contrib/xz/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 Stream Footer doesn't specify something
 207                 // that we don't support. This can only happen if the xz
 208                 // version is older than liblzma and liblzma supports
 209                 // something new.
 210                 //
 211                 // It is enough to check Stream Footer. Stream Header must
 212                 // match when it is compared against Stream Footer with
 213                 // lzma_stream_flags_compare().
 214                 if (footer_flags.version != 0) {
 215                         message_error("%s: %s", pair->src_name,
 216                                         message_strm(LZMA_OPTIONS_ERROR));
 217                         goto error;
 218                 }
 219
 220                 // Check that the size of the Index field looks sane.
 221                 lzma_vli index_size = footer_flags.backward_size;
 222                 if ((lzma_vli)(pos) < index_size + LZMA_STREAM_HEADER_SIZE) {
 223                         message_error("%s: %s", pair->src_name,
 224                                         message_strm(LZMA_DATA_ERROR));
 225                         goto error;
 226                 }
 227
 228                 // Set pos to the beginning of the Index.
 229                 pos -= index_size;
 230
 231                 // See how much memory we can use for decoding this Index.
 232                 uint64_t memlimit = hardware_memlimit_get(MODE_LIST);
 233                 uint64_t memused = 0;
 234                 if (combined_index != NULL) {
 235                         memused = lzma_index_memused(combined_index);
 236                         if (memused > memlimit)
 237                                 message_bug();
 238
 239                         memlimit -= memused;
 240                 }
 241
 242                 // Decode the Index.
 243                 ret = lzma_index_decoder(&strm, &this_index, memlimit);
 244                 if (ret != LZMA_OK) {
 245                         message_error("%s: %s", pair->src_name,
 246                                         message_strm(ret));
 247                         goto error;
 248                 }
 249
 250                 do {
 251                         // Don't give the decoder more input than the
 252                         // Index size.
 253                         strm.avail_in = my_min(IO_BUFFER_SIZE, index_size);
 254                         if (io_pread(pair, &buf, strm.avail_in, pos))
 255                                 goto error;
 256
 257                         pos += strm.avail_in;
 258                         index_size -= strm.avail_in;
 259
 260                         strm.next_in = buf.u8;
 261                         ret = lzma_code(&strm, LZMA_RUN);
 262
 263                 } while (ret == LZMA_OK);
 264
 265                 // If the decoding seems to be successful, check also that
 266                 // the Index decoder consumed as much input as indicated
 267                 // by the Backward Size field.
 268                 if (ret == LZMA_STREAM_END)
 269                         if (index_size != 0 || strm.avail_in != 0)
 270                                 ret = LZMA_DATA_ERROR;
 271
 272                 if (ret != LZMA_STREAM_END) {
 273                         // LZMA_BUFFER_ERROR means that the Index decoder
 274                         // would have liked more input than what the Index
 275                         // size should be according to Stream Footer.
 276                         // The message for LZMA_DATA_ERROR makes more
 277                         // sense in that case.
 278                         if (ret == LZMA_BUF_ERROR)
 279                                 ret = LZMA_DATA_ERROR;
 280
 281                         message_error("%s: %s", pair->src_name,
 282                                         message_strm(ret));
 283
 284                         // If the error was too low memory usage limit,
 285                         // show also how much memory would have been needed.
 286                         if (ret == LZMA_MEMLIMIT_ERROR) {
 287                                 uint64_t needed = lzma_memusage(&strm);
 288                                 if (UINT64_MAX - needed < memused)
 289                                         needed = UINT64_MAX;
 290                                 else
 291                                         needed += memused;
 292
 293                                 message_mem_needed(V_ERROR, needed);
 294                         }
 295
 296                         goto error;
 297                 }
 298
 299                 // Decode the Stream Header and check that its Stream Flags
 300                 // match the Stream Footer.
 301                 pos -= footer_flags.backward_size + LZMA_STREAM_HEADER_SIZE;
 302                 if ((lzma_vli)(pos) < lzma_index_total_size(this_index)) {
 303                         message_error("%s: %s", pair->src_name,
 304                                         message_strm(LZMA_DATA_ERROR));
 305                         goto error;
 306                 }
 307
 308                 pos -= lzma_index_total_size(this_index);
 309                 if (io_pread(pair, &buf, LZMA_STREAM_HEADER_SIZE, pos))
 310                         goto error;
 311
 312                 ret = lzma_stream_header_decode(&header_flags, buf.u8);
 313                 if (ret != LZMA_OK) {
 314                         message_error("%s: %s", pair->src_name,
 315                                         message_strm(ret));
 316                         goto error;
 317                 }
 318
 319                 ret = lzma_stream_flags_compare(&header_flags, &footer_flags);
 320                 if (ret != LZMA_OK) {
 321                         message_error("%s: %s", pair->src_name,
 322                                         message_strm(ret));
 323                         goto error;
 324                 }
 325
 326                 // Store the decoded Stream Flags into this_index. This is
 327                 // needed so that we can print which Check is used in each
 328                 // Stream.
 329                 ret = lzma_index_stream_flags(this_index, &footer_flags);
 330                 if (ret != LZMA_OK)
 331                         message_bug();
 332
 333                 // Store also the size of the Stream Padding field. It is
 334                 // needed to show the offsets of the Streams correctly.
 335                 ret = lzma_index_stream_padding(this_index, stream_padding);
 336                 if (ret != LZMA_OK)
 337                         message_bug();
 338
 339                 if (combined_index != NULL) {
 340                         // Append the earlier decoded Indexes
 341                         // after this_index.
 342                         ret = lzma_index_cat(
 343                                         this_index, combined_index, NULL);
 344                         if (ret != LZMA_OK) {
 345                                 message_error("%s: %s", pair->src_name,
 346                                                 message_strm(ret));
 347                                 goto error;
 348                         }
 349                 }
 350
 351                 combined_index = this_index;
 352                 this_index = NULL;
 353
 354                 xfi->stream_padding += stream_padding;
 355
 356         } while (pos > 0);
 357
 358         lzma_end(&strm);
 359
 360         // All OK. Make combined_index available to the caller.
 361         xfi->idx = combined_index;
 362         return false;
 363
 364 error:
 365         // Something went wrong, free the allocated memory.
 366         lzma_end(&strm);
 367         lzma_index_end(combined_index, NULL);
 368         lzma_index_end(this_index, NULL);
 369         return true;
 370 }
 371
 372
 373 /// \brief      Parse the Block Header
 374 ///
 375 /// The result is stored into *bhi. The caller takes care of initializing it.
 376 ///
 377 /// \return     False on success, true on error.
 378 static bool
 379 parse_block_header(file_pair *pair, const lzma_index_iter *iter,
 380                 block_header_info *bhi, xz_file_info *xfi)
 381 {
 382 #if IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
 383 #       error IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
 384 #endif
 385
 386         // Get the whole Block Header with one read, but don't read past
 387         // the end of the Block (or even its Check field).
 388         const uint32_t size = my_min(iter->block.total_size
 389                                 - lzma_check_size(iter->stream.flags->check),
 390                         LZMA_BLOCK_HEADER_SIZE_MAX);
 391         io_buf buf;
 392         if (io_pread(pair, &buf, size, iter->block.compressed_file_offset))
 393                 return true;
 394
 395         // Zero would mean Index Indicator and thus not a valid Block.
 396         if (buf.u8[0] == 0)
 397                 goto data_error;
 398
 399         // Initialize the block structure and decode Block Header Size.
 400         lzma_filter filters[LZMA_FILTERS_MAX + 1];
 401         lzma_block block;
 402         block.version = 0;
 403         block.check = iter->stream.flags->check;
 404         block.filters = filters;
 405
 406         block.header_size = lzma_block_header_size_decode(buf.u8[0]);
 407         if (block.header_size > size)
 408                 goto data_error;
 409
 410         // Decode the Block Header.
 411         switch (lzma_block_header_decode(&block, NULL, buf.u8)) {
 412         case LZMA_OK:
 413                 break;
 414
 415         case LZMA_OPTIONS_ERROR:
 416                 message_error("%s: %s", pair->src_name,
 417                                 message_strm(LZMA_OPTIONS_ERROR));
 418                 return true;
 419
 420         case LZMA_DATA_ERROR:
 421                 goto data_error;
 422
 423         default:
 424                 message_bug();
 425         }
 426
 427         // Check the Block Flags. These must be done before calling
 428         // lzma_block_compressed_size(), because it overwrites
 429         // block.compressed_size.
 430         bhi->flags[0] = block.compressed_size != LZMA_VLI_UNKNOWN
 431                         ? 'c' : '-';
 432         bhi->flags[1] = block.uncompressed_size != LZMA_VLI_UNKNOWN
 433                         ? 'u' : '-';
 434         bhi->flags[2] = '\0';
 435
 436         // Collect information if all Blocks have both Compressed Size
 437         // and Uncompressed Size fields. They can be useful e.g. for
 438         // multi-threaded decompression so it can be useful to know it.
 439         xfi->all_have_sizes &= block.compressed_size != LZMA_VLI_UNKNOWN
 440                         && block.uncompressed_size != LZMA_VLI_UNKNOWN;
 441
 442         // Validate or set block.compressed_size.
 443         switch (lzma_block_compressed_size(&block,
 444                         iter->block.unpadded_size)) {
 445         case LZMA_OK:
 446                 // Validate also block.uncompressed_size if it is present.
 447                 // If it isn't present, there's no need to set it since
 448                 // we aren't going to actually decompress the Block; if
 449                 // we were decompressing, then we should set it so that
 450                 // the Block decoder could validate the Uncompressed Size
 451                 // that was stored in the Index.
 452                 if (block.uncompressed_size == LZMA_VLI_UNKNOWN
 453                                 || block.uncompressed_size
 454                                         == iter->block.uncompressed_size)
 455                         break;
 456
 457                 // If the above fails, the file is corrupt so
 458                 // LZMA_DATA_ERROR is a good error code.
 459
 460         case LZMA_DATA_ERROR:
 461                 // Free the memory allocated by lzma_block_header_decode().
 462                 for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 463                         free(filters[i].options);
 464
 465                 goto data_error;
 466
 467         default:
 468                 message_bug();
 469         }
 470
 471         // Copy the known sizes.
 472         bhi->header_size = block.header_size;
 473         bhi->compressed_size = block.compressed_size;
 474
 475         // Calculate the decoder memory usage and update the maximum
 476         // memory usage of this Block.
 477         bhi->memusage = lzma_raw_decoder_memusage(filters);
 478         if (xfi->memusage_max < bhi->memusage)
 479                 xfi->memusage_max = bhi->memusage;
 480
 481         // Convert the filter chain to human readable form.
 482         message_filters_to_str(bhi->filter_chain, filters, false);
 483
 484         // Free the memory allocated by lzma_block_header_decode().
 485         for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 486                 free(filters[i].options);
 487
 488         return false;
 489
 490 data_error:
 491         // Show the error message.
 492         message_error("%s: %s", pair->src_name,
 493                         message_strm(LZMA_DATA_ERROR));
 494         return true;
 495 }
 496
 497
 498 /// \brief      Parse the Check field and put it into check_value[]
 499 ///
 500 /// \return     False on success, true on error.
 501 static bool
 502 parse_check_value(file_pair *pair, const lzma_index_iter *iter)
 503 {
 504         // Don't read anything from the file if there is no integrity Check.
 505         if (iter->stream.flags->check == LZMA_CHECK_NONE) {
 506                 snprintf(check_value, sizeof(check_value), "---");
 507                 return false;
 508         }
 509
 510         // Locate and read the Check field.
 511         const uint32_t size = lzma_check_size(iter->stream.flags->check);
 512         const off_t offset = iter->block.compressed_file_offset
 513                         + iter->block.total_size - size;
 514         io_buf buf;
 515         if (io_pread(pair, &buf, size, offset))
 516                 return true;
 517
 518         // CRC32 and CRC64 are in little endian. Guess that all the future
 519         // 32-bit and 64-bit Check values are little endian too. It shouldn't
 520         // be a too big problem if this guess is wrong.
 521         if (size == 4)
 522                 snprintf(check_value, sizeof(check_value),
 523                                 "%08" PRIx32, conv32le(buf.u32[0]));
 524         else if (size == 8)
 525                 snprintf(check_value, sizeof(check_value),
 526                                 "%016" PRIx64, conv64le(buf.u64[0]));
 527         else
 528                 for (size_t i = 0; i < size; ++i)
 529                         snprintf(check_value + i * 2, 3, "%02x", buf.u8[i]);
 530
 531         return false;
 532 }
 533
 534
 535 /// \brief      Parse detailed information about a Block
 536 ///
 537 /// Since this requires seek(s), listing information about all Blocks can
 538 /// be slow.
 539 ///
 540 /// \param      pair    Input file
 541 /// \param      iter    Location of the Block whose Check value should
 542 ///                     be printed.
 543 /// \param      bhi     Pointer to structure where to store the information
 544 ///                     about the Block Header field.
 545 ///
 546 /// \return     False on success, true on error. If an error occurs,
 547 ///             the error message is printed too so the caller doesn't
 548 ///             need to worry about that.
 549 static bool
 550 parse_details(file_pair *pair, const lzma_index_iter *iter,
 551                 block_header_info *bhi, xz_file_info *xfi)
 552 {
 553         if (parse_block_header(pair, iter, bhi, xfi))
 554                 return true;
 555
 556         if (parse_check_value(pair, iter))
 557                 return true;
 558
 559         return false;
 560 }
 561
 562
 563 /// \brief      Get the compression ratio
 564 ///
 565 /// This has slightly different format than that is used in message.c.
 566 static const char *
 567 get_ratio(uint64_t compressed_size, uint64_t uncompressed_size)
 568 {
 569         if (uncompressed_size == 0)
 570                 return "---";
 571
 572         const double ratio = (double)(compressed_size)
 573                         / (double)(uncompressed_size);
 574         if (ratio > 9.999)
 575                 return "---";
 576
 577         static char buf[16];
 578         snprintf(buf, sizeof(buf), "%.3f", ratio);
 579         return buf;
 580 }
 581
 582
 583 /// \brief      Get a comma-separated list of Check names
 584 ///
 585 /// The check names are translated with gettext except when in robot mode.
 586 ///
 587 /// \param      buf     Buffer to hold the resulting string
 588 /// \param      checks  Bit mask of Checks to print
 589 /// \param      space_after_comma
 590 ///                     It's better to not use spaces in table-like listings,
 591 ///                     but in more verbose formats a space after a comma
 592 ///                     is good for readability.
 593 static void
 594 get_check_names(char buf[CHECKS_STR_SIZE],
 595                 uint32_t checks, bool space_after_comma)
 596 {
 597         assert(checks != 0);
 598
 599         char *pos = buf;
 600         size_t left = CHECKS_STR_SIZE;
 601
 602         const char *sep = space_after_comma ? ", " : ",";
 603         bool comma = false;
 604
 605         for (size_t i = 0; i <= LZMA_CHECK_ID_MAX; ++i) {
 606                 if (checks & (UINT32_C(1) << i)) {
 607                         my_snprintf(&pos, &left, "%s%s",
 608                                         comma ? sep : "",
 609                                         opt_robot ? check_names[i]
 610                                                 : _(check_names[i]));
 611                         comma = true;
 612                 }
 613         }
 614
 615         return;
 616 }
 617
 618
 619 static bool
 620 print_info_basic(const xz_file_info *xfi, file_pair *pair)
 621 {
 622         static bool headings_displayed = false;
 623         if (!headings_displayed) {
 624                 headings_displayed = true;
 625                 // TRANSLATORS: These are column headings. From Strms (Streams)
 626                 // to Ratio, the columns are right aligned. Check and Filename
 627                 // are left aligned. If you need longer words, it's OK to
 628                 // use two lines here. Test with "xz -l foo.xz".
 629                 puts(_("Strms  Blocks   Compressed Uncompressed  Ratio  "
 630                                 "Check   Filename"));
 631         }
 632
 633         char checks[CHECKS_STR_SIZE];
 634         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 635
 636         const char *cols[7] = {
 637                 uint64_to_str(lzma_index_stream_count(xfi->idx), 0),
 638                 uint64_to_str(lzma_index_block_count(xfi->idx), 1),
 639                 uint64_to_nicestr(lzma_index_file_size(xfi->idx),
 640                         NICESTR_B, NICESTR_TIB, false, 2),
 641                 uint64_to_nicestr(lzma_index_uncompressed_size(xfi->idx),
 642                         NICESTR_B, NICESTR_TIB, false, 3),
 643                 get_ratio(lzma_index_file_size(xfi->idx),
 644                         lzma_index_uncompressed_size(xfi->idx)),
 645                 checks,
 646                 pair->src_name,
 647         };
 648         printf("%*s %*s  %*s  %*s  %*s  %-*s %s\n",
 649                         tuklib_mbstr_fw(cols[0], 5), cols[0],
 650                         tuklib_mbstr_fw(cols[1], 7), cols[1],
 651                         tuklib_mbstr_fw(cols[2], 11), cols[2],
 652                         tuklib_mbstr_fw(cols[3], 11), cols[3],
 653                         tuklib_mbstr_fw(cols[4], 5), cols[4],
 654                         tuklib_mbstr_fw(cols[5], 7), cols[5],
 655                         cols[6]);
 656
 657         return false;
 658 }
 659
 660
 661 static void
 662 print_adv_helper(uint64_t stream_count, uint64_t block_count,
 663                 uint64_t compressed_size, uint64_t uncompressed_size,
 664                 uint32_t checks, uint64_t stream_padding)
 665 {
 666         char checks_str[CHECKS_STR_SIZE];
 667         get_check_names(checks_str, checks, true);
 668
 669         printf(_("  Streams:            %s\n"),
 670                         uint64_to_str(stream_count, 0));
 671         printf(_("  Blocks:             %s\n"),
 672                         uint64_to_str(block_count, 0));
 673         printf(_("  Compressed size:    %s\n"),
 674                         uint64_to_nicestr(compressed_size,
 675                                 NICESTR_B, NICESTR_TIB, true, 0));
 676         printf(_("  Uncompressed size:  %s\n"),
 677                         uint64_to_nicestr(uncompressed_size,
 678                                 NICESTR_B, NICESTR_TIB, true, 0));
 679         printf(_("  Ratio:              %s\n"),
 680                         get_ratio(compressed_size, uncompressed_size));
 681         printf(_("  Check:              %s\n"), checks_str);
 682         printf(_("  Stream padding:     %s\n"),
 683                         uint64_to_nicestr(stream_padding,
 684                                 NICESTR_B, NICESTR_TIB, true, 0));
 685         return;
 686 }
 687
 688
 689 static bool
 690 print_info_adv(xz_file_info *xfi, file_pair *pair)
 691 {
 692         // Print the overall information.
 693         print_adv_helper(lzma_index_stream_count(xfi->idx),
 694                         lzma_index_block_count(xfi->idx),
 695                         lzma_index_file_size(xfi->idx),
 696                         lzma_index_uncompressed_size(xfi->idx),
 697                         lzma_index_checks(xfi->idx),
 698                         xfi->stream_padding);
 699
 700         // Size of the biggest Check. This is used to calculate the width
 701         // of the CheckVal field. The table would get insanely wide if
 702         // we always reserved space for 64-byte Check (128 chars as hex).
 703         uint32_t check_max = 0;
 704
 705         // Print information about the Streams.
 706         //
 707         // TRANSLATORS: The second line is column headings. All except
 708         // Check are right aligned; Check is left aligned. Test with
 709         // "xz -lv foo.xz".
 710         puts(_("  Streams:\n    Stream    Blocks"
 711                         "      CompOffset    UncompOffset"
 712                         "        CompSize      UncompSize  Ratio"
 713                         "  Check      Padding"));
 714
 715         lzma_index_iter iter;
 716         lzma_index_iter_init(&iter, xfi->idx);
 717
 718         while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM)) {
 719                 const char *cols1[4] = {
 720                         uint64_to_str(iter.stream.number, 0),
 721                         uint64_to_str(iter.stream.block_count, 1),
 722                         uint64_to_str(iter.stream.compressed_offset, 2),
 723                         uint64_to_str(iter.stream.uncompressed_offset, 3),
 724                 };
 725                 printf("    %*s %*s %*s %*s ",
 726                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 727                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 728                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 729                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 730
 731                 const char *cols2[5] = {
 732                         uint64_to_str(iter.stream.compressed_size, 0),
 733                         uint64_to_str(iter.stream.uncompressed_size, 1),
 734                         get_ratio(iter.stream.compressed_size,
 735                                 iter.stream.uncompressed_size),
 736                         _(check_names[iter.stream.flags->check]),
 737                         uint64_to_str(iter.stream.padding, 2),
 738                 };
 739                 printf("%*s %*s  %*s  %-*s %*s\n",
 740                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 741                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 742                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 743                                 tuklib_mbstr_fw(cols2[3], 10), cols2[3],
 744                                 tuklib_mbstr_fw(cols2[4], 7), cols2[4]);
 745
 746                 // Update the maximum Check size.
 747                 if (lzma_check_size(iter.stream.flags->check) > check_max)
 748                         check_max = lzma_check_size(iter.stream.flags->check);
 749         }
 750
 751         // Cache the verbosity level to a local variable.
 752         const bool detailed = message_verbosity_get() >= V_DEBUG;
 753
 754         // Information collected from Block Headers
 755         block_header_info bhi;
 756
 757         // Print information about the Blocks but only if there is
 758         // at least one Block.
 759         if (lzma_index_block_count(xfi->idx) > 0) {
 760                 // Calculate the width of the CheckVal field.
 761                 const int checkval_width = my_max(8, 2 * check_max);
 762
 763                 // TRANSLATORS: The second line is column headings. All
 764                 // except Check are right aligned; Check is left aligned.
 765                 printf(_("  Blocks:\n    Stream     Block"
 766                         "      CompOffset    UncompOffset"
 767                         "       TotalSize      UncompSize  Ratio  Check"));
 768
 769                 if (detailed) {
 770                         // TRANSLATORS: These are additional column headings
 771                         // for the most verbose listing mode. CheckVal
 772                         // (Check value), Flags, and Filters are left aligned.
 773                         // Header (Block Header Size), CompSize, and MemUsage
 774                         // are right aligned. %*s is replaced with 0-120
 775                         // spaces to make the CheckVal column wide enough.
 776                         // Test with "xz -lvv foo.xz".
 777                         printf(_("      CheckVal %*s Header  Flags        "
 778                                         "CompSize    MemUsage  Filters"),
 779                                         checkval_width - 8, "");
 780                 }
 781
 782                 putchar('\n');
 783
 784                 lzma_index_iter_init(&iter, xfi->idx);
 785
 786                 // Iterate over the Blocks.
 787                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 788                         if (detailed && parse_details(pair, &iter, &bhi, xfi))
 789                                         return true;
 790
 791                         const char *cols1[4] = {
 792                                 uint64_to_str(iter.stream.number, 0),
 793                                 uint64_to_str(
 794                                         iter.block.number_in_stream, 1),
 795                                 uint64_to_str(
 796                                         iter.block.compressed_file_offset, 2),
 797                                 uint64_to_str(
 798                                         iter.block.uncompressed_file_offset, 3)
 799                         };
 800                         printf("    %*s %*s %*s %*s ",
 801                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 802                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 803                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 804                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 805
 806                         const char *cols2[4] = {
 807                                 uint64_to_str(iter.block.total_size, 0),
 808                                 uint64_to_str(iter.block.uncompressed_size,
 809                                                 1),
 810                                 get_ratio(iter.block.total_size,
 811                                         iter.block.uncompressed_size),
 812                                 _(check_names[iter.stream.flags->check])
 813                         };
 814                         printf("%*s %*s  %*s  %-*s",
 815                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 816                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 817                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 818                                 tuklib_mbstr_fw(cols2[3], detailed ? 11 : 1),
 819                                         cols2[3]);
 820
 821                         if (detailed) {
 822                                 const lzma_vli compressed_size
 823                                                 = iter.block.unpadded_size
 824                                                 - bhi.header_size
 825                                                 - lzma_check_size(
 826                                                 iter.stream.flags->check);
 827
 828                                 const char *cols3[6] = {
 829                                         check_value,
 830                                         uint64_to_str(bhi.header_size, 0),
 831                                         bhi.flags,
 832                                         uint64_to_str(compressed_size, 1),
 833                                         uint64_to_str(
 834                                                 round_up_to_mib(bhi.memusage),
 835                                                 2),
 836                                         bhi.filter_chain
 837                                 };
 838                                 // Show MiB for memory usage, because it
 839                                 // is the only size which is not in bytes.
 840                                 printf("%-*s  %*s  %-5s %*s %*s MiB  %s",
 841                                         checkval_width, cols3[0],
 842                                         tuklib_mbstr_fw(cols3[1], 6), cols3[1],
 843                                         cols3[2],
 844                                         tuklib_mbstr_fw(cols3[3], 15),
 845                                                 cols3[3],
 846                                         tuklib_mbstr_fw(cols3[4], 7), cols3[4],
 847                                         cols3[5]);
 848                         }
 849
 850                         putchar('\n');
 851                 }
 852         }
 853
 854         if (detailed) {
 855                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
 856                                 round_up_to_mib(xfi->memusage_max), 0));
 857                 printf(_("  Sizes in headers:   %s\n"),
 858                                 xfi->all_have_sizes ? _("Yes") : _("No"));
 859         }
 860
 861         return false;
 862 }
 863
 864
 865 static bool
 866 print_info_robot(xz_file_info *xfi, file_pair *pair)
 867 {
 868         char checks[CHECKS_STR_SIZE];
 869         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 870
 871         printf("name\t%s\n", pair->src_name);
 872
 873         printf("file\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 874                         "\t%s\t%s\t%" PRIu64 "\n",
 875                         lzma_index_stream_count(xfi->idx),
 876                         lzma_index_block_count(xfi->idx),
 877                         lzma_index_file_size(xfi->idx),
 878                         lzma_index_uncompressed_size(xfi->idx),
 879                         get_ratio(lzma_index_file_size(xfi->idx),
 880                                 lzma_index_uncompressed_size(xfi->idx)),
 881                         checks,
 882                         xfi->stream_padding);
 883
 884         if (message_verbosity_get() >= V_VERBOSE) {
 885                 lzma_index_iter iter;
 886                 lzma_index_iter_init(&iter, xfi->idx);
 887
 888                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM))
 889                         printf("stream\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 890                                 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 891                                 "\t%s\t%s\t%" PRIu64 "\n",
 892                                 iter.stream.number,
 893                                 iter.stream.block_count,
 894                                 iter.stream.compressed_offset,
 895                                 iter.stream.uncompressed_offset,
 896                                 iter.stream.compressed_size,
 897                                 iter.stream.uncompressed_size,
 898                                 get_ratio(iter.stream.compressed_size,
 899                                         iter.stream.uncompressed_size),
 900                                 check_names[iter.stream.flags->check],
 901                                 iter.stream.padding);
 902
 903                 lzma_index_iter_rewind(&iter);
 904                 block_header_info bhi;
 905
 906                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 907                         if (message_verbosity_get() >= V_DEBUG
 908                                         && parse_details(
 909                                                 pair, &iter, &bhi, xfi))
 910                                 return true;
 911
 912                         printf("block\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 913                                         "\t%" PRIu64 "\t%" PRIu64
 914                                         "\t%" PRIu64 "\t%" PRIu64 "\t%s\t%s",
 915                                         iter.stream.number,
 916                                         iter.block.number_in_stream,
 917                                         iter.block.number_in_file,
 918                                         iter.block.compressed_file_offset,
 919                                         iter.block.uncompressed_file_offset,
 920                                         iter.block.total_size,
 921                                         iter.block.uncompressed_size,
 922                                         get_ratio(iter.block.total_size,
 923                                                 iter.block.uncompressed_size),
 924                                         check_names[iter.stream.flags->check]);
 925
 926                         if (message_verbosity_get() >= V_DEBUG)
 927                                 printf("\t%s\t%" PRIu32 "\t%s\t%" PRIu64
 928                                                 "\t%" PRIu64 "\t%s",
 929                                                 check_value,
 930                                                 bhi.header_size,
 931                                                 bhi.flags,
 932                                                 bhi.compressed_size,
 933                                                 bhi.memusage,
 934                                                 bhi.filter_chain);
 935
 936                         putchar('\n');
 937                 }
 938         }
 939
 940         if (message_verbosity_get() >= V_DEBUG)
 941                 printf("summary\t%" PRIu64 "\t%s\n",
 942                                 xfi->memusage_max,
 943                                 xfi->all_have_sizes ? "yes" : "no");
 944
 945         return false;
 946 }
 947
 948
 949 static void
 950 update_totals(const xz_file_info *xfi)
 951 {
 952         // TODO: Integer overflow checks
 953         ++totals.files;
 954         totals.streams += lzma_index_stream_count(xfi->idx);
 955         totals.blocks += lzma_index_block_count(xfi->idx);
 956         totals.compressed_size += lzma_index_file_size(xfi->idx);
 957         totals.uncompressed_size += lzma_index_uncompressed_size(xfi->idx);
 958         totals.stream_padding += xfi->stream_padding;
 959         totals.checks |= lzma_index_checks(xfi->idx);
 960
 961         if (totals.memusage_max < xfi->memusage_max)
 962                 totals.memusage_max = xfi->memusage_max;
 963
 964         totals.all_have_sizes &= xfi->all_have_sizes;
 965
 966         return;
 967 }
 968
 969
 970 static void
 971 print_totals_basic(void)
 972 {
 973         // Print a separator line.
 974         char line[80];
 975         memset(line, '-', sizeof(line));
 976         line[sizeof(line) - 1] = '\0';
 977         puts(line);
 978
 979         // Get the check names.
 980         char checks[CHECKS_STR_SIZE];
 981         get_check_names(checks, totals.checks, false);
 982
 983         // Print the totals except the file count, which needs
 984         // special handling.
 985         printf("%5s %7s  %11s  %11s  %5s  %-7s ",
 986                         uint64_to_str(totals.streams, 0),
 987                         uint64_to_str(totals.blocks, 1),
 988                         uint64_to_nicestr(totals.compressed_size,
 989                                 NICESTR_B, NICESTR_TIB, false, 2),
 990                         uint64_to_nicestr(totals.uncompressed_size,
 991                                 NICESTR_B, NICESTR_TIB, false, 3),
 992                         get_ratio(totals.compressed_size,
 993                                 totals.uncompressed_size),
 994                         checks);
 995
 996         // Since we print totals only when there are at least two files,
 997         // the English message will always use "%s files". But some other
 998         // languages need different forms for different plurals so we
 999         // have to translate this with ngettext().
1000         //
1001         // TRANSLATORS: %s is an integer. Only the plural form of this
1002         // message is used (e.g. "2 files"). Test with "xz -l foo.xz bar.xz".
1003         printf(ngettext("%s file\n", "%s files\n",
1004                         totals.files <= ULONG_MAX ? totals.files
1005                                 : (totals.files % 1000000) + 1000000),
1006                         uint64_to_str(totals.files, 0));
1007
1008         return;
1009 }
1010
1011
1012 static void
1013 print_totals_adv(void)
1014 {
1015         putchar('\n');
1016         puts(_("Totals:"));
1017         printf(_("  Number of files:    %s\n"),
1018                         uint64_to_str(totals.files, 0));
1019         print_adv_helper(totals.streams, totals.blocks,
1020                         totals.compressed_size, totals.uncompressed_size,
1021                         totals.checks, totals.stream_padding);
1022
1023         if (message_verbosity_get() >= V_DEBUG) {
1024                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
1025                                 round_up_to_mib(totals.memusage_max), 0));
1026                 printf(_("  Sizes in headers:   %s\n"),
1027                                 totals.all_have_sizes ? _("Yes") : _("No"));
1028         }
1029
1030         return;
1031 }
1032
1033
1034 static void
1035 print_totals_robot(void)
1036 {
1037         char checks[CHECKS_STR_SIZE];
1038         get_check_names(checks, totals.checks, false);
1039
1040         printf("totals\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
1041                         "\t%s\t%s\t%" PRIu64 "\t%" PRIu64,
1042                         totals.streams,
1043                         totals.blocks,
1044                         totals.compressed_size,
1045                         totals.uncompressed_size,
1046                         get_ratio(totals.compressed_size,
1047                                 totals.uncompressed_size),
1048                         checks,
1049                         totals.stream_padding,
1050                         totals.files);
1051
1052         if (message_verbosity_get() >= V_DEBUG)
1053                 printf("\t%" PRIu64 "\t%s",
1054                                 totals.memusage_max,
1055                                 totals.all_have_sizes ? "yes" : "no");
1056
1057         putchar('\n');
1058
1059         return;
1060 }
1061
1062
1063 extern void
1064 list_totals(void)
1065 {
1066         if (opt_robot) {
1067                 // Always print totals in --robot mode. It can be convenient
1068                 // in some cases and doesn't complicate usage of the
1069                 // single-file case much.
1070                 print_totals_robot();
1071
1072         } else if (totals.files > 1) {
1073                 // For non-robot mode, totals are printed only if there
1074                 // is more than one file.
1075                 if (message_verbosity_get() <= V_WARNING)
1076                         print_totals_basic();
1077                 else
1078                         print_totals_adv();
1079         }
1080
1081         return;
1082 }
1083
1084
1085 extern void
1086 list_file(const char *filename)
1087 {
1088         if (opt_format != FORMAT_XZ && opt_format != FORMAT_AUTO)
1089                 message_fatal(_("--list works only on .xz files "
1090                                 "(--format=xz or --format=auto)"));
1091
1092         message_filename(filename);
1093
1094         if (filename == stdin_filename) {
1095                 message_error(_("--list does not support reading from "
1096                                 "standard input"));
1097                 return;
1098         }
1099
1100         // Unset opt_stdout so that io_open_src() won't accept special files.
1101         // Set opt_force so that io_open_src() will follow symlinks.
1102         opt_stdout = false;
1103         opt_force = true;
1104         file_pair *pair = io_open_src(filename);
1105         if (pair == NULL)
1106                 return;
1107
1108         xz_file_info xfi = XZ_FILE_INFO_INIT;
1109         if (!parse_indexes(&xfi, pair)) {
1110                 bool fail;
1111
1112                 // We have three main modes:
1113                 //  - --robot, which has submodes if --verbose is specified
1114                 //    once or twice
1115                 //  - Normal --list without --verbose
1116                 //  - --list with one or two --verbose
1117                 if (opt_robot)
1118                         fail = print_info_robot(&xfi, pair);
1119                 else if (message_verbosity_get() <= V_WARNING)
1120                         fail = print_info_basic(&xfi, pair);
1121                 else
1122                         fail = print_info_adv(&xfi, pair);
1123
1124                 // Update the totals that are displayed after all
1125                 // the individual files have been listed. Don't count
1126                 // broken files.
1127                 if (!fail)
1128                         update_totals(&xfi);
1129
1130                 lzma_index_end(xfi.idx, NULL);
1131         }
1132
1133         io_close(pair, false);
1134         return;
1135 }