usr.bin/sed/main.c

   1 /*-
   2  * SPDX-License-Identifier: BSD-3-Clause
   3  *
   4  * Copyright (c) 2013 Johann 'Myrkraverk' Oskarsson.
   5  * Copyright (c) 1992 Diomidis Spinellis.
   6  * Copyright (c) 1992, 1993
   7  *      The Regents of the University of California.  All rights reserved.
   8  *
   9  * This code is derived from software contributed to Berkeley by
  10  * Diomidis Spinellis of Imperial College, University of London.
  11  *
  12  * Redistribution and use in source and binary forms, with or without
  13  * modification, are permitted provided that the following conditions
  14  * are met:
  15  * 1. Redistributions of source code must retain the above copyright
  16  *    notice, this list of conditions and the following disclaimer.
  17  * 2. Redistributions in binary form must reproduce the above copyright
  18  *    notice, this list of conditions and the following disclaimer in the
  19  *    documentation and/or other materials provided with the distribution.
  20  * 3. Neither the name of the University nor the names of its contributors
  21  *    may be used to endorse or promote products derived from this software
  22  *    without specific prior written permission.
  23  *
  24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  34  * SUCH DAMAGE.
  35  */
  36
  37 #include <sys/types.h>
  38 #include <sys/mman.h>
  39 #include <sys/param.h>
  40 #include <sys/stat.h>
  41
  42 #include <err.h>
  43 #include <errno.h>
  44 #include <fcntl.h>
  45 #include <libgen.h>
  46 #include <limits.h>
  47 #include <locale.h>
  48 #include <regex.h>
  49 #include <stddef.h>
  50 #include <stdio.h>
  51 #include <stdlib.h>
  52 #include <string.h>
  53 #include <unistd.h>
  54
  55 #include "defs.h"
  56 #include "extern.h"
  57
  58 /*
  59  * Linked list of units (strings and files) to be compiled
  60  */
  61 struct s_compunit {
  62         struct s_compunit *next;
  63         enum e_cut {CU_FILE, CU_STRING} type;
  64         char *s;                        /* Pointer to string or fname */
  65 };
  66
  67 /*
  68  * Linked list pointer to compilation units and pointer to current
  69  * next pointer.
  70  */
  71 static struct s_compunit *script, **cu_nextp = &script;
  72
  73 /*
  74  * Linked list of files to be processed
  75  */
  76 struct s_flist {
  77         char *fname;
  78         struct s_flist *next;
  79 };
  80
  81 /*
  82  * Linked list pointer to files and pointer to current
  83  * next pointer.
  84  */
  85 static struct s_flist *files, **fl_nextp = &files;
  86
  87 FILE *infile;                   /* Current input file */
  88 FILE *outfile;                  /* Current output file */
  89
  90 int aflag, eflag, nflag;
  91 int rflags = 0;
  92 int quit = 0;
  93 static int rval;                /* Exit status */
  94
  95 static int ispan;               /* Whether inplace editing spans across files */
  96
  97 /*
  98  * Current file and line number; line numbers restart across compilation
  99  * units, but span across input files.  The latter is optional if editing
 100  * in place.
 101  */
 102 const char *fname;              /* File name. */
 103 const char *outfname;           /* Output file name */
 104 static char oldfname[PATH_MAX]; /* Old file name (for in-place editing) */
 105 static char tmpfname[PATH_MAX]; /* Temporary file name (for in-place editing) */
 106 const char *inplace;            /* Inplace edit file extension. */
 107 u_long linenum;
 108
 109 static void add_compunit(enum e_cut, char *);
 110 static void add_file(char *);
 111 static void usage(void) __dead2;
 112
 113 int
 114 main(int argc, char *argv[])
 115 {
 116         int c, fflag, fflagstdin;
 117         char *temp_arg;
 118
 119         (void) setlocale(LC_ALL, "");
 120
 121         fflag = 0;
 122         fflagstdin = 0;
 123         inplace = NULL;
 124
 125         while ((c = getopt(argc, argv, "EI:ae:f:i:lnru")) != -1)
 126                 switch (c) {
 127                 case 'r':               /* Gnu sed compat */
 128                 case 'E':
 129                         rflags = REG_EXTENDED;
 130                         break;
 131                 case 'I':
 132                         inplace = optarg;
 133                         ispan = 1;      /* span across input files */
 134                         break;
 135                 case 'a':
 136                         aflag = 1;
 137                         break;
 138                 case 'e':
 139                         eflag = 1;
 140                         if ((temp_arg = malloc(strlen(optarg) + 2)) == NULL)
 141                                 err(1, "malloc");
 142                         strcpy(temp_arg, optarg);
 143                         strcat(temp_arg, "\n");
 144                         add_compunit(CU_STRING, temp_arg);
 145                         break;
 146                 case 'f':
 147                         fflag = 1;
 148                         if (strcmp(optarg, "-") == 0)
 149                                 fflagstdin = 1;
 150                         add_compunit(CU_FILE, optarg);
 151                         break;
 152                 case 'i':
 153                         inplace = optarg;
 154                         ispan = 0;      /* don't span across input files */
 155                         break;
 156                 case 'l':
 157                         if(setvbuf(stdout, NULL, _IOLBF, 0) != 0)
 158                                 warnx("setting line buffered output failed");
 159                         break;
 160                 case 'n':
 161                         nflag = 1;
 162                         break;
 163                 case 'u':
 164                         if(setvbuf(stdout, NULL, _IONBF, 0) != 0)
 165                                 warnx("setting unbuffered output failed");
 166                         break;
 167                 default:
 168                 case '?':
 169                         usage();
 170                 }
 171         argc -= optind;
 172         argv += optind;
 173
 174         /* First usage case; script is the first arg */
 175         if (!eflag && !fflag && *argv) {
 176                 add_compunit(CU_STRING, *argv);
 177                 argv++;
 178         }
 179
 180         compile();
 181
 182         /* Continue with first and start second usage */
 183         if (*argv)
 184                 for (; *argv; argv++)
 185                         add_file(*argv);
 186         else if (fflagstdin)
 187                 exit(rval);
 188         else
 189                 add_file(NULL);
 190         process();
 191         cfclose(prog, NULL);
 192         if (fclose(stdout))
 193                 err(1, "stdout");
 194         exit(rval);
 195 }
 196
 197 static void
 198 usage(void)
 199 {
 200         (void)fprintf(stderr,
 201             "usage: %s script [-Ealnru] [-i extension] [file ...]\n"
 202             "\t%s [-Ealnu] [-i extension] [-e script] ... [-f script_file]"
 203             " ... [file ...]\n", getprogname(), getprogname());
 204         exit(1);
 205 }
 206
 207 /*
 208  * Like fgets, but go through the chain of compilation units chaining them
 209  * together.  Empty strings and files are ignored.
 210  */
 211 char *
 212 cu_fgets(char *buf, int n, int *more)
 213 {
 214         static enum {ST_EOF, ST_FILE, ST_STRING} state = ST_EOF;
 215         static FILE *f;         /* Current open file */
 216         static char *s;         /* Current pointer inside string */
 217         static char string_ident[30];
 218         char *p;
 219
 220 again:
 221         switch (state) {
 222         case ST_EOF:
 223                 if (script == NULL) {
 224                         if (more != NULL)
 225                                 *more = 0;
 226                         return (NULL);
 227                 }
 228                 linenum = 0;
 229                 switch (script->type) {
 230                 case CU_FILE:
 231                         if (strcmp(script->s, "-") == 0) {
 232                                 f = stdin;
 233                                 fname = "stdin";
 234                         } else {
 235                                 if ((f = fopen(script->s, "r")) == NULL)
 236                                         err(1, "%s", script->s);
 237                                 fname = script->s;
 238                         }
 239                         state = ST_FILE;
 240                         goto again;
 241                 case CU_STRING:
 242                         if (((size_t)snprintf(string_ident,
 243                             sizeof(string_ident), "\"%s\"", script->s)) >=
 244                             sizeof(string_ident) - 1)
 245                                 (void)strcpy(string_ident +
 246                                     sizeof(string_ident) - 6, " ...\"");
 247                         fname = string_ident;
 248                         s = script->s;
 249                         state = ST_STRING;
 250                         goto again;
 251                 default:
 252                         __unreachable();
 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         char *dirbuf, *basebuf;
 320         static char *p = NULL;
 321         static size_t plen = 0;
 322         int c;
 323         static int firstfile;
 324
 325         if (infile == NULL) {
 326                 /* stdin? */
 327                 if (files->fname == NULL) {
 328                         if (inplace != NULL)
 329                                 errx(1, "-I or -i may not be used with stdin");
 330                         infile = stdin;
 331                         fname = "stdin";
 332                         outfile = stdout;
 333                         outfname = "stdout";
 334                 }
 335                 firstfile = 1;
 336         }
 337
 338         for (;;) {
 339                 if (infile != NULL && (c = getc(infile)) != EOF && !quit) {
 340                         (void)ungetc(c, infile);
 341                         break;
 342                 }
 343                 /* If we are here then either eof or no files are open yet */
 344                 if (infile == stdin) {
 345                         sp->len = 0;
 346                         return (0);
 347                 }
 348                 if (infile != NULL) {
 349                         fclose(infile);
 350                         if (*oldfname != '\0') {
 351                                 /* if there was a backup file, remove it */
 352                                 unlink(oldfname);
 353                                 /*
 354                                  * Backup the original.  Note that hard links
 355                                  * are not supported on all filesystems.
 356                                  */
 357                                 if ((link(fname, oldfname) != 0) &&
 358                                    (rename(fname, oldfname) != 0)) {
 359                                         warn("rename()");
 360                                         if (*tmpfname)
 361                                                 unlink(tmpfname);
 362                                         exit(1);
 363                                 }
 364                                 *oldfname = '\0';
 365                         }
 366                         if (*tmpfname != '\0') {
 367                                 if (outfile != NULL && outfile != stdout)
 368                                         if (fclose(outfile) != 0) {
 369                                                 warn("fclose()");
 370                                                 unlink(tmpfname);
 371                                                 exit(1);
 372                                         }
 373                                 outfile = NULL;
 374                                 if (rename(tmpfname, fname) != 0) {
 375                                         /* this should not happen really! */
 376                                         warn("rename()");
 377                                         unlink(tmpfname);
 378                                         exit(1);
 379                                 }
 380                                 *tmpfname = '\0';
 381                         }
 382                         outfname = NULL;
 383                 }
 384                 if (firstfile == 0)
 385                         files = files->next;
 386                 else
 387                         firstfile = 0;
 388                 if (files == NULL) {
 389                         sp->len = 0;
 390                         return (0);
 391                 }
 392                 fname = files->fname;
 393                 if (inplace != NULL) {
 394                         if (lstat(fname, &sb) != 0)
 395                                 err(1, "%s", fname);
 396                         if (!S_ISREG(sb.st_mode))
 397                                 errx(1, "%s: %s %s", fname,
 398                                     "in-place editing only",
 399                                     "works for regular files");
 400                         if (*inplace != '\0') {
 401                                 strlcpy(oldfname, fname,
 402                                     sizeof(oldfname));
 403                                 len = strlcat(oldfname, inplace,
 404                                     sizeof(oldfname));
 405                                 if (len > (ssize_t)sizeof(oldfname))
 406                                         errx(1, "%s: name too long", fname);
 407                         }
 408                         if ((dirbuf = strdup(fname)) == NULL ||
 409                             (basebuf = strdup(fname)) == NULL)
 410                                 err(1, "strdup");
 411                         len = snprintf(tmpfname, sizeof(tmpfname),
 412                             "%s/.!%ld!%s", dirname(dirbuf), (long)getpid(),
 413                             basename(basebuf));
 414                         free(dirbuf);
 415                         free(basebuf);
 416                         if (len >= (ssize_t)sizeof(tmpfname))
 417                                 errx(1, "%s: name too long", fname);
 418                         unlink(tmpfname);
 419                         if (outfile != NULL && outfile != stdout)
 420                                 fclose(outfile);
 421                         if ((outfile = fopen(tmpfname, "w")) == NULL)
 422                                 err(1, "%s", fname);
 423                         fchown(fileno(outfile), sb.st_uid, sb.st_gid);
 424                         fchmod(fileno(outfile), sb.st_mode & ALLPERMS);
 425                         outfname = tmpfname;
 426                         if (!ispan) {
 427                                 linenum = 0;
 428                                 resetstate();
 429                         }
 430                 } else {
 431                         outfile = stdout;
 432                         outfname = "stdout";
 433                 }
 434                 if ((infile = fopen(fname, "r")) == NULL) {
 435                         warn("%s", fname);
 436                         rval = 1;
 437                         continue;
 438                 }
 439         }
 440         /*
 441          * We are here only when infile is open and we still have something
 442          * to read from it.
 443          *
 444          * Use getline() so that we can handle essentially infinite input
 445          * data.  The p and plen are static so each invocation gives
 446          * getline() the same buffer which is expanded as needed.
 447          */
 448         len = getline(&p, &plen, infile);
 449         if (len == -1)
 450                 err(1, "%s", fname);
 451         if (len != 0 && p[len - 1] == '\n') {
 452                 sp->append_newline = 1;
 453                 len--;
 454         } else if (!lastline()) {
 455                 sp->append_newline = 1;
 456         } else {
 457                 sp->append_newline = 0;
 458         }
 459         cspace(sp, p, len, spflag);
 460
 461         linenum++;
 462
 463         return (1);
 464 }
 465
 466 /*
 467  * Add a compilation unit to the linked list
 468  */
 469 static void
 470 add_compunit(enum e_cut type, char *s)
 471 {
 472         struct s_compunit *cu;
 473
 474         if ((cu = malloc(sizeof(struct s_compunit))) == NULL)
 475                 err(1, "malloc");
 476         cu->type = type;
 477         cu->s = s;
 478         cu->next = NULL;
 479         *cu_nextp = cu;
 480         cu_nextp = &cu->next;
 481 }
 482
 483 /*
 484  * Add a file to the linked list
 485  */
 486 static void
 487 add_file(char *s)
 488 {
 489         struct s_flist *fp;
 490
 491         if ((fp = malloc(sizeof(struct s_flist))) == NULL)
 492                 err(1, "malloc");
 493         fp->next = NULL;
 494         *fl_nextp = fp;
 495         fp->fname = s;
 496         fl_nextp = &fp->next;
 497 }
 498
 499 static int
 500 next_files_have_lines(void)
 501 {
 502         struct s_flist *file;
 503         FILE *file_fd;
 504         int ch;
 505
 506         file = files;
 507         while ((file = file->next) != NULL) {
 508                 if ((file_fd = fopen(file->fname, "r")) == NULL)
 509                         continue;
 510
 511                 if ((ch = getc(file_fd)) != EOF) {
 512                         /*
 513                          * This next file has content, therefore current
 514                          * file doesn't contains the last line.
 515                          */
 516                         ungetc(ch, file_fd);
 517                         fclose(file_fd);
 518                         return (1);
 519                 }
 520
 521                 fclose(file_fd);
 522         }
 523
 524         return (0);
 525 }
 526
 527 int
 528 lastline(void)
 529 {
 530         int ch;
 531
 532         if (feof(infile)) {
 533                 return !(
 534                     (inplace == NULL || ispan) &&
 535                     next_files_have_lines());
 536         }
 537         if ((ch = getc(infile)) == EOF) {
 538                 return !(
 539                     (inplace == NULL || ispan) &&
 540                     next_files_have_lines());
 541         }
 542         ungetc(ch, infile);
 543         return (0);
 544 }