Upgrade xz from 5.2.2 to 5.2.4 on the vendor branch.
[dragonfly.git] / 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         /// Oldest XZ Utils version that will decompress the file
33         uint32_t min_version;
34
35 } xz_file_info;
36
37 #define XZ_FILE_INFO_INIT { NULL, 0, 0, true, 50000002 }
38
39
40 /// Information about a .xz Block
41 typedef struct {
42         /// Size of the Block Header
43         uint32_t header_size;
44
45         /// A few of the Block Flags as a string
46         char flags[3];
47
48         /// Size of the Compressed Data field in the Block
49         lzma_vli compressed_size;
50
51         /// Decoder memory usage for this Block
52         uint64_t memusage;
53
54         /// The filter chain of this Block in human-readable form
55         char filter_chain[FILTERS_STR_SIZE];
56
57 } block_header_info;
58
59
60 /// Check ID to string mapping
61 static const char check_names[LZMA_CHECK_ID_MAX + 1][12] = {
62         // TRANSLATORS: Indicates that there is no integrity check.
63         // This string is used in tables, so the width must not
64         // exceed ten columns with a fixed-width font.
65         N_("None"),
66         "CRC32",
67         // TRANSLATORS: Indicates that integrity check name is not known,
68         // but the Check ID is known (here 2). This and other "Unknown-N"
69         // strings are used in tables, so the width must not exceed ten
70         // columns with a fixed-width font. It's OK to omit the dash if
71         // you need space for one extra letter, but don't use spaces.
72         N_("Unknown-2"),
73         N_("Unknown-3"),
74         "CRC64",
75         N_("Unknown-5"),
76         N_("Unknown-6"),
77         N_("Unknown-7"),
78         N_("Unknown-8"),
79         N_("Unknown-9"),
80         "SHA-256",
81         N_("Unknown-11"),
82         N_("Unknown-12"),
83         N_("Unknown-13"),
84         N_("Unknown-14"),
85         N_("Unknown-15"),
86 };
87
88 /// Buffer size for get_check_names(). This may be a bit ridiculous,
89 /// but at least it's enough if some language needs many multibyte chars.
90 #define CHECKS_STR_SIZE 1024
91
92
93 /// Value of the Check field as hexadecimal string.
94 /// This is set by parse_check_value().
95 static char check_value[2 * LZMA_CHECK_SIZE_MAX + 1];
96
97
98 /// Totals that are displayed if there was more than one file.
99 /// The "files" counter is also used in print_info_adv() to show
100 /// the file number.
101 static struct {
102         uint64_t files;
103         uint64_t streams;
104         uint64_t blocks;
105         uint64_t compressed_size;
106         uint64_t uncompressed_size;
107         uint64_t stream_padding;
108         uint64_t memusage_max;
109         uint32_t checks;
110         uint32_t min_version;
111         bool all_have_sizes;
112 } totals = { 0, 0, 0, 0, 0, 0, 0, 0, 50000002, true };
113
114
115 /// Convert XZ Utils version number to a string.
116 static const char *
117 xz_ver_to_str(uint32_t ver)
118 {
119         static char buf[32];
120
121         unsigned int major = ver / 10000000U;
122         ver -= major * 10000000U;
123
124         unsigned int minor = ver / 10000U;
125         ver -= minor * 10000U;
126
127         unsigned int patch = ver / 10U;
128         ver -= patch * 10U;
129
130         const char *stability = ver == 0 ? "alpha" : ver == 1 ? "beta" : "";
131
132         snprintf(buf, sizeof(buf), "%u.%u.%u%s",
133                         major, minor, patch, stability);
134         return buf;
135 }
136
137
138 /// \brief      Parse the Index(es) from the given .xz file
139 ///
140 /// \param      xfi     Pointer to structure where the decoded information
141 ///                     is stored.
142 /// \param      pair    Input file
143 ///
144 /// \return     On success, false is returned. On error, true is returned.
145 ///
146 // TODO: This function is pretty big. liblzma should have a function that
147 // takes a callback function to parse the Index(es) from a .xz file to make
148 // it easy for applications.
149 static bool
150 parse_indexes(xz_file_info *xfi, file_pair *pair)
151 {
152         if (pair->src_st.st_size <= 0) {
153                 message_error(_("%s: File is empty"), pair->src_name);
154                 return true;
155         }
156
157         if (pair->src_st.st_size < 2 * LZMA_STREAM_HEADER_SIZE) {
158                 message_error(_("%s: Too small to be a valid .xz file"),
159                                 pair->src_name);
160                 return true;
161         }
162
163         io_buf buf;
164         lzma_stream_flags header_flags;
165         lzma_stream_flags footer_flags;
166         lzma_ret ret;
167
168         // lzma_stream for the Index decoder
169         lzma_stream strm = LZMA_STREAM_INIT;
170
171         // All Indexes decoded so far
172         lzma_index *combined_index = NULL;
173
174         // The Index currently being decoded
175         lzma_index *this_index = NULL;
176
177         // Current position in the file. We parse the file backwards so
178         // initialize it to point to the end of the file.
179         off_t pos = pair->src_st.st_size;
180
181         // Each loop iteration decodes one Index.
182         do {
183                 // Check that there is enough data left to contain at least
184                 // the Stream Header and Stream Footer. This check cannot
185                 // fail in the first pass of this loop.
186                 if (pos < 2 * LZMA_STREAM_HEADER_SIZE) {
187                         message_error("%s: %s", pair->src_name,
188                                         message_strm(LZMA_DATA_ERROR));
189                         goto error;
190                 }
191
192                 pos -= LZMA_STREAM_HEADER_SIZE;
193                 lzma_vli stream_padding = 0;
194
195                 // Locate the Stream Footer. There may be Stream Padding which
196                 // we must skip when reading backwards.
197                 while (true) {
198                         if (pos < LZMA_STREAM_HEADER_SIZE) {
199                                 message_error("%s: %s", pair->src_name,
200                                                 message_strm(
201                                                         LZMA_DATA_ERROR));
202                                 goto error;
203                         }
204
205                         if (io_pread(pair, &buf,
206                                         LZMA_STREAM_HEADER_SIZE, pos))
207                                 goto error;
208
209                         // Stream Padding is always a multiple of four bytes.
210                         int i = 2;
211                         if (buf.u32[i] != 0)
212                                 break;
213
214                         // To avoid calling io_pread() for every four bytes
215                         // of Stream Padding, take advantage that we read
216                         // 12 bytes (LZMA_STREAM_HEADER_SIZE) already and
217                         // check them too before calling io_pread() again.
218                         do {
219                                 stream_padding += 4;
220                                 pos -= 4;
221                                 --i;
222                         } while (i >= 0 && buf.u32[i] == 0);
223                 }
224
225                 // Decode the Stream Footer.
226                 ret = lzma_stream_footer_decode(&footer_flags, buf.u8);
227                 if (ret != LZMA_OK) {
228                         message_error("%s: %s", pair->src_name,
229                                         message_strm(ret));
230                         goto error;
231                 }
232
233                 // Check that the Stream Footer doesn't specify something
234                 // that we don't support. This can only happen if the xz
235                 // version is older than liblzma and liblzma supports
236                 // something new.
237                 //
238                 // It is enough to check Stream Footer. Stream Header must
239                 // match when it is compared against Stream Footer with
240                 // lzma_stream_flags_compare().
241                 if (footer_flags.version != 0) {
242                         message_error("%s: %s", pair->src_name,
243                                         message_strm(LZMA_OPTIONS_ERROR));
244                         goto error;
245                 }
246
247                 // Check that the size of the Index field looks sane.
248                 lzma_vli index_size = footer_flags.backward_size;
249                 if ((lzma_vli)(pos) < index_size + LZMA_STREAM_HEADER_SIZE) {
250                         message_error("%s: %s", pair->src_name,
251                                         message_strm(LZMA_DATA_ERROR));
252                         goto error;
253                 }
254
255                 // Set pos to the beginning of the Index.
256                 pos -= index_size;
257
258                 // See how much memory we can use for decoding this Index.
259                 uint64_t memlimit = hardware_memlimit_get(MODE_LIST);
260                 uint64_t memused = 0;
261                 if (combined_index != NULL) {
262                         memused = lzma_index_memused(combined_index);
263                         if (memused > memlimit)
264                                 message_bug();
265
266                         memlimit -= memused;
267                 }
268
269                 // Decode the Index.
270                 ret = lzma_index_decoder(&strm, &this_index, memlimit);
271                 if (ret != LZMA_OK) {
272                         message_error("%s: %s", pair->src_name,
273                                         message_strm(ret));
274                         goto error;
275                 }
276
277                 do {
278                         // Don't give the decoder more input than the
279                         // Index size.
280                         strm.avail_in = my_min(IO_BUFFER_SIZE, index_size);
281                         if (io_pread(pair, &buf, strm.avail_in, pos))
282                                 goto error;
283
284                         pos += strm.avail_in;
285                         index_size -= strm.avail_in;
286
287                         strm.next_in = buf.u8;
288                         ret = lzma_code(&strm, LZMA_RUN);
289
290                 } while (ret == LZMA_OK);
291
292                 // If the decoding seems to be successful, check also that
293                 // the Index decoder consumed as much input as indicated
294                 // by the Backward Size field.
295                 if (ret == LZMA_STREAM_END)
296                         if (index_size != 0 || strm.avail_in != 0)
297                                 ret = LZMA_DATA_ERROR;
298
299                 if (ret != LZMA_STREAM_END) {
300                         // LZMA_BUFFER_ERROR means that the Index decoder
301                         // would have liked more input than what the Index
302                         // size should be according to Stream Footer.
303                         // The message for LZMA_DATA_ERROR makes more
304                         // sense in that case.
305                         if (ret == LZMA_BUF_ERROR)
306                                 ret = LZMA_DATA_ERROR;
307
308                         message_error("%s: %s", pair->src_name,
309                                         message_strm(ret));
310
311                         // If the error was too low memory usage limit,
312                         // show also how much memory would have been needed.
313                         if (ret == LZMA_MEMLIMIT_ERROR) {
314                                 uint64_t needed = lzma_memusage(&strm);
315                                 if (UINT64_MAX - needed < memused)
316                                         needed = UINT64_MAX;
317                                 else
318                                         needed += memused;
319
320                                 message_mem_needed(V_ERROR, needed);
321                         }
322
323                         goto error;
324                 }
325
326                 // Decode the Stream Header and check that its Stream Flags
327                 // match the Stream Footer.
328                 pos -= footer_flags.backward_size + LZMA_STREAM_HEADER_SIZE;
329                 if ((lzma_vli)(pos) < lzma_index_total_size(this_index)) {
330                         message_error("%s: %s", pair->src_name,
331                                         message_strm(LZMA_DATA_ERROR));
332                         goto error;
333                 }
334
335                 pos -= lzma_index_total_size(this_index);
336                 if (io_pread(pair, &buf, LZMA_STREAM_HEADER_SIZE, pos))
337                         goto error;
338
339                 ret = lzma_stream_header_decode(&header_flags, buf.u8);
340                 if (ret != LZMA_OK) {
341                         message_error("%s: %s", pair->src_name,
342                                         message_strm(ret));
343                         goto error;
344                 }
345
346                 ret = lzma_stream_flags_compare(&header_flags, &footer_flags);
347                 if (ret != LZMA_OK) {
348                         message_error("%s: %s", pair->src_name,
349                                         message_strm(ret));
350                         goto error;
351                 }
352
353                 // Store the decoded Stream Flags into this_index. This is
354                 // needed so that we can print which Check is used in each
355                 // Stream.
356                 ret = lzma_index_stream_flags(this_index, &footer_flags);
357                 if (ret != LZMA_OK)
358                         message_bug();
359
360                 // Store also the size of the Stream Padding field. It is
361                 // needed to show the offsets of the Streams correctly.
362                 ret = lzma_index_stream_padding(this_index, stream_padding);
363                 if (ret != LZMA_OK)
364                         message_bug();
365
366                 if (combined_index != NULL) {
367                         // Append the earlier decoded Indexes
368                         // after this_index.
369                         ret = lzma_index_cat(
370                                         this_index, combined_index, NULL);
371                         if (ret != LZMA_OK) {
372                                 message_error("%s: %s", pair->src_name,
373                                                 message_strm(ret));
374                                 goto error;
375                         }
376                 }
377
378                 combined_index = this_index;
379                 this_index = NULL;
380
381                 xfi->stream_padding += stream_padding;
382
383         } while (pos > 0);
384
385         lzma_end(&strm);
386
387         // All OK. Make combined_index available to the caller.
388         xfi->idx = combined_index;
389         return false;
390
391 error:
392         // Something went wrong, free the allocated memory.
393         lzma_end(&strm);
394         lzma_index_end(combined_index, NULL);
395         lzma_index_end(this_index, NULL);
396         return true;
397 }
398
399
400 /// \brief      Parse the Block Header
401 ///
402 /// The result is stored into *bhi. The caller takes care of initializing it.
403 ///
404 /// \return     False on success, true on error.
405 static bool
406 parse_block_header(file_pair *pair, const lzma_index_iter *iter,
407                 block_header_info *bhi, xz_file_info *xfi)
408 {
409 #if IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
410 #       error IO_BUFFER_SIZE < LZMA_BLOCK_HEADER_SIZE_MAX
411 #endif
412
413         // Get the whole Block Header with one read, but don't read past
414         // the end of the Block (or even its Check field).
415         const uint32_t size = my_min(iter->block.total_size
416                                 - lzma_check_size(iter->stream.flags->check),
417                         LZMA_BLOCK_HEADER_SIZE_MAX);
418         io_buf buf;
419         if (io_pread(pair, &buf, size, iter->block.compressed_file_offset))
420                 return true;
421
422         // Zero would mean Index Indicator and thus not a valid Block.
423         if (buf.u8[0] == 0)
424                 goto data_error;
425
426         // Initialize the block structure and decode Block Header Size.
427         lzma_filter filters[LZMA_FILTERS_MAX + 1];
428         lzma_block block;
429         block.version = 0;
430         block.check = iter->stream.flags->check;
431         block.filters = filters;
432
433         block.header_size = lzma_block_header_size_decode(buf.u8[0]);
434         if (block.header_size > size)
435                 goto data_error;
436
437         // Decode the Block Header.
438         switch (lzma_block_header_decode(&block, NULL, buf.u8)) {
439         case LZMA_OK:
440                 break;
441
442         case LZMA_OPTIONS_ERROR:
443                 message_error("%s: %s", pair->src_name,
444                                 message_strm(LZMA_OPTIONS_ERROR));
445                 return true;
446
447         case LZMA_DATA_ERROR:
448                 goto data_error;
449
450         default:
451                 message_bug();
452         }
453
454         // Check the Block Flags. These must be done before calling
455         // lzma_block_compressed_size(), because it overwrites
456         // block.compressed_size.
457         bhi->flags[0] = block.compressed_size != LZMA_VLI_UNKNOWN
458                         ? 'c' : '-';
459         bhi->flags[1] = block.uncompressed_size != LZMA_VLI_UNKNOWN
460                         ? 'u' : '-';
461         bhi->flags[2] = '\0';
462
463         // Collect information if all Blocks have both Compressed Size
464         // and Uncompressed Size fields. They can be useful e.g. for
465         // multi-threaded decompression so it can be useful to know it.
466         xfi->all_have_sizes &= block.compressed_size != LZMA_VLI_UNKNOWN
467                         && block.uncompressed_size != LZMA_VLI_UNKNOWN;
468
469         // Validate or set block.compressed_size.
470         switch (lzma_block_compressed_size(&block,
471                         iter->block.unpadded_size)) {
472         case LZMA_OK:
473                 // Validate also block.uncompressed_size if it is present.
474                 // If it isn't present, there's no need to set it since
475                 // we aren't going to actually decompress the Block; if
476                 // we were decompressing, then we should set it so that
477                 // the Block decoder could validate the Uncompressed Size
478                 // that was stored in the Index.
479                 if (block.uncompressed_size == LZMA_VLI_UNKNOWN
480                                 || block.uncompressed_size
481                                         == iter->block.uncompressed_size)
482                         break;
483
484                 // If the above fails, the file is corrupt so
485                 // LZMA_DATA_ERROR is a good error code.
486
487         // Fall through
488
489         case LZMA_DATA_ERROR:
490                 // Free the memory allocated by lzma_block_header_decode().
491                 for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
492                         free(filters[i].options);
493
494                 goto data_error;
495
496         default:
497                 message_bug();
498         }
499
500         // Copy the known sizes.
501         bhi->header_size = block.header_size;
502         bhi->compressed_size = block.compressed_size;
503
504         // Calculate the decoder memory usage and update the maximum
505         // memory usage of this Block.
506         bhi->memusage = lzma_raw_decoder_memusage(filters);
507         if (xfi->memusage_max < bhi->memusage)
508                 xfi->memusage_max = bhi->memusage;
509
510         // Determine the minimum XZ Utils version that supports this Block.
511         //
512         // Currently the only thing that 5.0.0 doesn't support is empty
513         // LZMA2 Block. This decoder bug was fixed in 5.0.2.
514         {
515                 size_t i = 0;
516                 while (filters[i + 1].id != LZMA_VLI_UNKNOWN)
517                         ++i;
518
519                 if (filters[i].id == LZMA_FILTER_LZMA2
520                                 && iter->block.uncompressed_size == 0
521                                 && xfi->min_version < 50000022U)
522                         xfi->min_version = 50000022U;
523         }
524
525         // Convert the filter chain to human readable form.
526         message_filters_to_str(bhi->filter_chain, filters, false);
527
528         // Free the memory allocated by lzma_block_header_decode().
529         for (size_t i = 0; filters[i].id != LZMA_VLI_UNKNOWN; ++i)
530                 free(filters[i].options);
531
532         return false;
533
534 data_error:
535         // Show the error message.
536         message_error("%s: %s", pair->src_name,
537                         message_strm(LZMA_DATA_ERROR));
538         return true;
539 }
540
541
542 /// \brief      Parse the Check field and put it into check_value[]
543 ///
544 /// \return     False on success, true on error.
545 static bool
546 parse_check_value(file_pair *pair, const lzma_index_iter *iter)
547 {
548         // Don't read anything from the file if there is no integrity Check.
549         if (iter->stream.flags->check == LZMA_CHECK_NONE) {
550                 snprintf(check_value, sizeof(check_value), "---");
551                 return false;
552         }
553
554         // Locate and read the Check field.
555         const uint32_t size = lzma_check_size(iter->stream.flags->check);
556         const off_t offset = iter->block.compressed_file_offset
557                         + iter->block.total_size - size;
558         io_buf buf;
559         if (io_pread(pair, &buf, size, offset))
560                 return true;
561
562         // CRC32 and CRC64 are in little endian. Guess that all the future
563         // 32-bit and 64-bit Check values are little endian too. It shouldn't
564         // be a too big problem if this guess is wrong.
565         if (size == 4)
566                 snprintf(check_value, sizeof(check_value),
567                                 "%08" PRIx32, conv32le(buf.u32[0]));
568         else if (size == 8)
569                 snprintf(check_value, sizeof(check_value),
570                                 "%016" PRIx64, conv64le(buf.u64[0]));
571         else
572                 for (size_t i = 0; i < size; ++i)
573                         snprintf(check_value + i * 2, 3, "%02x", buf.u8[i]);
574
575         return false;
576 }
577
578
579 /// \brief      Parse detailed information about a Block
580 ///
581 /// Since this requires seek(s), listing information about all Blocks can
582 /// be slow.
583 ///
584 /// \param      pair    Input file
585 /// \param      iter    Location of the Block whose Check value should
586 ///                     be printed.
587 /// \param      bhi     Pointer to structure where to store the information
588 ///                     about the Block Header field.
589 ///
590 /// \return     False on success, true on error. If an error occurs,
591 ///             the error message is printed too so the caller doesn't
592 ///             need to worry about that.
593 static bool
594 parse_details(file_pair *pair, const lzma_index_iter *iter,
595                 block_header_info *bhi, xz_file_info *xfi)
596 {
597         if (parse_block_header(pair, iter, bhi, xfi))
598                 return true;
599
600         if (parse_check_value(pair, iter))
601                 return true;
602
603         return false;
604 }
605
606
607 /// \brief      Get the compression ratio
608 ///
609 /// This has slightly different format than that is used in message.c.
610 static const char *
611 get_ratio(uint64_t compressed_size, uint64_t uncompressed_size)
612 {
613         if (uncompressed_size == 0)
614                 return "---";
615
616         const double ratio = (double)(compressed_size)
617                         / (double)(uncompressed_size);
618         if (ratio > 9.999)
619                 return "---";
620
621         static char buf[16];
622         snprintf(buf, sizeof(buf), "%.3f", ratio);
623         return buf;
624 }
625
626
627 /// \brief      Get a comma-separated list of Check names
628 ///
629 /// The check names are translated with gettext except when in robot mode.
630 ///
631 /// \param      buf     Buffer to hold the resulting string
632 /// \param      checks  Bit mask of Checks to print
633 /// \param      space_after_comma
634 ///                     It's better to not use spaces in table-like listings,
635 ///                     but in more verbose formats a space after a comma
636 ///                     is good for readability.
637 static void
638 get_check_names(char buf[CHECKS_STR_SIZE],
639                 uint32_t checks, bool space_after_comma)
640 {
641         // If we get called when there are no Checks to print, set checks
642         // to 1 so that we print "None". This can happen in the robot mode
643         // when printing the totals line if there are no valid input files.
644         if (checks == 0)
645                 checks = 1;
646
647         char *pos = buf;
648         size_t left = CHECKS_STR_SIZE;
649
650         const char *sep = space_after_comma ? ", " : ",";
651         bool comma = false;
652
653         for (size_t i = 0; i <= LZMA_CHECK_ID_MAX; ++i) {
654                 if (checks & (UINT32_C(1) << i)) {
655                         my_snprintf(&pos, &left, "%s%s",
656                                         comma ? sep : "",
657                                         opt_robot ? check_names[i]
658                                                 : _(check_names[i]));
659                         comma = true;
660                 }
661         }
662
663         return;
664 }
665
666
667 static bool
668 print_info_basic(const xz_file_info *xfi, file_pair *pair)
669 {
670         static bool headings_displayed = false;
671         if (!headings_displayed) {
672                 headings_displayed = true;
673                 // TRANSLATORS: These are column headings. From Strms (Streams)
674                 // to Ratio, the columns are right aligned. Check and Filename
675                 // are left aligned. If you need longer words, it's OK to
676                 // use two lines here. Test with "xz -l foo.xz".
677                 puts(_("Strms  Blocks   Compressed Uncompressed  Ratio  "
678                                 "Check   Filename"));
679         }
680
681         char checks[CHECKS_STR_SIZE];
682         get_check_names(checks, lzma_index_checks(xfi->idx), false);
683
684         const char *cols[7] = {
685                 uint64_to_str(lzma_index_stream_count(xfi->idx), 0),
686                 uint64_to_str(lzma_index_block_count(xfi->idx), 1),
687                 uint64_to_nicestr(lzma_index_file_size(xfi->idx),
688                         NICESTR_B, NICESTR_TIB, false, 2),
689                 uint64_to_nicestr(lzma_index_uncompressed_size(xfi->idx),
690                         NICESTR_B, NICESTR_TIB, false, 3),
691                 get_ratio(lzma_index_file_size(xfi->idx),
692                         lzma_index_uncompressed_size(xfi->idx)),
693                 checks,
694                 pair->src_name,
695         };
696         printf("%*s %*s  %*s  %*s  %*s  %-*s %s\n",
697                         tuklib_mbstr_fw(cols[0], 5), cols[0],
698                         tuklib_mbstr_fw(cols[1], 7), cols[1],
699                         tuklib_mbstr_fw(cols[2], 11), cols[2],
700                         tuklib_mbstr_fw(cols[3], 11), cols[3],
701                         tuklib_mbstr_fw(cols[4], 5), cols[4],
702                         tuklib_mbstr_fw(cols[5], 7), cols[5],
703                         cols[6]);
704
705         return false;
706 }
707
708
709 static void
710 print_adv_helper(uint64_t stream_count, uint64_t block_count,
711                 uint64_t compressed_size, uint64_t uncompressed_size,
712                 uint32_t checks, uint64_t stream_padding)
713 {
714         char checks_str[CHECKS_STR_SIZE];
715         get_check_names(checks_str, checks, true);
716
717         printf(_("  Streams:            %s\n"),
718                         uint64_to_str(stream_count, 0));
719         printf(_("  Blocks:             %s\n"),
720                         uint64_to_str(block_count, 0));
721         printf(_("  Compressed size:    %s\n"),
722                         uint64_to_nicestr(compressed_size,
723                                 NICESTR_B, NICESTR_TIB, true, 0));
724         printf(_("  Uncompressed size:  %s\n"),
725                         uint64_to_nicestr(uncompressed_size,
726                                 NICESTR_B, NICESTR_TIB, true, 0));
727         printf(_("  Ratio:              %s\n"),
728                         get_ratio(compressed_size, uncompressed_size));
729         printf(_("  Check:              %s\n"), checks_str);
730         printf(_("  Stream padding:     %s\n"),
731                         uint64_to_nicestr(stream_padding,
732                                 NICESTR_B, NICESTR_TIB, true, 0));
733         return;
734 }
735
736
737 static bool
738 print_info_adv(xz_file_info *xfi, file_pair *pair)
739 {
740         // Print the overall information.
741         print_adv_helper(lzma_index_stream_count(xfi->idx),
742                         lzma_index_block_count(xfi->idx),
743                         lzma_index_file_size(xfi->idx),
744                         lzma_index_uncompressed_size(xfi->idx),
745                         lzma_index_checks(xfi->idx),
746                         xfi->stream_padding);
747
748         // Size of the biggest Check. This is used to calculate the width
749         // of the CheckVal field. The table would get insanely wide if
750         // we always reserved space for 64-byte Check (128 chars as hex).
751         uint32_t check_max = 0;
752
753         // Print information about the Streams.
754         //
755         // TRANSLATORS: The second line is column headings. All except
756         // Check are right aligned; Check is left aligned. Test with
757         // "xz -lv foo.xz".
758         puts(_("  Streams:\n    Stream    Blocks"
759                         "      CompOffset    UncompOffset"
760                         "        CompSize      UncompSize  Ratio"
761                         "  Check      Padding"));
762
763         lzma_index_iter iter;
764         lzma_index_iter_init(&iter, xfi->idx);
765
766         while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM)) {
767                 const char *cols1[4] = {
768                         uint64_to_str(iter.stream.number, 0),
769                         uint64_to_str(iter.stream.block_count, 1),
770                         uint64_to_str(iter.stream.compressed_offset, 2),
771                         uint64_to_str(iter.stream.uncompressed_offset, 3),
772                 };
773                 printf("    %*s %*s %*s %*s ",
774                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
775                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
776                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
777                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
778
779                 const char *cols2[5] = {
780                         uint64_to_str(iter.stream.compressed_size, 0),
781                         uint64_to_str(iter.stream.uncompressed_size, 1),
782                         get_ratio(iter.stream.compressed_size,
783                                 iter.stream.uncompressed_size),
784                         _(check_names[iter.stream.flags->check]),
785                         uint64_to_str(iter.stream.padding, 2),
786                 };
787                 printf("%*s %*s  %*s  %-*s %*s\n",
788                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
789                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
790                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
791                                 tuklib_mbstr_fw(cols2[3], 10), cols2[3],
792                                 tuklib_mbstr_fw(cols2[4], 7), cols2[4]);
793
794                 // Update the maximum Check size.
795                 if (lzma_check_size(iter.stream.flags->check) > check_max)
796                         check_max = lzma_check_size(iter.stream.flags->check);
797         }
798
799         // Cache the verbosity level to a local variable.
800         const bool detailed = message_verbosity_get() >= V_DEBUG;
801
802         // Information collected from Block Headers
803         block_header_info bhi;
804
805         // Print information about the Blocks but only if there is
806         // at least one Block.
807         if (lzma_index_block_count(xfi->idx) > 0) {
808                 // Calculate the width of the CheckVal field.
809                 const int checkval_width = my_max(8, 2 * check_max);
810
811                 // TRANSLATORS: The second line is column headings. All
812                 // except Check are right aligned; Check is left aligned.
813                 printf(_("  Blocks:\n    Stream     Block"
814                         "      CompOffset    UncompOffset"
815                         "       TotalSize      UncompSize  Ratio  Check"));
816
817                 if (detailed) {
818                         // TRANSLATORS: These are additional column headings
819                         // for the most verbose listing mode. CheckVal
820                         // (Check value), Flags, and Filters are left aligned.
821                         // Header (Block Header Size), CompSize, and MemUsage
822                         // are right aligned. %*s is replaced with 0-120
823                         // spaces to make the CheckVal column wide enough.
824                         // Test with "xz -lvv foo.xz".
825                         printf(_("      CheckVal %*s Header  Flags        "
826                                         "CompSize    MemUsage  Filters"),
827                                         checkval_width - 8, "");
828                 }
829
830                 putchar('\n');
831
832                 lzma_index_iter_init(&iter, xfi->idx);
833
834                 // Iterate over the Blocks.
835                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
836                         if (detailed && parse_details(pair, &iter, &bhi, xfi))
837                                         return true;
838
839                         const char *cols1[4] = {
840                                 uint64_to_str(iter.stream.number, 0),
841                                 uint64_to_str(
842                                         iter.block.number_in_stream, 1),
843                                 uint64_to_str(
844                                         iter.block.compressed_file_offset, 2),
845                                 uint64_to_str(
846                                         iter.block.uncompressed_file_offset, 3)
847                         };
848                         printf("    %*s %*s %*s %*s ",
849                                 tuklib_mbstr_fw(cols1[0], 6), cols1[0],
850                                 tuklib_mbstr_fw(cols1[1], 9), cols1[1],
851                                 tuklib_mbstr_fw(cols1[2], 15), cols1[2],
852                                 tuklib_mbstr_fw(cols1[3], 15), cols1[3]);
853
854                         const char *cols2[4] = {
855                                 uint64_to_str(iter.block.total_size, 0),
856                                 uint64_to_str(iter.block.uncompressed_size,
857                                                 1),
858                                 get_ratio(iter.block.total_size,
859                                         iter.block.uncompressed_size),
860                                 _(check_names[iter.stream.flags->check])
861                         };
862                         printf("%*s %*s  %*s  %-*s",
863                                 tuklib_mbstr_fw(cols2[0], 15), cols2[0],
864                                 tuklib_mbstr_fw(cols2[1], 15), cols2[1],
865                                 tuklib_mbstr_fw(cols2[2], 5), cols2[2],
866                                 tuklib_mbstr_fw(cols2[3], detailed ? 11 : 1),
867                                         cols2[3]);
868
869                         if (detailed) {
870                                 const lzma_vli compressed_size
871                                                 = iter.block.unpadded_size
872                                                 - bhi.header_size
873                                                 - lzma_check_size(
874                                                 iter.stream.flags->check);
875
876                                 const char *cols3[6] = {
877                                         check_value,
878                                         uint64_to_str(bhi.header_size, 0),
879                                         bhi.flags,
880                                         uint64_to_str(compressed_size, 1),
881                                         uint64_to_str(
882                                                 round_up_to_mib(bhi.memusage),
883                                                 2),
884                                         bhi.filter_chain
885                                 };
886                                 // Show MiB for memory usage, because it
887                                 // is the only size which is not in bytes.
888                                 printf("%-*s  %*s  %-5s %*s %*s MiB  %s",
889                                         checkval_width, cols3[0],
890                                         tuklib_mbstr_fw(cols3[1], 6), cols3[1],
891                                         cols3[2],
892                                         tuklib_mbstr_fw(cols3[3], 15),
893                                                 cols3[3],
894                                         tuklib_mbstr_fw(cols3[4], 7), cols3[4],
895                                         cols3[5]);
896                         }
897
898                         putchar('\n');
899                 }
900         }
901
902         if (detailed) {
903                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
904                                 round_up_to_mib(xfi->memusage_max), 0));
905                 printf(_("  Sizes in headers:   %s\n"),
906                                 xfi->all_have_sizes ? _("Yes") : _("No"));
907                 printf(_("  Minimum XZ Utils version: %s\n"),
908                                 xz_ver_to_str(xfi->min_version));
909         }
910
911         return false;
912 }
913
914
915 static bool
916 print_info_robot(xz_file_info *xfi, file_pair *pair)
917 {
918         char checks[CHECKS_STR_SIZE];
919         get_check_names(checks, lzma_index_checks(xfi->idx), false);
920
921         printf("name\t%s\n", pair->src_name);
922
923         printf("file\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
924                         "\t%s\t%s\t%" PRIu64 "\n",
925                         lzma_index_stream_count(xfi->idx),
926                         lzma_index_block_count(xfi->idx),
927                         lzma_index_file_size(xfi->idx),
928                         lzma_index_uncompressed_size(xfi->idx),
929                         get_ratio(lzma_index_file_size(xfi->idx),
930                                 lzma_index_uncompressed_size(xfi->idx)),
931                         checks,
932                         xfi->stream_padding);
933
934         if (message_verbosity_get() >= V_VERBOSE) {
935                 lzma_index_iter iter;
936                 lzma_index_iter_init(&iter, xfi->idx);
937
938                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_STREAM))
939                         printf("stream\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
940                                 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
941                                 "\t%s\t%s\t%" PRIu64 "\n",
942                                 iter.stream.number,
943                                 iter.stream.block_count,
944                                 iter.stream.compressed_offset,
945                                 iter.stream.uncompressed_offset,
946                                 iter.stream.compressed_size,
947                                 iter.stream.uncompressed_size,
948                                 get_ratio(iter.stream.compressed_size,
949                                         iter.stream.uncompressed_size),
950                                 check_names[iter.stream.flags->check],
951                                 iter.stream.padding);
952
953                 lzma_index_iter_rewind(&iter);
954                 block_header_info bhi;
955
956                 while (!lzma_index_iter_next(&iter, LZMA_INDEX_ITER_BLOCK)) {
957                         if (message_verbosity_get() >= V_DEBUG
958                                         && parse_details(
959                                                 pair, &iter, &bhi, xfi))
960                                 return true;
961
962                         printf("block\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
963                                         "\t%" PRIu64 "\t%" PRIu64
964                                         "\t%" PRIu64 "\t%" PRIu64 "\t%s\t%s",
965                                         iter.stream.number,
966                                         iter.block.number_in_stream,
967                                         iter.block.number_in_file,
968                                         iter.block.compressed_file_offset,
969                                         iter.block.uncompressed_file_offset,
970                                         iter.block.total_size,
971                                         iter.block.uncompressed_size,
972                                         get_ratio(iter.block.total_size,
973                                                 iter.block.uncompressed_size),
974                                         check_names[iter.stream.flags->check]);
975
976                         if (message_verbosity_get() >= V_DEBUG)
977                                 printf("\t%s\t%" PRIu32 "\t%s\t%" PRIu64
978                                                 "\t%" PRIu64 "\t%s",
979                                                 check_value,
980                                                 bhi.header_size,
981                                                 bhi.flags,
982                                                 bhi.compressed_size,
983                                                 bhi.memusage,
984                                                 bhi.filter_chain);
985
986                         putchar('\n');
987                 }
988         }
989
990         if (message_verbosity_get() >= V_DEBUG)
991                 printf("summary\t%" PRIu64 "\t%s\t%" PRIu32 "\n",
992                                 xfi->memusage_max,
993                                 xfi->all_have_sizes ? "yes" : "no",
994                                 xfi->min_version);
995
996         return false;
997 }
998
999
1000 static void
1001 update_totals(const xz_file_info *xfi)
1002 {
1003         // TODO: Integer overflow checks
1004         ++totals.files;
1005         totals.streams += lzma_index_stream_count(xfi->idx);
1006         totals.blocks += lzma_index_block_count(xfi->idx);
1007         totals.compressed_size += lzma_index_file_size(xfi->idx);
1008         totals.uncompressed_size += lzma_index_uncompressed_size(xfi->idx);
1009         totals.stream_padding += xfi->stream_padding;
1010         totals.checks |= lzma_index_checks(xfi->idx);
1011
1012         if (totals.memusage_max < xfi->memusage_max)
1013                 totals.memusage_max = xfi->memusage_max;
1014
1015         if (totals.min_version < xfi->min_version)
1016                 totals.min_version = xfi->min_version;
1017
1018         totals.all_have_sizes &= xfi->all_have_sizes;
1019
1020         return;
1021 }
1022
1023
1024 static void
1025 print_totals_basic(void)
1026 {
1027         // Print a separator line.
1028         char line[80];
1029         memset(line, '-', sizeof(line));
1030         line[sizeof(line) - 1] = '\0';
1031         puts(line);
1032
1033         // Get the check names.
1034         char checks[CHECKS_STR_SIZE];
1035         get_check_names(checks, totals.checks, false);
1036
1037         // Print the totals except the file count, which needs
1038         // special handling.
1039         printf("%5s %7s  %11s  %11s  %5s  %-7s ",
1040                         uint64_to_str(totals.streams, 0),
1041                         uint64_to_str(totals.blocks, 1),
1042                         uint64_to_nicestr(totals.compressed_size,
1043                                 NICESTR_B, NICESTR_TIB, false, 2),
1044                         uint64_to_nicestr(totals.uncompressed_size,
1045                                 NICESTR_B, NICESTR_TIB, false, 3),
1046                         get_ratio(totals.compressed_size,
1047                                 totals.uncompressed_size),
1048                         checks);
1049
1050         // Since we print totals only when there are at least two files,
1051         // the English message will always use "%s files". But some other
1052         // languages need different forms for different plurals so we
1053         // have to translate this with ngettext().
1054         //
1055         // TRANSLATORS: %s is an integer. Only the plural form of this
1056         // message is used (e.g. "2 files"). Test with "xz -l foo.xz bar.xz".
1057         printf(ngettext("%s file\n", "%s files\n",
1058                         totals.files <= ULONG_MAX ? totals.files
1059                                 : (totals.files % 1000000) + 1000000),
1060                         uint64_to_str(totals.files, 0));
1061
1062         return;
1063 }
1064
1065
1066 static void
1067 print_totals_adv(void)
1068 {
1069         putchar('\n');
1070         puts(_("Totals:"));
1071         printf(_("  Number of files:    %s\n"),
1072                         uint64_to_str(totals.files, 0));
1073         print_adv_helper(totals.streams, totals.blocks,
1074                         totals.compressed_size, totals.uncompressed_size,
1075                         totals.checks, totals.stream_padding);
1076
1077         if (message_verbosity_get() >= V_DEBUG) {
1078                 printf(_("  Memory needed:      %s MiB\n"), uint64_to_str(
1079                                 round_up_to_mib(totals.memusage_max), 0));
1080                 printf(_("  Sizes in headers:   %s\n"),
1081                                 totals.all_have_sizes ? _("Yes") : _("No"));
1082                 printf(_("  Minimum XZ Utils version: %s\n"),
1083                                 xz_ver_to_str(totals.min_version));
1084         }
1085
1086         return;
1087 }
1088
1089
1090 static void
1091 print_totals_robot(void)
1092 {
1093         char checks[CHECKS_STR_SIZE];
1094         get_check_names(checks, totals.checks, false);
1095
1096         printf("totals\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64 "\t%" PRIu64
1097                         "\t%s\t%s\t%" PRIu64 "\t%" PRIu64,
1098                         totals.streams,
1099                         totals.blocks,
1100                         totals.compressed_size,
1101                         totals.uncompressed_size,
1102                         get_ratio(totals.compressed_size,
1103                                 totals.uncompressed_size),
1104                         checks,
1105                         totals.stream_padding,
1106                         totals.files);
1107
1108         if (message_verbosity_get() >= V_DEBUG)
1109                 printf("\t%" PRIu64 "\t%s\t%" PRIu32,
1110                                 totals.memusage_max,
1111                                 totals.all_have_sizes ? "yes" : "no",
1112                                 totals.min_version);
1113
1114         putchar('\n');
1115
1116         return;
1117 }
1118
1119
1120 extern void
1121 list_totals(void)
1122 {
1123         if (opt_robot) {
1124                 // Always print totals in --robot mode. It can be convenient
1125                 // in some cases and doesn't complicate usage of the
1126                 // single-file case much.
1127                 print_totals_robot();
1128
1129         } else if (totals.files > 1) {
1130                 // For non-robot mode, totals are printed only if there
1131                 // is more than one file.
1132                 if (message_verbosity_get() <= V_WARNING)
1133                         print_totals_basic();
1134                 else
1135                         print_totals_adv();
1136         }
1137
1138         return;
1139 }
1140
1141
1142 extern void
1143 list_file(const char *filename)
1144 {
1145         if (opt_format != FORMAT_XZ && opt_format != FORMAT_AUTO)
1146                 message_fatal(_("--list works only on .xz files "
1147                                 "(--format=xz or --format=auto)"));
1148
1149         message_filename(filename);
1150
1151         if (filename == stdin_filename) {
1152                 message_error(_("--list does not support reading from "
1153                                 "standard input"));
1154                 return;
1155         }
1156
1157         // Unset opt_stdout so that io_open_src() won't accept special files.
1158         // Set opt_force so that io_open_src() will follow symlinks.
1159         opt_stdout = false;
1160         opt_force = true;
1161         file_pair *pair = io_open_src(filename);
1162         if (pair == NULL)
1163                 return;
1164
1165         xz_file_info xfi = XZ_FILE_INFO_INIT;
1166         if (!parse_indexes(&xfi, pair)) {
1167                 bool fail;
1168
1169                 // We have three main modes:
1170                 //  - --robot, which has submodes if --verbose is specified
1171                 //    once or twice
1172                 //  - Normal --list without --verbose
1173                 //  - --list with one or two --verbose
1174                 if (opt_robot)
1175                         fail = print_info_robot(&xfi, pair);
1176                 else if (message_verbosity_get() <= V_WARNING)
1177                         fail = print_info_basic(&xfi, pair);
1178                 else
1179                         fail = print_info_adv(&xfi, pair);
1180
1181                 // Update the totals that are displayed after all
1182                 // the individual files have been listed. Don't count
1183                 // broken files.
1184                 if (!fail)
1185                         update_totals(&xfi);
1186
1187                 lzma_index_end(xfi.idx, NULL);
1188         }
1189
1190         io_close(pair, false);
1191         return;
1192 }