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