src/xz/list.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       list.c
   4 /// \brief      Listing information about .xz files
   5 //
   6 //  Author:     Lasse Collin
   7 //
   8 //  This file has been put into the public domain.
   9 //  You can do whatever you want with this file.
  10 //
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 #include "private.h"
  14 #include "tuklib_integer.h"
  15
  16
  17 /// Information about a .xz file
  18 typedef struct {
  19         /// Combined Index of all Streams in the file
  20         lzma_index *idx;
  21
  22         /// Total amount of Stream Padding
  23         uint64_t stream_padding;
  24
  25         /// Highest memory usage so far
  26         uint64_t memusage_max;
  27
  28         /// True if all Blocks so far have Compressed Size and
  29         /// Uncompressed Size fields
  30         bool all_have_sizes;
  31
  32 } xz_file_info;
  33
  34 #define XZ_FILE_INFO_INIT { NULL, 0, 0, true }
  35
  36
  37 /// Information about a .xz Block
  38 typedef struct {
  39         /// Size of the Block Header
  40         uint32_t header_size;
  41
  42         /// A few of the Block Flags as a string
  43         char flags[3];
  44
  45         /// Size of the Compressed Data field in the Block
  46         lzma_vli compressed_size;
  47
  48         /// Decoder memory usage for this Block
  49         uint64_t memusage;
  50
  51         /// The filter chain of this Block in human-readable form
  52         char filter_chain[FILTERS_STR_SIZE];
  53
  54 } block_header_info;
  55
  56
  57 /// Check ID to string mapping
  58 static const char check_names[LZMA_CHECK_ID_MAX + 1][12] = {
  59         // TRANSLATORS: Indicates that there is no integrity check.
  60         // This string is used in tables, so the width must not
  61         // exceed ten columns with a fixed-width font.
  62         N_("None"),
  63         "CRC32",
  64         // TRANSLATORS: Indicates that integrity check name is not known,
  65         // but the Check ID is known (here 2). This and other "Unknown-N"
  66         // strings are used in tables, so the width must not exceed ten
  67         // columns with a fixed-width font. It's OK to omit the dash if
  68         // you need space for one extra letter, but don't use spaces.
  69         N_("Unknown-2"),
  70         N_("Unknown-3"),
  71         "CRC64",
  72         N_("Unknown-5"),
  73         N_("Unknown-6"),
  74         N_("Unknown-7"),
  75         N_("Unknown-8"),
  76         N_("Unknown-9"),
  77         "SHA-256",
  78         N_("Unknown-11"),
  79         N_("Unknown-12"),
  80         N_("Unknown-13"),
  81         N_("Unknown-14"),
  82         N_("Unknown-15"),
  83 };
  84
  85 /// Buffer size for get_check_names(). This may be a bit ridiculous,
  86 /// but at least it's enough if some language needs many multibyte chars.
  87 #define CHECKS_STR_SIZE 1024
  88
  89
  90 /// Value of the Check field as hexadecimal string.
  91 /// This is set by parse_check_value().
  92 static char check_value[2 * LZMA_CHECK_SIZE_MAX + 1];
  93
  94
  95 /// Totals that are displayed if there was more than one file.
  96 /// The "files" counter is also used in print_info_adv() to show
  97 /// the file number.
  98 static struct {
  99         uint64_t files;
 100         uint64_t streams;
 101         uint64_t blocks;
 102         uint64_t compressed_size;
 103         uint64_t uncompressed_size;
 104         uint64_t stream_padding;
 105         uint64_t memusage_max;
 106         uint32_t checks;
 107         bool all_have_sizes;
 108 } totals = { 0, 0, 0, 0, 0, 0, 0, 0, true };
 109
 110
 111 /// \brief      Parse the Index(es) from the given .xz file
 112 ///
 113 /// \param      xfi     Pointer to structure where the decoded information
 114 ///                     is stored.
 115 /// \param      pair    Input file
 116 ///
 117 /// \return     On success, false is returned. On error, true is returned.
 118 ///
 119 // TODO: This function is pretty big. liblzma should have a function that
 120 // takes a callback function to parse the Index(es) from a .xz file to make
 121 // it easy for applications.
 122 static bool
 123 parse_indexes(xz_file_info *xfi, file_pair *pair)
 124 {
 125         if (pair->src_st.st_size <= 0) {
 126                 message_error(_("%s: File is empty"), pair->src_name);
 127                 return true;
 128         }
 129
 130         if (pair->src_st.st_size < 2 * LZMA_STREAM_HEADER_SIZE) {
 131                 message_error(_("%s: Too small to be a valid .xz file"),
 132                                 pair->src_name);
 133                 return true;
 134         }
 135
 136         io_buf buf;
 137         lzma_stream_flags header_flags;
 138         lzma_stream_flags footer_flags;
 139         lzma_ret ret;
 140
 141         // lzma_stream for the Index decoder
 142         lzma_stream strm = LZMA_STREAM_INIT;
 143
 144         // All Indexes decoded so far
 145         lzma_index *combined_index = NULL;
 146
 147         // The Index currently being decoded
 148         lzma_index *this_index = NULL;
 149
 150         // Current position in the file. We parse the file backwards so
 151         // initialize it to point to the end of the file.
 152         off_t pos = pair->src_st.st_size;
 153
 154         // Each loop iteration decodes one Index.
 155         do {
 156                 // Check that there is enough data left to contain at least
 157                 // the Stream Header and Stream Footer. This check cannot
 158                 // fail in the first pass of this loop.
 159                 if (pos < 2 * LZMA_STREAM_HEADER_SIZE) {
 160                         message_error("%s: %s", pair->src_name,
 161                                         message_strm(LZMA_DATA_ERROR));
 162                         goto error;
 163                 }
 164
 165                 pos -= LZMA_STREAM_HEADER_SIZE;
 166                 lzma_vli stream_padding = 0;
 167
 168                 // Locate the Stream Footer. There may be Stream Padding which
 169                 // we must skip when reading backwards.
 170                 while (true) {
 171                         if (pos < LZMA_STREAM_HEADER_SIZE) {
 172                                 message_error("%s: %s", pair->src_name,
 173                                                 message_strm(
 174                                                         LZMA_DATA_ERROR));
 175                                 goto error;
 176                         }
 177
 178                         if (io_pread(pair, &buf,
 179                                         LZMA_STREAM_HEADER_SIZE, pos))
 180                                 goto error;
 181
 182                         // Stream Padding is always a multiple of four bytes.
 183                         int i = 2;
 184                         if (buf.u32[i] != 0)
 185                                 break;
 186
 187                         // To avoid calling io_pread() for every four bytes
 188                         // of Stream Padding, take advantage that we read
 189                         // 12 bytes (LZMA_STREAM_HEADER_SIZE) already and
 190                         // check them too before calling io_pread() again.
 191                         do {
 192                                 stream_padding += 4;
 193                                 pos -= 4;
 194                                 --i;
 195                         } while (i >= 0 && buf.u32[i] == 0);
 196                 }
 197
 198                 // Decode the Stream Footer.
 199                 ret = lzma_stream_footer_decode(&footer_flags, buf.u8);
 200                 if (ret != LZMA_OK) {
 201                         message_error("%s: %s", pair->src_name,
 202                                         message_strm(ret));
 203                         goto error;
 204                 }
 205
 206                 // Check that the size of the Index field looks sane.
 207                 lzma_vli index_size = footer_flags.backward_size;
 208                 if ((lzma_vli)(pos) < index_size + LZMA_STREAM_HEADER_SIZE) {
 209                         message_error("%s: %s", pair->src_name,
 210                                         message_strm(LZMA_DATA_ERROR));
 211                         goto error;
 212                 }
 213
 214                 // Set pos to the beginning of the Index.
 215                 pos -= index_size;
 216
 217                 // See how much memory we can use for decoding this Index.
 218                 uint64_t memlimit = hardware_memlimit_get(MODE_LIST);
 219                 uint64_t memused = 0;
 220                 if (combined_index != NULL) {
 221                         memused = lzma_index_memused(combined_index);
 222                         if (memused > memlimit)
 223                                 message_bug();
 224
 225                         memlimit -= memused;
 226                 }
 227
 228                 // Decode the Index.
 229                 ret = lzma_index_decoder(&strm, &this_index, memlimit);
 230                 if (ret != LZMA_OK) {
 231                         message_error("%s: %s", pair->src_name,
 232                                         message_strm(ret));
 233                         goto error;
 234                 }
 235
 236                 do {
 237                         // Don't give the decoder more input than the
 238                         // Index size.
 239                         strm.avail_in = my_min(IO_BUFFER_SIZE, index_size);
 240                         if (io_pread(pair, &buf, strm.avail_in, pos))
 241                                 goto error;
 242
 243                         pos += strm.avail_in;
 244                         index_size -= strm.avail_in;
 245
 246                         strm.next_in = buf.u8;
 247                         ret = lzma_code(&strm, LZMA_RUN);
 248
 249                 } while (ret == LZMA_OK);
 250
 251                 // If the decoding seems to be successful, check also that
 252                 // the Index decoder consumed as much input as indicated
 253                 // by the Backward Size field.
 254                 if (ret == LZMA_STREAM_END)
 255                         if (index_size != 0 || strm.avail_in != 0)
 256                                 ret = LZMA_DATA_ERROR;
 257
 258                 if (ret != LZMA_STREAM_END) {
 259                         // LZMA_BUFFER_ERROR means that the Index decoder
 260                         // would have liked more input than what the Index
 261                         // size should be according to Stream Footer.
 262                         // The message for LZMA_DATA_ERROR makes more
 263                         // sense in that case.
 264                         if (ret == LZMA_BUF_ERROR)
 265                                 ret = LZMA_DATA_ERROR;
 266
 267                         message_error("%s: %s", pair->src_name,
 268                                         message_strm(ret));
 269
 270                         // If the error was too low memory usage limit,
 271                         // show also how much memory would have been needed.
 272                         if (ret == LZMA_MEMLIMIT_ERROR) {
 273                                 uint64_t needed = lzma_memusage(&strm);
 274                                 if (UINT64_MAX - needed < memused)
 275                                         needed = UINT64_MAX;
 276                                 else
 277                                         needed += memused;
 278
 279                                 message_mem_needed(V_ERROR, needed);
 280                         }
 281
 282                         goto error;
 283                 }
 284
 285                 // Decode the Stream Header and check that its Stream Flags
 286                 // match the Stream Footer.
 287                 pos -= footer_flags.backward_size + LZMA_STREAM_HEADER_SIZE;
 288                 if ((lzma_vli)(pos) < lzma_index_total_size(this_index)) {
 289                         message_error("%s: %s", pair->src_name,
 290                                         message_strm(LZMA_DATA_ERROR));
 291                         goto error;
 292                 }
 293
 294                 pos -= lzma_index_total_size(this_index);
 295                 if (io_pread(pair, &buf, LZMA_STREAM_HEADER_SIZE, pos))
 296                         goto error;
 297
 298                 ret = lzma_stream_header_decode(&header_flags, buf.u8);
 299                 if (ret != LZMA_OK) {
 300                         message_error("%s: %s", pair->src_name,
 301                                         message_strm(ret));
 302                         goto error;
 303                 }
 304
 305                 ret = lzma_stream_flags_compare(&header_flags, &footer_flags);
 306                 if (ret != LZMA_OK) {
 307                         message_error("%s: %s", pair->src_name,
 308                                         message_strm(ret));
 309                         goto error;
 310                 }
 311
 312                 // Store the decoded Stream Flags into this_index. This is
 313                 // needed so that we can print which Check is used in each
 314                 // Stream.
 315                 ret = lzma_index_stream_flags(this_index, &footer_flags);
 316                 if (ret != LZMA_OK)
 317                         message_bug();
 318
 319                 // Store also the size of the Stream Padding field. It is
 320                 // needed to show the offsets of the Streams correctly.
 321                 ret = lzma_index_stream_padding(this_index, stream_padding);
 322                 if (ret != LZMA_OK)
 323                         message_bug();
 324
 325                 if (combined_index != NULL) {
 326                         // Append the earlier decoded Indexes
 327                         // after this_index.
 328                         ret = lzma_index_cat(
 329                                         this_index, combined_index, NULL);
 330                         if (ret != LZMA_OK) {
 331                                 message_error("%s: %s", pair->src_name,
 332                                                 message_strm(ret));
 333                                 goto error;
 334                         }
 335                 }
 336
 337                 combined_index = this_index;
 338                 this_index = NULL;
 339
 340                 xfi->stream_padding += stream_padding;
 341
 342         } while (pos > 0);
 343
 344         lzma_end(&strm);
 345
 346         // All OK. Make combined_index available to the caller.
 347         xfi->idx = combined_index;
 348         return false;
 349
 350 error:
 351         // Something went wrong, free the allocated memory.
 352         lzma_end(&strm);
 353         lzma_index_end(combined_index, NULL);
 354         lzma_index_end(this_index, NULL);
 355         return true;
 356 }
 357
 358
 359 /// \brief      Parse the Block Header
 360 ///
 361 /// The result is stored into *bhi. The caller takes care of initializing it.
 362 ///
 363 /// \return     False on success, true on error.
 364 static bool
 365 parse_block_header(file_pair *pair, const lzma_index_iter *iter,
 366                 block_header_info *bhi, xz_file_info *xfi)
 367 {
 368 #if IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
 369 #       error IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
 370 #endif
 371
 372         // Get the whole Block Header with one read, but don't read past
 373         // the end of the Block (or even its Check field).
 374         const uint32_t size = my_min(iter->block.total_size
 375                                 - lzma_check_size(iter->stream.flags->check),
 376                         LZMA_BLOCK_HEADER_SIZE_MAX);
 377         io_buf buf;
 378         if (io_pread(pair, &buf, size, iter->block.compressed_file_offset))
 379                 return true;
 380
 381         // Zero would mean Index Indicator and thus not a valid Block.
 382         if (buf.u8[0] == 0)
 383                 goto data_error;
 384
 385         lzma_block block;
 386         lzma_filter filters[LZMA_FILTERS_MAX + 1];
 387
 388         // Initialize the pointers so that they can be passed to free().
 389         for (size_t i = 0; i < ARRAY_SIZE(filters); ++i)
 390                 filters[i].options = NULL;
 391
 392         // Initialize the block structure and decode Block Header Size.
 393         block.version = 0;
 394         block.check = iter->stream.flags->check;
 395         block.filters = filters;
 396
 397         block.header_size = lzma_block_header_size_decode(buf.u8[0]);
 398         if (block.header_size > size)
 399                 goto data_error;
 400
 401         // Decode the Block Header.
 402         switch (lzma_block_header_decode(&block, NULL, buf.u8)) {
 403         case LZMA_OK:
 404                 break;
 405
 406         case LZMA_OPTIONS_ERROR:
 407                 message_error("%s: %s", pair->src_name,
 408                                 message_strm(LZMA_OPTIONS_ERROR));
 409                 return true;
 410
 411         case LZMA_DATA_ERROR:
 412                 goto data_error;
 413
 414         default:
 415                 message_bug();
 416         }
 417
 418         // Check the Block Flags. These must be done before calling
 419         // lzma_block_compressed_size(), because it overwrites
 420         // block.compressed_size.
 421         bhi->flags[0] = block.compressed_size != LZMA_VLI_UNKNOWN
 422                         ? 'c' : '-';
 423         bhi->flags[1] = block.uncompressed_size != LZMA_VLI_UNKNOWN
 424                         ? 'u' : '-';
 425         bhi->flags[2] = '\0';
 426
 427         // Collect information if all Blocks have both Compressed Size
 428         // and Uncompressed Size fields. They can be useful e.g. for
 429         // multi-threaded decompression so it can be useful to know it.
 430         xfi->all_have_sizes &= block.compressed_size != LZMA_VLI_UNKNOWN
 431                         && block.uncompressed_size != LZMA_VLI_UNKNOWN;
 432
 433         // Validate or set block.compressed_size.
 434         switch (lzma_block_compressed_size(&block,
 435                         iter->block.unpadded_size)) {
 436         case LZMA_OK:
 437                 break;
 438
 439         case LZMA_DATA_ERROR:
 440                 goto data_error;
 441
 442         default:
 443                 message_bug();
 444         }
 445
 446         // Copy the known sizes.
 447         bhi->header_size = block.header_size;
 448         bhi->compressed_size = block.compressed_size;
 449
 450         // Calculate the decoder memory usage and update the maximum
 451         // memory usage of this Block.
 452         bhi->memusage = lzma_raw_decoder_memusage(filters);
 453         if (xfi->memusage_max < bhi->memusage)
 454                 xfi->memusage_max = bhi->memusage;
 455
 456         // Convert the filter chain to human readable form.
 457         message_filters_to_str(bhi->filter_chain, filters, false);
 458
 459         // Free the memory allocated by lzma_block_header_decode().
 460         for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 461                 free(filters[i].options);
 462
 463         return false;
 464
 465 data_error:
 466         // Show the error message.
 467         message_error("%s: %s", pair->src_name,
 468                         message_strm(LZMA_DATA_ERROR));
 469
 470         // Free the memory allocated by lzma_block_header_decode().
 471         // This is truly needed only if we get here after a succcessful
 472         // call to lzma_block_header_decode() but it doesn't hurt to
 473         // always do it.
 474         for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 475                 free(filters[i].options);
 476
 477         return true;
 478 }
 479
 480
 481 /// \brief      Parse the Check field and put it into check_value[]
 482 ///
 483 /// \return     False on success, true on error.
 484 static bool
 485 parse_check_value(file_pair *pair, const lzma_index_iter *iter)
 486 {
 487         // Don't read anything from the file if there is no integrity Check.
 488         if (iter->stream.flags->check == LZMA_CHECK_NONE) {
 489                 snprintf(check_value, sizeof(check_value), "---");
 490                 return false;
 491         }
 492
 493         // Locate and read the Check field.
 494         const uint32_t size = lzma_check_size(iter->stream.flags->check);
 495         const off_t offset = iter->block.compressed_file_offset
 496                         + iter->block.total_size - size;
 497         io_buf buf;
 498         if (io_pread(pair, &buf, size, offset))
 499                 return true;
 500
 501         // CRC32 and CRC64 are in little endian. Guess that all the future
 502         // 32-bit and 64-bit Check values are little endian too. It shouldn't
 503         // be a too big problem if this guess is wrong.
 504         if (size == 4)
 505                 snprintf(check_value, sizeof(check_value),
 506                                 "%08" PRIx32, conv32le(buf.u32[0]));
 507         else if (size == 8)
 508                 snprintf(check_value, sizeof(check_value),
 509                                 "%016" PRIx64, conv64le(buf.u64[0]));
 510         else
 511                 for (size_t i = 0; i < size; ++i)
 512                         snprintf(check_value + i * 2, 3, "%02x", buf.u8[i]);
 513
 514         return false;
 515 }
 516
 517
 518 /// \brief      Parse detailed information about a Block
 519 ///
 520 /// Since this requires seek(s), listing information about all Blocks can
 521 /// be slow.
 522 ///
 523 /// \param      pair    Input file
 524 /// \param      iter    Location of the Block whose Check value should
 525 ///                     be printed.
 526 /// \param      bhi     Pointer to structure where to store the information
 527 ///                     about the Block Header field.
 528 ///
 529 /// \return     False on success, true on error. If an error occurs,
 530 ///             the error message is printed too so the caller doesn't
 531 ///             need to worry about that.
 532 static bool
 533 parse_details(file_pair *pair, const lzma_index_iter *iter,
 534                 block_header_info *bhi, xz_file_info *xfi)
 535 {
 536         if (parse_block_header(pair, iter, bhi, xfi))
 537                 return true;
 538
 539         if (parse_check_value(pair, iter))
 540                 return true;
 541
 542         return false;
 543 }
 544
 545
 546 /// \brief      Get the compression ratio
 547 ///
 548 /// This has slightly different format than that is used in message.c.
 549 static const char *
 550 get_ratio(uint64_t compressed_size, uint64_t uncompressed_size)
 551 {
 552         if (uncompressed_size == 0)
 553                 return "---";
 554
 555         const double ratio = (double)(compressed_size)
 556                         / (double)(uncompressed_size);
 557         if (ratio > 9.999)
 558                 return "---";
 559
 560         static char buf[16];
 561         snprintf(buf, sizeof(buf), "%.3f", ratio);
 562         return buf;
 563 }
 564
 565
 566 /// \brief      Get a comma-separated list of Check names
 567 ///
 568 /// The check names are translated with gettext except when in robot mode.
 569 ///
 570 /// \param      buf     Buffer to hold the resulting string
 571 /// \param      checks  Bit mask of Checks to print
 572 /// \param      space_after_comma
 573 ///                     It's better to not use spaces in table-like listings,
 574 ///                     but in more verbose formats a space after a comma
 575 ///                     is good for readability.
 576 static void
 577 get_check_names(char buf[CHECKS_STR_SIZE],
 578                 uint32_t checks, bool space_after_comma)
 579 {
 580         assert(checks != 0);
 581
 582         char *pos = buf;
 583         size_t left = CHECKS_STR_SIZE;
 584
 585         const char *sep = space_after_comma ? ", " : ",";
 586         bool comma = false;
 587
 588         for (size_t i = 0; i <= LZMA_CHECK_ID_MAX; ++i) {
 589                 if (checks & (UINT32_C(1) << i)) {
 590                         my_snprintf(&pos, &left, "%s%s",
 591                                         comma ? sep : "",
 592                                         opt_robot ? check_names[i]
 593                                                 : _(check_names[i]));
 594                         comma = true;
 595                 }
 596         }
 597
 598         return;
 599 }
 600
 601
 602 static bool
 603 print_info_basic(const xz_file_info *xfi, file_pair *pair)
 604 {
 605         static bool headings_displayed = false;
 606         if (!headings_displayed) {
 607                 headings_displayed = true;
 608                 // TRANSLATORS: These are column headings. From Strms (Streams)
 609                 // to Ratio, the columns are right aligned. Check and Filename
 610                 // are left aligned. If you need longer words, it's OK to
 611                 // use two lines here. Test with "xz -l foo.xz".
 612                 puts(_("Strms  Blocks   Compressed Uncompressed  Ratio  "
 613                                 "Check   Filename"));
 614         }
 615
 616         char checks[CHECKS_STR_SIZE];
 617         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 618
 619         const char *cols[7] = {
 620                 uint64_to_str(lzma_index_stream_count(xfi->idx), 0),
 621                 uint64_to_str(lzma_index_block_count(xfi->idx), 1),
 622                 uint64_to_nicestr(lzma_index_file_size(xfi->idx),
 623                         NICESTR_B, NICESTR_TIB, false, 2),
 624                 uint64_to_nicestr(lzma_index_uncompressed_size(xfi->idx),
 625                         NICESTR_B, NICESTR_TIB, false, 3),
 626                 get_ratio(lzma_index_file_size(xfi->idx),
 627                         lzma_index_uncompressed_size(xfi->idx)),
 628                 checks,
 629                 pair->src_name,
 630         };
 631         printf("%*s %*s  %*s  %*s  %*s  %-*s %s\n",
 632                         tuklib_mbstr_fw(cols[0], 5), cols[0],
 633                         tuklib_mbstr_fw(cols[1], 7), cols[1],
 634                         tuklib_mbstr_fw(cols[2], 11), cols[2],
 635                         tuklib_mbstr_fw(cols[3], 11), cols[3],
 636                         tuklib_mbstr_fw(cols[4], 5), cols[4],
 637                         tuklib_mbstr_fw(cols[5], 7), cols[5],
 638                         cols[6]);
 639
 640         return false;
 641 }
 642
 643
 644 static void
 645 print_adv_helper(uint64_t stream_count, uint64_t block_count,
 646                 uint64_t compressed_size, uint64_t uncompressed_size,
 647                 uint32_t checks, uint64_t stream_padding)
 648 {
 649         char checks_str[CHECKS_STR_SIZE];
 650         get_check_names(checks_str, checks, true);
 651
 652         printf(_("  Streams:            %s\n"),
 653                         uint64_to_str(stream_count, 0));
 654         printf(_("  Blocks:             %s\n"),
 655                         uint64_to_str(block_count, 0));
 656         printf(_("  Compressed size:    %s\n"),
 657                         uint64_to_nicestr(compressed_size,
 658                                 NICESTR_B, NICESTR_TIB, true, 0));
 659         printf(_("  Uncompressed size:  %s\n"),
 660                         uint64_to_nicestr(uncompressed_size,
 661                                 NICESTR_B, NICESTR_TIB, true, 0));
 662         printf(_("  Ratio:              %s\n"),
 663                         get_ratio(compressed_size, uncompressed_size));
 664         printf(_("  Check:              %s\n"), checks_str);
 665         printf(_("  Stream padding:     %s\n"),
 666                         uint64_to_nicestr(stream_padding,
 667                                 NICESTR_B, NICESTR_TIB, true, 0));
 668         return;
 669 }
 670
 671
 672 static bool
 673 print_info_adv(xz_file_info *xfi, file_pair *pair)
 674 {
 675         // Print the overall information.
 676         print_adv_helper(lzma_index_stream_count(xfi->idx),
 677                         lzma_index_block_count(xfi->idx),
 678                         lzma_index_file_size(xfi->idx),
 679                         lzma_index_uncompressed_size(xfi->idx),
 680                         lzma_index_checks(xfi->idx),
 681                         xfi->stream_padding);
 682
 683         // Size of the biggest Check. This is used to calculate the width
 684         // of the CheckVal field. The table would get insanely wide if
 685         // we always reserved space for 64-byte Check (128 chars as hex).
 686         uint32_t check_max = 0;
 687
 688         // Print information about the Streams.
 689         //
 690         // TRANSLATORS: The second line is column headings. All except
 691         // Check are right aligned; Check is left aligned. Test with
 692         // "xz -lv foo.xz".
 693         puts(_("  Streams:\n    Stream    Blocks"
 694                         "      CompOffset    UncompOffset"
 695                         "        CompSize      UncompSize  Ratio"
 696                         "  Check      Padding"));
 697
 698         lzma_index_iter iter;
 699         lzma_index_iter_init(&iter, xfi->idx);
 700
 701         while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM)) {
 702                 const char *cols1[4] = {
 703                         uint64_to_str(iter.stream.number, 0),
 704                         uint64_to_str(iter.stream.block_count, 1),
 705                         uint64_to_str(iter.stream.compressed_offset, 2),
 706                         uint64_to_str(iter.stream.uncompressed_offset, 3),
 707                 };
 708                 printf("    %*s %*s %*s %*s ",
 709                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 710                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 711                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 712                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 713
 714                 const char *cols2[5] = {
 715                         uint64_to_str(iter.stream.compressed_size, 0),
 716                         uint64_to_str(iter.stream.uncompressed_size, 1),
 717                         get_ratio(iter.stream.compressed_size,
 718                                 iter.stream.uncompressed_size),
 719                         _(check_names[iter.stream.flags->check]),
 720                         uint64_to_str(iter.stream.padding, 2),
 721                 };
 722                 printf("%*s %*s  %*s  %-*s %*s\n",
 723                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 724                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 725                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 726                                 tuklib_mbstr_fw(cols2[3], 10), cols2[3],
 727                                 tuklib_mbstr_fw(cols2[4], 7), cols2[4]);
 728
 729                 // Update the maximum Check size.
 730                 if (lzma_check_size(iter.stream.flags->check) > check_max)
 731                         check_max = lzma_check_size(iter.stream.flags->check);
 732         }
 733
 734         // Cache the verbosity level to a local variable.
 735         const bool detailed = message_verbosity_get() >= V_DEBUG;
 736
 737         // Information collected from Block Headers
 738         block_header_info bhi;
 739
 740         // Print information about the Blocks but only if there is
 741         // at least one Block.
 742         if (lzma_index_block_count(xfi->idx) > 0) {
 743                 // Calculate the width of the CheckVal field.
 744                 const int checkval_width = my_max(8, 2 * check_max);
 745
 746                 // TRANSLATORS: The second line is column headings. All
 747                 // except Check are right aligned; Check is left aligned.
 748                 printf(_("  Blocks:\n    Stream     Block"
 749                         "      CompOffset    UncompOffset"
 750                         "       TotalSize      UncompSize  Ratio  Check"));
 751
 752                 if (detailed) {
 753                         // TRANSLATORS: These are additional column headings
 754                         // for the most verbose listing mode. CheckVal
 755                         // (Check value), Flags, and Filters are left aligned.
 756                         // Header (Block Header Size), CompSize, and MemUsage
 757                         // are right aligned. %*s is replaced with 0-120
 758                         // spaces to make the CheckVal column wide enough.
 759                         // Test with "xz -lvv foo.xz".
 760                         printf(_("      CheckVal %*s Header  Flags        "
 761                                         "CompSize    MemUsage  Filters"),
 762                                         checkval_width - 8, "");
 763                 }
 764
 765                 putchar('\n');
 766
 767                 lzma_index_iter_init(&iter, xfi->idx);
 768
 769                 // Iterate over the Blocks.
 770                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 771                         if (detailed && parse_details(pair, &iter, &bhi, xfi))
 772                                         return true;
 773
 774                         const char *cols1[4] = {
 775                                 uint64_to_str(iter.stream.number, 0),
 776                                 uint64_to_str(
 777                                         iter.block.number_in_stream, 1),
 778                                 uint64_to_str(
 779                                         iter.block.compressed_file_offset, 2),
 780                                 uint64_to_str(
 781                                         iter.block.uncompressed_file_offset, 3)
 782                         };
 783                         printf("    %*s %*s %*s %*s ",
 784                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 785                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 786                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 787                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 788
 789                         const char *cols2[4] = {
 790                                 uint64_to_str(iter.block.total_size, 0),
 791                                 uint64_to_str(iter.block.uncompressed_size,
 792                                                 1),
 793                                 get_ratio(iter.block.total_size,
 794                                         iter.block.uncompressed_size),
 795                                 _(check_names[iter.stream.flags->check])
 796                         };
 797                         printf("%*s %*s  %*s  %-*s",
 798                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 799                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 800                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 801                                 tuklib_mbstr_fw(cols2[3], detailed ? 11 : 1),
 802                                         cols2[3]);
 803
 804                         if (detailed) {
 805                                 const lzma_vli compressed_size
 806                                                 = iter.block.unpadded_size
 807                                                 - bhi.header_size
 808                                                 - lzma_check_size(
 809                                                 iter.stream.flags->check);
 810
 811                                 const char *cols3[6] = {
 812                                         check_value,
 813                                         uint64_to_str(bhi.header_size, 0),
 814                                         bhi.flags,
 815                                         uint64_to_str(compressed_size, 1),
 816                                         uint64_to_str(
 817                                                 round_up_to_mib(bhi.memusage),
 818                                                 2),
 819                                         bhi.filter_chain
 820                                 };
 821                                 // Show MiB for memory usage, because it
 822                                 // is the only size which is not in bytes.
 823                                 printf("%-*s  %*s  %-5s %*s %*s MiB  %s",
 824                                         checkval_width, cols3[0],
 825                                         tuklib_mbstr_fw(cols3[1], 6), cols3[1],
 826                                         cols3[2],
 827                                         tuklib_mbstr_fw(cols3[3], 15),
 828                                                 cols3[3],
 829                                         tuklib_mbstr_fw(cols3[4], 7), cols3[4],
 830                                         cols3[5]);
 831                         }
 832
 833                         putchar('\n');
 834                 }
 835         }
 836
 837         if (detailed) {
 838                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
 839                                 round_up_to_mib(xfi->memusage_max), 0));
 840                 printf(_("  Sizes in headers:   %s\n"),
 841                                 xfi->all_have_sizes ? _("Yes") : _("No"));
 842         }
 843
 844         return false;
 845 }
 846
 847
 848 static bool
 849 print_info_robot(xz_file_info *xfi, file_pair *pair)
 850 {
 851         char checks[CHECKS_STR_SIZE];
 852         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 853
 854         printf("name\t%s\n", pair->src_name);
 855
 856         printf("file\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 857                         "\t%s\t%s\t%" PRIu64 "\n",
 858                         lzma_index_stream_count(xfi->idx),
 859                         lzma_index_block_count(xfi->idx),
 860                         lzma_index_file_size(xfi->idx),
 861                         lzma_index_uncompressed_size(xfi->idx),
 862                         get_ratio(lzma_index_file_size(xfi->idx),
 863                                 lzma_index_uncompressed_size(xfi->idx)),
 864                         checks,
 865                         xfi->stream_padding);
 866
 867         if (message_verbosity_get() >= V_VERBOSE) {
 868                 lzma_index_iter iter;
 869                 lzma_index_iter_init(&iter, xfi->idx);
 870
 871                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM))
 872                         printf("stream\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 873                                 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 874                                 "\t%s\t%s\t%" PRIu64 "\n",
 875                                 iter.stream.number,
 876                                 iter.stream.block_count,
 877                                 iter.stream.compressed_offset,
 878                                 iter.stream.uncompressed_offset,
 879                                 iter.stream.compressed_size,
 880                                 iter.stream.uncompressed_size,
 881                                 get_ratio(iter.stream.compressed_size,
 882                                         iter.stream.uncompressed_size),
 883                                 check_names[iter.stream.flags->check],
 884                                 iter.stream.padding);
 885
 886                 lzma_index_iter_rewind(&iter);
 887                 block_header_info bhi;
 888
 889                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 890                         if (message_verbosity_get() >= V_DEBUG
 891                                         && parse_details(
 892                                                 pair, &iter, &bhi, xfi))
 893                                 return true;
 894
 895                         printf("block\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 896                                         "\t%" PRIu64 "\t%" PRIu64
 897                                         "\t%" PRIu64 "\t%" PRIu64 "\t%s\t%s",
 898                                         iter.stream.number,
 899                                         iter.block.number_in_stream,
 900                                         iter.block.number_in_file,
 901                                         iter.block.compressed_file_offset,
 902                                         iter.block.uncompressed_file_offset,
 903                                         iter.block.total_size,
 904                                         iter.block.uncompressed_size,
 905                                         get_ratio(iter.block.total_size,
 906                                                 iter.block.uncompressed_size),
 907                                         check_names[iter.stream.flags->check]);
 908
 909                         if (message_verbosity_get() >= V_DEBUG)
 910                                 printf("\t%s\t%" PRIu32 "\t%s\t%" PRIu64
 911                                                 "\t%" PRIu64 "\t%s",
 912                                                 check_value,
 913                                                 bhi.header_size,
 914                                                 bhi.flags,
 915                                                 bhi.compressed_size,
 916                                                 bhi.memusage,
 917                                                 bhi.filter_chain);
 918
 919                         putchar('\n');
 920                 }
 921         }
 922
 923         if (message_verbosity_get() >= V_DEBUG)
 924                 printf("summary\t%" PRIu64 "\t%s\n",
 925                                 xfi->memusage_max,
 926                                 xfi->all_have_sizes ? "yes" : "no");
 927
 928         return false;
 929 }
 930
 931
 932 static void
 933 update_totals(const xz_file_info *xfi)
 934 {
 935         // TODO: Integer overflow checks
 936         ++totals.files;
 937         totals.streams += lzma_index_stream_count(xfi->idx);
 938         totals.blocks += lzma_index_block_count(xfi->idx);
 939         totals.compressed_size += lzma_index_file_size(xfi->idx);
 940         totals.uncompressed_size += lzma_index_uncompressed_size(xfi->idx);
 941         totals.stream_padding += xfi->stream_padding;
 942         totals.checks |= lzma_index_checks(xfi->idx);
 943
 944         if (totals.memusage_max < xfi->memusage_max)
 945                 totals.memusage_max = xfi->memusage_max;
 946
 947         totals.all_have_sizes &= xfi->all_have_sizes;
 948
 949         return;
 950 }
 951
 952
 953 static void
 954 print_totals_basic(void)
 955 {
 956         // Print a separator line.
 957         char line[80];
 958         memset(line, '-', sizeof(line));
 959         line[sizeof(line) - 1] = '\0';
 960         puts(line);
 961
 962         // Get the check names.
 963         char checks[CHECKS_STR_SIZE];
 964         get_check_names(checks, totals.checks, false);
 965
 966         // Print the totals except the file count, which needs
 967         // special handling.
 968         printf("%5s %7s  %11s  %11s  %5s  %-7s ",
 969                         uint64_to_str(totals.streams, 0),
 970                         uint64_to_str(totals.blocks, 1),
 971                         uint64_to_nicestr(totals.compressed_size,
 972                                 NICESTR_B, NICESTR_TIB, false, 2),
 973                         uint64_to_nicestr(totals.uncompressed_size,
 974                                 NICESTR_B, NICESTR_TIB, false, 3),
 975                         get_ratio(totals.compressed_size,
 976                                 totals.uncompressed_size),
 977                         checks);
 978
 979         // Since we print totals only when there are at least two files,
 980         // the English message will always use "%s files". But some other
 981         // languages need different forms for different plurals so we
 982         // have to translate this with ngettext().
 983         //
 984         // TRANSLATORS: %s is an integer. Only the plural form of this
 985         // message is used (e.g. "2 files"). Test with "xz -l foo.xz bar.xz".
 986         printf(ngettext("%s file\n", "%s files\n",
 987                         totals.files <= ULONG_MAX ? totals.files
 988                                 : (totals.files % 1000000) + 1000000),
 989                         uint64_to_str(totals.files, 0));
 990
 991         return;
 992 }
 993
 994
 995 static void
 996 print_totals_adv(void)
 997 {
 998         putchar('\n');
 999         puts(_("Totals:"));
1000         printf(_("  Number of files:    %s\n"),
1001                         uint64_to_str(totals.files, 0));
1002         print_adv_helper(totals.streams, totals.blocks,
1003                         totals.compressed_size, totals.uncompressed_size,
1004                         totals.checks, totals.stream_padding);
1005
1006         if (message_verbosity_get() >= V_DEBUG) {
1007                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
1008                                 round_up_to_mib(totals.memusage_max), 0));
1009                 printf(_("  Sizes in headers:   %s\n"),
1010                                 totals.all_have_sizes ? _("Yes") : _("No"));
1011         }
1012
1013         return;
1014 }
1015
1016
1017 static void
1018 print_totals_robot(void)
1019 {
1020         char checks[CHECKS_STR_SIZE];
1021         get_check_names(checks, totals.checks, false);
1022
1023         printf("totals\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
1024                         "\t%s\t%s\t%" PRIu64 "\t%" PRIu64,
1025                         totals.streams,
1026                         totals.blocks,
1027                         totals.compressed_size,
1028                         totals.uncompressed_size,
1029                         get_ratio(totals.compressed_size,
1030                                 totals.uncompressed_size),
1031                         checks,
1032                         totals.stream_padding,
1033                         totals.files);
1034
1035         if (message_verbosity_get() >= V_DEBUG)
1036                 printf("\t%" PRIu64 "\t%s",
1037                                 totals.memusage_max,
1038                                 totals.all_have_sizes ? "yes" : "no");
1039
1040         putchar('\n');
1041
1042         return;
1043 }
1044
1045
1046 extern void
1047 list_totals(void)
1048 {
1049         if (opt_robot) {
1050                 // Always print totals in --robot mode. It can be convenient
1051                 // in some cases and doesn't complicate usage of the
1052                 // single-file case much.
1053                 print_totals_robot();
1054
1055         } else if (totals.files > 1) {
1056                 // For non-robot mode, totals are printed only if there
1057                 // is more than one file.
1058                 if (message_verbosity_get() <= V_WARNING)
1059                         print_totals_basic();
1060                 else
1061                         print_totals_adv();
1062         }
1063
1064         return;
1065 }
1066
1067
1068 extern void
1069 list_file(const char *filename)
1070 {
1071         if (opt_format != FORMAT_XZ && opt_format != FORMAT_AUTO)
1072                 message_fatal(_("--list works only on .xz files "
1073                                 "(--format=xz or --format=auto)"));
1074
1075         message_filename(filename);
1076
1077         if (filename == stdin_filename) {
1078                 message_error(_("--list does not support reading from "
1079                                 "standard input"));
1080                 return;
1081         }
1082
1083         // Unset opt_stdout so that io_open_src() won't accept special files.
1084         // Set opt_force so that io_open_src() will follow symlinks.
1085         opt_stdout = false;
1086         opt_force = true;
1087         file_pair *pair = io_open_src(filename);
1088         if (pair == NULL)
1089                 return;
1090
1091         xz_file_info xfi = XZ_FILE_INFO_INIT;
1092         if (!parse_indexes(&xfi, pair)) {
1093                 bool fail;
1094
1095                 // We have three main modes:
1096                 //  - --robot, which has submodes if --verbose is specified
1097                 //    once or twice
1098                 //  - Normal --list without --verbose
1099                 //  - --list with one or two --verbose
1100                 if (opt_robot)
1101                         fail = print_info_robot(&xfi, pair);
1102                 else if (message_verbosity_get() <= V_WARNING)
1103                         fail = print_info_basic(&xfi, pair);
1104                 else
1105                         fail = print_info_adv(&xfi, pair);
1106
1107                 // Update the totals that are displayed after all
1108                 // the individual files have been listed. Don't count
1109                 // broken files.
1110                 if (!fail)
1111                         update_totals(&xfi);
1112
1113                 lzma_index_end(xfi.idx, NULL);
1114         }
1115
1116         io_close(pair, false);
1117         return;
1118 }