usr.bin/sed/main.c

   1 /*-
   2  * Copyright (c) 2013 Johann 'Myrkraverk' Oskarsson.
   3  * Copyright (c) 1992 Diomidis Spinellis.
   4  * Copyright (c) 1992, 1993
   5  *      The Regents of the University of California.  All rights reserved.
   6  *
   7  * This code is derived from software contributed to Berkeley by
   8  * Diomidis Spinellis of Imperial College, University of London.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  * 4. Neither the name of the University nor the names of its contributors
  19  *    may be used to endorse or promote products derived from this software
  20  *    without specific prior written permission.
  21  *
  22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  32  * SUCH DAMAGE.
  33  */
  34
  35 #include <sys/cdefs.h>
  36 __FBSDID("$FreeBSD$");
  37
  38 #ifndef lint
  39 static const char copyright[] =
  40 "@(#) Copyright (c) 1992, 1993\n\
  41         The Regents of the University of California.  All rights reserved.\n";
  42 #endif
  43
  44 #ifndef lint
  45 static const char sccsid[] = "@(#)main.c        8.2 (Berkeley) 1/3/94";
  46 #endif
  47
  48 #include <sys/types.h>
  49 #include <sys/mman.h>
  50 #include <sys/param.h>
  51 #include <sys/stat.h>
  52
  53 #include <err.h>
  54 #include <errno.h>
  55 #include <fcntl.h>
  56 #include <libgen.h>
  57 #include <limits.h>
  58 #include <locale.h>
  59 #include <regex.h>
  60 #include <stddef.h>
  61 #define _WITH_GETLINE
  62 #include <stdio.h>
  63 #include <stdlib.h>
  64 #include <string.h>
  65 #include <unistd.h>
  66
  67 #include "defs.h"
  68 #include "extern.h"
  69
  70 /*
  71  * Linked list of units (strings and files) to be compiled
  72  */
  73 struct s_compunit {
  74         struct s_compunit *next;
  75         enum e_cut {CU_FILE, CU_STRING} type;
  76         char *s;                        /* Pointer to string or fname */
  77 };
  78
  79 /*
  80  * Linked list pointer to compilation units and pointer to current
  81  * next pointer.
  82  */
  83 static struct s_compunit *script, **cu_nextp = &script;
  84
  85 /*
  86  * Linked list of files to be processed
  87  */
  88 struct s_flist {
  89         char *fname;
  90         struct s_flist *next;
  91 };
  92
  93 /*
  94  * Linked list pointer to files and pointer to current
  95  * next pointer.
  96  */
  97 static struct s_flist *files, **fl_nextp = &files;
  98
  99 FILE *infile;                   /* Current input file */
 100 FILE *outfile;                  /* Current output file */
 101
 102 int aflag, eflag, nflag;
 103 int rflags = 0;
 104 int quit = 0;
 105 static int rval;                /* Exit status */
 106
 107 static int ispan;               /* Whether inplace editing spans across files */
 108
 109 /*
 110  * Current file and line number; line numbers restart across compilation
 111  * units, but span across input files.  The latter is optional if editing
 112  * in place.
 113  */
 114 const char *fname;              /* File name. */
 115 const char *outfname;           /* Output file name */
 116 static char oldfname[PATH_MAX]; /* Old file name (for in-place editing) */
 117 static char tmpfname[PATH_MAX]; /* Temporary file name (for in-place editing) */
 118 const char *inplace;            /* Inplace edit file extension. */
 119 u_long linenum;
 120
 121 static void add_compunit(enum e_cut, char *);
 122 static void add_file(char *);
 123 static void usage(void);
 124
 125 int
 126 main(int argc, char *argv[])
 127 {
 128         int c, fflag;
 129         char *temp_arg;
 130
 131         (void) setlocale(LC_ALL, "");
 132
 133         fflag = 0;
 134         inplace = NULL;
 135
 136         while ((c = getopt(argc, argv, "EI:ae:f:i:lnru")) != -1)
 137                 switch (c) {
 138                 case 'r':               /* Gnu sed compat */
 139                 case 'E':
 140                         rflags = REG_EXTENDED;
 141                         break;
 142                 case 'I':
 143                         inplace = optarg;
 144                         ispan = 1;      /* span across input files */
 145                         break;
 146                 case 'a':
 147                         aflag = 1;
 148                         break;
 149                 case 'e':
 150                         eflag = 1;
 151                         if ((temp_arg = malloc(strlen(optarg) + 2)) == NULL)
 152                                 err(1, "malloc");
 153                         strcpy(temp_arg, optarg);
 154                         strcat(temp_arg, "\n");
 155                         add_compunit(CU_STRING, temp_arg);
 156                         break;
 157                 case 'f':
 158                         fflag = 1;
 159                         add_compunit(CU_FILE, optarg);
 160                         break;
 161                 case 'i':
 162                         inplace = optarg;
 163                         ispan = 0;      /* don't span across input files */
 164                         break;
 165                 case 'l':
 166                         if(setvbuf(stdout, NULL, _IOLBF, 0) != 0)
 167                                 warnx("setting line buffered output failed");
 168                         break;
 169                 case 'n':
 170                         nflag = 1;
 171                         break;
 172                 case 'u':
 173                         if(setvbuf(stdout, NULL, _IONBF, 0) != 0)
 174                                 warnx("setting unbuffered output failed");
 175                         break;
 176                 default:
 177                 case '?':
 178                         usage();
 179                 }
 180         argc -= optind;
 181         argv += optind;
 182
 183         /* First usage case; script is the first arg */
 184         if (!eflag && !fflag && *argv) {
 185                 add_compunit(CU_STRING, *argv);
 186                 argv++;
 187         }
 188
 189         compile();
 190
 191         /* Continue with first and start second usage */
 192         if (*argv)
 193                 for (; *argv; argv++)
 194                         add_file(*argv);
 195         else
 196                 add_file(NULL);
 197         process();
 198         cfclose(prog, NULL);
 199         if (fclose(stdout))
 200                 err(1, "stdout");
 201         exit(rval);
 202 }
 203
 204 static void
 205 usage(void)
 206 {
 207         (void)fprintf(stderr,
 208             "usage: %s script [-Ealnru] [-i extension] [file ...]\n"
 209             "\t%s [-Ealnu] [-i extension] [-e script] ... [-f script_file]"
 210             " ... [file ...]\n", getprogname(), getprogname());
 211         exit(1);
 212 }
 213
 214 /*
 215  * Like fgets, but go through the chain of compilation units chaining them
 216  * together.  Empty strings and files are ignored.
 217  */
 218 char *
 219 cu_fgets(char *buf, int n, int *more)
 220 {
 221         static enum {ST_EOF, ST_FILE, ST_STRING} state = ST_EOF;
 222         static FILE *f;         /* Current open file */
 223         static char *s;         /* Current pointer inside string */
 224         static char string_ident[30];
 225         char *p;
 226
 227 again:
 228         switch (state) {
 229         case ST_EOF:
 230                 if (script == NULL) {
 231                         if (more != NULL)
 232                                 *more = 0;
 233                         return (NULL);
 234                 }
 235                 linenum = 0;
 236                 switch (script->type) {
 237                 case CU_FILE:
 238                         if ((f = fopen(script->s, "r")) == NULL)
 239                                 err(1, "%s", script->s);
 240                         fname = script->s;
 241                         state = ST_FILE;
 242                         goto again;
 243                 case CU_STRING:
 244                         if (((size_t)snprintf(string_ident,
 245                             sizeof(string_ident), "\"%s\"", script->s)) >=
 246                             sizeof(string_ident) - 1)
 247                                 (void)strcpy(string_ident +
 248                                     sizeof(string_ident) - 6, " ...\"");
 249                         fname = string_ident;
 250                         s = script->s;
 251                         state = ST_STRING;
 252                         goto again;
 253                 }
 254         case ST_FILE:
 255                 if ((p = fgets(buf, n, f)) != NULL) {
 256                         linenum++;
 257                         if (linenum == 1 && buf[0] == '#' && buf[1] == 'n')
 258                                 nflag = 1;
 259                         if (more != NULL)
 260                                 *more = !feof(f);
 261                         return (p);
 262                 }
 263                 script = script->next;
 264                 (void)fclose(f);
 265                 state = ST_EOF;
 266                 goto again;
 267         case ST_STRING:
 268                 if (linenum == 0 && s[0] == '#' && s[1] == 'n')
 269                         nflag = 1;
 270                 p = buf;
 271                 for (;;) {
 272                         if (n-- <= 1) {
 273                                 *p = '\0';
 274                                 linenum++;
 275                                 if (more != NULL)
 276                                         *more = 1;
 277                                 return (buf);
 278                         }
 279                         switch (*s) {
 280                         case '\0':
 281                                 state = ST_EOF;
 282                                 if (s == script->s) {
 283                                         script = script->next;
 284                                         goto again;
 285                                 } else {
 286                                         script = script->next;
 287                                         *p = '\0';
 288                                         linenum++;
 289                                         if (more != NULL)
 290                                                 *more = 0;
 291                                         return (buf);
 292                                 }
 293                         case '\n':
 294                                 *p++ = '\n';
 295                                 *p = '\0';
 296                                 s++;
 297                                 linenum++;
 298                                 if (more != NULL)
 299                                         *more = 0;
 300                                 return (buf);
 301                         default:
 302                                 *p++ = *s++;
 303                         }
 304                 }
 305         }
 306         /* NOTREACHED */
 307         return (NULL);
 308 }
 309
 310 /*
 311  * Like fgets, but go through the list of files chaining them together.
 312  * Set len to the length of the line.
 313  */
 314 int
 315 mf_fgets(SPACE *sp, enum e_spflag spflag)
 316 {
 317         struct stat sb;
 318         ssize_t len;
 319         static char *p = NULL;
 320         static size_t plen = 0;
 321         int c;
 322         static int firstfile;
 323
 324         if (infile == NULL) {
 325                 /* stdin? */
 326                 if (files->fname == NULL) {
 327                         if (inplace != NULL)
 328                                 errx(1, "-I or -i may not be used with stdin");
 329                         infile = stdin;
 330                         fname = "stdin";
 331                         outfile = stdout;
 332                         outfname = "stdout";
 333                 }
 334                 firstfile = 1;
 335         }
 336
 337         for (;;) {
 338                 if (infile != NULL && (c = getc(infile)) != EOF && !quit) {
 339                         (void)ungetc(c, infile);
 340                         break;
 341                 }
 342                 /* If we are here then either eof or no files are open yet */
 343                 if (infile == stdin) {
 344                         sp->len = 0;
 345                         return (0);
 346                 }
 347                 if (infile != NULL) {
 348                         fclose(infile);
 349                         if (*oldfname != '\0') {
 350                                 /* if there was a backup file, remove it */
 351                                 unlink(oldfname);
 352                                 /*
 353                                  * Backup the original.  Note that hard links
 354                                  * are not supported on all filesystems.
 355                                  */
 356                                 if ((link(fname, oldfname) != 0) &&
 357                                    (rename(fname, oldfname) != 0)) {
 358                                         warn("rename()");
 359                                         if (*tmpfname)
 360                                                 unlink(tmpfname);
 361                                         exit(1);
 362                                 }
 363                                 *oldfname = '\0';
 364                         }
 365                         if (*tmpfname != '\0') {
 366                                 if (outfile != NULL && outfile != stdout)
 367                                         if (fclose(outfile) != 0) {
 368                                                 warn("fclose()");
 369                                                 unlink(tmpfname);
 370                                                 exit(1);
 371                                         }
 372                                 outfile = NULL;
 373                                 if (rename(tmpfname, fname) != 0) {
 374                                         /* this should not happen really! */
 375                                         warn("rename()");
 376                                         unlink(tmpfname);
 377                                         exit(1);
 378                                 }
 379                                 *tmpfname = '\0';
 380                         }
 381                         outfname = NULL;
 382                 }
 383                 if (firstfile == 0)
 384                         files = files->next;
 385                 else
 386                         firstfile = 0;
 387                 if (files == NULL) {
 388                         sp->len = 0;
 389                         return (0);
 390                 }
 391                 fname = files->fname;
 392                 if (inplace != NULL) {
 393                         if (lstat(fname, &sb) != 0)
 394                                 err(1, "%s", fname);
 395                         if (!S_ISREG(sb.st_mode))
 396                                 errx(1, "%s: %s %s", fname,
 397                                     "in-place editing only",
 398                                     "works for regular files");
 399                         if (*inplace != '\0') {
 400                                 strlcpy(oldfname, fname,
 401                                     sizeof(oldfname));
 402                                 len = strlcat(oldfname, inplace,
 403                                     sizeof(oldfname));
 404                                 if (len > (ssize_t)sizeof(oldfname))
 405                                         errx(1, "%s: name too long", fname);
 406                         }
 407                         len = snprintf(tmpfname, sizeof(tmpfname),
 408                             "%s/.!%ld!%s", dirname(fname), (long)getpid(),
 409                             basename(fname));
 410                         if (len >= (ssize_t)sizeof(tmpfname))
 411                                 errx(1, "%s: name too long", fname);
 412                         unlink(tmpfname);
 413                         if (outfile != NULL && outfile != stdout)
 414                                 fclose(outfile);
 415                         if ((outfile = fopen(tmpfname, "w")) == NULL)
 416                                 err(1, "%s", fname);
 417                         fchown(fileno(outfile), sb.st_uid, sb.st_gid);
 418                         fchmod(fileno(outfile), sb.st_mode & ALLPERMS);
 419                         outfname = tmpfname;
 420                         if (!ispan) {
 421                                 linenum = 0;
 422                                 resetstate();
 423                         }
 424                 } else {
 425                         outfile = stdout;
 426                         outfname = "stdout";
 427                 }
 428                 if ((infile = fopen(fname, "r")) == NULL) {
 429                         warn("%s", fname);
 430                         rval = 1;
 431                         continue;
 432                 }
 433         }
 434         /*
 435          * We are here only when infile is open and we still have something
 436          * to read from it.
 437          *
 438          * Use getline() so that we can handle essentially infinite input
 439          * data.  The p and plen are static so each invocation gives
 440          * getline() the same buffer which is expanded as needed.
 441          */
 442         len = getline(&p, &plen, infile);
 443         if (len == -1)
 444                 err(1, "%s", fname);
 445         if (len != 0 && p[len - 1] == '\n') {
 446                 sp->append_newline = 1;
 447                 len--;
 448         } else if (!lastline()) {
 449                 sp->append_newline = 1;
 450         } else {
 451                 sp->append_newline = 0;
 452         }
 453         cspace(sp, p, len, spflag);
 454
 455         linenum++;
 456
 457         return (1);
 458 }
 459
 460 /*
 461  * Add a compilation unit to the linked list
 462  */
 463 static void
 464 add_compunit(enum e_cut type, char *s)
 465 {
 466         struct s_compunit *cu;
 467
 468         if ((cu = malloc(sizeof(struct s_compunit))) == NULL)
 469                 err(1, "malloc");
 470         cu->type = type;
 471         cu->s = s;
 472         cu->next = NULL;
 473         *cu_nextp = cu;
 474         cu_nextp = &cu->next;
 475 }
 476
 477 /*
 478  * Add a file to the linked list
 479  */
 480 static void
 481 add_file(char *s)
 482 {
 483         struct s_flist *fp;
 484
 485         if ((fp = malloc(sizeof(struct s_flist))) == NULL)
 486                 err(1, "malloc");
 487         fp->next = NULL;
 488         *fl_nextp = fp;
 489         fp->fname = s;
 490         fl_nextp = &fp->next;
 491 }
 492
 493 static int
 494 next_files_have_lines(void)
 495 {
 496         struct s_flist *file;
 497         FILE *file_fd;
 498         int ch;
 499
 500         file = files;
 501         while ((file = file->next) != NULL) {
 502                 if ((file_fd = fopen(file->fname, "r")) == NULL)
 503                         continue;
 504
 505                 if ((ch = getc(file_fd)) != EOF) {
 506                         /*
 507                          * This next file has content, therefore current
 508                          * file doesn't contains the last line.
 509                          */
 510                         ungetc(ch, file_fd);
 511                         fclose(file_fd);
 512                         return (1);
 513                 }
 514
 515                 fclose(file_fd);
 516         }
 517
 518         return (0);
 519 }
 520
 521 int
 522 lastline(void)
 523 {
 524         int ch;
 525
 526         if (feof(infile)) {
 527                 return !(
 528                     (inplace == NULL || ispan) &&
 529                     next_files_have_lines());
 530         }
 531         if ((ch = getc(infile)) == EOF) {
 532                 return !(
 533                     (inplace == NULL || ispan) &&
 534                     next_files_have_lines());
 535         }
 536         ungetc(ch, infile);
 537         return (0);
 538 }