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 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         // Initialize the block structure and decode Block Header Size.
 386         lzma_filter filters[LZMA_FILTERS_MAX + 1];
 387         lzma_block block;
 388         block.version = 0;
 389         block.check = iter->stream.flags->check;
 390         block.filters = filters;
 391
 392         block.header_size = lzma_block_header_size_decode(buf.u8[0]);
 393         if (block.header_size > size)
 394                 goto data_error;
 395
 396         // Decode the Block Header.
 397         switch (lzma_block_header_decode(&block, NULL, buf.u8)) {
 398         case LZMA_OK:
 399                 break;
 400
 401         case LZMA_OPTIONS_ERROR:
 402                 message_error("%s: %s", pair->src_name,
 403                                 message_strm(LZMA_OPTIONS_ERROR));
 404                 return true;
 405
 406         case LZMA_DATA_ERROR:
 407                 goto data_error;
 408
 409         default:
 410                 message_bug();
 411         }
 412
 413         // Check the Block Flags. These must be done before calling
 414         // lzma_block_compressed_size(), because it overwrites
 415         // block.compressed_size.
 416         bhi->flags[0] = block.compressed_size != LZMA_VLI_UNKNOWN
 417                         ? 'c' : '-';
 418         bhi->flags[1] = block.uncompressed_size != LZMA_VLI_UNKNOWN
 419                         ? 'u' : '-';
 420         bhi->flags[2] = '\0';
 421
 422         // Collect information if all Blocks have both Compressed Size
 423         // and Uncompressed Size fields. They can be useful e.g. for
 424         // multi-threaded decompression so it can be useful to know it.
 425         xfi->all_have_sizes &= block.compressed_size != LZMA_VLI_UNKNOWN
 426                         && block.uncompressed_size != LZMA_VLI_UNKNOWN;
 427
 428         // Validate or set block.compressed_size.
 429         switch (lzma_block_compressed_size(&block,
 430                         iter->block.unpadded_size)) {
 431         case LZMA_OK:
 432                 break;
 433
 434         case LZMA_DATA_ERROR:
 435                 // Free the memory allocated by lzma_block_header_decode().
 436                 for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 437                         free(filters[i].options);
 438
 439                 goto data_error;
 440
 441         default:
 442                 message_bug();
 443         }
 444
 445         // Copy the known sizes.
 446         bhi->header_size = block.header_size;
 447         bhi->compressed_size = block.compressed_size;
 448
 449         // Calculate the decoder memory usage and update the maximum
 450         // memory usage of this Block.
 451         bhi->memusage = lzma_raw_decoder_memusage(filters);
 452         if (xfi->memusage_max < bhi->memusage)
 453                 xfi->memusage_max = bhi->memusage;
 454
 455         // Convert the filter chain to human readable form.
 456         message_filters_to_str(bhi->filter_chain, filters, false);
 457
 458         // Free the memory allocated by lzma_block_header_decode().
 459         for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 460                 free(filters[i].options);
 461
 462         return false;
 463
 464 data_error:
 465         // Show the error message.
 466         message_error("%s: %s", pair->src_name,
 467                         message_strm(LZMA_DATA_ERROR));
 468         return true;
 469 }
 470
 471
 472 /// \brief      Parse the Check field and put it into check_value[]
 473 ///
 474 /// \return     False on success, true on error.
 475 static bool
 476 parse_check_value(file_pair *pair, const lzma_index_iter *iter)
 477 {
 478         // Don't read anything from the file if there is no integrity Check.
 479         if (iter->stream.flags->check == LZMA_CHECK_NONE) {
 480                 snprintf(check_value, sizeof(check_value), "---");
 481                 return false;
 482         }
 483
 484         // Locate and read the Check field.
 485         const uint32_t size = lzma_check_size(iter->stream.flags->check);
 486         const off_t offset = iter->block.compressed_file_offset
 487                         + iter->block.total_size - size;
 488         io_buf buf;
 489         if (io_pread(pair, &buf, size, offset))
 490                 return true;
 491
 492         // CRC32 and CRC64 are in little endian. Guess that all the future
 493         // 32-bit and 64-bit Check values are little endian too. It shouldn't
 494         // be a too big problem if this guess is wrong.
 495         if (size == 4)
 496                 snprintf(check_value, sizeof(check_value),
 497                                 "%08" PRIx32, conv32le(buf.u32[0]));
 498         else if (size == 8)
 499                 snprintf(check_value, sizeof(check_value),
 500                                 "%016" PRIx64, conv64le(buf.u64[0]));
 501         else
 502                 for (size_t i = 0; i < size; ++i)
 503                         snprintf(check_value + i * 2, 3, "%02x", buf.u8[i]);
 504
 505         return false;
 506 }
 507
 508
 509 /// \brief      Parse detailed information about a Block
 510 ///
 511 /// Since this requires seek(s), listing information about all Blocks can
 512 /// be slow.
 513 ///
 514 /// \param      pair    Input file
 515 /// \param      iter    Location of the Block whose Check value should
 516 ///                     be printed.
 517 /// \param      bhi     Pointer to structure where to store the information
 518 ///                     about the Block Header field.
 519 ///
 520 /// \return     False on success, true on error. If an error occurs,
 521 ///             the error message is printed too so the caller doesn't
 522 ///             need to worry about that.
 523 static bool
 524 parse_details(file_pair *pair, const lzma_index_iter *iter,
 525                 block_header_info *bhi, xz_file_info *xfi)
 526 {
 527         if (parse_block_header(pair, iter, bhi, xfi))
 528                 return true;
 529
 530         if (parse_check_value(pair, iter))
 531                 return true;
 532
 533         return false;
 534 }
 535
 536
 537 /// \brief      Get the compression ratio
 538 ///
 539 /// This has slightly different format than that is used in message.c.
 540 static const char *
 541 get_ratio(uint64_t compressed_size, uint64_t uncompressed_size)
 542 {
 543         if (uncompressed_size == 0)
 544                 return "---";
 545
 546         const double ratio = (double)(compressed_size)
 547                         / (double)(uncompressed_size);
 548         if (ratio > 9.999)
 549                 return "---";
 550
 551         static char buf[16];
 552         snprintf(buf, sizeof(buf), "%.3f", ratio);
 553         return buf;
 554 }
 555
 556
 557 /// \brief      Get a comma-separated list of Check names
 558 ///
 559 /// The check names are translated with gettext except when in robot mode.
 560 ///
 561 /// \param      buf     Buffer to hold the resulting string
 562 /// \param      checks  Bit mask of Checks to print
 563 /// \param      space_after_comma
 564 ///                     It's better to not use spaces in table-like listings,
 565 ///                     but in more verbose formats a space after a comma
 566 ///                     is good for readability.
 567 static void
 568 get_check_names(char buf[CHECKS_STR_SIZE],
 569                 uint32_t checks, bool space_after_comma)
 570 {
 571         assert(checks != 0);
 572
 573         char *pos = buf;
 574         size_t left = CHECKS_STR_SIZE;
 575
 576         const char *sep = space_after_comma ? ", " : ",";
 577         bool comma = false;
 578
 579         for (size_t i = 0; i <= LZMA_CHECK_ID_MAX; ++i) {
 580                 if (checks & (UINT32_C(1) << i)) {
 581                         my_snprintf(&pos, &left, "%s%s",
 582                                         comma ? sep : "",
 583                                         opt_robot ? check_names[i]
 584                                                 : _(check_names[i]));
 585                         comma = true;
 586                 }
 587         }
 588
 589         return;
 590 }
 591
 592
 593 static bool
 594 print_info_basic(const xz_file_info *xfi, file_pair *pair)
 595 {
 596         static bool headings_displayed = false;
 597         if (!headings_displayed) {
 598                 headings_displayed = true;
 599                 // TRANSLATORS: These are column headings. From Strms (Streams)
 600                 // to Ratio, the columns are right aligned. Check and Filename
 601                 // are left aligned. If you need longer words, it's OK to
 602                 // use two lines here. Test with "xz -l foo.xz".
 603                 puts(_("Strms  Blocks   Compressed Uncompressed  Ratio  "
 604                                 "Check   Filename"));
 605         }
 606
 607         char checks[CHECKS_STR_SIZE];
 608         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 609
 610         const char *cols[7] = {
 611                 uint64_to_str(lzma_index_stream_count(xfi->idx), 0),
 612                 uint64_to_str(lzma_index_block_count(xfi->idx), 1),
 613                 uint64_to_nicestr(lzma_index_file_size(xfi->idx),
 614                         NICESTR_B, NICESTR_TIB, false, 2),
 615                 uint64_to_nicestr(lzma_index_uncompressed_size(xfi->idx),
 616                         NICESTR_B, NICESTR_TIB, false, 3),
 617                 get_ratio(lzma_index_file_size(xfi->idx),
 618                         lzma_index_uncompressed_size(xfi->idx)),
 619                 checks,
 620                 pair->src_name,
 621         };
 622         printf("%*s %*s  %*s  %*s  %*s  %-*s %s\n",
 623                         tuklib_mbstr_fw(cols[0], 5), cols[0],
 624                         tuklib_mbstr_fw(cols[1], 7), cols[1],
 625                         tuklib_mbstr_fw(cols[2], 11), cols[2],
 626                         tuklib_mbstr_fw(cols[3], 11), cols[3],
 627                         tuklib_mbstr_fw(cols[4], 5), cols[4],
 628                         tuklib_mbstr_fw(cols[5], 7), cols[5],
 629                         cols[6]);
 630
 631         return false;
 632 }
 633
 634
 635 static void
 636 print_adv_helper(uint64_t stream_count, uint64_t block_count,
 637                 uint64_t compressed_size, uint64_t uncompressed_size,
 638                 uint32_t checks, uint64_t stream_padding)
 639 {
 640         char checks_str[CHECKS_STR_SIZE];
 641         get_check_names(checks_str, checks, true);
 642
 643         printf(_("  Streams:            %s\n"),
 644                         uint64_to_str(stream_count, 0));
 645         printf(_("  Blocks:             %s\n"),
 646                         uint64_to_str(block_count, 0));
 647         printf(_("  Compressed size:    %s\n"),
 648                         uint64_to_nicestr(compressed_size,
 649                                 NICESTR_B, NICESTR_TIB, true, 0));
 650         printf(_("  Uncompressed size:  %s\n"),
 651                         uint64_to_nicestr(uncompressed_size,
 652                                 NICESTR_B, NICESTR_TIB, true, 0));
 653         printf(_("  Ratio:              %s\n"),
 654                         get_ratio(compressed_size, uncompressed_size));
 655         printf(_("  Check:              %s\n"), checks_str);
 656         printf(_("  Stream padding:     %s\n"),
 657                         uint64_to_nicestr(stream_padding,
 658                                 NICESTR_B, NICESTR_TIB, true, 0));
 659         return;
 660 }
 661
 662
 663 static bool
 664 print_info_adv(xz_file_info *xfi, file_pair *pair)
 665 {
 666         // Print the overall information.
 667         print_adv_helper(lzma_index_stream_count(xfi->idx),
 668                         lzma_index_block_count(xfi->idx),
 669                         lzma_index_file_size(xfi->idx),
 670                         lzma_index_uncompressed_size(xfi->idx),
 671                         lzma_index_checks(xfi->idx),
 672                         xfi->stream_padding);
 673
 674         // Size of the biggest Check. This is used to calculate the width
 675         // of the CheckVal field. The table would get insanely wide if
 676         // we always reserved space for 64-byte Check (128 chars as hex).
 677         uint32_t check_max = 0;
 678
 679         // Print information about the Streams.
 680         //
 681         // TRANSLATORS: The second line is column headings. All except
 682         // Check are right aligned; Check is left aligned. Test with
 683         // "xz -lv foo.xz".
 684         puts(_("  Streams:\n    Stream    Blocks"
 685                         "      CompOffset    UncompOffset"
 686                         "        CompSize      UncompSize  Ratio"
 687                         "  Check      Padding"));
 688
 689         lzma_index_iter iter;
 690         lzma_index_iter_init(&iter, xfi->idx);
 691
 692         while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM)) {
 693                 const char *cols1[4] = {
 694                         uint64_to_str(iter.stream.number, 0),
 695                         uint64_to_str(iter.stream.block_count, 1),
 696                         uint64_to_str(iter.stream.compressed_offset, 2),
 697                         uint64_to_str(iter.stream.uncompressed_offset, 3),
 698                 };
 699                 printf("    %*s %*s %*s %*s ",
 700                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 701                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 702                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 703                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 704
 705                 const char *cols2[5] = {
 706                         uint64_to_str(iter.stream.compressed_size, 0),
 707                         uint64_to_str(iter.stream.uncompressed_size, 1),
 708                         get_ratio(iter.stream.compressed_size,
 709                                 iter.stream.uncompressed_size),
 710                         _(check_names[iter.stream.flags->check]),
 711                         uint64_to_str(iter.stream.padding, 2),
 712                 };
 713                 printf("%*s %*s  %*s  %-*s %*s\n",
 714                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 715                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 716                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 717                                 tuklib_mbstr_fw(cols2[3], 10), cols2[3],
 718                                 tuklib_mbstr_fw(cols2[4], 7), cols2[4]);
 719
 720                 // Update the maximum Check size.
 721                 if (lzma_check_size(iter.stream.flags->check) > check_max)
 722                         check_max = lzma_check_size(iter.stream.flags->check);
 723         }
 724
 725         // Cache the verbosity level to a local variable.
 726         const bool detailed = message_verbosity_get() >= V_DEBUG;
 727
 728         // Information collected from Block Headers
 729         block_header_info bhi;
 730
 731         // Print information about the Blocks but only if there is
 732         // at least one Block.
 733         if (lzma_index_block_count(xfi->idx) > 0) {
 734                 // Calculate the width of the CheckVal field.
 735                 const int checkval_width = my_max(8, 2 * check_max);
 736
 737                 // TRANSLATORS: The second line is column headings. All
 738                 // except Check are right aligned; Check is left aligned.
 739                 printf(_("  Blocks:\n    Stream     Block"
 740                         "      CompOffset    UncompOffset"
 741                         "       TotalSize      UncompSize  Ratio  Check"));
 742
 743                 if (detailed) {
 744                         // TRANSLATORS: These are additional column headings
 745                         // for the most verbose listing mode. CheckVal
 746                         // (Check value), Flags, and Filters are left aligned.
 747                         // Header (Block Header Size), CompSize, and MemUsage
 748                         // are right aligned. %*s is replaced with 0-120
 749                         // spaces to make the CheckVal column wide enough.
 750                         // Test with "xz -lvv foo.xz".
 751                         printf(_("      CheckVal %*s Header  Flags        "
 752                                         "CompSize    MemUsage  Filters"),
 753                                         checkval_width - 8, "");
 754                 }
 755
 756                 putchar('\n');
 757
 758                 lzma_index_iter_init(&iter, xfi->idx);
 759
 760                 // Iterate over the Blocks.
 761                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 762                         if (detailed && parse_details(pair, &iter, &bhi, xfi))
 763                                         return true;
 764
 765                         const char *cols1[4] = {
 766                                 uint64_to_str(iter.stream.number, 0),
 767                                 uint64_to_str(
 768                                         iter.block.number_in_stream, 1),
 769                                 uint64_to_str(
 770                                         iter.block.compressed_file_offset, 2),
 771                                 uint64_to_str(
 772                                         iter.block.uncompressed_file_offset, 3)
 773                         };
 774                         printf("    %*s %*s %*s %*s ",
 775                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
 776                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
 777                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
 778                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
 779
 780                         const char *cols2[4] = {
 781                                 uint64_to_str(iter.block.total_size, 0),
 782                                 uint64_to_str(iter.block.uncompressed_size,
 783                                                 1),
 784                                 get_ratio(iter.block.total_size,
 785                                         iter.block.uncompressed_size),
 786                                 _(check_names[iter.stream.flags->check])
 787                         };
 788                         printf("%*s %*s  %*s  %-*s",
 789                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
 790                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
 791                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
 792                                 tuklib_mbstr_fw(cols2[3], detailed ? 11 : 1),
 793                                         cols2[3]);
 794
 795                         if (detailed) {
 796                                 const lzma_vli compressed_size
 797                                                 = iter.block.unpadded_size
 798                                                 - bhi.header_size
 799                                                 - lzma_check_size(
 800                                                 iter.stream.flags->check);
 801
 802                                 const char *cols3[6] = {
 803                                         check_value,
 804                                         uint64_to_str(bhi.header_size, 0),
 805                                         bhi.flags,
 806                                         uint64_to_str(compressed_size, 1),
 807                                         uint64_to_str(
 808                                                 round_up_to_mib(bhi.memusage),
 809                                                 2),
 810                                         bhi.filter_chain
 811                                 };
 812                                 // Show MiB for memory usage, because it
 813                                 // is the only size which is not in bytes.
 814                                 printf("%-*s  %*s  %-5s %*s %*s MiB  %s",
 815                                         checkval_width, cols3[0],
 816                                         tuklib_mbstr_fw(cols3[1], 6), cols3[1],
 817                                         cols3[2],
 818                                         tuklib_mbstr_fw(cols3[3], 15),
 819                                                 cols3[3],
 820                                         tuklib_mbstr_fw(cols3[4], 7), cols3[4],
 821                                         cols3[5]);
 822                         }
 823
 824                         putchar('\n');
 825                 }
 826         }
 827
 828         if (detailed) {
 829                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
 830                                 round_up_to_mib(xfi->memusage_max), 0));
 831                 printf(_("  Sizes in headers:   %s\n"),
 832                                 xfi->all_have_sizes ? _("Yes") : _("No"));
 833         }
 834
 835         return false;
 836 }
 837
 838
 839 static bool
 840 print_info_robot(xz_file_info *xfi, file_pair *pair)
 841 {
 842         char checks[CHECKS_STR_SIZE];
 843         get_check_names(checks, lzma_index_checks(xfi->idx), false);
 844
 845         printf("name\t%s\n", pair->src_name);
 846
 847         printf("file\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 848                         "\t%s\t%s\t%" PRIu64 "\n",
 849                         lzma_index_stream_count(xfi->idx),
 850                         lzma_index_block_count(xfi->idx),
 851                         lzma_index_file_size(xfi->idx),
 852                         lzma_index_uncompressed_size(xfi->idx),
 853                         get_ratio(lzma_index_file_size(xfi->idx),
 854                                 lzma_index_uncompressed_size(xfi->idx)),
 855                         checks,
 856                         xfi->stream_padding);
 857
 858         if (message_verbosity_get() >= V_VERBOSE) {
 859                 lzma_index_iter iter;
 860                 lzma_index_iter_init(&iter, xfi->idx);
 861
 862                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM))
 863                         printf("stream\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 864                                 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 865                                 "\t%s\t%s\t%" PRIu64 "\n",
 866                                 iter.stream.number,
 867                                 iter.stream.block_count,
 868                                 iter.stream.compressed_offset,
 869                                 iter.stream.uncompressed_offset,
 870                                 iter.stream.compressed_size,
 871                                 iter.stream.uncompressed_size,
 872                                 get_ratio(iter.stream.compressed_size,
 873                                         iter.stream.uncompressed_size),
 874                                 check_names[iter.stream.flags->check],
 875                                 iter.stream.padding);
 876
 877                 lzma_index_iter_rewind(&iter);
 878                 block_header_info bhi;
 879
 880                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
 881                         if (message_verbosity_get() >= V_DEBUG
 882                                         && parse_details(
 883                                                 pair, &iter, &bhi, xfi))
 884                                 return true;
 885
 886                         printf("block\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
 887                                         "\t%" PRIu64 "\t%" PRIu64
 888                                         "\t%" PRIu64 "\t%" PRIu64 "\t%s\t%s",
 889                                         iter.stream.number,
 890                                         iter.block.number_in_stream,
 891                                         iter.block.number_in_file,
 892                                         iter.block.compressed_file_offset,
 893                                         iter.block.uncompressed_file_offset,
 894                                         iter.block.total_size,
 895                                         iter.block.uncompressed_size,
 896                                         get_ratio(iter.block.total_size,
 897                                                 iter.block.uncompressed_size),
 898                                         check_names[iter.stream.flags->check]);
 899
 900                         if (message_verbosity_get() >= V_DEBUG)
 901                                 printf("\t%s\t%" PRIu32 "\t%s\t%" PRIu64
 902                                                 "\t%" PRIu64 "\t%s",
 903                                                 check_value,
 904                                                 bhi.header_size,
 905                                                 bhi.flags,
 906                                                 bhi.compressed_size,
 907                                                 bhi.memusage,
 908                                                 bhi.filter_chain);
 909
 910                         putchar('\n');
 911                 }
 912         }
 913
 914         if (message_verbosity_get() >= V_DEBUG)
 915                 printf("summary\t%" PRIu64 "\t%s\n",
 916                                 xfi->memusage_max,
 917                                 xfi->all_have_sizes ? "yes" : "no");
 918
 919         return false;
 920 }
 921
 922
 923 static void
 924 update_totals(const xz_file_info *xfi)
 925 {
 926         // TODO: Integer overflow checks
 927         ++totals.files;
 928         totals.streams += lzma_index_stream_count(xfi->idx);
 929         totals.blocks += lzma_index_block_count(xfi->idx);
 930         totals.compressed_size += lzma_index_file_size(xfi->idx);
 931         totals.uncompressed_size += lzma_index_uncompressed_size(xfi->idx);
 932         totals.stream_padding += xfi->stream_padding;
 933         totals.checks |= lzma_index_checks(xfi->idx);
 934
 935         if (totals.memusage_max < xfi->memusage_max)
 936                 totals.memusage_max = xfi->memusage_max;
 937
 938         totals.all_have_sizes &= xfi->all_have_sizes;
 939
 940         return;
 941 }
 942
 943
 944 static void
 945 print_totals_basic(void)
 946 {
 947         // Print a separator line.
 948         char line[80];
 949         memset(line, '-', sizeof(line));
 950         line[sizeof(line) - 1] = '\0';
 951         puts(line);
 952
 953         // Get the check names.
 954         char checks[CHECKS_STR_SIZE];
 955         get_check_names(checks, totals.checks, false);
 956
 957         // Print the totals except the file count, which needs
 958         // special handling.
 959         printf("%5s %7s  %11s  %11s  %5s  %-7s ",
 960                         uint64_to_str(totals.streams, 0),
 961                         uint64_to_str(totals.blocks, 1),
 962                         uint64_to_nicestr(totals.compressed_size,
 963                                 NICESTR_B, NICESTR_TIB, false, 2),
 964                         uint64_to_nicestr(totals.uncompressed_size,
 965                                 NICESTR_B, NICESTR_TIB, false, 3),
 966                         get_ratio(totals.compressed_size,
 967                                 totals.uncompressed_size),
 968                         checks);
 969
 970         // Since we print totals only when there are at least two files,
 971         // the English message will always use "%s files". But some other
 972         // languages need different forms for different plurals so we
 973         // have to translate this with ngettext().
 974         //
 975         // TRANSLATORS: %s is an integer. Only the plural form of this
 976         // message is used (e.g. "2 files"). Test with "xz -l foo.xz bar.xz".
 977         printf(ngettext("%s file\n", "%s files\n",
 978                         totals.files <= ULONG_MAX ? totals.files
 979                                 : (totals.files % 1000000) + 1000000),
 980                         uint64_to_str(totals.files, 0));
 981
 982         return;
 983 }
 984
 985
 986 static void
 987 print_totals_adv(void)
 988 {
 989         putchar('\n');
 990         puts(_("Totals:"));
 991         printf(_("  Number of files:    %s\n"),
 992                         uint64_to_str(totals.files, 0));
 993         print_adv_helper(totals.streams, totals.blocks,
 994                         totals.compressed_size, totals.uncompressed_size,
 995                         totals.checks, totals.stream_padding);
 996
 997         if (message_verbosity_get() >= V_DEBUG) {
 998                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
 999                                 round_up_to_mib(totals.memusage_max), 0));
1000                 printf(_("  Sizes in headers:   %s\n"),
1001                                 totals.all_have_sizes ? _("Yes") : _("No"));
1002         }
1003
1004         return;
1005 }
1006
1007
1008 static void
1009 print_totals_robot(void)
1010 {
1011         char checks[CHECKS_STR_SIZE];
1012         get_check_names(checks, totals.checks, false);
1013
1014         printf("totals\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
1015                         "\t%s\t%s\t%" PRIu64 "\t%" PRIu64,
1016                         totals.streams,
1017                         totals.blocks,
1018                         totals.compressed_size,
1019                         totals.uncompressed_size,
1020                         get_ratio(totals.compressed_size,
1021                                 totals.uncompressed_size),
1022                         checks,
1023                         totals.stream_padding,
1024                         totals.files);
1025
1026         if (message_verbosity_get() >= V_DEBUG)
1027                 printf("\t%" PRIu64 "\t%s",
1028                                 totals.memusage_max,
1029                                 totals.all_have_sizes ? "yes" : "no");
1030
1031         putchar('\n');
1032
1033         return;
1034 }
1035
1036
1037 extern void
1038 list_totals(void)
1039 {
1040         if (opt_robot) {
1041                 // Always print totals in --robot mode. It can be convenient
1042                 // in some cases and doesn't complicate usage of the
1043                 // single-file case much.
1044                 print_totals_robot();
1045
1046         } else if (totals.files > 1) {
1047                 // For non-robot mode, totals are printed only if there
1048                 // is more than one file.
1049                 if (message_verbosity_get() <= V_WARNING)
1050                         print_totals_basic();
1051                 else
1052                         print_totals_adv();
1053         }
1054
1055         return;
1056 }
1057
1058
1059 extern void
1060 list_file(const char *filename)
1061 {
1062         if (opt_format != FORMAT_XZ && opt_format != FORMAT_AUTO)
1063                 message_fatal(_("--list works only on .xz files "
1064                                 "(--format=xz or --format=auto)"));
1065
1066         message_filename(filename);
1067
1068         if (filename == stdin_filename) {
1069                 message_error(_("--list does not support reading from "
1070                                 "standard input"));
1071                 return;
1072         }
1073
1074         // Unset opt_stdout so that io_open_src() won't accept special files.
1075         // Set opt_force so that io_open_src() will follow symlinks.
1076         opt_stdout = false;
1077         opt_force = true;
1078         file_pair *pair = io_open_src(filename);
1079         if (pair == NULL)
1080                 return;
1081
1082         xz_file_info xfi = XZ_FILE_INFO_INIT;
1083         if (!parse_indexes(&xfi, pair)) {
1084                 bool fail;
1085
1086                 // We have three main modes:
1087                 //  - --robot, which has submodes if --verbose is specified
1088                 //    once or twice
1089                 //  - Normal --list without --verbose
1090                 //  - --list with one or two --verbose
1091                 if (opt_robot)
1092                         fail = print_info_robot(&xfi, pair);
1093                 else if (message_verbosity_get() <= V_WARNING)
1094                         fail = print_info_basic(&xfi, pair);
1095                 else
1096                         fail = print_info_adv(&xfi, pair);
1097
1098                 // Update the totals that are displayed after all
1099                 // the individual files have been listed. Don't count
1100                 // broken files.
1101                 if (!fail)
1102                         update_totals(&xfi);
1103
1104                 lzma_index_end(xfi.idx, NULL);
1105         }
1106
1107         io_close(pair, false);
1108         return;
1109 }