modules/macbinary.c

   1 // This file is part of Deark.
   2 // Copyright (C) 2018 Jason Summers
   3 // See the file COPYING for terms of use.
   4
   5 // MacBinary
   6
   7 #include <deark-config.h>
   8 #include <deark-private.h>
   9 #include <deark-fmtutil.h>
  10 DE_DECLARE_MODULE(de_module_macbinary);
  11
  12 typedef struct localctx_struct {
  13         u8 extract_files;
  14         u8 oldver;
  15         u8 extver;
  16         u8 extver_minneeded;
  17         int is_v23;
  18         i64 dfpos, rfpos;
  19         i64 dflen, rflen;
  20         struct de_stringreaderdata *filename_srd;
  21         struct de_timestamp create_time;
  22         struct de_timestamp mod_time;
  23 } lctx;
  24
  25 struct fork_info {
  26         u8 is_rsrc;
  27         u8 extract_error_flag;
  28         i64 pos;
  29 };
  30
  31 struct extract_ctx {
  32         lctx *d;
  33         struct fork_info fki_data;
  34         struct fork_info fki_rsrc;
  35 };
  36
  37 static const char *fork_name(int is_rsrc, int capitalize)
  38 {
  39         if(is_rsrc) {
  40                 return capitalize?"Resource":"resource";
  41         }
  42         return capitalize?"Data":"data";
  43 }
  44
  45 static void do_header(deark *c, lctx *d, struct de_advfile *advf)
  46 {
  47         u8 b;
  48         i64 namelen;
  49         i64 pos = 0;
  50         i64 n, n2;
  51         i64 mod_time_raw;
  52         u32 crc_reported, crc_calc;
  53         struct de_fourcc type4cc;
  54         struct de_fourcc creator4cc;
  55         char timestamp_buf[64];
  56
  57         d->oldver = de_getbyte_p(&pos);
  58         de_dbg(c, "original version: %u", (unsigned int)d->oldver);
  59         if(d->oldver!=0) {
  60                 de_warn(c, "Unsupported MacBinary version");
  61                 goto done;
  62         }
  63
  64         d->extver = de_getbyte(122);
  65         de_dbg(c, "extended version: %u", (unsigned int)d->extver);
  66         if(d->extver==129 || d->extver==130) {
  67                 d->is_v23 = 1;
  68         }
  69         if(d->extver >= 129) {
  70                 d->extver_minneeded = de_getbyte(123);
  71                 de_dbg(c, "extended version, min needed: %u", (unsigned int)d->extver_minneeded);
  72         }
  73
  74         namelen = (i64)de_getbyte_p(&pos);
  75         if(namelen>=1 && namelen<=63) {
  76                 // Required to be 1-63 by MacBinary II spec.
  77                 // Original spec has no written requirements.
  78                 // Not supposed to be NUL terminated, but such files exist.
  79                 d->filename_srd = dbuf_read_string(c->infile, pos, namelen, namelen,
  80                         DE_CONVFLAG_STOP_AT_NUL, DE_ENCODING_MACROMAN);
  81                 de_dbg(c, "filename: \"%s\"", ucstring_getpsz(d->filename_srd->str));
  82         }
  83         else {
  84                 de_warn(c, "Bad MacBinary filename length (%d)", (int)namelen);
  85         }
  86         pos += 63;
  87
  88         de_dbg(c, "finder info:");
  89         de_dbg_indent(c, 1);
  90
  91         dbuf_read_fourcc(c->infile, pos, &type4cc, 4, 0x0);
  92         de_dbg(c, "type: '%s'", type4cc.id_dbgstr);
  93         de_memcpy(advf->typecode, type4cc.bytes, 4);
  94         advf->has_typecode = 1;
  95         pos += 4;
  96         dbuf_read_fourcc(c->infile, pos, &creator4cc, 4, 0x0);
  97         de_dbg(c, "creator: '%s'", creator4cc.id_dbgstr);
  98         de_memcpy(advf->creatorcode, creator4cc.bytes, 4);
  99         advf->has_creatorcode = 1;
 100         pos += 4;
 101
 102         advf->has_finderflags = 1;
 103         if(d->is_v23) {
 104                 u8 fflags_hibyte;
 105
 106                 fflags_hibyte = de_getbyte_p(&pos);
 107                 de_dbg(c, "finder flags (high byte): 0x%02x__", (unsigned int)fflags_hibyte);
 108                 pos++;
 109                 advf->finderflags = (u16)fflags_hibyte << 8;
 110         }
 111         else {
 112                 advf->finderflags = (u16)de_getu16be_p(&pos);
 113                 de_dbg(c, "finder flags: 0x%04x", (unsigned int)advf->finderflags);
 114         }
 115
 116         n = de_geti16be_p(&pos);
 117         n2 = de_geti16be_p(&pos);
 118         de_dbg(c, "position in window: %d,%d", (int)n2, (int)n);
 119
 120         n = de_getu16be_p(&pos);
 121         de_dbg(c, "window/folder id: %d", (int)n);
 122         de_dbg_indent(c, -1);
 123
 124         b = de_getbyte_p(&pos);
 125         de_dbg(c, "protected: 0x%02x", (unsigned int)b);
 126
 127         pos++;
 128
 129         d->dflen = de_getu32be_p(&pos);
 130         de_dbg(c, "data fork len: %u", (unsigned int)d->dflen);
 131         d->rflen = de_getu32be_p(&pos);
 132         de_dbg(c, "resource fork len: %u", (unsigned int)d->rflen);
 133
 134         n = de_getu32be_p(&pos);
 135         if(n==0) {
 136                 d->create_time.is_valid = 0;
 137                 de_strlcpy(timestamp_buf, "unknown", sizeof(timestamp_buf));
 138         }
 139         else {
 140                 de_mac_time_to_timestamp(n, &d->create_time);
 141                 d->create_time.tzcode = DE_TZCODE_LOCAL;
 142                 de_timestamp_to_string(&d->create_time, timestamp_buf, sizeof(timestamp_buf), 0);
 143         }
 144         de_dbg(c, "create date: %"I64_FMT" (%s)", n, timestamp_buf);
 145
 146         mod_time_raw = de_getu32be_p(&pos);
 147         if(mod_time_raw==0) {
 148                 d->mod_time.is_valid = 0;
 149                 de_strlcpy(timestamp_buf, "unknown", sizeof(timestamp_buf));
 150         }
 151         else {
 152                 de_mac_time_to_timestamp(mod_time_raw, &d->mod_time);
 153                 d->mod_time.tzcode = DE_TZCODE_LOCAL;
 154                 de_timestamp_to_string(&d->mod_time, timestamp_buf, sizeof(timestamp_buf), 0);
 155         }
 156         de_dbg(c, "mod date: %"I64_FMT" (%s)", mod_time_raw, timestamp_buf);
 157
 158         pos += 2; // length of Get Info comment
 159
 160         if(d->is_v23) {
 161                 u8 fflags_lobyte;
 162
 163                 fflags_lobyte = de_getbyte(pos);
 164                 de_dbg(c, "finder flags (low byte): 0x__%02x", (unsigned int)fflags_lobyte);
 165                 advf->finderflags |= (u16)fflags_lobyte;
 166         }
 167         pos += 1;
 168
 169         pos += 14; // unused
 170         pos += 4; // unpacked total length
 171
 172         if(d->is_v23) {
 173                 n = de_getu16be(pos);
 174                 de_dbg(c, "length of secondary header: %u", (unsigned int)n);
 175         }
 176         pos += 2;
 177
 178         pos += 1; // version number, already read
 179         pos += 1; // version number, already read
 180
 181         crc_reported = (u32)de_getu16be_p(&pos);
 182         if(d->is_v23 || crc_reported!=0) {
 183                 struct de_crcobj *crco;
 184
 185                 de_dbg(c, "crc of header (reported%s): 0x%04x",
 186                         (d->is_v23)?"":", hypothetical", (unsigned int)crc_reported);
 187                 crco = de_crcobj_create(c, DE_CRCOBJ_CRC16_XMODEM);
 188                 de_crcobj_addslice(crco, c->infile, 0, 124);
 189                 crc_calc = de_crcobj_getval(crco);
 190                 de_crcobj_destroy(crco);
 191                 de_dbg(c, "crc of header (calculated): 0x%04x", (unsigned int)crc_calc);
 192
 193                 if(d->is_v23 && crc_reported!=0 && crc_calc!=crc_reported) {
 194                         de_warn(c, "MacBinary header CRC check failed");
 195                 }
 196         }
 197
 198         pos += 2; // Reserved for computer type and OS ID
 199 done:
 200         ;
 201 }
 202
 203 // If a fork is going to be extracted, call this to set up some things.
 204 // Caller must first set advfki->fork_len, among other things.
 205 // Sets fki->extract_error_flag if there was a problem that would prevent the
 206 // fork from being extracted.
 207 static void do_prepare_one_fork(deark *c, lctx *d, struct de_advfile_forkinfo *advfki,
 208         struct fork_info *fki)
 209 {
 210         de_dbg(c, "%s fork at %"I64_FMT", len=%"I64_FMT, fork_name(fki->is_rsrc, 0),
 211                 fki->pos, advfki->fork_len);
 212
 213         if(fki->pos+advfki->fork_len>c->infile->len) {
 214                 de_err(c, "%s fork at %"I64_FMT" goes beyond end of file.",
 215                         fork_name(fki->is_rsrc, 1), fki->pos);
 216                 if(fki->pos+advfki->fork_len > c->infile->len+1024) {
 217                         fki->extract_error_flag = 1;
 218                         goto done;
 219                 }
 220         }
 221
 222 done:
 223         ;
 224 }
 225
 226 static int my_advfile_cbfn(deark *c, struct de_advfile *advf,
 227         struct de_advfile_cbparams *afp)
 228 {
 229         struct extract_ctx *ectx = (struct extract_ctx*)advf->userdata;
 230
 231         if(afp->whattodo == DE_ADVFILE_WRITEMAIN) {
 232                 dbuf_copy(c->infile, ectx->fki_data.pos, advf->mainfork.fork_len, afp->outf);
 233         }
 234         else if(afp->whattodo == DE_ADVFILE_WRITERSRC) {
 235                 dbuf_copy(c->infile, ectx->fki_rsrc.pos, advf->rsrcfork.fork_len, afp->outf);
 236         }
 237         return 1;
 238 }
 239
 240 static void run_macbinary_internal(deark *c, lctx *d)
 241 {
 242         i64 pos = 128;
 243         struct de_advfile *advf = NULL;
 244         struct extract_ctx *ectx = NULL;
 245
 246         ectx = de_malloc(c, sizeof(struct extract_ctx));
 247         advf = de_advfile_create(c);
 248
 249         do_header(c, d, advf);
 250         if(d->filename_srd && ucstring_isnonempty(d->filename_srd->str)) {
 251                 ucstring_append_ucstring(advf->filename, d->filename_srd->str);
 252                 advf->original_filename_flag = 1;
 253         }
 254         if(d->filename_srd) {
 255                 de_advfile_set_orig_filename(advf, d->filename_srd->sz,
 256                         d->filename_srd->sz_strlen);
 257         }
 258         advf->mainfork.fi->timestamp[DE_TIMESTAMPIDX_MODIFY] = d->mod_time;
 259         advf->mainfork.fi->timestamp[DE_TIMESTAMPIDX_CREATE] = d->create_time;
 260
 261         if(d->dflen>0) {
 262                 d->dfpos = pos;
 263                 ectx->fki_data.pos = d->dfpos;
 264                 advf->mainfork.fork_len = d->dflen;
 265
 266                 if(d->extract_files) {
 267                         do_prepare_one_fork(c, d, &advf->mainfork, &ectx->fki_data);
 268                         if(!ectx->fki_data.extract_error_flag) {
 269                                 advf->mainfork.fork_exists = 1;
 270                         }
 271                 }
 272
 273                 pos += de_pad_to_n(d->dflen, 128);
 274         }
 275
 276         if(d->rflen>0) {
 277                 d->rfpos = pos;
 278
 279                 ectx->fki_rsrc.is_rsrc = 1;
 280                 ectx->fki_rsrc.pos = d->rfpos;
 281                 advf->rsrcfork.fork_len = d->rflen;
 282
 283                 if(d->extract_files) {
 284                         do_prepare_one_fork(c, d, &advf->rsrcfork, &ectx->fki_rsrc);
 285                         if(!ectx->fki_rsrc.extract_error_flag) {
 286                                 advf->rsrcfork.fork_exists = 1;
 287                         }
 288                 }
 289         }
 290
 291         if(d->extract_files) {
 292                 advf->userdata = (void*)ectx;
 293                 advf->writefork_cbfn = my_advfile_cbfn;
 294                 de_advfile_run(advf);
 295         }
 296
 297         de_advfile_destroy(advf);
 298         de_free(c, ectx);
 299 }
 300
 301 static void de_run_macbinary(deark *c, de_module_params *mparams)
 302 {
 303         lctx *d = NULL;
 304
 305         d = de_malloc(c, sizeof(lctx));
 306
 307         d->extract_files = 1;
 308         if(de_havemodcode(c, mparams, 'D')) {
 309                 d->extract_files = 0;
 310         }
 311
 312         run_macbinary_internal(c, d);
 313
 314         if(mparams) {
 315                 mparams->out_params.uint1 = (u32)d->dfpos;
 316                 mparams->out_params.uint2 = (u32)d->dflen;
 317                 mparams->out_params.uint3 = (u32)d->rfpos;
 318                 mparams->out_params.uint4 = (u32)d->rflen;
 319
 320                 if(mparams->out_params.fi) {
 321                         // If caller created out_params.fi for us, save the mod time to it.
 322                         mparams->out_params.fi->timestamp[DE_TIMESTAMPIDX_MODIFY] = d->mod_time;
 323
 324                         // If caller created .fi->name_other, copy the filename to it.
 325                         if(d->filename_srd && d->filename_srd->str->len>0 && mparams->out_params.fi->name_other) {
 326                                 ucstring_append_ucstring(mparams->out_params.fi->name_other, d->filename_srd->str);
 327                         }
 328                 }
 329         }
 330
 331         if(d) {
 332                 de_destroy_stringreaderdata(c, d->filename_srd);
 333                 de_free(c, d);
 334         }
 335 }
 336
 337 // Detecting MacBinary format is important, but also very difficult.
 338 // Note: This must be coordinated with the macpaint detection routine.
 339 // It should never set the confidence to 100, because macpaint and
 340 // maybe other formats need to be able to have higher confidence.
 341 static int de_identify_macbinary(deark *c)
 342 {
 343         int conf = 0;
 344         int k;
 345         int has_sig;
 346         int is_v23 = 0; // v2 or v3
 347         int good_file_len = 0;
 348         int good_cc = 0;
 349         int bad_crc = 0;
 350         i64 n;
 351         i64 dflen, rflen;
 352         i64 min_expected_len;
 353         u32 crc_reported, crc_calc;
 354         u8 b[128];
 355
 356         // "old" version number is always 0.
 357         b[0] = de_getbyte(0);
 358         if(b[0]!=0) goto done;
 359
 360         // filename length
 361         b[1] = de_getbyte(1);
 362         if(b[1]<1 || b[1]>63) goto done;
 363
 364         de_read(&b[2], 2, sizeof(b)-2);
 365
 366         if(b[2]==0) goto done; // First filename byte
 367         if(b[74]!=0) goto done;
 368         if(b[82]!=0) goto done;
 369
 370         // Extended version number
 371         if(b[122]==129 && b[123]==129) {
 372                 // v2
 373                 is_v23 = 1;
 374         }
 375         else if(b[122]==130 && (b[123]==129 || b[123]==130)) {
 376                 // v3
 377                 is_v23 = 1;
 378         }
 379         // else v1.
 380         // Some v1 files have garbage in the last 29 bytes of the file,
 381         // so we can't assume the extended version number is 0.
 382
 383         // Ver.III signature, but possibly used in some files that have earlier
 384         // version numbers.
 385         has_sig = !de_memcmp(&b[102], (const void*)"mBIN", 4);
 386         if(has_sig) {
 387                 conf = 90;
 388                 goto done;
 389         }
 390
 391         // Check if filename characters are sensible
 392         for(k=0; k<(int)b[1]; k++) {
 393                 if(b[2+k]>0 && b[2+k]<32) goto done;
 394         }
 395
 396         // File type code. Expect ASCII.
 397         good_cc = 1;
 398         for(k=65; k<=68; k++) {
 399                 if(b[k]<32 || b[k]>127) good_cc = 0;
 400         }
 401
 402         dflen = de_getu32be_direct(&b[83]);
 403         rflen = de_getu32be_direct(&b[87]);
 404
 405         crc_reported = (u32)de_getu16be_direct(&b[124]);
 406
 407         // Check the file size.
 408
 409         // Resource forks that go beyond the end of file are too common to
 410         // disallow.
 411         if(128 + dflen > c->infile->len) goto done;
 412         if(128 + rflen + dflen > c->infile->len + 4096) goto done;
 413
 414         if(rflen>0) {
 415                 min_expected_len = 128 + de_pad_to_n(dflen, 128) + rflen;
 416         }
 417         else {
 418                 min_expected_len = 128 + dflen;
 419         }
 420
 421         // The file size really should be exactly min_expected_len, or that
 422         // number padded to the next multiple of 128. But I'm not bold
 423         // enough to require it.
 424         if((c->infile->len == min_expected_len) ||
 425                 (c->infile->len == de_pad_to_n(min_expected_len, 128)))
 426         {
 427                 good_file_len = 1;
 428         }
 429
 430         if(is_v23) {
 431                 // Most MacBinary II specific checks go here
 432
 433                 if(!de_is_all_zeroes(&b[102], 14)) {
 434                         if(!good_file_len) goto done;
 435                 }
 436
 437                 // Secondary header length. We don't support this.
 438                 n = de_getu16be_direct(&b[120]);
 439                 if(n!=0) goto done;
 440         }
 441         else {
 442                 // Most Original MacBinary format checks go here
 443
 444                 // An empty file is not illegal, but we need as many checks as possible
 445                 // that won't be passed by all 0 bytes.
 446                 if(dflen==0 && rflen==0) goto done;
 447
 448                 // Unused fields in this version should be all 0, though we'll allow
 449                 // nonzero values in some cases.
 450                 if(!de_is_all_zeroes(&b[99], 25)) {
 451                         if(!good_file_len) goto done;
 452                 }
 453         }
 454
 455         if(crc_reported!=0 || is_v23) {
 456                 struct de_crcobj *crco;
 457
 458                 crco = de_crcobj_create(c, DE_CRCOBJ_CRC16_XMODEM);
 459                 de_crcobj_addbuf(crco, b, 124);
 460                 crc_calc = de_crcobj_getval(crco);
 461                 de_crcobj_destroy(crco);
 462                 if(crc_calc!=crc_reported && is_v23 && crc_reported!=0) {
 463                         bad_crc = 1;
 464                 }
 465         }
 466
 467         if(is_v23 && good_file_len && good_cc) {
 468                 if(bad_crc) {
 469                         conf = 19;
 470                 }
 471                 else {
 472                         conf = 90;
 473                 }
 474         }
 475         else if(bad_crc) {
 476                 goto done;
 477         }
 478         else if(is_v23) {
 479                 conf = 49;
 480         }
 481         else if(good_cc) {
 482                 conf = 29;
 483         }
 484         else {
 485                 conf = 19;
 486         }
 487
 488 done:
 489         if(conf>0) {
 490                 c->detection_data->is_macbinary = 1;
 491         }
 492         return conf;
 493 }
 494
 495 void de_module_macbinary(deark *c, struct deark_module_info *mi)
 496 {
 497         mi->id = "macbinary";
 498         mi->desc = "MacBinary";
 499         mi->run_fn = de_run_macbinary;
 500         mi->identify_fn = de_identify_macbinary;
 501         mi->flags = DE_MODFLAG_SHAREDDETECTION;
 502 }