modules/pif.c

   1 // This file is part of Deark.
   2 // Copyright (C) 2022 Jason Summers
   3 // See the file COPYING for terms of use.
   4
   5 // Windows Program Information File (.PIF)
   6 // DESQview Program Information File (.DVP)
   7
   8 #include <deark-private.h>
   9 DE_DECLARE_MODULE(de_module_pif);
  10 DE_DECLARE_MODULE(de_module_desqview_dvp);
  11
  12 #define PIF_BASIC_SECTION_SIZE 369
  13 #define PIF_INVALID_HEADING_POS 0xffff
  14
  15 struct pif_ctx {
  16         de_encoding input_encoding_oem;
  17         de_encoding input_encoding_ansi;
  18         de_ucstring *tmpstr;
  19         i64 next_section_heading_pos;
  20         struct de_inthashtable *pos_seen;
  21         UI checksum_calc;
  22 };
  23
  24 static int pif_validate_pos(deark *c, struct pif_ctx *d, i64 pos)
  25 {
  26         if(de_inthashtable_add_item(c, d->pos_seen, pos, NULL)) {
  27                 return 1;
  28         }
  29         de_err(c, "Bad offset detected");
  30         return 0;
  31 }
  32
  33 static void do_pif_section_extract(deark *c, struct pif_ctx *d, i64 pos1, i64 len, const char *name)
  34 {
  35         dbuf_create_file_from_slice(c->infile, pos1, len, name, NULL, 0);
  36 }
  37
  38 static void do_pif_section_default(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
  39 {
  40         de_dbg_hexdump(c, c->infile, pos1, len, 256, NULL, 0x1);
  41 }
  42
  43 static void do_pif_section_basic(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
  44 {
  45         i64 pos = pos1;
  46         i64 n;
  47         UI checksum_reported;
  48
  49         pos++; // unused
  50
  51         checksum_reported = (UI)de_getbyte_p(&pos);
  52         de_dbg(c, "checksum (reported): 0x%02x", checksum_reported);
  53         d->checksum_calc = (UI)de_calccrc_oneshot(c->infile, 2, PIF_BASIC_SECTION_SIZE-2,
  54                 DE_CRCOBJ_SUM_BYTES);
  55         d->checksum_calc &= 0xff;
  56         // Note - Not all files set the checksum field. Often it's just set to 0x78,
  57         // but other wrong values are common.
  58         de_dbg(c, "checksum (calculated): 0x%02x", d->checksum_calc);
  59
  60         ucstring_empty(d->tmpstr);
  61         dbuf_read_to_ucstring(c->infile, pos, 30, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
  62                 d->input_encoding_oem);
  63         ucstring_strip_trailing_spaces(d->tmpstr);
  64         de_dbg(c, "title: \"%s\"", ucstring_getpsz_d(d->tmpstr));
  65         pos += 30;
  66
  67         n = de_getu16le_p(&pos);
  68         de_dbg(c, "max conventional mem: %"I64_FMT" kb", n);
  69         n = de_getu16le_p(&pos);
  70         de_dbg(c, "min conventional mem: %"I64_FMT" kb", n);
  71
  72         ucstring_empty(d->tmpstr);
  73         dbuf_read_to_ucstring(c->infile, pos, 63, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
  74                 d->input_encoding_oem);
  75         ucstring_strip_trailing_spaces(d->tmpstr);
  76         de_dbg(c, "target filename: \"%s\"", ucstring_getpsz_d(d->tmpstr));
  77         pos += 63;
  78
  79         // TODO: There's disagreement about what the next 2 bytes are.
  80         n = de_getu16le_p(&pos);
  81         de_dbg(c, "flags1: 0x%04x", (UI)n);
  82
  83         ucstring_empty(d->tmpstr);
  84         dbuf_read_to_ucstring(c->infile, pos, 64, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
  85                 d->input_encoding_oem);
  86         ucstring_strip_trailing_spaces(d->tmpstr);
  87         de_dbg(c, "work dir: \"%s\"", ucstring_getpsz_d(d->tmpstr));
  88         pos += 64;
  89
  90         ucstring_empty(d->tmpstr);
  91         dbuf_read_to_ucstring(c->infile, pos, 64, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
  92                 d->input_encoding_oem);
  93         ucstring_strip_trailing_spaces(d->tmpstr);
  94         de_dbg(c, "params: \"%s\"", ucstring_getpsz_d(d->tmpstr));
  95         pos += 64;
  96
  97         // TODO: More fields
  98 }
  99
 100 static void do_pif_section_win286(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
 101 {
 102         i64 pos = pos1;
 103         i64 n;
 104         UI flags;
 105
 106         if(len<6) return;
 107         n = de_getu16le_p(&pos);
 108         de_dbg(c, "max XMS: %"I64_FMT, n);
 109         n = de_getu16le_p(&pos);
 110         de_dbg(c, "min XMS: %"I64_FMT, n);
 111         flags = (UI)de_getu16le_p(&pos);
 112         de_dbg(c, "flags: 0x%04x", (UI)flags);
 113 }
 114
 115 static void do_pif_section_win386(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
 116 {
 117         i64 pos = pos1;
 118         i64 n;
 119         UI flags;
 120
 121         if(len<104) return;
 122         n = de_getu16le_p(&pos);
 123         de_dbg(c, "max conventional mem: %"I64_FMT, n);
 124         n = de_getu16le_p(&pos);
 125         de_dbg(c, "min conventional mem: %"I64_FMT, n);
 126         pos += 2; // fg priority
 127         pos += 2; // bg priority
 128         n = de_getu16le_p(&pos);
 129         de_dbg(c, "max EMS: %"I64_FMT, n);
 130         n = de_getu16le_p(&pos);
 131         de_dbg(c, "min EMS: %"I64_FMT, n);
 132         n = de_getu16le_p(&pos);
 133         de_dbg(c, "max XMS: %"I64_FMT, n);
 134         n = de_getu16le_p(&pos);
 135         de_dbg(c, "min XMS: %"I64_FMT, n);
 136         flags = (UI)de_getu32le_p(&pos);
 137         de_dbg(c, "flags1: 0x%08x", (UI)flags);
 138         flags = (UI)de_getu16le_p(&pos);
 139         de_dbg(c, "flags2: 0x%04x", (UI)flags);
 140
 141         // TODO: More fields
 142
 143         pos = pos1 + 40;
 144         ucstring_empty(d->tmpstr);
 145         // TODO: There are questions about whether this is OEM or ANSI
 146         dbuf_read_to_ucstring(c->infile, pos, 64, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 147                 d->input_encoding_oem);
 148         ucstring_strip_trailing_spaces(d->tmpstr);
 149         de_dbg(c, "params: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 150 }
 151
 152 static void do_pif_section_winvmm(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
 153 {
 154         i64 pos = pos1;
 155         i64 n;
 156
 157         if(len<428) return;
 158         pos += 88;
 159
 160         ucstring_empty(d->tmpstr);
 161         dbuf_read_to_ucstring(c->infile, pos, 80, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 162                 d->input_encoding_ansi);
 163         de_dbg(c, "icon file: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 164         pos += 80;
 165
 166         n = de_getu16le_p(&pos);
 167         de_dbg(c, "icon #: %"I64_FMT, n);
 168
 169         pos = pos1 + 234;
 170         ucstring_empty(d->tmpstr);
 171         dbuf_read_to_ucstring(c->infile, pos, 32, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 172                 d->input_encoding_ansi);
 173         de_dbg(c, "raster font: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 174         pos += 32;
 175
 176         ucstring_empty(d->tmpstr);
 177         dbuf_read_to_ucstring(c->infile, pos, 32, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 178                 d->input_encoding_ansi);
 179         de_dbg(c, "TrueType font: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 180
 181         pos = pos1 + 342;
 182         ucstring_empty(d->tmpstr);
 183         dbuf_read_to_ucstring(c->infile, pos, 80, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 184                 d->input_encoding_oem);
 185         de_dbg(c, "BAT file: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 186
 187         // TODO: More fields
 188 }
 189
 190 static void do_pif_section_winnt31(deark *c, struct pif_ctx *d, i64 pos1, i64 len)
 191 {
 192         i64 pos = pos1;
 193
 194         if(len<140) return;
 195         pos += 12;
 196
 197         ucstring_empty(d->tmpstr);
 198         dbuf_read_to_ucstring(c->infile, pos, 64, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 199                 d->input_encoding_ansi);
 200         de_dbg(c, "alt config.sys: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 201         pos += 64;
 202
 203         ucstring_empty(d->tmpstr);
 204         dbuf_read_to_ucstring(c->infile, pos, 64, d->tmpstr, DE_CONVFLAG_STOP_AT_NUL,
 205                 d->input_encoding_ansi);
 206         de_dbg(c, "alt autoexec.bat: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 207 }
 208
 209 // Returns nonzero if we should look for more sections after this.
 210 // Sets d->next_section_heading_pos.
 211 static int do_pif_section(deark *c, struct pif_ctx *d, i64 pos1)
 212 {
 213         int saved_indent_level;
 214         int retval = 0;
 215         i64 pos = pos1;
 216         i64 dpos;
 217         i64 dlen;
 218         struct de_stringreaderdata *secname = NULL;
 219
 220         de_dbg_indent_save(c, &saved_indent_level);
 221         de_dbg(c, "section at %"I64_FMT, pos1);
 222         de_dbg_indent(c, 1);
 223
 224         secname = dbuf_read_string(c->infile, pos,
 225                 16, 16, DE_CONVFLAG_STOP_AT_NUL, d->input_encoding_ansi);
 226         de_dbg(c, "name: \"%s\"", ucstring_getpsz_d(secname->str));
 227         pos += 16;
 228
 229         d->next_section_heading_pos = de_getu16le_p(&pos);
 230         de_dbg(c, "next section pos: %"I64_FMT, d->next_section_heading_pos);
 231         dpos = de_getu16le_p(&pos);
 232         de_dbg(c, "data pos: %"I64_FMT, dpos);
 233         dlen = de_getu16le_p(&pos);
 234         de_dbg(c, "data len: %"I64_FMT, dlen);
 235         if(d->next_section_heading_pos != PIF_INVALID_HEADING_POS) {
 236                 retval = 1;
 237         }
 238
 239         if(dlen != 0) {
 240                 if(!pif_validate_pos(c, d, dpos)) goto done;
 241         }
 242
 243         de_dbg(c, "section data at %"I64_FMT", len=%"I64_FMT, dpos, dlen);
 244         de_dbg_indent(c, 1);
 245         if(!de_strcmp(secname->sz, "MICROSOFT PIFEX")) {
 246                 do_pif_section_basic(c, d, dpos, dlen);
 247         }
 248         else if(!de_strcmp(secname->sz, "WINDOWS 286 3.0")) {
 249                 do_pif_section_win286(c, d, dpos, dlen);
 250         }
 251         else if(!de_strcmp(secname->sz, "WINDOWS 386 3.0")) {
 252                 do_pif_section_win386(c, d, dpos, dlen);
 253         }
 254         else if(!de_strcmp(secname->sz, "WINDOWS VMM 4.0")) {
 255                 do_pif_section_winvmm(c, d, dpos, dlen);
 256         }
 257         else if(!de_strcmp(secname->sz, "WINDOWS NT  3.1")) {
 258                 do_pif_section_winnt31(c, d, pos, dlen);
 259         }
 260         else if(!de_strcmp(secname->sz, "AUTOEXECBAT 4.0")) {
 261                 do_pif_section_extract(c, d, dpos, dlen, "autoexec.bat");
 262         }
 263         else if(!de_strcmp(secname->sz, "CONFIG  SYS 4.0")) {
 264                 do_pif_section_extract(c, d, dpos, dlen, "config.sys");
 265         }
 266         else {
 267                 do_pif_section_default(c, d, dpos, dlen);
 268         }
 269         // TODO:
 270         //  WINDOWS NT  4.0
 271         //  WINDOWS PIF.402
 272         //  WINDOWS PIF.403
 273         //  WINDOWS ICO.001
 274
 275 done:
 276         de_destroy_stringreaderdata(c, secname);
 277         de_dbg_indent_restore(c, saved_indent_level);
 278         return retval;
 279 }
 280
 281 static void do_pif_sections(deark *c, struct pif_ctx *d)
 282 {
 283         d->next_section_heading_pos = PIF_BASIC_SECTION_SIZE;
 284
 285         while(1) {
 286                 i64 this_section_heading_pos;
 287
 288                 this_section_heading_pos = d->next_section_heading_pos;
 289                 if(this_section_heading_pos == PIF_INVALID_HEADING_POS) break;
 290                 d->next_section_heading_pos = PIF_INVALID_HEADING_POS;
 291                 if(this_section_heading_pos+22 > c->infile->len) break;
 292                 if(!pif_validate_pos(c, d, this_section_heading_pos)) break;
 293
 294                 if(!do_pif_section(c, d, this_section_heading_pos)) break;
 295         }
 296 }
 297
 298 static void do_dvp_extensions(deark *c, struct pif_ctx *d)
 299 {
 300         i64 pos;
 301         u8 dvextver;
 302
 303         pos = PIF_BASIC_SECTION_SIZE;
 304         ucstring_empty(d->tmpstr);
 305         dbuf_read_to_ucstring(c->infile, pos, 2, d->tmpstr, 0, DE_ENCODING_ASCII);
 306         de_dbg(c, "keys: \"%s\"", ucstring_getpsz_d(d->tmpstr));
 307
 308         pos = PIF_BASIC_SECTION_SIZE + 13;
 309         dvextver = de_getbyte(pos);
 310         de_dbg(c, "DV extensions ver: %u", (UI)dvextver);
 311 }
 312
 313 static void do_pif_main(deark *c, de_module_params *mparams, int is_dvp)
 314 {
 315         const char *tmps;
 316         struct pif_ctx *d = NULL;
 317         int is_oldfmt = 0;
 318
 319         d = de_malloc(c, sizeof(struct pif_ctx));
 320         d->tmpstr = ucstring_create(c);
 321
 322         d->input_encoding_ansi = de_get_input_encoding(c, NULL, DE_ENCODING_WINDOWS1252);
 323
 324         d->input_encoding_oem = DE_ENCODING_CP437; // default
 325         tmps = de_get_ext_option(c, "oemenc");
 326         if(tmps) {
 327                 d->input_encoding_oem = de_encoding_name_to_code(tmps);
 328                 if(d->input_encoding_oem == DE_ENCODING_UNKNOWN) {
 329                         d->input_encoding_oem = DE_ENCODING_CP437;
 330                 }
 331         }
 332
 333         d->pos_seen = de_inthashtable_create(c);
 334
 335         if(!is_dvp && (c->infile->len < PIF_BASIC_SECTION_SIZE+22)) {
 336                 is_oldfmt = 1;
 337         }
 338
 339         if(is_oldfmt) {
 340                 do_pif_section_basic(c, d, 0, PIF_BASIC_SECTION_SIZE);
 341         }
 342         else if(is_dvp) {
 343                 do_pif_section_basic(c, d, 0, PIF_BASIC_SECTION_SIZE);
 344                 do_dvp_extensions(c, d);
 345         }
 346         else {
 347                 do_pif_sections(c, d);
 348         }
 349
 350         if(d) {
 351                 ucstring_destroy(d->tmpstr);
 352                 de_inthashtable_destroy(c, d->pos_seen);
 353                 de_free(c, d);
 354         }
 355 }
 356
 357 static void de_run_pif(deark *c, de_module_params *mparams)
 358 {
 359         do_pif_main(c, mparams, 0);
 360 }
 361
 362 static int de_identify_pif(deark *c)
 363 {
 364         int maybe_oldfmt = 0;
 365         int has_id = 0;
 366         int has_ext;
 367
 368         if(c->infile->len == PIF_BASIC_SECTION_SIZE) {
 369                 maybe_oldfmt = 1;
 370         }
 371         else if(c->infile->len >= PIF_BASIC_SECTION_SIZE+22) {
 372                 has_id = !dbuf_memcmp(c->infile, PIF_BASIC_SECTION_SIZE,
 373                         (const u8*)"MICROSOFT PIFEX\0", 16);
 374         }
 375
 376         if(!maybe_oldfmt && !has_id) return 0;
 377         has_ext = de_input_file_has_ext(c, "pif");
 378         if(maybe_oldfmt) {
 379                 return has_ext ? 24 : 0;
 380         }
 381         return has_ext ? 100 : 35;
 382 }
 383
 384 static void de_help_pif(deark *c)
 385 {
 386         de_msg(c, "-opt oemenc=... : The encoding for OEM Text items");
 387 }
 388
 389 void de_module_pif(deark *c, struct deark_module_info *mi)
 390 {
 391         mi->id = "pif";
 392         mi->desc = "Windows Program Information File";
 393         mi->identify_fn = de_identify_pif;
 394         mi->run_fn = de_run_pif;
 395         mi->help_fn = de_help_pif;
 396 }
 397
 398 static void de_run_desqview_dvp(deark *c, de_module_params *mparams)
 399 {
 400         do_pif_main(c, mparams, 1);
 401 }
 402
 403 static int de_identify_desqview_dvp(deark *c)
 404 {
 405         int has_ext;
 406
 407         if(c->infile->len!=416) return 0;
 408         if(de_getbyte(0) != 0x00) return 0;
 409         has_ext = de_input_file_has_ext(c, "dvp");
 410         if(has_ext) return 40;
 411         return 0;
 412 }
 413
 414 void de_module_desqview_dvp(deark *c, struct deark_module_info *mi)
 415 {
 416         mi->id = "desqview_dvp";
 417         mi->desc = "DESQview Program Information File";
 418         mi->identify_fn = de_identify_desqview_dvp;
 419         mi->run_fn = de_run_desqview_dvp;
 420 }