archival/ar.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini ar implementation for busybox
   4  *
   5  * Copyright (C) 2000 by Glenn McGrath
   6  * Written by Glenn McGrath <bug1@netconnect.com.au> 1 June 2000
   7  *
   8  * Based in part on BusyBox tar, Debian dpkg-deb and GNU ar.
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18  * General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with this program; if not, write to the Free Software
  22  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  23  *
  24  * Last modified 20 September 2000
  25  */
  26 #include <stdio.h>
  27 #include <string.h>
  28 #include <fcntl.h>
  29 #include <errno.h>
  30 #include <ctype.h>
  31 #include <time.h>
  32 #include <utime.h>
  33 #include <unistd.h>
  34 #include <stdlib.h>
  35 #include <sys/types.h>
  36 #include <sys/stat.h>
  37 #include <malloc.h>
  38 #include "busybox.h"
  39
  40 #define BLOCK_SIZE 60
  41 #define PRESERVE_DATE 1 /* preserve original dates */
  42 #define VERBOSE       2 /* be verbose */
  43 #define DISPLAY       4 /* display contents */
  44 #define EXT_TO_FILE   8 /* extract contents of archive */
  45 #define EXT_TO_STDOUT 16        /* extract to stdout */
  46 #define RECURSIVE     32
  47
  48 #define MAX_NAME_LENGTH 100
  49
  50 //#define BB_DECLARE_EXTERN
  51 //#define bb_need_io_error
  52 //#include "messages.c"
  53
  54 //#define BB_AR_EXPERIMENTAL_UNTAR
  55
  56 #if defined BB_AR_EXPERIMENTAL_UNTAR
  57 typedef struct rawTarHeader {
  58         char name[100];               /*   0-99 */
  59         char mode[8];                 /* 100-107 */
  60         char uid[8];                  /* 108-115 */
  61         char gid[8];                  /* 116-123 */
  62         char size[12];                /* 124-135 */
  63         char mtime[12];               /* 136-147 */
  64         char chksum[8];               /* 148-155 */
  65         char typeflag;                /* 156-156 */
  66         char linkname[100];           /* 157-256 */
  67         char magic[6];                /* 257-262 */
  68         char version[2];              /* 263-264 */
  69         char uname[32];               /* 265-296 */
  70         char gname[32];               /* 297-328 */
  71         char devmajor[8];             /* 329-336 */
  72         char devminor[8];             /* 337-344 */
  73         char prefix[155];             /* 345-499 */
  74         char padding[12];             /* 500-512 */
  75 } rawTarHeader_t;
  76 #endif
  77
  78 typedef struct rawArHeader {    /* Byte Offset */
  79         char name[16];          /*  0-15 */
  80         char date[12];          /* 16-27 */
  81         char uid[6], gid[6];    /* 28-39 */
  82         char mode[8];           /* 40-47 */
  83         char size[10];          /* 48-57 */
  84         char fmag[2];           /* 58-59 */
  85 } rawArHeader_t;
  86
  87 typedef struct headerL {
  88         char name[MAX_NAME_LENGTH];
  89         size_t size;
  90         uid_t uid;
  91         gid_t gid;
  92         mode_t mode;
  93         time_t mtime;
  94         off_t offset;
  95         struct headerL *next;
  96 } headerL_t;
  97
  98 #if defined BB_AR_EXPERIMENTAL_UNTAR
  99 /*
 100  * identify Tar header (magic field) and reset srcFd to entry position
 101  */
 102 static int checkTarMagic(int srcFd)
 103 {
 104         off_t headerStart;
 105         char magic[6];
 106
 107         headerStart = lseek(srcFd, 0, SEEK_CUR);
 108         lseek(srcFd, (off_t) 257, SEEK_CUR);
 109         fullRead(srcFd, magic, 6);
 110         lseek(srcFd, headerStart, SEEK_SET);
 111         if (strncmp(magic, "ustar", 5)!=0)
 112                 return(FALSE);
 113         return(TRUE);
 114 }
 115
 116
 117 static int readTarHeader(int srcFd, headerL_t *current)
 118 {
 119         rawTarHeader_t rawTarHeader;
 120         unsigned char *temp = (unsigned char *) &rawTarHeader;
 121         long sum = 0;
 122         int i;
 123         off_t initialOffset;
 124
 125         initialOffset = lseek(srcFd, 0, SEEK_CUR);
 126         if (fullRead(srcFd, (char *) &rawTarHeader, 512) != 512) {
 127                 lseek(srcFd, initialOffset, SEEK_SET);
 128                 return(FALSE);
 129         }
 130         for (i =  0; i < 148 ; i++)
 131         sum += temp[i];
 132         sum += ' ' * 8;
 133         for (i =  156; i < 512 ; i++)
 134                 sum += temp[i];
 135         if (sum!= strtol(rawTarHeader.chksum, NULL, 8))
 136                 return(FALSE);
 137         sscanf(rawTarHeader.name, "%s", current->name);
 138         current->size = strtol(rawTarHeader.size, NULL, 8);
 139         current->uid = strtol(rawTarHeader.uid, NULL, 8);
 140         current->gid = strtol(rawTarHeader.gid, NULL, 8);
 141         current->mode = strtol(rawTarHeader.mode, NULL, 8);
 142         current->mtime = strtol(rawTarHeader.mtime, NULL, 8);
 143         current->offset = lseek(srcFd, 0 , SEEK_CUR);
 144
 145         current->next = (headerL_t *) xmalloc(sizeof(headerL_t));
 146         current = current->next;
 147         return(TRUE);
 148 }
 149 #endif
 150
 151 /*
 152  * identify Ar header (magic) and reset srcFd to entry position
 153  */
 154 static int checkArMagic(int srcFd)
 155 {
 156         off_t headerStart;
 157         char arMagic[8];
 158
 159         headerStart = lseek(srcFd, 0, SEEK_CUR);
 160         if (fullRead(srcFd, arMagic, 8) != 8) {
 161                 errorMsg("fatal error\n");
 162                 return (FALSE);
 163         }
 164         lseek(srcFd, headerStart, SEEK_SET);
 165
 166         if (strncmp(arMagic,"!<arch>",7) != 0)
 167                 return(FALSE);
 168         return(TRUE);
 169 }
 170
 171 /*
 172  * get, check and correct the converted header
 173  */
 174 static int readArEntry(int srcFd, headerL_t *entry)
 175 {
 176         size_t nameLength;
 177         rawArHeader_t rawArHeader;
 178         off_t   initialOffset;
 179
 180         initialOffset = lseek(srcFd, 0, SEEK_CUR);
 181         if (fullRead(srcFd, (char *) &rawArHeader, 60) != 60) {
 182                 lseek(srcFd, initialOffset, SEEK_SET);
 183                 return(FALSE);
 184         }
 185         if ((rawArHeader.fmag[0]!='`') || (rawArHeader.fmag[1]!='\n')) {
 186                 lseek(srcFd, initialOffset, SEEK_SET);
 187                 return(FALSE);
 188         }
 189
 190         strncpy(entry->name, rawArHeader.name, 16);
 191         nameLength=strcspn(entry->name, " \\");
 192         entry->name[nameLength]='\0';
 193         parse_mode(rawArHeader.mode, &entry->mode);
 194         entry->mtime = atoi(rawArHeader.date);
 195         entry->uid = atoi(rawArHeader.uid);
 196         entry->gid = atoi(rawArHeader.gid);
 197         entry->size = (size_t) atoi(rawArHeader.size);
 198         entry->offset = initialOffset + (off_t) 60;
 199
 200         nameLength = strcspn(entry->name, "/");
 201
 202         /* handle GNU style short filenames, strip trailing '/' */
 203         if (nameLength > 0)
 204                 entry->name[nameLength]='\0';
 205
 206         /* handle GNU style long filenames */
 207         if (nameLength == 0) {
 208                 /* escape from recursive call */
 209                 if (entry->name[1]=='0')
 210                         return(TRUE);
 211
 212                 /* the data section contains the real filename */
 213                 if (entry->name[1]=='/') {
 214                         char tempName[MAX_NAME_LENGTH];
 215
 216                         if (entry->size > MAX_NAME_LENGTH)
 217                                 entry->size = MAX_NAME_LENGTH;
 218                         fullRead(srcFd, tempName, entry->size);
 219                         tempName[entry->size-3]='\0';
 220
 221                         /* read the second header for this entry */
 222                         /* be carefull, this is recursive */
 223                         if (readArEntry(srcFd, entry)==FALSE)
 224                                 return(FALSE);
 225
 226                         if ((entry->name[0]='/') && (entry->name[1]='0'))
 227                                 strcpy(entry->name, tempName);
 228                         else {
 229                                 errorMsg("Invalid long filename\n");
 230                                 return(FALSE);
 231                         }
 232                 }
 233         }
 234         return(TRUE);
 235 }
 236
 237 /*
 238  * return the headerL_t struct for the specified filename
 239  */
 240 static headerL_t *getHeaders(int srcFd, headerL_t *head, int funct)
 241 {
 242 #if defined BB_AR_EXPERIMENTAL_UNTAR
 243         int tar=FALSE;
 244 #endif
 245         int ar=FALSE;
 246         headerL_t *list;
 247         off_t initialOffset;
 248
 249         list = (headerL_t *) xmalloc(sizeof(headerL_t));
 250         initialOffset=lseek(srcFd, 0, SEEK_CUR);
 251         if (checkArMagic(srcFd)==TRUE)
 252                 ar=TRUE;
 253
 254 #if defined BB_AR_EXPERIMENTAL_UNTAR
 255         if (checkTarMagic(srcFd)==TRUE)
 256                 tar=TRUE;
 257
 258         if (tar==TRUE) {
 259                 while(readTarHeader(srcFd, list)==TRUE) {
 260                         off_t tarOffset;
 261                         list->next = (headerL_t *) xmalloc(sizeof(headerL_t));
 262                         *list->next = *head;
 263                         *head = *list;
 264
 265                         /* recursive check for sub-archives */
 266                         if ((funct & RECURSIVE) == RECURSIVE)
 267                                 head = getHeaders(srcFd, head, funct);
 268                         tarOffset = (off_t) head->size/512;
 269                         if ( head->size % 512 > 0)
 270                                 tarOffset++;
 271                         tarOffset=tarOffset*512;
 272                         lseek(srcFd, head->offset + tarOffset, SEEK_SET);
 273                 }
 274         }
 275 #endif
 276
 277         if (ar==TRUE) {
 278                 lseek(srcFd, 8, SEEK_CUR);
 279                 while(1) {
 280                         if (readArEntry(srcFd, list) == FALSE) {
 281                                 lseek(srcFd, ++initialOffset, SEEK_CUR);
 282                                 if (readArEntry(srcFd, list) == FALSE)
 283                                         return(head);
 284                         }
 285                         list->next = (headerL_t *) xmalloc(sizeof(headerL_t));
 286                         *list->next = *head;
 287                         *head = *list;
 288                         /* recursive check for sub-archives */
 289                         if (funct & RECURSIVE)
 290                                 head = getHeaders(srcFd, head, funct);
 291                         lseek(srcFd, head->offset + head->size, SEEK_SET);
 292                 }
 293         }
 294         return(head);
 295 }
 296
 297 /*
 298  * find an entry in the linked list matching the filename
 299  */
 300 static headerL_t *findEntry(headerL_t *head, const char *filename)
 301 {
 302         while(head->next != NULL) {
 303                 if (strcmp(filename, head->name)==0)
 304                         return(head);
 305                 head=head->next;
 306         }
 307         return(NULL);
 308 }
 309
 310 extern int ar_main(int argc, char **argv)
 311 {
 312         int funct = 0, opt=0;
 313         int srcFd=0, dstFd=0;
 314         headerL_t *header, *entry, *extractList;
 315
 316         while ((opt = getopt(argc, argv, "ovtpxR")) != -1) {
 317                 switch (opt) {
 318                 case 'o':
 319                         funct |= PRESERVE_DATE;
 320                         break;
 321                 case 'v':
 322                         funct |= VERBOSE;
 323                         break;
 324                 case 't':
 325                         funct |= DISPLAY;
 326                         break;
 327                 case 'x':
 328                         funct |= EXT_TO_FILE;
 329                         break;
 330                 case 'p':
 331                         funct |= EXT_TO_STDOUT;
 332                         break;
 333                 case 'R':
 334                         funct |= RECURSIVE;
 335                         break;
 336                 default:
 337                         usage(ar_usage);
 338                 }
 339         }
 340
 341         /* check the src filename was specified */
 342         if (optind == argc)
 343                 usage(ar_usage);
 344
 345         if ( (srcFd = open(argv[optind], O_RDONLY)) < 0)
 346                 fatalError("Cannot read %s\n", argv[optind]);
 347
 348         optind++;
 349         entry = (headerL_t *) xmalloc(sizeof(headerL_t));
 350         header = (headerL_t *) xmalloc(sizeof(headerL_t));
 351         extractList = (headerL_t *) xmalloc(sizeof(headerL_t));
 352
 353         header = getHeaders(srcFd, header, funct);
 354         /* find files to extract or display */
 355         if (optind<argc) {
 356                 /* only handle specified files */
 357                 while(optind < argc) {
 358                         if ( (entry = findEntry(header, argv[optind])) != NULL) {
 359                                 entry->next = (headerL_t *) xmalloc(sizeof(headerL_t));
 360                                 *entry->next = *extractList;
 361                                 *extractList = *entry;
 362                         }
 363                         optind++;
 364                 }
 365         }
 366         else
 367                 extractList = header;
 368
 369         while(extractList->next != NULL) {
 370                 if (funct & EXT_TO_FILE) {
 371                         if (isDirectory(extractList->name, TRUE, NULL)==FALSE)
 372                                 createPath(extractList->name, 0666);
 373                         dstFd = open(extractList->name, O_WRONLY | O_CREAT, extractList->mode);
 374                         lseek(srcFd, extractList->offset, SEEK_SET);
 375                         copy_file_chunk(srcFd, dstFd, (size_t) extractList->size);
 376                 }
 377                 if (funct & EXT_TO_STDOUT) {
 378                         lseek(srcFd, extractList->offset, SEEK_SET);
 379                         copy_file_chunk(srcFd, fileno(stdout), (size_t) extractList->size);
 380                 }
 381                 if ( (funct & DISPLAY) || (funct & VERBOSE)) {
 382                         if (funct & VERBOSE)
 383                                 printf("%s %d/%d %8d %s ", modeString(extractList->mode),
 384                                         extractList->uid, extractList->gid,
 385                                         extractList->size, timeString(extractList->mtime));
 386                         printf("%s\n", extractList->name);
 387                 }
 388                 extractList=extractList->next;
 389         }
 390         return EXIT_SUCCESS;
 391 }