all the commands */
 775                 for (sed_cmd = bbg.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next)
 776                 {
 777                         int old_matched, matched;
 778
 779                         old_matched = sed_cmd->in_match;
 780
 781                         /* Determine if this command matches this line: */
 782
 783                         /* Are we continuing a previous multi-line match? */
 784
 785                         sed_cmd->in_match = sed_cmd->in_match
 786
 787                         /* Or is no range necessary? */
 788                                 || (!sed_cmd->beg_line && !sed_cmd->end_line
 789                                         && !sed_cmd->beg_match && !sed_cmd->end_match)
 790
 791                         /* Or did we match the start of a numerical range? */
 792                                 || (sed_cmd->beg_line > 0 && (sed_cmd->beg_line == linenum))
 793
 794                         /* Or does this line match our begin address regex? */
 795                                 || (sed_cmd->beg_match &&
 796                                     !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0))
 797
 798                         /* Or did we match last line of input? */
 799                                 || (sed_cmd->beg_line == -1 && next_line == NULL);
 800
 801                         /* Snapshot the value */
 802
 803                         matched = sed_cmd->in_match;
 804
 805                         /* Is this line the end of the current match? */
 806
 807                         if(matched) {
 808                                 sed_cmd->in_match = !(
 809                                         /* has the ending line come, or is this a single address command? */
 810                                         (sed_cmd->end_line ?
 811                                                 sed_cmd->end_line==-1 ?
 812                                                         !next_line
 813                                                         : sed_cmd->end_line<=linenum
 814                                                 : !sed_cmd->end_match)
 815                                         /* or does this line matches our last address regex */
 816                                         || (sed_cmd->end_match && old_matched && (regexec(sed_cmd->end_match, pattern_space, 0, NULL, 0) == 0))
 817                                 );
 818                         }
 819
 820                         /* Skip blocks of commands we didn't match. */
 821                         if (sed_cmd->cmd == '{') {
 822                                 if(sed_cmd->invert ? matched : !matched)
 823                                         while(sed_cmd && sed_cmd->cmd!='}') sed_cmd=sed_cmd->next;
 824                                 if(!sed_cmd) bb_error_msg_and_die("Unterminated {");
 825                                 continue;
 826                         }
 827
 828                         /* Okay, so did this line match? */
 829                         if (sed_cmd->invert ? !matched : matched) {
 830                                 /* Update last used regex in case a blank substitute BRE is found */
 831                                 if (sed_cmd->beg_match) {
 832                                         bbg.previous_regex_ptr = sed_cmd->beg_match;
 833                                 }
 834
 835                                 /* actual sedding */
 836                                 switch (sed_cmd->cmd) {
 837
 838                                         /* Print line number */
 839                                         case '=':
 840                                                 fprintf(bbg.nonstdout,"%d\n", linenum);
 841                                                 break;
 842
 843                                         /* Write the current pattern space up to the first newline */
 844                                         case 'P':
 845                                         {
 846                                                 char *tmp = strchr(pattern_space, '\n');
 847
 848                                                 if (tmp) {
 849                                                         *tmp = '\0';
 850                                                         sed_puts(pattern_space,1);
 851                                                         *tmp = '\n';
 852                                                         break;
 853                                                 }
 854                                                 /* Fall Through */
 855                                         }
 856
 857                                         /* Write the current pattern space to output */
 858                                         case 'p':
 859                                                 sed_puts(pattern_space,no_newline);
 860                                                 break;
 861                                         /* Delete up through first newline */
 862                                         case 'D':
 863                                         {
 864                                                 char *tmp = strchr(pattern_space,'\n');
 865
 866                                                 if(tmp) {
 867                                                         tmp=bb_xstrdup(tmp+1);
 868                                                         free(pattern_space);
 869                                                         pattern_space=tmp;
 870                                                         goto restart;
 871                                                 }
 872                                         }
 873                                         /* discard this line. */
 874                                         case 'd':
 875                                                 goto discard_line;
 876
 877                                         /* Substitute with regex */
 878                                         case 's':
 879                                                 if(do_subst_command(sed_cmd, &pattern_space)) {
 880                                                         substituted|=1;
 881
 882                                                         /* handle p option */
 883                                                         if(sed_cmd->sub_p)
 884                                                                 sed_puts(pattern_space,no_newline);
 885                                                         /* handle w option */
 886                                                         if(sed_cmd->file)
 887                                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space, sed_cmd->file, sed_cmd->no_newline, no_newline);
 888
 889                                                 }
 890                                                 break;
 891
 892                                         /* Append line to linked list to be printed later */
 893                                         case 'a':
 894                                         {
 895                                                 append(sed_cmd->string);
 896                                                 break;
 897                                         }
 898
 899                                         /* Insert text before this line */
 900                                         case 'i':
 901                                                 sed_puts(sed_cmd->string,1);
 902                                                 break;
 903
 904                                         /* Cut and paste text (replace) */
 905                                         case 'c':
 906                                                 /* Only triggers on last line of a matching range. */
 907                                                 if (!sed_cmd->in_match) sed_puts(sed_cmd->string,0);
 908                                                 goto discard_line;
 909
 910                                         /* Read file, append contents to output */
 911                                         case 'r':
 912                                         {
 913                                                 FILE *rfile;
 914
 915                                                 rfile = fopen(sed_cmd->string, "r");
 916                                                 if (rfile) {
 917                                                         char *line;
 918
 919                                                         while ((line = bb_get_chomped_line_from_file(rfile))
 920                                                                         != NULL)
 921                                                                 append(line);
 922                                                         bb_xprint_and_close_file(rfile);
 923                                                 }
 924
 925                                                 break;
 926                                         }
 927
 928                                         /* Write pattern space to file. */
 929                                         case 'w':
 930                                                 sed_cmd->no_newline=puts_maybe_newline(pattern_space,sed_cmd->file, sed_cmd->no_newline,no_newline);
 931                                                 break;
 932
 933                                         /* Read next line from input */
 934                                         case 'n':
 935                                                 if (!bbg.be_quiet)
 936                                                         sed_puts(pattern_space,no_newline);
 937                                                 if (next_line) {
 938                                                         free(pattern_space);
 939                                                         pattern_space = next_line;
 940                                                         no_newline=next_no_newline;
 941                                                         next_line = get_next_line(&next_no_newline);
 942                                                         linenum++;
 943                                                         break;
 944                                                 }
 945                                                 /* fall through */
 946
 947                                         /* Quit.  End of script, end of input. */
 948                                         case 'q':
 949                                                 /* Exit the outer while loop */
 950                                                 free(next_line);
 951                                                 next_line = NULL;
 952                                                 goto discard_commands;
 953
 954                                         /* Append the next line to the current line */
 955                                         case 'N':
 956                                         {
 957                                                 /* If no next line, jump to end of script and exit. */
 958                                                 if (next_line == NULL) {
 959                                                         /* Jump to end of script and exit */
 960                                                         free(next_line);
 961                                                         next_line = NULL;
 962                                                         goto discard_line;
 963                                                 /* append next_line, read new next_line. */
 964                                                 } else {
 965                                                         int len=strlen(pattern_space);
 966
 967                                                         pattern_space = realloc(pattern_space, len + strlen(next_line) + 2);
 968                                                         pattern_space[len]='\n';
 969                                                         strcpy(pattern_space+len+1, next_line);
 970                                                         no_newline=next_no_newline;
 971                                                         next_line = get_next_line(&next_no_newline);
 972                                                         linenum++;
 973                                                 }
 974                                                 break;
 975                                         }
 976
 977                                         /* Test/branch if substitution occurred */
 978                                         case 't':
 979                                                 if(!substituted) break;
 980                                                 substituted=0;
 981                                                 /* Fall through */
 982                                         /* Test/branch if substitution didn't occur */
 983                                         case 'T':
 984                                                 if (substituted) break;
 985                                                 /* Fall through */
 986                                         /* Branch to label */
 987                                         case 'b':
 988                                                 if (!sed_cmd->string) goto discard_commands;
 989                                                 else sed_cmd = branch_to(sed_cmd->string);
 990                                                 break;
 991                                         /* Transliterate characters */
 992                                         case 'y':
 993                                         {
 994                                                 int i;
 995
 996                                                 for (i = 0; pattern_space[i]; i++) {
 997                                                         int j;
 998
 999                                                         for (j = 0; sed_cmd->string[j]; j += 2) {
1000                                                                 if (pattern_space[i] == sed_cmd->string[j]) {
1001                                                                         pattern_space[i] = sed_cmd->string[j + 1];
1002                                                                         break;
1003                                                                 }
1004                                                         }
1005                                                 }
1006
1007                                                 break;
1008                                         }
1009                                         case 'g':       /* Replace pattern space with hold space */
1010                                                 free(pattern_space);
1011                                                 pattern_space = bb_xstrdup(bbg.hold_space ? bbg.hold_space : "");
1012                                                 break;
1013                                         case 'G':       /* Append newline and hold space to pattern space */
1014                                         {
1015                                                 int pattern_space_size = 2;
1016                                                 int hold_space_size = 0;
1017
1018                                                 if (pattern_space)
1019                                                         pattern_space_size += strlen(pattern_space);
1020                                                 if (bbg.hold_space)
1021                                                         hold_space_size = strlen(bbg.hold_space);
1022                                                 pattern_space = xrealloc(pattern_space,
1023                                                                 pattern_space_size + hold_space_size);
1024                                                 if (pattern_space_size == 2) pattern_space[0]=0;
1025                                                 strcat(pattern_space, "\n");
1026                                                 if (bbg.hold_space)
1027                                                         strcat(pattern_space, bbg.hold_space);
1028                                                 no_newline=0;
1029
1030                                                 break;
1031                                         }
1032                                         case 'h':       /* Replace hold space with pattern space */
1033                                                 free(bbg.hold_space);
1034                                                 bbg.hold_space = bb_xstrdup(pattern_space);
1035                                                 break;
1036                                         case 'H':       /* Append newline and pattern space to hold space */
1037                                         {
1038                                                 int hold_space_size = 2;
1039                                                 int pattern_space_size = 0;
1040
1041                                                 if (bbg.hold_space)
1042                                                         hold_space_size += strlen(bbg.hold_space);
1043                                                 if (pattern_space)
1044                                                         pattern_space_size = strlen(pattern_space);
1045                                                 bbg.hold_space = xrealloc(bbg.hold_space,
1046                                                                                         hold_space_size + pattern_space_size);
1047
1048                                                 if (hold_space_size == 2) *bbg.hold_space=0;
1049                                                 strcat(bbg.hold_space, "\n");
1050                                                 if (pattern_space) strcat(bbg.hold_space, pattern_space);
1051
1052                                                 break;
1053                                         }
1054                                         case 'x': /* Exchange hold and pattern space */
1055                                         {
1056                                                 char *tmp = pattern_space;
1057                                                 pattern_space = bbg.hold_space;
1058                                                 no_newline=0;
1059                                                 bbg.hold_space = tmp;
1060                                                 break;
1061                                         }
1062                                 }
1063                         }
1064                 }
1065
1066                 /*
1067                  * exit point from sedding...
1068                  */
1069 discard_commands:
1070                 /* we will print the line unless we were told to be quiet ('-n')
1071                    or if the line was suppressed (ala 'd'elete) */
1072                 if (!bbg.be_quiet) sed_puts(pattern_space,no_newline);
1073
1074                 /* Delete and such jump here. */
1075 discard_line:
1076                 flush_append();
1077                 free(pattern_space);
1078         }
1079 }
1080
1081 /* It is possible to have a command line argument with embedded
1082    newlines.  This counts as multiple command lines. */
1083
1084 static void add_cmd_block(char *cmdstr)
1085 {
1086         int go=1;
1087         char *temp=bb_xstrdup(cmdstr),*temp2=temp;
1088
1089         while(go) {
1090                 int len=strcspn(temp2,"\n");
1091                 if(!temp2[len]) go=0;
1092                 else temp2[len]=0;
1093                 add_cmd(temp2);
1094                 temp2+=len+1;
1095         }
1096         free(temp);
1097 }
1098
1099 int sed_main(int argc, char **argv)
1100 {
1101         int status = EXIT_SUCCESS, opt, getpat = 1;
1102
1103         bbg.sed_cmd_tail=&bbg.sed_cmd_head;
1104
1105         /* destroy command strings on exit */
1106         if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff);
1107
1108         /* Lie to autoconf when it starts asking stupid questions. */
1109         if(argc==2 && !strcmp(argv[1],"--version")) {
1110                 printf("This is not GNU sed version 4.0\n");
1111                 exit(0);
1112         }
1113
1114         /* do normal option parsing */
1115         while ((opt = getopt(argc, argv, "irne:f:")) > 0) {
1116                 switch (opt) {
1117                 case 'i':
1118                         bbg.in_place++;
1119                         atexit(cleanup_outname);
1120                         break;
1121                 case 'r':
1122                         bbg.regex_type|=REG_EXTENDED;
1123                         break;
1124                 case 'n':
1125                         bbg.be_quiet++;
1126                         break;
1127                 case 'e':
1128                         add_cmd_block(optarg);
1129                         getpat=0;
1130                         break;
1131                 case 'f':
1132                 {
1133                         FILE *cmdfile;
1134                         char *line;
1135
1136                         cmdfile = bb_xfopen(optarg, "r");
1137
1138                         while ((line = bb_get_chomped_line_from_file(cmdfile)) != NULL) {
1139                                 add_cmd(line);
1140                                 getpat=0;
1141                                 free(line);
1142                         }
1143                         bb_xprint_and_close_file(cmdfile);
1144
1145                         break;
1146                 }
1147                 default:
1148                         bb_show_usage();
1149                 }
1150         }
1151
1152         /* if we didn't get a pattern from -e or -f, use argv[optind] */
1153         if(getpat) {
1154                 if (argv[optind] == NULL)
1155                         bb_show_usage();
1156                 else
1157                         add_cmd_block(argv[optind++]);
1158         }
1159         /* Flush any unfinished commands. */
1160         add_cmd("");
1161
1162         /* By default, we write to stdout */
1163         bbg.nonstdout=stdout;
1164
1165         /* argv[(optind)..(argc-1)] should be names of file to process. If no
1166          * files were specified or '-' was specified, take input from stdin.
1167          * Otherwise, we process all the files specified. */
1168         if (argv[optind] == NULL) {
1169                 if(bbg.in_place) bb_error_msg_and_die("Filename required for -i");
1170                 add_input_file(stdin);
1171                 process_files();
1172         } else {
1173                 int i;
1174                 FILE *file;
1175
1176                 for (i = optind; i < argc; i++) {
1177                         if(!strcmp(argv[i], "-") && !bbg.in_place) {
1178                                 add_input_file(stdin);
1179                                 process_files();
1180                         } else {
1181                                 file = bb_wfopen(argv[i], "r");
1182                                 if (file) {
1183                                         if(bbg.in_place) {
1184                                                 struct stat statbuf;
1185                                                 int nonstdoutfd;
1186
1187                                                 bbg.outname=bb_xstrndup(argv[i],strlen(argv[i])+6);
1188                                                 strcat(bbg.outname,"XXXXXX");
1189                                                 if(-1==(nonstdoutfd=mkstemp(bbg.outname)))
1190                                                         bb_error_msg_and_die("no temp file");
1191                                                 bbg.nonstdout=fdopen(nonstdoutfd,"w");
1192
1193                                                 /* Set permissions of output file */
1194
1195                                                 fstat(fileno(file),&statbuf);
1196                                                 fchmod(nonstdoutfd,statbuf.st_mode);
1197                                                 add_input_file(file);
1198                                                 process_files();
1199                                                 fclose(bbg.nonstdout);
1200
1201                                                 bbg.nonstdout=stdout;
1202                                                 unlink(argv[i]);
1203                                                 rename(bbg.outname,argv[i]);
1204                                                 free(bbg.outname);
1205                                                 bbg.outname=0;
1206                                         } else add_input_file(file);
1207                                 } else {
1208                                         status = EXIT_FAILURE;
1209                                 }
1210                         }
1211                 }
1212                 if(bbg.input_file_count>bbg.current_input_file) process_files();
1213         }
1214
1215         return status;
1216 }