/src/gdal/port/cpl_vsil_gzip.cpp
Line | Count | Source |
1 | | /****************************************************************************** |
2 | | * |
3 | | * Project: CPL - Common Portability Library |
4 | | * Purpose: Implement VSI large file api for gz/zip files (.gz and .zip). |
5 | | * Author: Even Rouault, even.rouault at spatialys.com |
6 | | * |
7 | | ****************************************************************************** |
8 | | * Copyright (c) 2008-2014, Even Rouault <even dot rouault at spatialys.com> |
9 | | * |
10 | | * SPDX-License-Identifier: MIT |
11 | | ****************************************************************************/ |
12 | | |
13 | | //! @cond Doxygen_Suppress |
14 | | |
15 | | /* gzio.c -- IO on .gz files |
16 | | Copyright (C) 1995-2005 Jean-loup Gailly. |
17 | | |
18 | | This software is provided 'as-is', without any express or implied |
19 | | warranty. In no event will the authors be held liable for any damages |
20 | | arising from the use of this software. |
21 | | |
22 | | Permission is granted to anyone to use this software for any purpose, |
23 | | including commercial applications, and to alter it and redistribute it |
24 | | freely, subject to the following restrictions: |
25 | | |
26 | | 1. The origin of this software must not be misrepresented; you must not |
27 | | claim that you wrote the original software. If you use this software |
28 | | in a product, an acknowledgment in the product documentation would be |
29 | | appreciated but is not required. |
30 | | 2. Altered source versions must be plainly marked as such, and must not be |
31 | | misrepresented as being the original software. |
32 | | 3. This notice may not be removed or altered from any source distribution. |
33 | | |
34 | | Jean-loup Gailly Mark Adler |
35 | | jloup@gzip.org madler@alumni.caltech.edu |
36 | | |
37 | | The data format used by the zlib library is described by RFCs (Request for |
38 | | Comments) 1950 to 1952 in the files http://www.ietf.org/rfc/rfc1950.txt |
39 | | (zlib format), rfc1951.txt (deflate format) and rfc1952.txt (gzip format). |
40 | | */ |
41 | | |
42 | | /* This file contains a refactoring of gzio.c from zlib project. |
43 | | |
44 | | It replaces classical calls operating on FILE* by calls to the VSI large file |
45 | | API. It also adds the capability to seek at the end of the file, which is not |
46 | | implemented in original gzSeek. It also implements a concept of in-memory |
47 | | "snapshots", that are a way of improving efficiency while seeking GZip |
48 | | files. Snapshots are created regularly when decompressing the data a snapshot |
49 | | of the gzip state. Later we can seek directly in the compressed data to the |
50 | | closest snapshot in order to reduce the amount of data to uncompress again. |
51 | | |
52 | | For .gz files, an effort is done to cache the size of the uncompressed data |
53 | | in a .gz.properties file, so that we don't need to seek at the end of the |
54 | | file each time a Stat() is done. |
55 | | |
56 | | For .zip and .gz, both reading and writing are supported, but just one mode |
57 | | at a time (read-only or write-only). |
58 | | */ |
59 | | |
60 | | #include "cpl_port.h" |
61 | | #include "cpl_conv.h" |
62 | | #include "cpl_vsi.h" |
63 | | |
64 | | #include <cerrno> |
65 | | #include <cinttypes> |
66 | | #include <climits> |
67 | | #include <cstddef> |
68 | | #include <cstdio> |
69 | | #include <cstdlib> |
70 | | #include <cstring> |
71 | | #include <ctime> |
72 | | |
73 | | #include <fcntl.h> |
74 | | |
75 | | #include "cpl_zlib_header.h" // to avoid warnings when including zlib.h |
76 | | |
77 | | #ifdef HAVE_LIBDEFLATE |
78 | | #include "libdeflate.h" |
79 | | #endif |
80 | | |
81 | | #include <algorithm> |
82 | | #include <iterator> |
83 | | #include <limits> |
84 | | #include <list> |
85 | | #include <map> |
86 | | #include <memory> |
87 | | #include <mutex> |
88 | | #include <string> |
89 | | #include <utility> |
90 | | #include <vector> |
91 | | |
92 | | #include "cpl_error.h" |
93 | | #include "cpl_minizip_ioapi.h" |
94 | | #include "cpl_minizip_unzip.h" |
95 | | #include "cpl_multiproc.h" |
96 | | #include "cpl_string.h" |
97 | | #include "cpl_time.h" |
98 | | #include "cpl_vsi_virtual.h" |
99 | | #include "cpl_worker_thread_pool.h" |
100 | | |
101 | | constexpr int Z_BUFSIZE = 65536; // Original size is 16384 |
102 | | constexpr int gz_magic[2] = {0x1f, 0x8b}; // gzip magic header |
103 | | |
104 | | // gzip flag byte. |
105 | | #define ASCII_FLAG 0x01 // bit 0 set: file probably ascii text |
106 | 0 | #define HEAD_CRC 0x02 // bit 1 set: header CRC present |
107 | 0 | #define EXTRA_FIELD 0x04 // bit 2 set: extra field present |
108 | 0 | #define ORIG_NAME 0x08 // bit 3 set: original file name present |
109 | 0 | #define COMMENT 0x10 // bit 4 set: file comment present |
110 | 0 | #define RESERVED 0xE0 // bits 5..7: reserved |
111 | | |
112 | 0 | #define ALLOC(size) malloc(size) |
113 | | #define TRYFREE(p) \ |
114 | 0 | { \ |
115 | 0 | if (p) \ |
116 | 0 | free(p); \ |
117 | 0 | } |
118 | | |
119 | | #define CPL_VSIL_GZ_RETURN(ret) \ |
120 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "In file %s, at line %d, return %d", \ |
121 | 0 | __FILE__, __LINE__, ret) |
122 | | |
123 | | // To avoid aliasing to CopyFile to CopyFileA on Windows |
124 | | #ifdef CopyFile |
125 | | #undef CopyFile |
126 | | #endif |
127 | | |
128 | | // #define ENABLE_DEBUG 1 |
129 | | |
130 | | /************************************************************************/ |
131 | | /* ==================================================================== */ |
132 | | /* VSIGZipHandle */ |
133 | | /* ==================================================================== */ |
134 | | /************************************************************************/ |
135 | | |
136 | | typedef struct |
137 | | { |
138 | | vsi_l_offset posInBaseHandle; |
139 | | z_stream stream; |
140 | | uLong crc; |
141 | | int transparent; |
142 | | vsi_l_offset in; |
143 | | vsi_l_offset out; |
144 | | } GZipSnapshot; |
145 | | |
146 | | class VSIGZipHandle final : public VSIVirtualHandle |
147 | | { |
148 | | VSIVirtualHandleUniquePtr m_poBaseHandle{}; |
149 | | #ifdef DEBUG |
150 | | vsi_l_offset m_offset = 0; |
151 | | #endif |
152 | | vsi_l_offset m_compressed_size = 0; |
153 | | vsi_l_offset m_uncompressed_size = 0; |
154 | | vsi_l_offset offsetEndCompressedData = 0; |
155 | | uLong m_expected_crc = 0; |
156 | | char *m_pszBaseFileName = nullptr; /* optional */ |
157 | | bool m_bWriteProperties = false; |
158 | | bool m_bCanSaveInfo = false; |
159 | | |
160 | | /* Fields from gz_stream structure */ |
161 | | z_stream stream; |
162 | | int z_err = Z_OK; /* error code for last stream operation */ |
163 | | int z_eof = 0; /* set if end of input file (but not necessarily of the |
164 | | uncompressed stream !) */ |
165 | | bool m_bEOF = false; /* EOF flag for uncompressed stream */ |
166 | | Byte *inbuf = nullptr; /* input buffer */ |
167 | | Byte *outbuf = nullptr; /* output buffer */ |
168 | | uLong crc = 0; /* crc32 of uncompressed data */ |
169 | | int m_transparent = 0; /* 1 if input file is not a .gz file */ |
170 | | vsi_l_offset startOff = |
171 | | 0; /* startOff of compressed data in file (header skipped) */ |
172 | | vsi_l_offset in = 0; /* bytes into deflate or inflate */ |
173 | | vsi_l_offset out = 0; /* bytes out of deflate or inflate */ |
174 | | vsi_l_offset m_nLastReadOffset = 0; |
175 | | |
176 | | GZipSnapshot *snapshots = nullptr; |
177 | | vsi_l_offset snapshot_byte_interval = |
178 | | 0; /* number of compressed bytes at which we create a "snapshot" */ |
179 | | |
180 | | void check_header(); |
181 | | int get_byte(); |
182 | | bool gzseek(vsi_l_offset nOffset, int nWhence); |
183 | | int gzrewind(); |
184 | | uLong getLong(); |
185 | | |
186 | | CPL_DISALLOW_COPY_ASSIGN(VSIGZipHandle) |
187 | | |
188 | | public: |
189 | | VSIGZipHandle(VSIVirtualHandleUniquePtr poBaseHandleIn, |
190 | | const char *pszBaseFileName, vsi_l_offset offset = 0, |
191 | | vsi_l_offset compressed_size = 0, |
192 | | vsi_l_offset uncompressed_size = 0, uLong expected_crc = 0, |
193 | | int transparent = 0); |
194 | | ~VSIGZipHandle() override; |
195 | | |
196 | | bool IsInitOK() const |
197 | 0 | { |
198 | 0 | return inbuf != nullptr; |
199 | 0 | } |
200 | | |
201 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
202 | | vsi_l_offset Tell() override; |
203 | | size_t Read(void *pBuffer, size_t nSize, size_t nMemb) override; |
204 | | size_t Write(const void *pBuffer, size_t nSize, size_t nMemb) override; |
205 | | void ClearErr() override; |
206 | | int Eof() override; |
207 | | int Error() override; |
208 | | int Flush() override; |
209 | | int Close() override; |
210 | | |
211 | | VSIGZipHandle *Duplicate(); |
212 | | bool CloseBaseHandle(); |
213 | | |
214 | | vsi_l_offset GetLastReadOffset() |
215 | 0 | { |
216 | 0 | return m_nLastReadOffset; |
217 | 0 | } |
218 | | |
219 | | const char *GetBaseFileName() |
220 | 0 | { |
221 | 0 | return m_pszBaseFileName; |
222 | 0 | } |
223 | | |
224 | | void SetUncompressedSize(vsi_l_offset nUncompressedSize) |
225 | 0 | { |
226 | 0 | m_uncompressed_size = nUncompressedSize; |
227 | 0 | } |
228 | | |
229 | | vsi_l_offset GetUncompressedSize() |
230 | 0 | { |
231 | 0 | return m_uncompressed_size; |
232 | 0 | } |
233 | | |
234 | | void SaveInfo_unlocked(); |
235 | | |
236 | | void UnsetCanSaveInfo() |
237 | 0 | { |
238 | 0 | m_bCanSaveInfo = false; |
239 | 0 | } |
240 | | }; |
241 | | |
242 | | #ifdef ENABLE_DEFLATE64 |
243 | | |
244 | | /************************************************************************/ |
245 | | /* ==================================================================== */ |
246 | | /* VSIDeflate64Handle */ |
247 | | /* ==================================================================== */ |
248 | | /************************************************************************/ |
249 | | |
250 | | struct VSIDeflate64Snapshot |
251 | | { |
252 | | vsi_l_offset posInBaseHandle = 0; |
253 | | z_stream stream{}; |
254 | | uLong crc = 0; |
255 | | vsi_l_offset in = 0; |
256 | | vsi_l_offset out = 0; |
257 | | std::vector<GByte> extraOutput{}; |
258 | | bool m_bStreamEndReached = false; |
259 | | }; |
260 | | |
261 | | class VSIDeflate64Handle final : public VSIVirtualHandle |
262 | | { |
263 | | VSIVirtualHandleUniquePtr m_poBaseHandle{}; |
264 | | #ifdef DEBUG |
265 | | vsi_l_offset m_offset = 0; |
266 | | #endif |
267 | | vsi_l_offset m_compressed_size = 0; |
268 | | vsi_l_offset m_uncompressed_size = 0; |
269 | | vsi_l_offset offsetEndCompressedData = 0; |
270 | | uLong m_expected_crc = 0; |
271 | | char *m_pszBaseFileName = nullptr; /* optional */ |
272 | | |
273 | | /* Fields from gz_stream structure */ |
274 | | z_stream stream; |
275 | | int z_err = Z_OK; /* error code for last stream operation */ |
276 | | int z_eof = 0; /* set if end of input file (but not necessarily of the |
277 | | uncompressed stream ! ) */ |
278 | | bool m_bEOF = false; /* EOF flag for uncompressed stream */ |
279 | | Byte *inbuf = nullptr; /* input buffer */ |
280 | | Byte *outbuf = nullptr; /* output buffer */ |
281 | | std::vector<GByte> extraOutput{}; |
282 | | bool m_bStreamEndReached = false; |
283 | | uLong crc = 0; /* crc32 of uncompressed data */ |
284 | | vsi_l_offset startOff = |
285 | | 0; /* startOff of compressed data in file (header skipped) */ |
286 | | vsi_l_offset in = 0; /* bytes into deflate or inflate */ |
287 | | vsi_l_offset out = 0; /* bytes out of deflate or inflate */ |
288 | | |
289 | | std::vector<VSIDeflate64Snapshot> snapshots{}; |
290 | | vsi_l_offset snapshot_byte_interval = |
291 | | 0; /* number of compressed bytes at which we create a "snapshot" */ |
292 | | |
293 | | bool gzseek(vsi_l_offset nOffset, int nWhence); |
294 | | int gzrewind(); |
295 | | |
296 | | CPL_DISALLOW_COPY_ASSIGN(VSIDeflate64Handle) |
297 | | |
298 | | public: |
299 | | VSIDeflate64Handle(VSIVirtualHandleUniquePtr poBaseHandleIn, |
300 | | const char *pszBaseFileName, vsi_l_offset offset = 0, |
301 | | vsi_l_offset compressed_size = 0, |
302 | | vsi_l_offset uncompressed_size = 0, |
303 | | uLong expected_crc = 0); |
304 | | ~VSIDeflate64Handle() override; |
305 | | |
306 | | bool IsInitOK() const |
307 | 0 | { |
308 | 0 | return inbuf != nullptr; |
309 | 0 | } |
310 | | |
311 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
312 | | vsi_l_offset Tell() override; |
313 | | size_t Read(void *pBuffer, size_t nSize, size_t nMemb) override; |
314 | | size_t Write(const void *pBuffer, size_t nSize, size_t nMemb) override; |
315 | | void ClearErr() override; |
316 | | int Eof() override; |
317 | | int Error() override; |
318 | | int Flush() override; |
319 | | int Close() override; |
320 | | |
321 | | VSIDeflate64Handle *Duplicate(); |
322 | | bool CloseBaseHandle(); |
323 | | |
324 | | const char *GetBaseFileName() |
325 | 0 | { |
326 | 0 | return m_pszBaseFileName; |
327 | 0 | } |
328 | | |
329 | | void SetUncompressedSize(vsi_l_offset nUncompressedSize) |
330 | 0 | { |
331 | 0 | m_uncompressed_size = nUncompressedSize; |
332 | 0 | } |
333 | | |
334 | | vsi_l_offset GetUncompressedSize() |
335 | 0 | { |
336 | 0 | return m_uncompressed_size; |
337 | 0 | } |
338 | | }; |
339 | | #endif |
340 | | |
341 | | class VSIGZipFilesystemHandler final : public VSIFilesystemHandler |
342 | | { |
343 | | CPL_DISALLOW_COPY_ASSIGN(VSIGZipFilesystemHandler) |
344 | | |
345 | | std::recursive_mutex oMutex{}; |
346 | | std::unique_ptr<VSIGZipHandle> poHandleLastGZipFile{}; |
347 | | bool m_bInSaveInfo = false; |
348 | | |
349 | | public: |
350 | 3 | VSIGZipFilesystemHandler() = default; |
351 | | ~VSIGZipFilesystemHandler() override; |
352 | | |
353 | | VSIVirtualHandleUniquePtr Open(const char *pszFilename, |
354 | | const char *pszAccess, bool bSetError, |
355 | | CSLConstList /* papszOptions */) override; |
356 | | VSIGZipHandle *OpenGZipReadOnly(const char *pszFilename, |
357 | | const char *pszAccess); |
358 | | int Stat(const char *pszFilename, VSIStatBufL *pStatBuf, |
359 | | int nFlags) override; |
360 | | char **ReadDirEx(const char *pszDirname, int nMaxFiles) override; |
361 | | |
362 | | const char *GetOptions() override; |
363 | | |
364 | | virtual bool SupportsSequentialWrite(const char *pszPath, |
365 | | bool bAllowLocalTempFile) override; |
366 | | |
367 | | virtual bool SupportsRandomWrite(const char * /* pszPath */, |
368 | | bool /* bAllowLocalTempFile */) override |
369 | 0 | { |
370 | 0 | return false; |
371 | 0 | } |
372 | | |
373 | | void SaveInfo(VSIGZipHandle *poHandle); |
374 | | void SaveInfo_unlocked(VSIGZipHandle *poHandle); |
375 | | }; |
376 | | |
377 | | /************************************************************************/ |
378 | | /* Duplicate() */ |
379 | | /************************************************************************/ |
380 | | |
381 | | VSIGZipHandle *VSIGZipHandle::Duplicate() |
382 | 0 | { |
383 | 0 | CPLAssert(m_offset == 0); |
384 | 0 | CPLAssert(m_compressed_size != 0); |
385 | 0 | CPLAssert(m_pszBaseFileName != nullptr); |
386 | | |
387 | 0 | VSIFilesystemHandler *poFSHandler = |
388 | 0 | VSIFileManager::GetHandler(m_pszBaseFileName); |
389 | |
|
390 | 0 | auto poNewBaseHandle = poFSHandler->Open(m_pszBaseFileName, "rb"); |
391 | |
|
392 | 0 | if (poNewBaseHandle == nullptr) |
393 | 0 | return nullptr; |
394 | | |
395 | 0 | auto poHandle = std::make_unique<VSIGZipHandle>( |
396 | 0 | std::move(poNewBaseHandle), m_pszBaseFileName, 0, m_compressed_size, |
397 | 0 | m_uncompressed_size); |
398 | 0 | if (!(poHandle->IsInitOK())) |
399 | 0 | { |
400 | 0 | return nullptr; |
401 | 0 | } |
402 | | |
403 | 0 | poHandle->m_nLastReadOffset = m_nLastReadOffset; |
404 | | |
405 | | // Most important: duplicate the snapshots! |
406 | |
|
407 | 0 | for (unsigned int i = 0; i < m_compressed_size / snapshot_byte_interval + 1; |
408 | 0 | i++) |
409 | 0 | { |
410 | 0 | if (snapshots[i].posInBaseHandle == 0) |
411 | 0 | break; |
412 | | |
413 | 0 | poHandle->snapshots[i].posInBaseHandle = snapshots[i].posInBaseHandle; |
414 | 0 | inflateCopy(&poHandle->snapshots[i].stream, &snapshots[i].stream); |
415 | 0 | poHandle->snapshots[i].crc = snapshots[i].crc; |
416 | 0 | poHandle->snapshots[i].transparent = snapshots[i].transparent; |
417 | 0 | poHandle->snapshots[i].in = snapshots[i].in; |
418 | 0 | poHandle->snapshots[i].out = snapshots[i].out; |
419 | 0 | } |
420 | |
|
421 | 0 | return poHandle.release(); |
422 | 0 | } |
423 | | |
424 | | /************************************************************************/ |
425 | | /* CloseBaseHandle() */ |
426 | | /************************************************************************/ |
427 | | |
428 | | bool VSIGZipHandle::CloseBaseHandle() |
429 | 0 | { |
430 | 0 | bool bRet = true; |
431 | 0 | if (m_poBaseHandle) |
432 | 0 | { |
433 | 0 | bRet = m_poBaseHandle->Close() == 0; |
434 | 0 | m_poBaseHandle.reset(); |
435 | 0 | } |
436 | 0 | return bRet; |
437 | 0 | } |
438 | | |
439 | | /************************************************************************/ |
440 | | /* VSIGZipHandle() */ |
441 | | /************************************************************************/ |
442 | | |
443 | | VSIGZipHandle::VSIGZipHandle(VSIVirtualHandleUniquePtr poBaseHandleIn, |
444 | | const char *pszBaseFileName, vsi_l_offset offset, |
445 | | vsi_l_offset compressed_size, |
446 | | vsi_l_offset uncompressed_size, uLong expected_crc, |
447 | | int transparent) |
448 | 0 | : m_poBaseHandle(std::move(poBaseHandleIn)), |
449 | | #ifdef DEBUG |
450 | 0 | m_offset(offset), |
451 | | #endif |
452 | 0 | m_uncompressed_size(uncompressed_size), m_expected_crc(expected_crc), |
453 | 0 | m_pszBaseFileName(pszBaseFileName ? CPLStrdup(pszBaseFileName) : nullptr), |
454 | 0 | m_bWriteProperties(CPLTestBool( |
455 | 0 | CPLGetConfigOption("CPL_VSIL_GZIP_WRITE_PROPERTIES", "YES"))), |
456 | | m_bCanSaveInfo( |
457 | 0 | CPLTestBool(CPLGetConfigOption("CPL_VSIL_GZIP_SAVE_INFO", "YES"))), |
458 | 0 | stream(), crc(0), m_transparent(transparent) |
459 | 0 | { |
460 | 0 | if (compressed_size || transparent) |
461 | 0 | { |
462 | 0 | m_compressed_size = compressed_size; |
463 | 0 | } |
464 | 0 | else |
465 | 0 | { |
466 | 0 | if (m_poBaseHandle->Seek(0, SEEK_END) != 0) |
467 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
468 | 0 | m_compressed_size = m_poBaseHandle->Tell() - offset; |
469 | 0 | compressed_size = m_compressed_size; |
470 | 0 | } |
471 | 0 | offsetEndCompressedData = offset + compressed_size; |
472 | |
|
473 | 0 | if (m_poBaseHandle->Seek(offset, SEEK_SET) != 0) |
474 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
475 | |
|
476 | 0 | stream.zalloc = nullptr; |
477 | 0 | stream.zfree = nullptr; |
478 | 0 | stream.opaque = nullptr; |
479 | 0 | stream.next_in = inbuf = nullptr; |
480 | 0 | stream.next_out = outbuf = nullptr; |
481 | 0 | stream.avail_in = stream.avail_out = 0; |
482 | |
|
483 | 0 | inbuf = static_cast<Byte *>(ALLOC(Z_BUFSIZE)); |
484 | 0 | stream.next_in = inbuf; |
485 | |
|
486 | 0 | int err = inflateInit2(&(stream), -MAX_WBITS); |
487 | | // windowBits is passed < 0 to tell that there is no zlib header. |
488 | | // Note that in this case inflate *requires* an extra "dummy" byte |
489 | | // after the compressed stream in order to complete decompression and |
490 | | // return Z_STREAM_END. Here the gzip CRC32 ensures that 4 bytes are |
491 | | // present after the compressed stream. |
492 | 0 | if (err != Z_OK || inbuf == nullptr) |
493 | 0 | { |
494 | 0 | CPLError(CE_Failure, CPLE_NotSupported, "inflateInit2 init failed"); |
495 | 0 | TRYFREE(inbuf); |
496 | 0 | inbuf = nullptr; |
497 | 0 | return; |
498 | 0 | } |
499 | 0 | stream.avail_out = static_cast<uInt>(Z_BUFSIZE); |
500 | |
|
501 | 0 | if (offset == 0) |
502 | 0 | check_header(); // Skip the .gz header. |
503 | 0 | startOff = m_poBaseHandle->Tell() - stream.avail_in; |
504 | |
|
505 | 0 | if (transparent == 0) |
506 | 0 | { |
507 | 0 | snapshot_byte_interval = std::max(static_cast<vsi_l_offset>(Z_BUFSIZE), |
508 | 0 | compressed_size / 100); |
509 | 0 | snapshots = static_cast<GZipSnapshot *>(CPLCalloc( |
510 | 0 | sizeof(GZipSnapshot), |
511 | 0 | static_cast<size_t>(compressed_size / snapshot_byte_interval + 1))); |
512 | 0 | } |
513 | 0 | } |
514 | | |
515 | | /************************************************************************/ |
516 | | /* SaveInfo_unlocked() */ |
517 | | /************************************************************************/ |
518 | | |
519 | | void VSIGZipHandle::SaveInfo_unlocked() |
520 | 0 | { |
521 | 0 | if (m_pszBaseFileName && m_bCanSaveInfo) |
522 | 0 | { |
523 | 0 | VSIFilesystemHandler *poFSHandler = |
524 | 0 | VSIFileManager::GetHandler("/vsigzip/"); |
525 | 0 | cpl::down_cast<VSIGZipFilesystemHandler *>(poFSHandler) |
526 | 0 | ->SaveInfo_unlocked(this); |
527 | 0 | m_bCanSaveInfo = false; |
528 | 0 | } |
529 | 0 | } |
530 | | |
531 | | /************************************************************************/ |
532 | | /* ~VSIGZipHandle() */ |
533 | | /************************************************************************/ |
534 | | |
535 | | VSIGZipHandle::~VSIGZipHandle() |
536 | 0 | { |
537 | 0 | if (m_pszBaseFileName && m_bCanSaveInfo) |
538 | 0 | { |
539 | 0 | VSIFilesystemHandler *poFSHandler = |
540 | 0 | VSIFileManager::GetHandler("/vsigzip/"); |
541 | 0 | cpl::down_cast<VSIGZipFilesystemHandler *>(poFSHandler)->SaveInfo(this); |
542 | 0 | } |
543 | |
|
544 | 0 | if (stream.state != nullptr) |
545 | 0 | { |
546 | 0 | inflateEnd(&(stream)); |
547 | 0 | } |
548 | |
|
549 | 0 | TRYFREE(inbuf); |
550 | 0 | TRYFREE(outbuf); |
551 | |
|
552 | 0 | if (snapshots != nullptr) |
553 | 0 | { |
554 | 0 | for (size_t i = 0; i < m_compressed_size / snapshot_byte_interval + 1; |
555 | 0 | i++) |
556 | 0 | { |
557 | 0 | if (snapshots[i].posInBaseHandle) |
558 | 0 | { |
559 | 0 | inflateEnd(&(snapshots[i].stream)); |
560 | 0 | } |
561 | 0 | } |
562 | 0 | CPLFree(snapshots); |
563 | 0 | } |
564 | 0 | CPLFree(m_pszBaseFileName); |
565 | |
|
566 | 0 | CloseBaseHandle(); |
567 | 0 | } |
568 | | |
569 | | /************************************************************************/ |
570 | | /* check_header() */ |
571 | | /************************************************************************/ |
572 | | |
573 | | void VSIGZipHandle::check_header() |
574 | 0 | { |
575 | | // Assure two bytes in the buffer so we can peek ahead -- handle case |
576 | | // where first byte of header is at the end of the buffer after the last |
577 | | // gzip segment. |
578 | 0 | uInt len = stream.avail_in; |
579 | 0 | if (len < 2) |
580 | 0 | { |
581 | 0 | if (len) |
582 | 0 | inbuf[0] = stream.next_in[0]; |
583 | 0 | errno = 0; |
584 | 0 | size_t nToRead = static_cast<size_t>(Z_BUFSIZE - len); |
585 | 0 | CPLAssert(m_poBaseHandle->Tell() <= offsetEndCompressedData); |
586 | 0 | if (m_poBaseHandle->Tell() + nToRead > offsetEndCompressedData) |
587 | 0 | nToRead = static_cast<size_t>(offsetEndCompressedData - |
588 | 0 | m_poBaseHandle->Tell()); |
589 | |
|
590 | 0 | len = static_cast<uInt>(m_poBaseHandle->Read(inbuf + len, 1, nToRead)); |
591 | | #ifdef ENABLE_DEBUG |
592 | | CPLDebug("GZIP", CPL_FRMT_GUIB " " CPL_FRMT_GUIB, |
593 | | m_poBaseHandle->Tell(), offsetEndCompressedData); |
594 | | #endif |
595 | 0 | if (len == 0) // && ferror(file) |
596 | 0 | { |
597 | 0 | if (m_poBaseHandle->Tell() != offsetEndCompressedData) |
598 | 0 | z_err = Z_ERRNO; |
599 | 0 | } |
600 | 0 | stream.avail_in += len; |
601 | 0 | stream.next_in = inbuf; |
602 | 0 | if (stream.avail_in < 2) |
603 | 0 | { |
604 | 0 | m_transparent = stream.avail_in; |
605 | 0 | return; |
606 | 0 | } |
607 | 0 | } |
608 | | |
609 | | // Peek ahead to check the gzip magic header. |
610 | 0 | if (stream.next_in[0] != gz_magic[0] || stream.next_in[1] != gz_magic[1]) |
611 | 0 | { |
612 | 0 | m_transparent = 1; |
613 | 0 | return; |
614 | 0 | } |
615 | 0 | stream.avail_in -= 2; |
616 | 0 | stream.next_in += 2; |
617 | | |
618 | | // Check the rest of the gzip header. |
619 | 0 | const int method = get_byte(); |
620 | 0 | const int flags = get_byte(); |
621 | 0 | if (method != Z_DEFLATED || (flags & RESERVED) != 0) |
622 | 0 | { |
623 | 0 | z_err = Z_DATA_ERROR; |
624 | 0 | return; |
625 | 0 | } |
626 | | |
627 | | // Discard time, xflags and OS code: |
628 | 0 | for (len = 0; len < 6; len++) |
629 | 0 | CPL_IGNORE_RET_VAL(get_byte()); |
630 | |
|
631 | 0 | if ((flags & EXTRA_FIELD) != 0) |
632 | 0 | { |
633 | | // Skip the extra field. |
634 | 0 | len = static_cast<uInt>(get_byte()) & 0xFF; |
635 | 0 | len += (static_cast<uInt>(get_byte()) & 0xFF) << 8; |
636 | | // len is garbage if EOF but the loop below will quit anyway. |
637 | 0 | while (len != 0 && get_byte() != EOF) |
638 | 0 | { |
639 | 0 | --len; |
640 | 0 | } |
641 | 0 | } |
642 | |
|
643 | 0 | if ((flags & ORIG_NAME) != 0) |
644 | 0 | { |
645 | | // Skip the original file name. |
646 | 0 | int c; |
647 | 0 | while ((c = get_byte()) != 0 && c != EOF) |
648 | 0 | { |
649 | 0 | } |
650 | 0 | } |
651 | 0 | if ((flags & COMMENT) != 0) |
652 | 0 | { |
653 | | // skip the .gz file comment. |
654 | 0 | int c; |
655 | 0 | while ((c = get_byte()) != 0 && c != EOF) |
656 | 0 | { |
657 | 0 | } |
658 | 0 | } |
659 | 0 | if ((flags & HEAD_CRC) != 0) |
660 | 0 | { |
661 | | // Skip the header crc. |
662 | 0 | for (len = 0; len < 2; len++) |
663 | 0 | CPL_IGNORE_RET_VAL(get_byte()); |
664 | 0 | } |
665 | 0 | z_err = z_eof ? Z_DATA_ERROR : Z_OK; |
666 | 0 | } |
667 | | |
668 | | /************************************************************************/ |
669 | | /* get_byte() */ |
670 | | /************************************************************************/ |
671 | | |
672 | | int VSIGZipHandle::get_byte() |
673 | 0 | { |
674 | 0 | if (z_eof) |
675 | 0 | return EOF; |
676 | 0 | if (stream.avail_in == 0) |
677 | 0 | { |
678 | 0 | errno = 0; |
679 | 0 | size_t nToRead = static_cast<size_t>(Z_BUFSIZE); |
680 | 0 | CPLAssert(m_poBaseHandle->Tell() <= offsetEndCompressedData); |
681 | 0 | if (m_poBaseHandle->Tell() + nToRead > offsetEndCompressedData) |
682 | 0 | nToRead = static_cast<size_t>(offsetEndCompressedData - |
683 | 0 | m_poBaseHandle->Tell()); |
684 | 0 | stream.avail_in = |
685 | 0 | static_cast<uInt>(m_poBaseHandle->Read(inbuf, 1, nToRead)); |
686 | | #ifdef ENABLE_DEBUG |
687 | | CPLDebug("GZIP", CPL_FRMT_GUIB " " CPL_FRMT_GUIB, |
688 | | m_poBaseHandle->Tell(), offsetEndCompressedData); |
689 | | #endif |
690 | 0 | if (stream.avail_in == 0) |
691 | 0 | { |
692 | 0 | z_eof = 1; |
693 | 0 | if (m_poBaseHandle->Tell() != offsetEndCompressedData) |
694 | 0 | z_err = Z_ERRNO; |
695 | | // if( ferror(file) ) z_err = Z_ERRNO; |
696 | 0 | return EOF; |
697 | 0 | } |
698 | 0 | stream.next_in = inbuf; |
699 | 0 | } |
700 | 0 | stream.avail_in--; |
701 | 0 | return *(stream.next_in)++; |
702 | 0 | } |
703 | | |
704 | | /************************************************************************/ |
705 | | /* gzrewind() */ |
706 | | /************************************************************************/ |
707 | | |
708 | | int VSIGZipHandle::gzrewind() |
709 | 0 | { |
710 | 0 | z_err = Z_OK; |
711 | 0 | z_eof = 0; |
712 | 0 | m_bEOF = false; |
713 | 0 | stream.avail_in = 0; |
714 | 0 | stream.next_in = inbuf; |
715 | 0 | crc = 0; |
716 | 0 | if (!m_transparent) |
717 | 0 | CPL_IGNORE_RET_VAL(inflateReset(&stream)); |
718 | 0 | in = 0; |
719 | 0 | out = 0; |
720 | 0 | return m_poBaseHandle->Seek(startOff, SEEK_SET); |
721 | 0 | } |
722 | | |
723 | | /************************************************************************/ |
724 | | /* Seek() */ |
725 | | /************************************************************************/ |
726 | | |
727 | | int VSIGZipHandle::Seek(vsi_l_offset nOffset, int nWhence) |
728 | 0 | { |
729 | 0 | m_bEOF = false; |
730 | |
|
731 | 0 | return gzseek(nOffset, nWhence) ? 0 : -1; |
732 | 0 | } |
733 | | |
734 | | /************************************************************************/ |
735 | | /* gzseek() */ |
736 | | /************************************************************************/ |
737 | | |
738 | | bool VSIGZipHandle::gzseek(vsi_l_offset offset, int whence) |
739 | 0 | { |
740 | 0 | const vsi_l_offset original_offset = offset; |
741 | 0 | const int original_nWhence = whence; |
742 | |
|
743 | 0 | z_eof = 0; |
744 | | #ifdef ENABLE_DEBUG |
745 | | CPLDebug("GZIP", "Seek(" CPL_FRMT_GUIB ",%d)", offset, whence); |
746 | | #endif |
747 | |
|
748 | 0 | if (m_transparent) |
749 | 0 | { |
750 | 0 | stream.avail_in = 0; |
751 | 0 | stream.next_in = inbuf; |
752 | 0 | if (whence == SEEK_CUR) |
753 | 0 | { |
754 | 0 | if (out + offset > m_compressed_size) |
755 | 0 | { |
756 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
757 | 0 | return false; |
758 | 0 | } |
759 | | |
760 | 0 | offset = startOff + out + offset; |
761 | 0 | } |
762 | 0 | else if (whence == SEEK_SET) |
763 | 0 | { |
764 | 0 | if (offset > m_compressed_size) |
765 | 0 | { |
766 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
767 | 0 | return false; |
768 | 0 | } |
769 | | |
770 | 0 | offset = startOff + offset; |
771 | 0 | } |
772 | 0 | else if (whence == SEEK_END) |
773 | 0 | { |
774 | | // Commented test: because vsi_l_offset is unsigned (for the moment) |
775 | | // so no way to seek backward. See #1590 */ |
776 | 0 | if (offset > 0) // || -offset > compressed_size |
777 | 0 | { |
778 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
779 | 0 | return false; |
780 | 0 | } |
781 | | |
782 | 0 | offset = startOff + m_compressed_size - offset; |
783 | 0 | } |
784 | 0 | else |
785 | 0 | { |
786 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
787 | 0 | return false; |
788 | 0 | } |
789 | | |
790 | 0 | if (m_poBaseHandle->Seek(offset, SEEK_SET) < 0) |
791 | 0 | { |
792 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
793 | 0 | return false; |
794 | 0 | } |
795 | | |
796 | 0 | out = offset - startOff; |
797 | 0 | in = out; |
798 | 0 | return true; |
799 | 0 | } |
800 | | |
801 | | // whence == SEEK_END is unsuppored in original gzseek. |
802 | 0 | if (whence == SEEK_END) |
803 | 0 | { |
804 | | // If we known the uncompressed size, we can fake a jump to |
805 | | // the end of the stream. |
806 | 0 | if (offset == 0 && m_uncompressed_size != 0) |
807 | 0 | { |
808 | 0 | out = m_uncompressed_size; |
809 | 0 | return true; |
810 | 0 | } |
811 | | |
812 | | // We don't know the uncompressed size. This is unfortunate. |
813 | | // Do the slow version. |
814 | 0 | static int firstWarning = 1; |
815 | 0 | if (m_compressed_size > 10 * 1024 * 1024 && firstWarning) |
816 | 0 | { |
817 | 0 | CPLError(CE_Warning, CPLE_AppDefined, |
818 | 0 | "VSIFSeekL(xxx, SEEK_END) may be really slow " |
819 | 0 | "on GZip streams."); |
820 | 0 | firstWarning = 0; |
821 | 0 | } |
822 | |
|
823 | 0 | whence = SEEK_CUR; |
824 | 0 | offset = 1024 * 1024 * 1024; |
825 | 0 | offset *= 1024 * 1024; |
826 | 0 | } |
827 | | |
828 | | // Rest of function is for reading only. |
829 | | |
830 | | // Compute absolute position. |
831 | 0 | if (whence == SEEK_CUR) |
832 | 0 | { |
833 | 0 | offset += out; |
834 | 0 | } |
835 | | |
836 | | // For a negative seek, rewind and use positive seek. |
837 | 0 | if (offset >= out) |
838 | 0 | { |
839 | 0 | offset -= out; |
840 | 0 | } |
841 | 0 | else if (gzrewind() < 0) |
842 | 0 | { |
843 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
844 | 0 | return false; |
845 | 0 | } |
846 | | |
847 | 0 | if (z_err != Z_OK && z_err != Z_STREAM_END) |
848 | 0 | { |
849 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
850 | 0 | return false; |
851 | 0 | } |
852 | | |
853 | 0 | for (unsigned int i = 0; i < m_compressed_size / snapshot_byte_interval + 1; |
854 | 0 | i++) |
855 | 0 | { |
856 | 0 | if (snapshots[i].posInBaseHandle == 0) |
857 | 0 | break; |
858 | 0 | if (snapshots[i].out <= out + offset && |
859 | 0 | (i == m_compressed_size / snapshot_byte_interval || |
860 | 0 | snapshots[i + 1].out == 0 || snapshots[i + 1].out > out + offset)) |
861 | 0 | { |
862 | 0 | if (out >= snapshots[i].out) |
863 | 0 | break; |
864 | | |
865 | | #ifdef ENABLE_DEBUG |
866 | | CPLDebug("SNAPSHOT", |
867 | | "using snapshot %d : " |
868 | | "posInBaseHandle(snapshot)=" CPL_FRMT_GUIB |
869 | | " in(snapshot)=" CPL_FRMT_GUIB |
870 | | " out(snapshot)=" CPL_FRMT_GUIB " out=" CPL_FRMT_GUIB |
871 | | " offset=" CPL_FRMT_GUIB, |
872 | | i, snapshots[i].posInBaseHandle, snapshots[i].in, |
873 | | snapshots[i].out, out, offset); |
874 | | #endif |
875 | 0 | offset = out + offset - snapshots[i].out; |
876 | 0 | if (m_poBaseHandle->Seek(snapshots[i].posInBaseHandle, SEEK_SET) != |
877 | 0 | 0) |
878 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
879 | |
|
880 | 0 | inflateEnd(&stream); |
881 | 0 | inflateCopy(&stream, &snapshots[i].stream); |
882 | 0 | crc = snapshots[i].crc; |
883 | 0 | m_transparent = snapshots[i].transparent; |
884 | 0 | in = snapshots[i].in; |
885 | 0 | out = snapshots[i].out; |
886 | 0 | break; |
887 | 0 | } |
888 | 0 | } |
889 | | |
890 | | // Offset is now the number of bytes to skip. |
891 | |
|
892 | 0 | if (offset != 0 && outbuf == nullptr) |
893 | 0 | { |
894 | 0 | outbuf = static_cast<Byte *>(ALLOC(Z_BUFSIZE)); |
895 | 0 | if (outbuf == nullptr) |
896 | 0 | { |
897 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
898 | 0 | return false; |
899 | 0 | } |
900 | 0 | } |
901 | | |
902 | 0 | if (original_nWhence == SEEK_END && z_err == Z_STREAM_END) |
903 | 0 | { |
904 | 0 | return true; |
905 | 0 | } |
906 | | |
907 | 0 | while (offset > 0) |
908 | 0 | { |
909 | 0 | int size = Z_BUFSIZE; |
910 | 0 | if (offset < static_cast<vsi_l_offset>(Z_BUFSIZE)) |
911 | 0 | size = static_cast<int>(offset); |
912 | |
|
913 | 0 | int read_size = |
914 | 0 | static_cast<int>(Read(outbuf, 1, static_cast<uInt>(size))); |
915 | 0 | if (original_nWhence == SEEK_END) |
916 | 0 | { |
917 | 0 | if (size != read_size) |
918 | 0 | { |
919 | 0 | z_err = Z_STREAM_END; |
920 | 0 | break; |
921 | 0 | } |
922 | 0 | } |
923 | 0 | else if (read_size == 0) |
924 | 0 | { |
925 | | // CPL_VSIL_GZ_RETURN(FALSE); |
926 | 0 | return false; |
927 | 0 | } |
928 | 0 | offset -= read_size; |
929 | 0 | } |
930 | | #ifdef ENABLE_DEBUG |
931 | | CPLDebug("GZIP", "gzseek at offset " CPL_FRMT_GUIB, out); |
932 | | #endif |
933 | | |
934 | 0 | if (original_offset == 0 && original_nWhence == SEEK_END) |
935 | 0 | { |
936 | 0 | m_uncompressed_size = out; |
937 | |
|
938 | 0 | if (m_pszBaseFileName && !STARTS_WITH(m_pszBaseFileName, "/vsicurl/") && |
939 | 0 | !STARTS_WITH(m_pszBaseFileName, "/vsitar/") && |
940 | 0 | !STARTS_WITH(m_pszBaseFileName, "/vsizip/") && m_bWriteProperties) |
941 | 0 | { |
942 | 0 | CPLErrorStateBackuper oErrorStateBackuper(CPLQuietErrorHandler); |
943 | |
|
944 | 0 | CPLString osCacheFilename(m_pszBaseFileName); |
945 | 0 | osCacheFilename += ".properties"; |
946 | | |
947 | | // Write a .properties file to avoid seeking next time. |
948 | 0 | VSILFILE *fpCacheLength = VSIFOpenL(osCacheFilename.c_str(), "wb"); |
949 | 0 | if (fpCacheLength) |
950 | 0 | { |
951 | 0 | char szBuffer[32] = {}; |
952 | |
|
953 | 0 | CPLPrintUIntBig(szBuffer, m_compressed_size, 31); |
954 | 0 | char *pszFirstNonSpace = szBuffer; |
955 | 0 | while (*pszFirstNonSpace == ' ') |
956 | 0 | pszFirstNonSpace++; |
957 | 0 | CPL_IGNORE_RET_VAL(VSIFPrintfL( |
958 | 0 | fpCacheLength, "compressed_size=%s\n", pszFirstNonSpace)); |
959 | |
|
960 | 0 | CPLPrintUIntBig(szBuffer, m_uncompressed_size, 31); |
961 | 0 | pszFirstNonSpace = szBuffer; |
962 | 0 | while (*pszFirstNonSpace == ' ') |
963 | 0 | pszFirstNonSpace++; |
964 | 0 | CPL_IGNORE_RET_VAL(VSIFPrintfL( |
965 | 0 | fpCacheLength, "uncompressed_size=%s\n", pszFirstNonSpace)); |
966 | |
|
967 | 0 | CPL_IGNORE_RET_VAL(VSIFCloseL(fpCacheLength)); |
968 | 0 | } |
969 | 0 | } |
970 | 0 | } |
971 | |
|
972 | 0 | return true; |
973 | 0 | } |
974 | | |
975 | | /************************************************************************/ |
976 | | /* Tell() */ |
977 | | /************************************************************************/ |
978 | | |
979 | | vsi_l_offset VSIGZipHandle::Tell() |
980 | 0 | { |
981 | | #ifdef ENABLE_DEBUG |
982 | | CPLDebug("GZIP", "Tell() = " CPL_FRMT_GUIB, out); |
983 | | #endif |
984 | 0 | return out; |
985 | 0 | } |
986 | | |
987 | | /************************************************************************/ |
988 | | /* Read() */ |
989 | | /************************************************************************/ |
990 | | |
991 | | size_t VSIGZipHandle::Read(void *const buf, size_t const nSize, |
992 | | size_t const nMemb) |
993 | 0 | { |
994 | | #ifdef ENABLE_DEBUG |
995 | | CPLDebug("GZIP", "Read(%p, %d, %d)", buf, static_cast<int>(nSize), |
996 | | static_cast<int>(nMemb)); |
997 | | #endif |
998 | |
|
999 | 0 | if (m_bEOF || z_err != Z_OK) |
1000 | 0 | { |
1001 | 0 | if (z_err == Z_STREAM_END && nSize > 0 && nMemb > 0) |
1002 | 0 | m_bEOF = true; |
1003 | 0 | return 0; |
1004 | 0 | } |
1005 | | |
1006 | 0 | if (nSize > 0 && nMemb > UINT32_MAX / nSize) |
1007 | 0 | { |
1008 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Too many bytes to read at once"); |
1009 | 0 | return 0; |
1010 | 0 | } |
1011 | | |
1012 | 0 | const unsigned len = |
1013 | 0 | static_cast<unsigned int>(nSize) * static_cast<unsigned int>(nMemb); |
1014 | 0 | Bytef *pStart = |
1015 | 0 | static_cast<Bytef *>(buf); // Start off point for crc computation. |
1016 | | // == stream.next_out but not forced far (for MSDOS). |
1017 | 0 | Byte *next_out = static_cast<Byte *>(buf); |
1018 | 0 | stream.next_out = static_cast<Bytef *>(buf); |
1019 | 0 | stream.avail_out = len; |
1020 | |
|
1021 | 0 | while (stream.avail_out != 0) |
1022 | 0 | { |
1023 | 0 | if (m_transparent) |
1024 | 0 | { |
1025 | | // Copy first the lookahead bytes: |
1026 | 0 | uInt nRead = 0; |
1027 | 0 | uInt n = stream.avail_in; |
1028 | 0 | if (n > stream.avail_out) |
1029 | 0 | n = stream.avail_out; |
1030 | 0 | if (n > 0) |
1031 | 0 | { |
1032 | 0 | memcpy(stream.next_out, stream.next_in, n); |
1033 | 0 | next_out += n; |
1034 | 0 | stream.next_out = next_out; |
1035 | 0 | stream.next_in += n; |
1036 | 0 | stream.avail_out -= n; |
1037 | 0 | stream.avail_in -= n; |
1038 | 0 | nRead += n; |
1039 | 0 | } |
1040 | 0 | if (stream.avail_out > 0) |
1041 | 0 | { |
1042 | 0 | const uInt nToRead = static_cast<uInt>( |
1043 | 0 | std::min(m_compressed_size - (in + nRead), |
1044 | 0 | static_cast<vsi_l_offset>(stream.avail_out))); |
1045 | 0 | const uInt nReadFromFile = static_cast<uInt>( |
1046 | 0 | m_poBaseHandle->Read(next_out, 1, nToRead)); |
1047 | 0 | if (nReadFromFile < nToRead && m_poBaseHandle->Error()) |
1048 | 0 | z_err = Z_ERRNO; |
1049 | 0 | stream.avail_out -= nReadFromFile; |
1050 | 0 | nRead += nReadFromFile; |
1051 | 0 | } |
1052 | 0 | in += nRead; |
1053 | 0 | out += nRead; |
1054 | 0 | if (nRead < len) |
1055 | 0 | { |
1056 | 0 | m_bEOF = true; |
1057 | 0 | z_eof = 1; |
1058 | 0 | } |
1059 | | #ifdef ENABLE_DEBUG |
1060 | | CPLDebug("GZIP", "Read return %d", static_cast<int>(nRead / nSize)); |
1061 | | #endif |
1062 | 0 | return static_cast<int>(nRead) / nSize; |
1063 | 0 | } |
1064 | 0 | if (stream.avail_in == 0 && !z_eof) |
1065 | 0 | { |
1066 | 0 | vsi_l_offset posInBaseHandle = m_poBaseHandle->Tell(); |
1067 | 0 | if (posInBaseHandle - startOff > m_compressed_size) |
1068 | 0 | { |
1069 | | // If we reach here, file size has changed (because at |
1070 | | // construction time startOff + m_compressed_size marked the |
1071 | | // end of file). |
1072 | | // We should probably have a better fix than that, by detecting |
1073 | | // at open time that the saved snapshot is not valid and |
1074 | | // discarding it. |
1075 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1076 | 0 | "File size of underlying /vsigzip/ file has changed"); |
1077 | 0 | z_err = Z_ERRNO; |
1078 | 0 | CPL_VSIL_GZ_RETURN(0); |
1079 | 0 | return 0; |
1080 | 0 | } |
1081 | 0 | GZipSnapshot *snapshot = &snapshots[(posInBaseHandle - startOff) / |
1082 | 0 | snapshot_byte_interval]; |
1083 | 0 | if (snapshot->posInBaseHandle == 0) |
1084 | 0 | { |
1085 | 0 | snapshot->crc = crc32( |
1086 | 0 | crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1087 | | #ifdef ENABLE_DEBUG |
1088 | | CPLDebug("SNAPSHOT", |
1089 | | "creating snapshot %d : " |
1090 | | "posInBaseHandle=" CPL_FRMT_GUIB " in=" CPL_FRMT_GUIB |
1091 | | " out=" CPL_FRMT_GUIB " crc=%X", |
1092 | | static_cast<int>((posInBaseHandle - startOff) / |
1093 | | snapshot_byte_interval), |
1094 | | posInBaseHandle, in, out, |
1095 | | static_cast<unsigned int>(snapshot->crc)); |
1096 | | #endif |
1097 | 0 | snapshot->posInBaseHandle = posInBaseHandle; |
1098 | 0 | inflateCopy(&snapshot->stream, &stream); |
1099 | 0 | snapshot->transparent = m_transparent; |
1100 | 0 | snapshot->in = in; |
1101 | 0 | snapshot->out = out; |
1102 | |
|
1103 | 0 | if (out > m_nLastReadOffset) |
1104 | 0 | m_nLastReadOffset = out; |
1105 | 0 | } |
1106 | |
|
1107 | 0 | errno = 0; |
1108 | 0 | stream.avail_in = |
1109 | 0 | static_cast<uInt>(m_poBaseHandle->Read(inbuf, 1, Z_BUFSIZE)); |
1110 | | #ifdef ENABLE_DEBUG |
1111 | | CPLDebug("GZIP", CPL_FRMT_GUIB " " CPL_FRMT_GUIB, |
1112 | | m_poBaseHandle->Tell(), offsetEndCompressedData); |
1113 | | #endif |
1114 | 0 | if (m_poBaseHandle->Tell() > offsetEndCompressedData) |
1115 | 0 | { |
1116 | | #ifdef ENABLE_DEBUG |
1117 | | CPLDebug("GZIP", "avail_in before = %d", stream.avail_in); |
1118 | | #endif |
1119 | 0 | stream.avail_in = stream.avail_in - |
1120 | 0 | static_cast<uInt>(m_poBaseHandle->Tell() - |
1121 | 0 | offsetEndCompressedData); |
1122 | 0 | if (m_poBaseHandle->Seek(offsetEndCompressedData, SEEK_SET) != |
1123 | 0 | 0) |
1124 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
1125 | | #ifdef ENABLE_DEBUG |
1126 | | CPLDebug("GZIP", "avail_in after = %d", stream.avail_in); |
1127 | | #endif |
1128 | 0 | } |
1129 | 0 | if (stream.avail_in == 0) |
1130 | 0 | { |
1131 | 0 | z_eof = 1; |
1132 | 0 | if (m_poBaseHandle->Error() || |
1133 | 0 | m_poBaseHandle->Tell() != offsetEndCompressedData) |
1134 | 0 | { |
1135 | 0 | z_err = Z_ERRNO; |
1136 | 0 | break; |
1137 | 0 | } |
1138 | 0 | } |
1139 | 0 | stream.next_in = inbuf; |
1140 | 0 | } |
1141 | 0 | in += stream.avail_in; |
1142 | 0 | out += stream.avail_out; |
1143 | 0 | z_err = inflate(&(stream), Z_NO_FLUSH); |
1144 | 0 | in -= stream.avail_in; |
1145 | 0 | out -= stream.avail_out; |
1146 | |
|
1147 | 0 | if (z_err == Z_STREAM_END && m_compressed_size != 2) |
1148 | 0 | { |
1149 | | // Check CRC and original size. |
1150 | 0 | crc = |
1151 | 0 | crc32(crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1152 | 0 | pStart = stream.next_out; |
1153 | 0 | if (m_expected_crc) |
1154 | 0 | { |
1155 | | #ifdef ENABLE_DEBUG |
1156 | | CPLDebug("GZIP", "Computed CRC = %X. Expected CRC = %X", |
1157 | | static_cast<unsigned int>(crc), |
1158 | | static_cast<unsigned int>(m_expected_crc)); |
1159 | | #endif |
1160 | 0 | } |
1161 | 0 | if (m_expected_crc != 0 && m_expected_crc != crc) |
1162 | 0 | { |
1163 | 0 | CPLError(CE_Failure, CPLE_FileIO, |
1164 | 0 | "CRC error. Got %X instead of %X", |
1165 | 0 | static_cast<unsigned int>(crc), |
1166 | 0 | static_cast<unsigned int>(m_expected_crc)); |
1167 | 0 | z_err = Z_DATA_ERROR; |
1168 | 0 | } |
1169 | 0 | else if (m_expected_crc == 0) |
1170 | 0 | { |
1171 | 0 | const uLong read_crc = static_cast<unsigned long>(getLong()); |
1172 | 0 | if (read_crc != crc) |
1173 | 0 | { |
1174 | 0 | CPLError(CE_Failure, CPLE_FileIO, |
1175 | 0 | "CRC error. Got %X instead of %X", |
1176 | 0 | static_cast<unsigned int>(crc), |
1177 | 0 | static_cast<unsigned int>(read_crc)); |
1178 | 0 | z_err = Z_DATA_ERROR; |
1179 | 0 | } |
1180 | 0 | else |
1181 | 0 | { |
1182 | 0 | CPL_IGNORE_RET_VAL(getLong()); |
1183 | | // The uncompressed length returned by above getlong() may |
1184 | | // be different from out in case of concatenated .gz files. |
1185 | | // Check for such files: |
1186 | 0 | check_header(); |
1187 | 0 | if (z_err == Z_OK) |
1188 | 0 | { |
1189 | 0 | inflateReset(&(stream)); |
1190 | 0 | crc = 0; |
1191 | 0 | } |
1192 | 0 | } |
1193 | 0 | } |
1194 | 0 | } |
1195 | 0 | if (z_err != Z_OK || z_eof) |
1196 | 0 | break; |
1197 | 0 | } |
1198 | 0 | crc = crc32(crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1199 | |
|
1200 | 0 | size_t ret = (len - stream.avail_out) / nSize; |
1201 | 0 | if (z_err != Z_OK && z_err != Z_STREAM_END) |
1202 | 0 | { |
1203 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1204 | 0 | "In file %s, at line %d, decompression failed with " |
1205 | 0 | "z_err = %d, return = %d", |
1206 | 0 | __FILE__, __LINE__, z_err, static_cast<int>(ret)); |
1207 | 0 | } |
1208 | 0 | else if (ret < nMemb) |
1209 | 0 | { |
1210 | 0 | m_bEOF = true; |
1211 | 0 | } |
1212 | |
|
1213 | | #ifdef ENABLE_DEBUG |
1214 | | CPLDebug("GZIP", "Read return %d (z_err=%d, z_eof=%d)", |
1215 | | static_cast<int>(ret), z_err, z_eof); |
1216 | | #endif |
1217 | 0 | return ret; |
1218 | 0 | } |
1219 | | |
1220 | | /************************************************************************/ |
1221 | | /* getLong() */ |
1222 | | /************************************************************************/ |
1223 | | |
1224 | | uLong VSIGZipHandle::getLong() |
1225 | 0 | { |
1226 | 0 | uLong x = static_cast<uLong>(get_byte()) & 0xFF; |
1227 | |
|
1228 | 0 | x += (static_cast<uLong>(get_byte()) & 0xFF) << 8; |
1229 | 0 | x += (static_cast<uLong>(get_byte()) & 0xFF) << 16; |
1230 | 0 | const int c = get_byte(); |
1231 | 0 | if (c == EOF) |
1232 | 0 | { |
1233 | 0 | z_err = Z_DATA_ERROR; |
1234 | 0 | return 0; |
1235 | 0 | } |
1236 | 0 | x += static_cast<uLong>(c) << 24; |
1237 | | // coverity[overflow_sink] |
1238 | 0 | return x; |
1239 | 0 | } |
1240 | | |
1241 | | /************************************************************************/ |
1242 | | /* Write() */ |
1243 | | /************************************************************************/ |
1244 | | |
1245 | | size_t VSIGZipHandle::Write(const void * /* pBuffer */, size_t /* nSize */, |
1246 | | size_t /* nMemb */) |
1247 | 0 | { |
1248 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
1249 | 0 | "VSIFWriteL is not supported on GZip streams"); |
1250 | 0 | return 0; |
1251 | 0 | } |
1252 | | |
1253 | | /************************************************************************/ |
1254 | | /* Eof() */ |
1255 | | /************************************************************************/ |
1256 | | |
1257 | | int VSIGZipHandle::Eof() |
1258 | 0 | { |
1259 | | #ifdef ENABLE_DEBUG |
1260 | | CPLDebug("GZIP", "Eof()"); |
1261 | | #endif |
1262 | 0 | return m_bEOF; |
1263 | 0 | } |
1264 | | |
1265 | | /************************************************************************/ |
1266 | | /* Error() */ |
1267 | | /************************************************************************/ |
1268 | | |
1269 | | int VSIGZipHandle::Error() |
1270 | 0 | { |
1271 | | #ifdef ENABLE_DEBUG |
1272 | | CPLDebug("GZIP", "Error()"); |
1273 | | #endif |
1274 | 0 | return z_err != Z_OK && z_err != Z_STREAM_END; |
1275 | 0 | } |
1276 | | |
1277 | | /************************************************************************/ |
1278 | | /* ClearErr() */ |
1279 | | /************************************************************************/ |
1280 | | |
1281 | | void VSIGZipHandle::ClearErr() |
1282 | 0 | { |
1283 | 0 | m_poBaseHandle->ClearErr(); |
1284 | 0 | z_eof = 0; |
1285 | 0 | m_bEOF = false; |
1286 | 0 | z_err = Z_OK; |
1287 | 0 | } |
1288 | | |
1289 | | /************************************************************************/ |
1290 | | /* Flush() */ |
1291 | | /************************************************************************/ |
1292 | | |
1293 | | int VSIGZipHandle::Flush() |
1294 | 0 | { |
1295 | 0 | return 0; |
1296 | 0 | } |
1297 | | |
1298 | | /************************************************************************/ |
1299 | | /* Close() */ |
1300 | | /************************************************************************/ |
1301 | | |
1302 | | int VSIGZipHandle::Close() |
1303 | 0 | { |
1304 | 0 | return 0; |
1305 | 0 | } |
1306 | | |
1307 | | #ifdef ENABLE_DEFLATE64 |
1308 | | |
1309 | | /************************************************************************/ |
1310 | | /* Duplicate() */ |
1311 | | /************************************************************************/ |
1312 | | |
1313 | | VSIDeflate64Handle *VSIDeflate64Handle::Duplicate() |
1314 | 0 | { |
1315 | 0 | CPLAssert(m_offset == 0); |
1316 | 0 | CPLAssert(m_compressed_size != 0); |
1317 | 0 | CPLAssert(m_pszBaseFileName != nullptr); |
1318 | | |
1319 | 0 | VSIFilesystemHandler *poFSHandler = |
1320 | 0 | VSIFileManager::GetHandler(m_pszBaseFileName); |
1321 | |
|
1322 | 0 | VSIVirtualHandleUniquePtr poNewBaseHandle( |
1323 | 0 | poFSHandler->Open(m_pszBaseFileName, "rb")); |
1324 | |
|
1325 | 0 | if (poNewBaseHandle == nullptr) |
1326 | 0 | return nullptr; |
1327 | | |
1328 | 0 | auto poHandle = std::make_unique<VSIDeflate64Handle>( |
1329 | 0 | std::move(poNewBaseHandle), m_pszBaseFileName, 0, m_compressed_size, |
1330 | 0 | m_uncompressed_size); |
1331 | 0 | if (!(poHandle->IsInitOK())) |
1332 | 0 | { |
1333 | 0 | return nullptr; |
1334 | 0 | } |
1335 | | |
1336 | | // Most important: duplicate the snapshots! |
1337 | | |
1338 | 0 | for (unsigned int i = 0; i < m_compressed_size / snapshot_byte_interval + 1; |
1339 | 0 | i++) |
1340 | 0 | { |
1341 | 0 | if (snapshots[i].posInBaseHandle == 0) |
1342 | 0 | break; |
1343 | | |
1344 | 0 | poHandle->snapshots[i].posInBaseHandle = snapshots[i].posInBaseHandle; |
1345 | 0 | if (inflateBack9Copy(&poHandle->snapshots[i].stream, |
1346 | 0 | &snapshots[i].stream) != Z_OK) |
1347 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "inflateBack9Copy() failed"); |
1348 | 0 | poHandle->snapshots[i].crc = snapshots[i].crc; |
1349 | 0 | poHandle->snapshots[i].in = snapshots[i].in; |
1350 | 0 | poHandle->snapshots[i].out = snapshots[i].out; |
1351 | 0 | poHandle->snapshots[i].extraOutput = snapshots[i].extraOutput; |
1352 | 0 | poHandle->snapshots[i].m_bStreamEndReached = |
1353 | 0 | snapshots[i].m_bStreamEndReached; |
1354 | 0 | } |
1355 | |
|
1356 | 0 | return poHandle.release(); |
1357 | 0 | } |
1358 | | |
1359 | | /************************************************************************/ |
1360 | | /* CloseBaseHandle() */ |
1361 | | /************************************************************************/ |
1362 | | |
1363 | | bool VSIDeflate64Handle::CloseBaseHandle() |
1364 | 0 | { |
1365 | 0 | bool bRet = true; |
1366 | 0 | if (m_poBaseHandle) |
1367 | 0 | { |
1368 | 0 | bRet = m_poBaseHandle->Close() == 0; |
1369 | 0 | m_poBaseHandle.reset(); |
1370 | 0 | } |
1371 | 0 | return bRet; |
1372 | 0 | } |
1373 | | |
1374 | | /************************************************************************/ |
1375 | | /* VSIDeflate64Handle() */ |
1376 | | /************************************************************************/ |
1377 | | |
1378 | | VSIDeflate64Handle::VSIDeflate64Handle(VSIVirtualHandleUniquePtr poBaseHandleIn, |
1379 | | const char *pszBaseFileName, |
1380 | | vsi_l_offset offset, |
1381 | | vsi_l_offset compressed_size, |
1382 | | vsi_l_offset uncompressed_size, |
1383 | | uLong expected_crc) |
1384 | 0 | : m_poBaseHandle(std::move(poBaseHandleIn)), |
1385 | | #ifdef DEBUG |
1386 | 0 | m_offset(offset), |
1387 | | #endif |
1388 | 0 | m_uncompressed_size(uncompressed_size), m_expected_crc(expected_crc), |
1389 | 0 | m_pszBaseFileName(pszBaseFileName ? CPLStrdup(pszBaseFileName) : nullptr), |
1390 | 0 | stream(), crc(0) |
1391 | 0 | { |
1392 | 0 | if (compressed_size) |
1393 | 0 | { |
1394 | 0 | m_compressed_size = compressed_size; |
1395 | 0 | } |
1396 | 0 | else |
1397 | 0 | { |
1398 | 0 | if (m_poBaseHandle->Seek(0, SEEK_END) != 0) |
1399 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
1400 | 0 | m_compressed_size = m_poBaseHandle->Tell() - offset; |
1401 | 0 | compressed_size = m_compressed_size; |
1402 | 0 | } |
1403 | 0 | offsetEndCompressedData = offset + compressed_size; |
1404 | |
|
1405 | 0 | if (m_poBaseHandle->Seek(offset, SEEK_SET) != 0) |
1406 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
1407 | |
|
1408 | 0 | stream.zalloc = nullptr; |
1409 | 0 | stream.zfree = nullptr; |
1410 | 0 | stream.opaque = nullptr; |
1411 | 0 | stream.next_in = inbuf = nullptr; |
1412 | 0 | stream.next_out = outbuf = nullptr; |
1413 | 0 | stream.avail_in = stream.avail_out = 0; |
1414 | |
|
1415 | 0 | inbuf = static_cast<Byte *>(ALLOC(Z_BUFSIZE)); |
1416 | 0 | stream.next_in = inbuf; |
1417 | |
|
1418 | 0 | int err = inflateBack9Init(&(stream), nullptr); |
1419 | | // Note that in this case inflate *requires* an extra "dummy" byte |
1420 | | // after the compressed stream in order to complete decompression and |
1421 | | // return Z_STREAM_END. Here the gzip CRC32 ensures that 4 bytes are |
1422 | | // present after the compressed stream. |
1423 | 0 | if (err != Z_OK || inbuf == nullptr) |
1424 | 0 | { |
1425 | 0 | CPLError(CE_Failure, CPLE_NotSupported, "inflateBack9Init init failed"); |
1426 | 0 | TRYFREE(inbuf); |
1427 | 0 | inbuf = nullptr; |
1428 | 0 | return; |
1429 | 0 | } |
1430 | 0 | startOff = m_poBaseHandle->Tell() - stream.avail_in; |
1431 | |
|
1432 | 0 | snapshot_byte_interval = |
1433 | 0 | std::max(static_cast<vsi_l_offset>(Z_BUFSIZE), compressed_size / 100); |
1434 | 0 | snapshots.resize( |
1435 | 0 | static_cast<size_t>(compressed_size / snapshot_byte_interval + 1)); |
1436 | 0 | } |
1437 | | |
1438 | | /************************************************************************/ |
1439 | | /* ~VSIDeflate64Handle() */ |
1440 | | /************************************************************************/ |
1441 | | |
1442 | | VSIDeflate64Handle::~VSIDeflate64Handle() |
1443 | 0 | { |
1444 | 0 | if (stream.state != nullptr) |
1445 | 0 | { |
1446 | 0 | inflateBack9End(&(stream)); |
1447 | 0 | } |
1448 | |
|
1449 | 0 | TRYFREE(inbuf); |
1450 | 0 | TRYFREE(outbuf); |
1451 | |
|
1452 | 0 | for (auto &snapshot : snapshots) |
1453 | 0 | { |
1454 | 0 | if (snapshot.posInBaseHandle) |
1455 | 0 | { |
1456 | 0 | inflateBack9End(&(snapshot.stream)); |
1457 | 0 | } |
1458 | 0 | } |
1459 | 0 | CPLFree(m_pszBaseFileName); |
1460 | |
|
1461 | 0 | CloseBaseHandle(); |
1462 | 0 | } |
1463 | | |
1464 | | /************************************************************************/ |
1465 | | /* gzrewind() */ |
1466 | | /************************************************************************/ |
1467 | | |
1468 | | int VSIDeflate64Handle::gzrewind() |
1469 | 0 | { |
1470 | 0 | m_bStreamEndReached = false; |
1471 | 0 | extraOutput.clear(); |
1472 | 0 | z_err = Z_OK; |
1473 | 0 | z_eof = 0; |
1474 | 0 | stream.avail_in = 0; |
1475 | 0 | stream.next_in = inbuf; |
1476 | 0 | crc = 0; |
1477 | 0 | CPL_IGNORE_RET_VAL(inflateBack9End(&stream)); |
1478 | 0 | CPL_IGNORE_RET_VAL(inflateBack9Init(&stream, nullptr)); |
1479 | 0 | in = 0; |
1480 | 0 | out = 0; |
1481 | 0 | return m_poBaseHandle->Seek(startOff, SEEK_SET); |
1482 | 0 | } |
1483 | | |
1484 | | /************************************************************************/ |
1485 | | /* Seek() */ |
1486 | | /************************************************************************/ |
1487 | | |
1488 | | int VSIDeflate64Handle::Seek(vsi_l_offset nOffset, int nWhence) |
1489 | 0 | { |
1490 | 0 | m_bEOF = false; |
1491 | 0 | return gzseek(nOffset, nWhence) ? 0 : -1; |
1492 | 0 | } |
1493 | | |
1494 | | /************************************************************************/ |
1495 | | /* gzseek() */ |
1496 | | /************************************************************************/ |
1497 | | |
1498 | | bool VSIDeflate64Handle::gzseek(vsi_l_offset offset, int whence) |
1499 | 0 | { |
1500 | 0 | const vsi_l_offset original_offset = offset; |
1501 | 0 | const int original_nWhence = whence; |
1502 | |
|
1503 | 0 | z_eof = 0; |
1504 | | #ifdef ENABLE_DEBUG |
1505 | | CPLDebug("GZIP", "Seek(" CPL_FRMT_GUIB ",%d)", offset, whence); |
1506 | | #endif |
1507 | | |
1508 | | // whence == SEEK_END is unsuppored in original gzseek. |
1509 | 0 | if (whence == SEEK_END) |
1510 | 0 | { |
1511 | | // If we known the uncompressed size, we can fake a jump to |
1512 | | // the end of the stream. |
1513 | 0 | if (offset == 0 && m_uncompressed_size != 0) |
1514 | 0 | { |
1515 | 0 | out = m_uncompressed_size; |
1516 | 0 | return true; |
1517 | 0 | } |
1518 | | |
1519 | | // We don't know the uncompressed size. This is unfortunate. |
1520 | | // Do the slow version. |
1521 | 0 | static int firstWarning = 1; |
1522 | 0 | if (m_compressed_size > 10 * 1024 * 1024 && firstWarning) |
1523 | 0 | { |
1524 | 0 | CPLError(CE_Warning, CPLE_AppDefined, |
1525 | 0 | "VSIFSeekL(xxx, SEEK_END) may be really slow " |
1526 | 0 | "on GZip streams."); |
1527 | 0 | firstWarning = 0; |
1528 | 0 | } |
1529 | |
|
1530 | 0 | whence = SEEK_CUR; |
1531 | 0 | offset = 1024 * 1024 * 1024; |
1532 | 0 | offset *= 1024 * 1024; |
1533 | 0 | } |
1534 | | |
1535 | | // Rest of function is for reading only. |
1536 | | |
1537 | | // Compute absolute position. |
1538 | 0 | if (whence == SEEK_CUR) |
1539 | 0 | { |
1540 | 0 | offset += out; |
1541 | 0 | } |
1542 | | |
1543 | | // For a negative seek, rewind and use positive seek. |
1544 | 0 | if (offset >= out) |
1545 | 0 | { |
1546 | 0 | offset -= out; |
1547 | 0 | } |
1548 | 0 | else if (gzrewind() < 0) |
1549 | 0 | { |
1550 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
1551 | 0 | return false; |
1552 | 0 | } |
1553 | | |
1554 | 0 | if (z_err != Z_OK && z_err != Z_STREAM_END) |
1555 | 0 | { |
1556 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
1557 | 0 | return false; |
1558 | 0 | } |
1559 | | |
1560 | 0 | for (unsigned int i = 0; i < m_compressed_size / snapshot_byte_interval + 1; |
1561 | 0 | i++) |
1562 | 0 | { |
1563 | 0 | if (snapshots[i].posInBaseHandle == 0) |
1564 | 0 | break; |
1565 | 0 | if (snapshots[i].out <= out + offset && |
1566 | 0 | (i == m_compressed_size / snapshot_byte_interval || |
1567 | 0 | snapshots[i + 1].out == 0 || snapshots[i + 1].out > out + offset)) |
1568 | 0 | { |
1569 | 0 | if (out >= snapshots[i].out) |
1570 | 0 | break; |
1571 | | |
1572 | | #ifdef ENABLE_DEBUG |
1573 | | CPLDebug("SNAPSHOT", |
1574 | | "using snapshot %d : " |
1575 | | "posInBaseHandle(snapshot)=" CPL_FRMT_GUIB |
1576 | | " in(snapshot)=" CPL_FRMT_GUIB |
1577 | | " out(snapshot)=" CPL_FRMT_GUIB " out=" CPL_FRMT_GUIB |
1578 | | " offset=" CPL_FRMT_GUIB, |
1579 | | i, snapshots[i].posInBaseHandle, snapshots[i].in, |
1580 | | snapshots[i].out, out, offset); |
1581 | | #endif |
1582 | 0 | offset = out + offset - snapshots[i].out; |
1583 | 0 | if (m_poBaseHandle->Seek(snapshots[i].posInBaseHandle, SEEK_SET) != |
1584 | 0 | 0) |
1585 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
1586 | |
|
1587 | 0 | inflateBack9End(&stream); |
1588 | 0 | if (inflateBack9Copy(&stream, &snapshots[i].stream) != Z_OK) |
1589 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1590 | 0 | "inflateBack9Copy() failed"); |
1591 | 0 | crc = snapshots[i].crc; |
1592 | 0 | in = snapshots[i].in; |
1593 | 0 | out = snapshots[i].out; |
1594 | 0 | extraOutput = snapshots[i].extraOutput; |
1595 | 0 | m_bStreamEndReached = snapshots[i].m_bStreamEndReached; |
1596 | 0 | break; |
1597 | 0 | } |
1598 | 0 | } |
1599 | | |
1600 | | // Offset is now the number of bytes to skip. |
1601 | |
|
1602 | 0 | if (offset != 0 && outbuf == nullptr) |
1603 | 0 | { |
1604 | 0 | outbuf = static_cast<Byte *>(ALLOC(Z_BUFSIZE)); |
1605 | 0 | if (outbuf == nullptr) |
1606 | 0 | { |
1607 | 0 | CPL_VSIL_GZ_RETURN(FALSE); |
1608 | 0 | return false; |
1609 | 0 | } |
1610 | 0 | } |
1611 | | |
1612 | 0 | if (original_nWhence == SEEK_END && z_err == Z_STREAM_END) |
1613 | 0 | { |
1614 | 0 | return true; |
1615 | 0 | } |
1616 | | |
1617 | 0 | while (offset > 0) |
1618 | 0 | { |
1619 | 0 | int size = Z_BUFSIZE; |
1620 | 0 | if (offset < static_cast<vsi_l_offset>(Z_BUFSIZE)) |
1621 | 0 | size = static_cast<int>(offset); |
1622 | |
|
1623 | 0 | int read_size = |
1624 | 0 | static_cast<int>(Read(outbuf, 1, static_cast<uInt>(size))); |
1625 | 0 | if (original_nWhence == SEEK_END) |
1626 | 0 | { |
1627 | 0 | if (size != read_size) |
1628 | 0 | { |
1629 | 0 | z_err = Z_STREAM_END; |
1630 | 0 | break; |
1631 | 0 | } |
1632 | 0 | } |
1633 | 0 | else if (read_size == 0) |
1634 | 0 | { |
1635 | | // CPL_VSIL_GZ_RETURN(FALSE); |
1636 | 0 | return false; |
1637 | 0 | } |
1638 | 0 | offset -= read_size; |
1639 | 0 | } |
1640 | | #ifdef ENABLE_DEBUG |
1641 | | CPLDebug("GZIP", "gzseek at offset " CPL_FRMT_GUIB, out); |
1642 | | #endif |
1643 | | |
1644 | 0 | if (original_offset == 0 && original_nWhence == SEEK_END) |
1645 | 0 | { |
1646 | 0 | m_uncompressed_size = out; |
1647 | 0 | } |
1648 | |
|
1649 | 0 | return true; |
1650 | 0 | } |
1651 | | |
1652 | | /************************************************************************/ |
1653 | | /* Tell() */ |
1654 | | /************************************************************************/ |
1655 | | |
1656 | | vsi_l_offset VSIDeflate64Handle::Tell() |
1657 | 0 | { |
1658 | | #ifdef ENABLE_DEBUG |
1659 | | CPLDebug("GZIP", "Tell() = " CPL_FRMT_GUIB, out); |
1660 | | #endif |
1661 | 0 | return out; |
1662 | 0 | } |
1663 | | |
1664 | | /************************************************************************/ |
1665 | | /* Read() */ |
1666 | | /************************************************************************/ |
1667 | | |
1668 | | size_t VSIDeflate64Handle::Read(void *const buf, size_t const nSize, |
1669 | | size_t const nMemb) |
1670 | 0 | { |
1671 | | #ifdef ENABLE_DEBUG |
1672 | | CPLDebug("GZIP", "Read(%p, %d, %d)", buf, static_cast<int>(nSize), |
1673 | | static_cast<int>(nMemb)); |
1674 | | #endif |
1675 | |
|
1676 | 0 | if (m_bEOF || z_err != Z_OK) |
1677 | 0 | { |
1678 | 0 | if (z_err == Z_STREAM_END && nSize > 0 && nMemb > 0) |
1679 | 0 | m_bEOF = true; |
1680 | 0 | return 0; |
1681 | 0 | } |
1682 | | |
1683 | 0 | if (nSize > 0 && nMemb > UINT32_MAX / nSize) |
1684 | 0 | { |
1685 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Too many bytes to read at once"); |
1686 | 0 | return 0; |
1687 | 0 | } |
1688 | | |
1689 | 0 | const unsigned len = |
1690 | 0 | static_cast<unsigned int>(nSize) * static_cast<unsigned int>(nMemb); |
1691 | 0 | Bytef *pStart = |
1692 | 0 | static_cast<Bytef *>(buf); // Start off point for crc computation. |
1693 | | // == stream.next_out but not forced far (for MSDOS). |
1694 | 0 | stream.next_out = static_cast<Bytef *>(buf); |
1695 | 0 | stream.avail_out = len; |
1696 | |
|
1697 | 0 | while (stream.avail_out != 0) |
1698 | 0 | { |
1699 | 0 | if (!extraOutput.empty()) |
1700 | 0 | { |
1701 | 0 | if (extraOutput.size() >= stream.avail_out) |
1702 | 0 | { |
1703 | 0 | memcpy(stream.next_out, extraOutput.data(), stream.avail_out); |
1704 | 0 | extraOutput.erase(extraOutput.begin(), |
1705 | 0 | extraOutput.begin() + stream.avail_out); |
1706 | 0 | out += stream.avail_out; |
1707 | 0 | stream.next_out += stream.avail_out; |
1708 | 0 | stream.avail_out = 0; |
1709 | 0 | } |
1710 | 0 | else |
1711 | 0 | { |
1712 | 0 | memcpy(stream.next_out, extraOutput.data(), extraOutput.size()); |
1713 | 0 | stream.next_out += extraOutput.size(); |
1714 | 0 | out += static_cast<uInt>(extraOutput.size()); |
1715 | 0 | stream.avail_out -= static_cast<uInt>(extraOutput.size()); |
1716 | 0 | CPLAssert(stream.avail_out > 0); |
1717 | 0 | extraOutput.clear(); |
1718 | 0 | } |
1719 | 0 | z_err = Z_OK; |
1720 | 0 | } |
1721 | | |
1722 | 0 | if (stream.avail_in == 0 && !z_eof) |
1723 | 0 | { |
1724 | 0 | vsi_l_offset posInBaseHandle = m_poBaseHandle->Tell(); |
1725 | 0 | if (posInBaseHandle - startOff > m_compressed_size) |
1726 | 0 | { |
1727 | | // If we reach here, file size has changed (because at |
1728 | | // construction time startOff + m_compressed_size marked the |
1729 | | // end of file). |
1730 | | // We should probably have a better fix than that, by detecting |
1731 | | // at open time that the saved snapshot is not valid and |
1732 | | // discarding it. |
1733 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1734 | 0 | "File size of underlying /vsigzip/ file has changed"); |
1735 | 0 | z_err = Z_ERRNO; |
1736 | 0 | CPL_VSIL_GZ_RETURN(0); |
1737 | 0 | return 0; |
1738 | 0 | } |
1739 | 0 | auto snapshot = &snapshots[static_cast<size_t>( |
1740 | 0 | (posInBaseHandle - startOff) / snapshot_byte_interval)]; |
1741 | 0 | if (snapshot->posInBaseHandle == 0) |
1742 | 0 | { |
1743 | 0 | snapshot->crc = crc32( |
1744 | 0 | crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1745 | | #ifdef ENABLE_DEBUG |
1746 | | CPLDebug("SNAPSHOT", |
1747 | | "creating snapshot %d : " |
1748 | | "posInBaseHandle=" CPL_FRMT_GUIB " in=" CPL_FRMT_GUIB |
1749 | | " out=" CPL_FRMT_GUIB " crc=%X", |
1750 | | static_cast<int>((posInBaseHandle - startOff) / |
1751 | | snapshot_byte_interval), |
1752 | | posInBaseHandle, in, out, |
1753 | | static_cast<unsigned int>(snapshot->crc)); |
1754 | | #endif |
1755 | 0 | snapshot->posInBaseHandle = posInBaseHandle; |
1756 | 0 | if (inflateBack9Copy(&snapshot->stream, &stream) != Z_OK) |
1757 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1758 | 0 | "inflateBack9Copy() failed"); |
1759 | 0 | snapshot->in = in; |
1760 | 0 | snapshot->out = out; |
1761 | 0 | snapshot->extraOutput = extraOutput; |
1762 | 0 | snapshot->m_bStreamEndReached = m_bStreamEndReached; |
1763 | 0 | } |
1764 | |
|
1765 | 0 | errno = 0; |
1766 | 0 | stream.avail_in = |
1767 | 0 | static_cast<uInt>(m_poBaseHandle->Read(inbuf, 1, Z_BUFSIZE)); |
1768 | | #ifdef ENABLE_DEBUG |
1769 | | CPLDebug("GZIP", CPL_FRMT_GUIB " " CPL_FRMT_GUIB, |
1770 | | m_poBaseHandle->Tell(), offsetEndCompressedData); |
1771 | | #endif |
1772 | 0 | if (m_poBaseHandle->Tell() > offsetEndCompressedData) |
1773 | 0 | { |
1774 | | #ifdef ENABLE_DEBUG |
1775 | | CPLDebug("GZIP", "avail_in before = %d", stream.avail_in); |
1776 | | #endif |
1777 | 0 | stream.avail_in = stream.avail_in - |
1778 | 0 | static_cast<uInt>(m_poBaseHandle->Tell() - |
1779 | 0 | offsetEndCompressedData); |
1780 | 0 | if (m_poBaseHandle->Seek(offsetEndCompressedData, SEEK_SET) != |
1781 | 0 | 0) |
1782 | 0 | CPLError(CE_Failure, CPLE_FileIO, "Seek() failed"); |
1783 | | #ifdef ENABLE_DEBUG |
1784 | | CPLDebug("GZIP", "avail_in after = %d", stream.avail_in); |
1785 | | #endif |
1786 | 0 | } |
1787 | 0 | if (stream.avail_in == 0) |
1788 | 0 | { |
1789 | 0 | z_eof = 1; |
1790 | 0 | if (m_poBaseHandle->Error() || |
1791 | 0 | m_poBaseHandle->Tell() != offsetEndCompressedData) |
1792 | 0 | { |
1793 | 0 | z_err = Z_ERRNO; |
1794 | 0 | break; |
1795 | 0 | } |
1796 | 0 | } |
1797 | 0 | stream.next_in = inbuf; |
1798 | 0 | } |
1799 | | |
1800 | 0 | struct InOutCallback |
1801 | 0 | { |
1802 | 0 | vsi_l_offset *pOut = nullptr; |
1803 | 0 | std::vector<GByte> *pExtraOutput = nullptr; |
1804 | 0 | z_stream *pStream = nullptr; |
1805 | |
|
1806 | 0 | static unsigned inCbk(void FAR *, z_const unsigned char FAR *FAR *) |
1807 | 0 | { |
1808 | 0 | return 0; |
1809 | 0 | } |
1810 | |
|
1811 | 0 | static int outCbk(void FAR *user_data, unsigned char FAR *data, |
1812 | 0 | unsigned len) |
1813 | 0 | { |
1814 | 0 | auto self = static_cast<InOutCallback *>(user_data); |
1815 | 0 | if (self->pStream->avail_out >= len) |
1816 | 0 | { |
1817 | 0 | memcpy(self->pStream->next_out, data, len); |
1818 | 0 | *(self->pOut) += len; |
1819 | 0 | self->pStream->next_out += len; |
1820 | 0 | self->pStream->avail_out -= len; |
1821 | 0 | } |
1822 | 0 | else |
1823 | 0 | { |
1824 | 0 | if (self->pStream->avail_out != 0) |
1825 | 0 | { |
1826 | 0 | memcpy(self->pStream->next_out, data, |
1827 | 0 | self->pStream->avail_out); |
1828 | 0 | *(self->pOut) += self->pStream->avail_out; |
1829 | 0 | data += self->pStream->avail_out; |
1830 | 0 | len -= self->pStream->avail_out; |
1831 | 0 | self->pStream->next_out += self->pStream->avail_out; |
1832 | 0 | self->pStream->avail_out = 0; |
1833 | 0 | } |
1834 | 0 | if (len > 0) |
1835 | 0 | { |
1836 | 0 | self->pExtraOutput->insert(self->pExtraOutput->end(), |
1837 | 0 | data, data + len); |
1838 | 0 | } |
1839 | 0 | } |
1840 | 0 | return 0; |
1841 | 0 | } |
1842 | 0 | }; |
1843 | |
|
1844 | 0 | InOutCallback cbkData; |
1845 | 0 | cbkData.pOut = &out; |
1846 | 0 | cbkData.pExtraOutput = &extraOutput; |
1847 | 0 | cbkData.pStream = &stream; |
1848 | |
|
1849 | 0 | if (stream.avail_out) |
1850 | 0 | { |
1851 | 0 | if (m_bStreamEndReached) |
1852 | 0 | z_err = Z_STREAM_END; |
1853 | 0 | else |
1854 | 0 | { |
1855 | 0 | in += stream.avail_in; |
1856 | 0 | z_err = inflateBack9(&(stream), InOutCallback::inCbk, &cbkData, |
1857 | 0 | InOutCallback::outCbk, &cbkData); |
1858 | 0 | in -= stream.avail_in; |
1859 | 0 | } |
1860 | 0 | } |
1861 | 0 | if (z_err == Z_BUF_ERROR && stream.next_in == Z_NULL) |
1862 | 0 | z_err = Z_OK; |
1863 | 0 | else if (!extraOutput.empty() && z_err == Z_STREAM_END) |
1864 | 0 | { |
1865 | 0 | m_bStreamEndReached = true; |
1866 | 0 | z_err = Z_OK; |
1867 | 0 | } |
1868 | |
|
1869 | 0 | if (z_err == Z_STREAM_END /*&& m_compressed_size != 2*/) |
1870 | 0 | { |
1871 | | // Check CRC and original size. |
1872 | 0 | crc = |
1873 | 0 | crc32(crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1874 | 0 | pStart = stream.next_out; |
1875 | 0 | if (m_expected_crc) |
1876 | 0 | { |
1877 | | #ifdef ENABLE_DEBUG |
1878 | | CPLDebug("GZIP", "Computed CRC = %X. Expected CRC = %X", |
1879 | | static_cast<unsigned int>(crc), |
1880 | | static_cast<unsigned int>(m_expected_crc)); |
1881 | | #endif |
1882 | 0 | } |
1883 | 0 | if (m_expected_crc != 0 && m_expected_crc != crc) |
1884 | 0 | { |
1885 | 0 | CPLError(CE_Failure, CPLE_FileIO, |
1886 | 0 | "CRC error. Got %X instead of %X", |
1887 | 0 | static_cast<unsigned int>(crc), |
1888 | 0 | static_cast<unsigned int>(m_expected_crc)); |
1889 | 0 | z_err = Z_DATA_ERROR; |
1890 | 0 | } |
1891 | 0 | } |
1892 | 0 | if (z_err != Z_OK || z_eof) |
1893 | 0 | break; |
1894 | 0 | } |
1895 | 0 | crc = crc32(crc, pStart, static_cast<uInt>(stream.next_out - pStart)); |
1896 | |
|
1897 | 0 | size_t ret = (len - stream.avail_out) / nSize; |
1898 | 0 | if (z_err != Z_OK && z_err != Z_STREAM_END) |
1899 | 0 | { |
1900 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
1901 | 0 | "In file %s, at line %d, decompression failed with " |
1902 | 0 | "z_err = %d, return = %d", |
1903 | 0 | __FILE__, __LINE__, z_err, static_cast<int>(ret)); |
1904 | 0 | } |
1905 | 0 | else if (ret < nMemb) |
1906 | 0 | { |
1907 | 0 | m_bEOF = true; |
1908 | 0 | } |
1909 | |
|
1910 | | #ifdef ENABLE_DEBUG |
1911 | | CPLDebug("GZIP", "Read return %d (z_err=%d, z_eof=%d)", |
1912 | | static_cast<int>(ret), z_err, z_eof); |
1913 | | #endif |
1914 | 0 | return ret; |
1915 | 0 | } |
1916 | | |
1917 | | /************************************************************************/ |
1918 | | /* Write() */ |
1919 | | /************************************************************************/ |
1920 | | |
1921 | | size_t VSIDeflate64Handle::Write(const void * /* pBuffer */, size_t /* nSize */, |
1922 | | size_t /* nMemb */) |
1923 | 0 | { |
1924 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
1925 | 0 | "VSIFWriteL is not supported on GZip streams"); |
1926 | 0 | return 0; |
1927 | 0 | } |
1928 | | |
1929 | | /************************************************************************/ |
1930 | | /* Eof() */ |
1931 | | /************************************************************************/ |
1932 | | |
1933 | | int VSIDeflate64Handle::Eof() |
1934 | 0 | { |
1935 | | #ifdef ENABLE_DEBUG |
1936 | | CPLDebug("GZIP", "Eof()"); |
1937 | | #endif |
1938 | 0 | return m_bEOF; |
1939 | 0 | } |
1940 | | |
1941 | | /************************************************************************/ |
1942 | | /* Error() */ |
1943 | | /************************************************************************/ |
1944 | | |
1945 | | int VSIDeflate64Handle::Error() |
1946 | 0 | { |
1947 | | #ifdef ENABLE_DEBUG |
1948 | | CPLDebug("GZIP", "Error()"); |
1949 | | #endif |
1950 | 0 | return z_err != Z_OK && z_err != Z_STREAM_END; |
1951 | 0 | } |
1952 | | |
1953 | | /************************************************************************/ |
1954 | | /* ClearErr() */ |
1955 | | /************************************************************************/ |
1956 | | |
1957 | | void VSIDeflate64Handle::ClearErr() |
1958 | 0 | { |
1959 | 0 | m_poBaseHandle->ClearErr(); |
1960 | 0 | z_eof = 0; |
1961 | 0 | m_bEOF = false; |
1962 | 0 | z_err = Z_OK; |
1963 | 0 | } |
1964 | | |
1965 | | /************************************************************************/ |
1966 | | /* Flush() */ |
1967 | | /************************************************************************/ |
1968 | | |
1969 | | int VSIDeflate64Handle::Flush() |
1970 | 0 | { |
1971 | 0 | return 0; |
1972 | 0 | } |
1973 | | |
1974 | | /************************************************************************/ |
1975 | | /* Close() */ |
1976 | | /************************************************************************/ |
1977 | | |
1978 | | int VSIDeflate64Handle::Close() |
1979 | 0 | { |
1980 | 0 | return 0; |
1981 | 0 | } |
1982 | | #endif |
1983 | | |
1984 | | /************************************************************************/ |
1985 | | /* ==================================================================== */ |
1986 | | /* VSIGZipWriteHandleMT */ |
1987 | | /* ==================================================================== */ |
1988 | | /************************************************************************/ |
1989 | | |
1990 | | class VSIGZipWriteHandleMT final : public VSIVirtualHandle |
1991 | | { |
1992 | | CPL_DISALLOW_COPY_ASSIGN(VSIGZipWriteHandleMT) |
1993 | | |
1994 | | VSIVirtualHandle *poBaseHandle_ = nullptr; |
1995 | | vsi_l_offset nCurOffset_ = 0; |
1996 | | uLong nCRC_ = 0; |
1997 | | int nDeflateType_ = CPL_DEFLATE_TYPE_GZIP; |
1998 | | bool bAutoCloseBaseHandle_ = false; |
1999 | | int nThreads_ = 0; |
2000 | | std::unique_ptr<CPLWorkerThreadPool> poPool_{}; |
2001 | | std::list<std::string *> aposBuffers_{}; |
2002 | | std::string *pCurBuffer_ = nullptr; |
2003 | | std::mutex sMutex_{}; |
2004 | | int nSeqNumberGenerated_ = 0; |
2005 | | int nSeqNumberExpected_ = 0; |
2006 | | int nSeqNumberExpectedCRC_ = 0; |
2007 | | size_t nChunkSize_ = 0; |
2008 | | bool bHasErrored_ = false; |
2009 | | |
2010 | | struct Job |
2011 | | { |
2012 | | VSIGZipWriteHandleMT *pParent_ = nullptr; |
2013 | | std::string *pBuffer_ = nullptr; |
2014 | | int nSeqNumber_ = 0; |
2015 | | bool bFinish_ = false; |
2016 | | bool bInCRCComputation_ = false; |
2017 | | |
2018 | | std::string sCompressedData_{}; |
2019 | | uLong nCRC_ = 0; |
2020 | | }; |
2021 | | |
2022 | | std::list<Job *> apoFinishedJobs_{}; |
2023 | | std::list<Job *> apoCRCFinishedJobs_{}; |
2024 | | std::list<Job *> apoFreeJobs_{}; |
2025 | | vsi_l_offset nStartOffset_ = 0; |
2026 | | size_t nSOZIPIndexEltSize_ = 0; |
2027 | | std::vector<uint8_t> *panSOZIPIndex_ = nullptr; |
2028 | | |
2029 | | static void DeflateCompress(void *inData); |
2030 | | static void CRCCompute(void *inData); |
2031 | | bool ProcessCompletedJobs(); |
2032 | | Job *GetJobObject(); |
2033 | | #ifdef DEBUG_VERBOSE |
2034 | | void DumpState(); |
2035 | | #endif |
2036 | | |
2037 | | public: |
2038 | | VSIGZipWriteHandleMT(VSIVirtualHandle *poBaseHandle, int nDeflateType, |
2039 | | bool bAutoCloseBaseHandleIn, int nThreads, |
2040 | | size_t nChunkSize, size_t nSOZIPIndexEltSize, |
2041 | | std::vector<uint8_t> *panSOZIPIndex); |
2042 | | |
2043 | | ~VSIGZipWriteHandleMT() override; |
2044 | | |
2045 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
2046 | | vsi_l_offset Tell() override; |
2047 | | size_t Read(void *pBuffer, size_t nSize, size_t nMemb) override; |
2048 | | size_t Write(const void *pBuffer, size_t nSize, size_t nMemb) override; |
2049 | | |
2050 | | int Eof() override |
2051 | 0 | { |
2052 | 0 | return 0; |
2053 | 0 | } |
2054 | | |
2055 | | int Error() override |
2056 | 0 | { |
2057 | 0 | return 0; |
2058 | 0 | } |
2059 | | |
2060 | | void ClearErr() override |
2061 | 0 | { |
2062 | 0 | } |
2063 | | |
2064 | | int Flush() override; |
2065 | | int Close() override; |
2066 | | }; |
2067 | | |
2068 | | /************************************************************************/ |
2069 | | /* VSIGZipWriteHandleMT() */ |
2070 | | /************************************************************************/ |
2071 | | |
2072 | | VSIGZipWriteHandleMT::VSIGZipWriteHandleMT(VSIVirtualHandle *poBaseHandle, |
2073 | | int nDeflateType, |
2074 | | bool bAutoCloseBaseHandleIn, |
2075 | | int nThreads, size_t nChunkSize, |
2076 | | size_t nSOZIPIndexEltSize, |
2077 | | std::vector<uint8_t> *panSOZIPIndex) |
2078 | 0 | : poBaseHandle_(poBaseHandle), nDeflateType_(nDeflateType), |
2079 | 0 | bAutoCloseBaseHandle_(bAutoCloseBaseHandleIn), nThreads_(nThreads), |
2080 | 0 | nChunkSize_(nChunkSize), nSOZIPIndexEltSize_(nSOZIPIndexEltSize), |
2081 | 0 | panSOZIPIndex_(panSOZIPIndex) |
2082 | 0 | { |
2083 | 0 | if (nChunkSize_ == 0) |
2084 | 0 | { |
2085 | 0 | const char *pszChunkSize = |
2086 | 0 | CPLGetConfigOption("CPL_VSIL_DEFLATE_CHUNK_SIZE", "1024K"); |
2087 | 0 | nChunkSize_ = static_cast<size_t>(atoi(pszChunkSize)); |
2088 | 0 | if (strchr(pszChunkSize, 'K')) |
2089 | 0 | nChunkSize_ *= 1024; |
2090 | 0 | else if (strchr(pszChunkSize, 'M')) |
2091 | 0 | nChunkSize_ *= 1024 * 1024; |
2092 | 0 | nChunkSize_ = |
2093 | 0 | std::max(static_cast<size_t>(4 * 1024), |
2094 | 0 | std::min(static_cast<size_t>(UINT_MAX), nChunkSize_)); |
2095 | 0 | } |
2096 | |
|
2097 | 0 | for (int i = 0; i < 1 + nThreads_; i++) |
2098 | 0 | aposBuffers_.emplace_back(new std::string()); |
2099 | |
|
2100 | 0 | nStartOffset_ = poBaseHandle_->Tell(); |
2101 | 0 | if (nDeflateType == CPL_DEFLATE_TYPE_GZIP) |
2102 | 0 | { |
2103 | 0 | char header[11] = {}; |
2104 | | |
2105 | | // Write a very simple .gz header: |
2106 | 0 | snprintf(header, sizeof(header), "%c%c%c%c%c%c%c%c%c%c", gz_magic[0], |
2107 | 0 | gz_magic[1], Z_DEFLATED, 0 /*flags*/, 0, 0, 0, 0 /*time*/, |
2108 | 0 | 0 /*xflags*/, 0x03); |
2109 | 0 | poBaseHandle_->Write(header, 1, 10); |
2110 | 0 | } |
2111 | 0 | } |
2112 | | |
2113 | | /************************************************************************/ |
2114 | | /* ~VSIGZipWriteHandleMT() */ |
2115 | | /************************************************************************/ |
2116 | | |
2117 | | VSIGZipWriteHandleMT::~VSIGZipWriteHandleMT() |
2118 | | |
2119 | 0 | { |
2120 | 0 | VSIGZipWriteHandleMT::Close(); |
2121 | 0 | for (auto &psJob : apoFinishedJobs_) |
2122 | 0 | { |
2123 | 0 | delete psJob->pBuffer_; |
2124 | 0 | delete psJob; |
2125 | 0 | } |
2126 | 0 | for (auto &psJob : apoCRCFinishedJobs_) |
2127 | 0 | { |
2128 | 0 | delete psJob->pBuffer_; |
2129 | 0 | delete psJob; |
2130 | 0 | } |
2131 | 0 | for (auto &psJob : apoFreeJobs_) |
2132 | 0 | { |
2133 | 0 | delete psJob->pBuffer_; |
2134 | 0 | delete psJob; |
2135 | 0 | } |
2136 | 0 | for (auto &pstr : aposBuffers_) |
2137 | 0 | { |
2138 | 0 | delete pstr; |
2139 | 0 | } |
2140 | 0 | delete pCurBuffer_; |
2141 | 0 | } |
2142 | | |
2143 | | /************************************************************************/ |
2144 | | /* Close() */ |
2145 | | /************************************************************************/ |
2146 | | |
2147 | | int VSIGZipWriteHandleMT::Close() |
2148 | | |
2149 | 0 | { |
2150 | 0 | if (!poBaseHandle_) |
2151 | 0 | return 0; |
2152 | | |
2153 | 0 | int nRet = 0; |
2154 | |
|
2155 | 0 | if (!pCurBuffer_) |
2156 | 0 | pCurBuffer_ = new std::string(); |
2157 | |
|
2158 | 0 | { |
2159 | 0 | auto psJob = GetJobObject(); |
2160 | 0 | psJob->bFinish_ = true; |
2161 | 0 | psJob->pParent_ = this; |
2162 | 0 | psJob->pBuffer_ = pCurBuffer_; |
2163 | 0 | pCurBuffer_ = nullptr; |
2164 | 0 | psJob->nSeqNumber_ = nSeqNumberGenerated_; |
2165 | 0 | VSIGZipWriteHandleMT::DeflateCompress(psJob); |
2166 | 0 | } |
2167 | |
|
2168 | 0 | if (poPool_) |
2169 | 0 | { |
2170 | 0 | poPool_->WaitCompletion(0); |
2171 | 0 | } |
2172 | 0 | if (!ProcessCompletedJobs()) |
2173 | 0 | { |
2174 | 0 | nRet = -1; |
2175 | 0 | } |
2176 | 0 | else |
2177 | 0 | { |
2178 | 0 | CPLAssert(apoFinishedJobs_.empty()); |
2179 | 0 | if (nDeflateType_ == CPL_DEFLATE_TYPE_GZIP) |
2180 | 0 | { |
2181 | 0 | if (poPool_) |
2182 | 0 | { |
2183 | 0 | poPool_->WaitCompletion(0); |
2184 | 0 | } |
2185 | 0 | ProcessCompletedJobs(); |
2186 | 0 | } |
2187 | 0 | CPLAssert(apoCRCFinishedJobs_.empty()); |
2188 | 0 | } |
2189 | | |
2190 | 0 | if (nDeflateType_ == CPL_DEFLATE_TYPE_GZIP) |
2191 | 0 | { |
2192 | 0 | const GUInt32 anTrailer[2] = { |
2193 | 0 | CPL_LSBWORD32(static_cast<GUInt32>(nCRC_)), |
2194 | 0 | CPL_LSBWORD32(static_cast<GUInt32>(nCurOffset_))}; |
2195 | |
|
2196 | 0 | if (poBaseHandle_->Write(anTrailer, 1, 8) < 8) |
2197 | 0 | { |
2198 | 0 | nRet = -1; |
2199 | 0 | } |
2200 | 0 | } |
2201 | |
|
2202 | 0 | if (bAutoCloseBaseHandle_) |
2203 | 0 | { |
2204 | 0 | int nRetClose = poBaseHandle_->Close(); |
2205 | 0 | if (nRet == 0) |
2206 | 0 | nRet = nRetClose; |
2207 | |
|
2208 | 0 | delete poBaseHandle_; |
2209 | 0 | } |
2210 | 0 | poBaseHandle_ = nullptr; |
2211 | |
|
2212 | 0 | return nRet; |
2213 | 0 | } |
2214 | | |
2215 | | /************************************************************************/ |
2216 | | /* Read() */ |
2217 | | /************************************************************************/ |
2218 | | |
2219 | | size_t VSIGZipWriteHandleMT::Read(void * /* pBuffer */, size_t /* nSize */, |
2220 | | size_t /* nMemb */) |
2221 | 0 | { |
2222 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
2223 | 0 | "VSIFReadL is not supported on GZip write streams"); |
2224 | 0 | return 0; |
2225 | 0 | } |
2226 | | |
2227 | | /************************************************************************/ |
2228 | | /* DeflateCompress() */ |
2229 | | /************************************************************************/ |
2230 | | |
2231 | | void VSIGZipWriteHandleMT::DeflateCompress(void *inData) |
2232 | 0 | { |
2233 | 0 | Job *psJob = static_cast<Job *>(inData); |
2234 | |
|
2235 | 0 | CPLAssert(psJob->pBuffer_); |
2236 | | |
2237 | 0 | z_stream sStream; |
2238 | 0 | memset(&sStream, 0, sizeof(sStream)); |
2239 | 0 | sStream.zalloc = nullptr; |
2240 | 0 | sStream.zfree = nullptr; |
2241 | 0 | sStream.opaque = nullptr; |
2242 | |
|
2243 | 0 | sStream.avail_in = static_cast<uInt>(psJob->pBuffer_->size()); |
2244 | 0 | sStream.next_in = reinterpret_cast<Bytef *>(&(*psJob->pBuffer_)[0]); |
2245 | |
|
2246 | 0 | int ret = deflateInit2( |
2247 | 0 | &sStream, Z_DEFAULT_COMPRESSION, Z_DEFLATED, |
2248 | 0 | (psJob->pParent_->nDeflateType_ == CPL_DEFLATE_TYPE_ZLIB) ? MAX_WBITS |
2249 | 0 | : -MAX_WBITS, |
2250 | 0 | 8, Z_DEFAULT_STRATEGY); |
2251 | 0 | CPLAssertAlwaysEval(ret == Z_OK); |
2252 | | |
2253 | 0 | size_t nRealSize = 0; |
2254 | |
|
2255 | 0 | while (sStream.avail_in > 0) |
2256 | 0 | { |
2257 | 0 | psJob->sCompressedData_.resize(nRealSize + Z_BUFSIZE); |
2258 | 0 | sStream.avail_out = static_cast<uInt>(Z_BUFSIZE); |
2259 | 0 | sStream.next_out = |
2260 | 0 | reinterpret_cast<Bytef *>(&psJob->sCompressedData_[0]) + nRealSize; |
2261 | |
|
2262 | 0 | const int zlibRet = deflate(&sStream, Z_NO_FLUSH); |
2263 | 0 | CPLAssertAlwaysEval(zlibRet == Z_OK); |
2264 | | |
2265 | 0 | nRealSize += static_cast<uInt>(Z_BUFSIZE) - sStream.avail_out; |
2266 | 0 | } |
2267 | | |
2268 | 0 | psJob->sCompressedData_.resize(nRealSize + Z_BUFSIZE); |
2269 | 0 | sStream.avail_out = static_cast<uInt>(Z_BUFSIZE); |
2270 | 0 | sStream.next_out = |
2271 | 0 | reinterpret_cast<Bytef *>(&psJob->sCompressedData_[0]) + nRealSize; |
2272 | |
|
2273 | 0 | if (psJob->bFinish_) |
2274 | 0 | { |
2275 | 0 | const int zlibRet = deflate(&sStream, Z_FINISH); |
2276 | 0 | CPLAssertAlwaysEval(zlibRet == Z_STREAM_END); |
2277 | 0 | } |
2278 | 0 | else |
2279 | 0 | { |
2280 | | // Do a Z_SYNC_FLUSH and Z_FULL_FLUSH, so as to have two markers when |
2281 | | // independent as pigz 2.3.4 or later. The following 9 byte sequence |
2282 | | // will be found: 0x00 0x00 0xff 0xff 0x00 0x00 0x00 0xff 0xff |
2283 | | // Z_FULL_FLUSH only is sufficient, but it is not obvious if a |
2284 | | // 0x00 0x00 0xff 0xff marker in the codestream is just a SYNC_FLUSH ( |
2285 | | // without dictionary reset) or a FULL_FLUSH (with dictionary reset) |
2286 | 0 | { |
2287 | 0 | const int zlibRet = deflate(&sStream, Z_SYNC_FLUSH); |
2288 | 0 | CPLAssertAlwaysEval(zlibRet == Z_OK); |
2289 | 0 | } |
2290 | | |
2291 | 0 | { |
2292 | 0 | const int zlibRet = deflate(&sStream, Z_FULL_FLUSH); |
2293 | 0 | CPLAssertAlwaysEval(zlibRet == Z_OK); |
2294 | 0 | } |
2295 | 0 | } |
2296 | | |
2297 | 0 | nRealSize += static_cast<uInt>(Z_BUFSIZE) - sStream.avail_out; |
2298 | 0 | psJob->sCompressedData_.resize(nRealSize); |
2299 | |
|
2300 | 0 | deflateEnd(&sStream); |
2301 | |
|
2302 | 0 | { |
2303 | 0 | std::lock_guard<std::mutex> oLock(psJob->pParent_->sMutex_); |
2304 | 0 | psJob->pParent_->apoFinishedJobs_.push_back(psJob); |
2305 | 0 | } |
2306 | 0 | } |
2307 | | |
2308 | | /************************************************************************/ |
2309 | | /* CRCCompute() */ |
2310 | | /************************************************************************/ |
2311 | | |
2312 | | void VSIGZipWriteHandleMT::CRCCompute(void *inData) |
2313 | 0 | { |
2314 | 0 | Job *psJob = static_cast<Job *>(inData); |
2315 | 0 | psJob->bInCRCComputation_ = true; |
2316 | 0 | psJob->nCRC_ = |
2317 | 0 | crc32(0U, reinterpret_cast<const Bytef *>(psJob->pBuffer_->data()), |
2318 | 0 | static_cast<uInt>(psJob->pBuffer_->size())); |
2319 | |
|
2320 | 0 | { |
2321 | 0 | std::lock_guard<std::mutex> oLock(psJob->pParent_->sMutex_); |
2322 | 0 | psJob->pParent_->apoCRCFinishedJobs_.push_back(psJob); |
2323 | 0 | } |
2324 | 0 | } |
2325 | | |
2326 | | /************************************************************************/ |
2327 | | /* DumpState() */ |
2328 | | /************************************************************************/ |
2329 | | |
2330 | | #ifdef DEBUG_VERBOSE |
2331 | | void VSIGZipWriteHandleMT::DumpState() |
2332 | | { |
2333 | | fprintf(stderr, "Finished jobs (expected = %d):\n", // ok |
2334 | | nSeqNumberExpected_); |
2335 | | for (const auto *psJob : apoFinishedJobs_) |
2336 | | { |
2337 | | fprintf(stderr, "seq number=%d, bInCRCComputation = %d\n", // ok |
2338 | | psJob->nSeqNumber_, psJob->bInCRCComputation_ ? 1 : 0); |
2339 | | } |
2340 | | fprintf(stderr, "Finished CRC jobs (expected = %d):\n", // ok |
2341 | | nSeqNumberExpectedCRC_); |
2342 | | for (const auto *psJob : apoFinishedJobs_) |
2343 | | { |
2344 | | fprintf(stderr, "seq number=%d\n", // ok |
2345 | | psJob->nSeqNumber_); |
2346 | | } |
2347 | | fprintf(stderr, "apoFreeJobs_.size() = %d\n", // ok |
2348 | | static_cast<int>(apoFreeJobs_.size())); |
2349 | | fprintf(stderr, "aposBuffers_.size() = %d\n", // ok |
2350 | | static_cast<int>(aposBuffers_.size())); |
2351 | | } |
2352 | | #endif |
2353 | | |
2354 | | /************************************************************************/ |
2355 | | /* ProcessCompletedJobs() */ |
2356 | | /************************************************************************/ |
2357 | | |
2358 | | bool VSIGZipWriteHandleMT::ProcessCompletedJobs() |
2359 | 0 | { |
2360 | 0 | std::lock_guard<std::mutex> oLock(sMutex_); |
2361 | 0 | bool do_it_again = true; |
2362 | 0 | while (do_it_again) |
2363 | 0 | { |
2364 | 0 | do_it_again = false; |
2365 | 0 | if (nDeflateType_ == CPL_DEFLATE_TYPE_GZIP) |
2366 | 0 | { |
2367 | 0 | for (auto iter = apoFinishedJobs_.begin(); |
2368 | 0 | iter != apoFinishedJobs_.end(); ++iter) |
2369 | 0 | { |
2370 | 0 | auto psJob = *iter; |
2371 | |
|
2372 | 0 | if (!psJob->bInCRCComputation_) |
2373 | 0 | { |
2374 | 0 | psJob->bInCRCComputation_ = true; |
2375 | 0 | sMutex_.unlock(); |
2376 | 0 | if (poPool_) |
2377 | 0 | { |
2378 | 0 | poPool_->SubmitJob(VSIGZipWriteHandleMT::CRCCompute, |
2379 | 0 | psJob); |
2380 | 0 | } |
2381 | 0 | else |
2382 | 0 | { |
2383 | 0 | CRCCompute(psJob); |
2384 | 0 | } |
2385 | 0 | sMutex_.lock(); |
2386 | 0 | } |
2387 | 0 | } |
2388 | 0 | } |
2389 | |
|
2390 | 0 | for (auto iter = apoFinishedJobs_.begin(); |
2391 | 0 | iter != apoFinishedJobs_.end(); ++iter) |
2392 | 0 | { |
2393 | 0 | auto psJob = *iter; |
2394 | 0 | if (psJob->nSeqNumber_ == nSeqNumberExpected_) |
2395 | 0 | { |
2396 | 0 | apoFinishedJobs_.erase(iter); |
2397 | |
|
2398 | 0 | const bool bIsSeqNumberExpectedZero = |
2399 | 0 | (nSeqNumberExpected_ == 0); |
2400 | 0 | sMutex_.unlock(); |
2401 | |
|
2402 | 0 | const size_t nToWrite = psJob->sCompressedData_.size(); |
2403 | 0 | if (panSOZIPIndex_ && !bIsSeqNumberExpectedZero && |
2404 | 0 | !psJob->pBuffer_->empty()) |
2405 | 0 | { |
2406 | 0 | uint64_t nOffset = poBaseHandle_->Tell() - nStartOffset_; |
2407 | 0 | if (nSOZIPIndexEltSize_ == 8) |
2408 | 0 | { |
2409 | 0 | CPL_LSBPTR64(&nOffset); |
2410 | 0 | std::copy(reinterpret_cast<const uint8_t *>(&nOffset), |
2411 | 0 | reinterpret_cast<const uint8_t *>(&nOffset) + |
2412 | 0 | sizeof(nOffset), |
2413 | 0 | std::back_inserter(*panSOZIPIndex_)); |
2414 | 0 | } |
2415 | 0 | else |
2416 | 0 | { |
2417 | 0 | if (nOffset > std::numeric_limits<uint32_t>::max()) |
2418 | 0 | { |
2419 | | // shouldn't happen normally... |
2420 | 0 | CPLError( |
2421 | 0 | CE_Failure, CPLE_AppDefined, |
2422 | 0 | "Too big offset for SOZIP_OFFSET_SIZE = 4"); |
2423 | 0 | panSOZIPIndex_->clear(); |
2424 | 0 | panSOZIPIndex_ = nullptr; |
2425 | 0 | } |
2426 | 0 | else |
2427 | 0 | { |
2428 | 0 | uint32_t nOffset32 = static_cast<uint32_t>(nOffset); |
2429 | 0 | CPL_LSBPTR32(&nOffset32); |
2430 | 0 | std::copy( |
2431 | 0 | reinterpret_cast<const uint8_t *>(&nOffset32), |
2432 | 0 | reinterpret_cast<const uint8_t *>(&nOffset32) + |
2433 | 0 | sizeof(nOffset32), |
2434 | 0 | std::back_inserter(*panSOZIPIndex_)); |
2435 | 0 | } |
2436 | 0 | } |
2437 | 0 | } |
2438 | 0 | bool bError = |
2439 | 0 | poBaseHandle_->Write(psJob->sCompressedData_.data(), 1, |
2440 | 0 | nToWrite) < nToWrite; |
2441 | 0 | sMutex_.lock(); |
2442 | 0 | nSeqNumberExpected_++; |
2443 | |
|
2444 | 0 | if (nDeflateType_ != CPL_DEFLATE_TYPE_GZIP) |
2445 | 0 | { |
2446 | 0 | aposBuffers_.push_back(psJob->pBuffer_); |
2447 | 0 | psJob->pBuffer_ = nullptr; |
2448 | |
|
2449 | 0 | apoFreeJobs_.push_back(psJob); |
2450 | 0 | } |
2451 | |
|
2452 | 0 | if (bError) |
2453 | 0 | { |
2454 | 0 | return false; |
2455 | 0 | } |
2456 | | |
2457 | 0 | do_it_again = true; |
2458 | 0 | break; |
2459 | 0 | } |
2460 | 0 | } |
2461 | | |
2462 | 0 | if (nDeflateType_ == CPL_DEFLATE_TYPE_GZIP) |
2463 | 0 | { |
2464 | 0 | for (auto iter = apoCRCFinishedJobs_.begin(); |
2465 | 0 | iter != apoCRCFinishedJobs_.end(); ++iter) |
2466 | 0 | { |
2467 | 0 | auto psJob = *iter; |
2468 | 0 | if (psJob->nSeqNumber_ == nSeqNumberExpectedCRC_) |
2469 | 0 | { |
2470 | 0 | apoCRCFinishedJobs_.erase(iter); |
2471 | |
|
2472 | 0 | nCRC_ = crc32_combine( |
2473 | 0 | nCRC_, psJob->nCRC_, |
2474 | 0 | static_cast<uLong>(psJob->pBuffer_->size())); |
2475 | |
|
2476 | 0 | nSeqNumberExpectedCRC_++; |
2477 | |
|
2478 | 0 | aposBuffers_.push_back(psJob->pBuffer_); |
2479 | 0 | psJob->pBuffer_ = nullptr; |
2480 | |
|
2481 | 0 | apoFreeJobs_.push_back(psJob); |
2482 | 0 | do_it_again = true; |
2483 | 0 | break; |
2484 | 0 | } |
2485 | 0 | } |
2486 | 0 | } |
2487 | 0 | } |
2488 | 0 | return true; |
2489 | 0 | } |
2490 | | |
2491 | | /************************************************************************/ |
2492 | | /* GetJobObject() */ |
2493 | | /************************************************************************/ |
2494 | | |
2495 | | VSIGZipWriteHandleMT::Job *VSIGZipWriteHandleMT::GetJobObject() |
2496 | 0 | { |
2497 | 0 | { |
2498 | 0 | std::lock_guard<std::mutex> oLock(sMutex_); |
2499 | 0 | if (!apoFreeJobs_.empty()) |
2500 | 0 | { |
2501 | 0 | auto job = apoFreeJobs_.back(); |
2502 | 0 | apoFreeJobs_.pop_back(); |
2503 | 0 | job->sCompressedData_.clear(); |
2504 | 0 | job->bInCRCComputation_ = false; |
2505 | 0 | return job; |
2506 | 0 | } |
2507 | 0 | } |
2508 | 0 | return new Job(); |
2509 | 0 | } |
2510 | | |
2511 | | /************************************************************************/ |
2512 | | /* Write() */ |
2513 | | /************************************************************************/ |
2514 | | |
2515 | | size_t VSIGZipWriteHandleMT::Write(const void *const pBuffer, |
2516 | | size_t const nSize, size_t const nMemb) |
2517 | | |
2518 | 0 | { |
2519 | 0 | if (bHasErrored_) |
2520 | 0 | return 0; |
2521 | | |
2522 | 0 | const char *pszBuffer = static_cast<const char *>(pBuffer); |
2523 | 0 | size_t nBytesToWrite = nSize * nMemb; |
2524 | 0 | while (nBytesToWrite > 0) |
2525 | 0 | { |
2526 | 0 | if (pCurBuffer_ == nullptr) |
2527 | 0 | { |
2528 | 0 | while (true) |
2529 | 0 | { |
2530 | | // We store in a local variable instead of pCurBuffer_ directly |
2531 | | // to avoid Coverity Scan to be confused by the fact that we |
2532 | | // have used above pCurBuffer_ outside of the mutex. But what |
2533 | | // is protected by the mutex is aposBuffers_, not pCurBuffer_. |
2534 | 0 | std::string *l_pCurBuffer = nullptr; |
2535 | 0 | { |
2536 | 0 | std::lock_guard<std::mutex> oLock(sMutex_); |
2537 | 0 | if (!aposBuffers_.empty()) |
2538 | 0 | { |
2539 | 0 | l_pCurBuffer = aposBuffers_.back(); |
2540 | 0 | aposBuffers_.pop_back(); |
2541 | 0 | } |
2542 | 0 | } |
2543 | 0 | pCurBuffer_ = l_pCurBuffer; |
2544 | 0 | if (pCurBuffer_) |
2545 | 0 | break; |
2546 | | |
2547 | 0 | if (poPool_) |
2548 | 0 | { |
2549 | 0 | poPool_->WaitEvent(); |
2550 | 0 | } |
2551 | 0 | if (!ProcessCompletedJobs()) |
2552 | 0 | { |
2553 | 0 | bHasErrored_ = true; |
2554 | 0 | return 0; |
2555 | 0 | } |
2556 | 0 | } |
2557 | 0 | pCurBuffer_->clear(); |
2558 | 0 | } |
2559 | 0 | size_t nConsumed = |
2560 | 0 | std::min(nBytesToWrite, nChunkSize_ - pCurBuffer_->size()); |
2561 | 0 | pCurBuffer_->append(pszBuffer, nConsumed); |
2562 | 0 | nCurOffset_ += nConsumed; |
2563 | 0 | pszBuffer += nConsumed; |
2564 | 0 | nBytesToWrite -= nConsumed; |
2565 | 0 | if (pCurBuffer_->size() == nChunkSize_) |
2566 | 0 | { |
2567 | 0 | if (poPool_ == nullptr) |
2568 | 0 | { |
2569 | 0 | poPool_.reset(new CPLWorkerThreadPool()); |
2570 | 0 | if (!poPool_->Setup(nThreads_, nullptr, nullptr, false)) |
2571 | 0 | { |
2572 | 0 | bHasErrored_ = true; |
2573 | 0 | poPool_.reset(); |
2574 | 0 | return 0; |
2575 | 0 | } |
2576 | 0 | } |
2577 | | |
2578 | 0 | auto psJob = GetJobObject(); |
2579 | 0 | psJob->pParent_ = this; |
2580 | 0 | psJob->pBuffer_ = pCurBuffer_; |
2581 | 0 | psJob->nSeqNumber_ = nSeqNumberGenerated_; |
2582 | 0 | nSeqNumberGenerated_++; |
2583 | 0 | pCurBuffer_ = nullptr; |
2584 | 0 | poPool_->SubmitJob(VSIGZipWriteHandleMT::DeflateCompress, psJob); |
2585 | 0 | } |
2586 | 0 | } |
2587 | | |
2588 | 0 | return nMemb; |
2589 | 0 | } |
2590 | | |
2591 | | /************************************************************************/ |
2592 | | /* Flush() */ |
2593 | | /************************************************************************/ |
2594 | | |
2595 | | int VSIGZipWriteHandleMT::Flush() |
2596 | | |
2597 | 0 | { |
2598 | | // we *could* do something for this but for now we choose not to. |
2599 | |
|
2600 | 0 | return 0; |
2601 | 0 | } |
2602 | | |
2603 | | /************************************************************************/ |
2604 | | /* Seek() */ |
2605 | | /************************************************************************/ |
2606 | | |
2607 | | int VSIGZipWriteHandleMT::Seek(vsi_l_offset nOffset, int nWhence) |
2608 | | |
2609 | 0 | { |
2610 | 0 | if (nOffset == 0 && (nWhence == SEEK_END || nWhence == SEEK_CUR)) |
2611 | 0 | return 0; |
2612 | 0 | else if (nWhence == SEEK_SET && nOffset == nCurOffset_) |
2613 | 0 | return 0; |
2614 | 0 | else |
2615 | 0 | { |
2616 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
2617 | 0 | "Seeking on writable compressed data streams not supported."); |
2618 | |
|
2619 | 0 | return -1; |
2620 | 0 | } |
2621 | 0 | } |
2622 | | |
2623 | | /************************************************************************/ |
2624 | | /* Tell() */ |
2625 | | /************************************************************************/ |
2626 | | |
2627 | | vsi_l_offset VSIGZipWriteHandleMT::Tell() |
2628 | | |
2629 | 0 | { |
2630 | 0 | return nCurOffset_; |
2631 | 0 | } |
2632 | | |
2633 | | /************************************************************************/ |
2634 | | /* ==================================================================== */ |
2635 | | /* VSIGZipWriteHandle */ |
2636 | | /* ==================================================================== */ |
2637 | | /************************************************************************/ |
2638 | | |
2639 | | class VSIGZipWriteHandle final : public VSIVirtualHandle |
2640 | | { |
2641 | | CPL_DISALLOW_COPY_ASSIGN(VSIGZipWriteHandle) |
2642 | | |
2643 | | VSIVirtualHandle *m_poBaseHandle = nullptr; |
2644 | | z_stream sStream; |
2645 | | Byte *pabyInBuf = nullptr; |
2646 | | Byte *pabyOutBuf = nullptr; |
2647 | | bool bCompressActive = false; |
2648 | | vsi_l_offset nCurOffset = 0; |
2649 | | uLong nCRC = 0; |
2650 | | int nDeflateType = CPL_DEFLATE_TYPE_GZIP; |
2651 | | bool bAutoCloseBaseHandle = false; |
2652 | | |
2653 | | public: |
2654 | | VSIGZipWriteHandle(VSIVirtualHandle *poBaseHandle, int nDeflateType, |
2655 | | bool bAutoCloseBaseHandleIn); |
2656 | | |
2657 | | ~VSIGZipWriteHandle() override; |
2658 | | |
2659 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
2660 | | vsi_l_offset Tell() override; |
2661 | | size_t Read(void *pBuffer, size_t nSize, size_t nMemb) override; |
2662 | | size_t Write(const void *pBuffer, size_t nSize, size_t nMemb) override; |
2663 | | |
2664 | | int Eof() override |
2665 | 0 | { |
2666 | 0 | return 0; |
2667 | 0 | } |
2668 | | |
2669 | | int Error() override |
2670 | 0 | { |
2671 | 0 | return 0; |
2672 | 0 | } |
2673 | | |
2674 | | void ClearErr() override |
2675 | 0 | { |
2676 | 0 | } |
2677 | | |
2678 | | int Flush() override; |
2679 | | int Close() override; |
2680 | | }; |
2681 | | |
2682 | | /************************************************************************/ |
2683 | | /* VSIGZipWriteHandle() */ |
2684 | | /************************************************************************/ |
2685 | | |
2686 | | VSIGZipWriteHandle::VSIGZipWriteHandle(VSIVirtualHandle *poBaseHandle, |
2687 | | int nDeflateTypeIn, |
2688 | | bool bAutoCloseBaseHandleIn) |
2689 | 0 | : m_poBaseHandle(poBaseHandle), sStream(), |
2690 | 0 | pabyInBuf(static_cast<Byte *>(CPLMalloc(Z_BUFSIZE))), |
2691 | 0 | pabyOutBuf(static_cast<Byte *>(CPLMalloc(Z_BUFSIZE))), |
2692 | 0 | nCRC(crc32(0L, nullptr, 0)), nDeflateType(nDeflateTypeIn), |
2693 | 0 | bAutoCloseBaseHandle(bAutoCloseBaseHandleIn) |
2694 | 0 | { |
2695 | 0 | sStream.zalloc = nullptr; |
2696 | 0 | sStream.zfree = nullptr; |
2697 | 0 | sStream.opaque = nullptr; |
2698 | 0 | sStream.next_in = nullptr; |
2699 | 0 | sStream.next_out = nullptr; |
2700 | 0 | sStream.avail_in = sStream.avail_out = 0; |
2701 | |
|
2702 | 0 | sStream.next_in = pabyInBuf; |
2703 | |
|
2704 | 0 | if (deflateInit2(&sStream, Z_DEFAULT_COMPRESSION, Z_DEFLATED, |
2705 | 0 | (nDeflateType == CPL_DEFLATE_TYPE_ZLIB) ? MAX_WBITS |
2706 | 0 | : -MAX_WBITS, |
2707 | 0 | 8, Z_DEFAULT_STRATEGY) != Z_OK) |
2708 | 0 | { |
2709 | 0 | bCompressActive = false; |
2710 | 0 | } |
2711 | 0 | else |
2712 | 0 | { |
2713 | 0 | if (nDeflateType == CPL_DEFLATE_TYPE_GZIP) |
2714 | 0 | { |
2715 | 0 | char header[11] = {}; |
2716 | | |
2717 | | // Write a very simple .gz header: |
2718 | 0 | snprintf(header, sizeof(header), "%c%c%c%c%c%c%c%c%c%c", |
2719 | 0 | gz_magic[0], gz_magic[1], Z_DEFLATED, 0 /*flags*/, 0, 0, 0, |
2720 | 0 | 0 /*time*/, 0 /*xflags*/, 0x03); |
2721 | 0 | m_poBaseHandle->Write(header, 1, 10); |
2722 | 0 | } |
2723 | |
|
2724 | 0 | bCompressActive = true; |
2725 | 0 | } |
2726 | 0 | } |
2727 | | |
2728 | | /************************************************************************/ |
2729 | | /* VSICreateGZipWritable() */ |
2730 | | /************************************************************************/ |
2731 | | |
2732 | | VSIVirtualHandle *VSICreateGZipWritable(VSIVirtualHandle *poBaseHandle, |
2733 | | int nDeflateTypeIn, |
2734 | | int bAutoCloseBaseHandle) |
2735 | 0 | { |
2736 | 0 | return VSICreateGZipWritable(poBaseHandle, nDeflateTypeIn, |
2737 | 0 | CPL_TO_BOOL(bAutoCloseBaseHandle), 0, 0, 0, |
2738 | 0 | nullptr); |
2739 | 0 | } |
2740 | | |
2741 | | VSIVirtualHandle *VSICreateGZipWritable(VSIVirtualHandle *poBaseHandle, |
2742 | | int nDeflateTypeIn, |
2743 | | bool bAutoCloseBaseHandle, int nThreads, |
2744 | | size_t nChunkSize, |
2745 | | size_t nSOZIPIndexEltSize, |
2746 | | std::vector<uint8_t> *panSOZIPIndex) |
2747 | 0 | { |
2748 | 0 | const char *pszThreads = CPLGetConfigOption("GDAL_NUM_THREADS", nullptr); |
2749 | 0 | if (pszThreads || nThreads > 0 || nChunkSize > 0) |
2750 | 0 | { |
2751 | 0 | if (nThreads == 0) |
2752 | 0 | { |
2753 | 0 | if (!pszThreads || EQUAL(pszThreads, "ALL_CPUS")) |
2754 | 0 | nThreads = CPLGetNumCPUs(); |
2755 | 0 | else |
2756 | 0 | nThreads = atoi(pszThreads); |
2757 | 0 | nThreads = std::max(1, std::min(128, nThreads)); |
2758 | 0 | } |
2759 | 0 | if (nThreads > 1 || nChunkSize > 0) |
2760 | 0 | { |
2761 | | // coverity[tainted_data] |
2762 | 0 | return new VSIGZipWriteHandleMT( |
2763 | 0 | poBaseHandle, nDeflateTypeIn, bAutoCloseBaseHandle, nThreads, |
2764 | 0 | nChunkSize, nSOZIPIndexEltSize, panSOZIPIndex); |
2765 | 0 | } |
2766 | 0 | } |
2767 | 0 | return new VSIGZipWriteHandle(poBaseHandle, nDeflateTypeIn, |
2768 | 0 | bAutoCloseBaseHandle); |
2769 | 0 | } |
2770 | | |
2771 | | /************************************************************************/ |
2772 | | /* ~VSIGZipWriteHandle() */ |
2773 | | /************************************************************************/ |
2774 | | |
2775 | | VSIGZipWriteHandle::~VSIGZipWriteHandle() |
2776 | | |
2777 | 0 | { |
2778 | 0 | if (bCompressActive) |
2779 | 0 | VSIGZipWriteHandle::Close(); |
2780 | |
|
2781 | 0 | CPLFree(pabyInBuf); |
2782 | 0 | CPLFree(pabyOutBuf); |
2783 | 0 | } |
2784 | | |
2785 | | /************************************************************************/ |
2786 | | /* Close() */ |
2787 | | /************************************************************************/ |
2788 | | |
2789 | | int VSIGZipWriteHandle::Close() |
2790 | | |
2791 | 0 | { |
2792 | 0 | int nRet = 0; |
2793 | 0 | if (bCompressActive) |
2794 | 0 | { |
2795 | 0 | sStream.next_out = pabyOutBuf; |
2796 | 0 | sStream.avail_out = static_cast<uInt>(Z_BUFSIZE); |
2797 | |
|
2798 | 0 | const int zlibRet = deflate(&sStream, Z_FINISH); |
2799 | 0 | CPLAssertAlwaysEval(zlibRet == Z_STREAM_END); |
2800 | | |
2801 | 0 | const size_t nOutBytes = |
2802 | 0 | static_cast<uInt>(Z_BUFSIZE) - sStream.avail_out; |
2803 | |
|
2804 | 0 | deflateEnd(&sStream); |
2805 | |
|
2806 | 0 | if (m_poBaseHandle->Write(pabyOutBuf, 1, nOutBytes) < nOutBytes) |
2807 | 0 | { |
2808 | 0 | nRet = -1; |
2809 | 0 | } |
2810 | |
|
2811 | 0 | if (nRet == 0 && nDeflateType == CPL_DEFLATE_TYPE_GZIP) |
2812 | 0 | { |
2813 | 0 | const GUInt32 anTrailer[2] = { |
2814 | 0 | CPL_LSBWORD32(static_cast<GUInt32>(nCRC)), |
2815 | 0 | CPL_LSBWORD32(static_cast<GUInt32>(nCurOffset))}; |
2816 | |
|
2817 | 0 | if (m_poBaseHandle->Write(anTrailer, 1, 8) < 8) |
2818 | 0 | { |
2819 | 0 | nRet = -1; |
2820 | 0 | } |
2821 | 0 | } |
2822 | |
|
2823 | 0 | if (bAutoCloseBaseHandle) |
2824 | 0 | { |
2825 | 0 | if (nRet == 0) |
2826 | 0 | nRet = m_poBaseHandle->Close(); |
2827 | |
|
2828 | 0 | delete m_poBaseHandle; |
2829 | 0 | } |
2830 | |
|
2831 | 0 | bCompressActive = false; |
2832 | 0 | } |
2833 | | |
2834 | 0 | return nRet; |
2835 | 0 | } |
2836 | | |
2837 | | /************************************************************************/ |
2838 | | /* Read() */ |
2839 | | /************************************************************************/ |
2840 | | |
2841 | | size_t VSIGZipWriteHandle::Read(void * /* pBuffer */, size_t /* nSize */, |
2842 | | size_t /* nMemb */) |
2843 | 0 | { |
2844 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
2845 | 0 | "VSIFReadL is not supported on GZip write streams"); |
2846 | 0 | return 0; |
2847 | 0 | } |
2848 | | |
2849 | | /************************************************************************/ |
2850 | | /* Write() */ |
2851 | | /************************************************************************/ |
2852 | | |
2853 | | size_t VSIGZipWriteHandle::Write(const void *const pBuffer, size_t const nSize, |
2854 | | size_t const nMemb) |
2855 | | |
2856 | 0 | { |
2857 | 0 | size_t nBytesToWrite = nSize * nMemb; |
2858 | |
|
2859 | 0 | { |
2860 | 0 | size_t nOffset = 0; |
2861 | 0 | while (nOffset < nBytesToWrite) |
2862 | 0 | { |
2863 | 0 | uInt nChunk = static_cast<uInt>(std::min( |
2864 | 0 | static_cast<size_t>(UINT_MAX), nBytesToWrite - nOffset)); |
2865 | 0 | nCRC = |
2866 | 0 | crc32(nCRC, reinterpret_cast<const Bytef *>(pBuffer) + nOffset, |
2867 | 0 | nChunk); |
2868 | 0 | nOffset += nChunk; |
2869 | 0 | } |
2870 | 0 | } |
2871 | |
|
2872 | 0 | if (!bCompressActive) |
2873 | 0 | return 0; |
2874 | | |
2875 | 0 | size_t nNextByte = 0; |
2876 | 0 | while (nNextByte < nBytesToWrite) |
2877 | 0 | { |
2878 | 0 | sStream.next_out = pabyOutBuf; |
2879 | 0 | sStream.avail_out = static_cast<uInt>(Z_BUFSIZE); |
2880 | |
|
2881 | 0 | if (sStream.avail_in > 0) |
2882 | 0 | memmove(pabyInBuf, sStream.next_in, sStream.avail_in); |
2883 | |
|
2884 | 0 | const uInt nNewBytesToWrite = static_cast<uInt>( |
2885 | 0 | std::min(static_cast<size_t>(Z_BUFSIZE - sStream.avail_in), |
2886 | 0 | nBytesToWrite - nNextByte)); |
2887 | 0 | memcpy(pabyInBuf + sStream.avail_in, |
2888 | 0 | reinterpret_cast<const Byte *>(pBuffer) + nNextByte, |
2889 | 0 | nNewBytesToWrite); |
2890 | |
|
2891 | 0 | sStream.next_in = pabyInBuf; |
2892 | 0 | sStream.avail_in += nNewBytesToWrite; |
2893 | |
|
2894 | 0 | const int zlibRet = deflate(&sStream, Z_NO_FLUSH); |
2895 | 0 | CPLAssertAlwaysEval(zlibRet == Z_OK); |
2896 | | |
2897 | 0 | const size_t nOutBytes = |
2898 | 0 | static_cast<uInt>(Z_BUFSIZE) - sStream.avail_out; |
2899 | |
|
2900 | 0 | if (nOutBytes > 0) |
2901 | 0 | { |
2902 | 0 | if (m_poBaseHandle->Write(pabyOutBuf, 1, nOutBytes) < nOutBytes) |
2903 | 0 | return 0; |
2904 | 0 | } |
2905 | | |
2906 | 0 | nNextByte += nNewBytesToWrite; |
2907 | 0 | nCurOffset += nNewBytesToWrite; |
2908 | 0 | } |
2909 | | |
2910 | 0 | return nMemb; |
2911 | 0 | } |
2912 | | |
2913 | | /************************************************************************/ |
2914 | | /* Flush() */ |
2915 | | /************************************************************************/ |
2916 | | |
2917 | | int VSIGZipWriteHandle::Flush() |
2918 | | |
2919 | 0 | { |
2920 | | // we *could* do something for this but for now we choose not to. |
2921 | |
|
2922 | 0 | return 0; |
2923 | 0 | } |
2924 | | |
2925 | | /************************************************************************/ |
2926 | | /* Seek() */ |
2927 | | /************************************************************************/ |
2928 | | |
2929 | | int VSIGZipWriteHandle::Seek(vsi_l_offset nOffset, int nWhence) |
2930 | | |
2931 | 0 | { |
2932 | 0 | if (nOffset == 0 && (nWhence == SEEK_END || nWhence == SEEK_CUR)) |
2933 | 0 | return 0; |
2934 | 0 | else if (nWhence == SEEK_SET && nOffset == nCurOffset) |
2935 | 0 | return 0; |
2936 | 0 | else |
2937 | 0 | { |
2938 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
2939 | 0 | "Seeking on writable compressed data streams not supported."); |
2940 | |
|
2941 | 0 | return -1; |
2942 | 0 | } |
2943 | 0 | } |
2944 | | |
2945 | | /************************************************************************/ |
2946 | | /* Tell() */ |
2947 | | /************************************************************************/ |
2948 | | |
2949 | | vsi_l_offset VSIGZipWriteHandle::Tell() |
2950 | | |
2951 | 0 | { |
2952 | 0 | return nCurOffset; |
2953 | 0 | } |
2954 | | |
2955 | | /************************************************************************/ |
2956 | | /* ==================================================================== */ |
2957 | | /* VSIGZipFilesystemHandler */ |
2958 | | /* ==================================================================== */ |
2959 | | /************************************************************************/ |
2960 | | |
2961 | | /************************************************************************/ |
2962 | | /* ~VSIGZipFilesystemHandler() */ |
2963 | | /************************************************************************/ |
2964 | | |
2965 | | VSIGZipFilesystemHandler::~VSIGZipFilesystemHandler() |
2966 | 0 | { |
2967 | 0 | if (poHandleLastGZipFile) |
2968 | 0 | { |
2969 | 0 | poHandleLastGZipFile->UnsetCanSaveInfo(); |
2970 | 0 | poHandleLastGZipFile.reset(); |
2971 | 0 | } |
2972 | 0 | } |
2973 | | |
2974 | | /************************************************************************/ |
2975 | | /* SaveInfo() */ |
2976 | | /************************************************************************/ |
2977 | | |
2978 | | void VSIGZipFilesystemHandler::SaveInfo(VSIGZipHandle *poHandle) |
2979 | 0 | { |
2980 | 0 | std::unique_lock oLock(oMutex); |
2981 | 0 | SaveInfo_unlocked(poHandle); |
2982 | 0 | } |
2983 | | |
2984 | | void VSIGZipFilesystemHandler::SaveInfo_unlocked(VSIGZipHandle *poHandle) |
2985 | 0 | { |
2986 | 0 | if (m_bInSaveInfo) |
2987 | 0 | return; |
2988 | 0 | m_bInSaveInfo = true; |
2989 | |
|
2990 | 0 | CPLAssert(poHandle != poHandleLastGZipFile.get()); |
2991 | 0 | CPLAssert(poHandle->GetBaseFileName() != nullptr); |
2992 | | |
2993 | 0 | if (poHandleLastGZipFile == nullptr || |
2994 | 0 | strcmp(poHandleLastGZipFile->GetBaseFileName(), |
2995 | 0 | poHandle->GetBaseFileName()) != 0 || |
2996 | 0 | poHandle->GetLastReadOffset() > |
2997 | 0 | poHandleLastGZipFile->GetLastReadOffset()) |
2998 | 0 | { |
2999 | 0 | std::unique_ptr<VSIGZipHandle> poTmp; |
3000 | 0 | std::swap(poTmp, poHandleLastGZipFile); |
3001 | 0 | if (poTmp) |
3002 | 0 | { |
3003 | 0 | poTmp->UnsetCanSaveInfo(); |
3004 | 0 | poTmp.reset(); |
3005 | 0 | } |
3006 | 0 | poHandleLastGZipFile.reset(poHandle->Duplicate()); |
3007 | 0 | if (poHandleLastGZipFile) |
3008 | 0 | poHandleLastGZipFile->CloseBaseHandle(); |
3009 | 0 | } |
3010 | 0 | m_bInSaveInfo = false; |
3011 | 0 | } |
3012 | | |
3013 | | /************************************************************************/ |
3014 | | /* Open() */ |
3015 | | /************************************************************************/ |
3016 | | |
3017 | | VSIVirtualHandleUniquePtr |
3018 | | VSIGZipFilesystemHandler::Open(const char *pszFilename, const char *pszAccess, |
3019 | | bool /* bSetError */, |
3020 | | CSLConstList /* papszOptions */) |
3021 | 1.51k | { |
3022 | 1.51k | if (!STARTS_WITH_CI(pszFilename, "/vsigzip/")) |
3023 | 1 | return nullptr; |
3024 | | |
3025 | 1.51k | VSIFilesystemHandler *poFSHandler = |
3026 | 1.51k | VSIFileManager::GetHandler(pszFilename + strlen("/vsigzip/")); |
3027 | | |
3028 | | /* -------------------------------------------------------------------- */ |
3029 | | /* Is this an attempt to write a new file without update (w+) */ |
3030 | | /* access? If so, create a writable handle for the underlying */ |
3031 | | /* filename. */ |
3032 | | /* -------------------------------------------------------------------- */ |
3033 | 1.51k | if (strchr(pszAccess, 'w') != nullptr) |
3034 | 0 | { |
3035 | 0 | if (strchr(pszAccess, '+') != nullptr) |
3036 | 0 | { |
3037 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
3038 | 0 | "Write+update (w+) not supported for /vsigzip, " |
3039 | 0 | "only read-only or write-only."); |
3040 | 0 | return nullptr; |
3041 | 0 | } |
3042 | | |
3043 | 0 | auto poVirtualHandle = |
3044 | 0 | poFSHandler->Open(pszFilename + strlen("/vsigzip/"), "wb"); |
3045 | |
|
3046 | 0 | if (poVirtualHandle == nullptr) |
3047 | 0 | return nullptr; |
3048 | | |
3049 | 0 | return VSIVirtualHandleUniquePtr( |
3050 | 0 | VSICreateGZipWritable(poVirtualHandle.release(), |
3051 | 0 | strchr(pszAccess, 'z') != nullptr, TRUE)); |
3052 | 0 | } |
3053 | | |
3054 | | /* -------------------------------------------------------------------- */ |
3055 | | /* Otherwise we are in the read access case. */ |
3056 | | /* -------------------------------------------------------------------- */ |
3057 | | |
3058 | 1.51k | VSIGZipHandle *poGZIPHandle = OpenGZipReadOnly(pszFilename, pszAccess); |
3059 | 1.51k | if (poGZIPHandle) |
3060 | | // Wrap the VSIGZipHandle inside a buffered reader that will |
3061 | | // improve dramatically performance when doing small backward |
3062 | | // seeks. |
3063 | 0 | return VSIVirtualHandleUniquePtr( |
3064 | 0 | VSICreateBufferedReaderHandle(poGZIPHandle)); |
3065 | | |
3066 | 1.51k | return nullptr; |
3067 | 1.51k | } |
3068 | | |
3069 | | /************************************************************************/ |
3070 | | /* SupportsSequentialWrite() */ |
3071 | | /************************************************************************/ |
3072 | | |
3073 | | bool VSIGZipFilesystemHandler::SupportsSequentialWrite(const char *pszPath, |
3074 | | bool bAllowLocalTempFile) |
3075 | 0 | { |
3076 | 0 | if (!STARTS_WITH_CI(pszPath, "/vsigzip/")) |
3077 | 0 | return false; |
3078 | 0 | const char *pszBaseFileName = pszPath + strlen("/vsigzip/"); |
3079 | 0 | VSIFilesystemHandler *poFSHandler = |
3080 | 0 | VSIFileManager::GetHandler(pszBaseFileName); |
3081 | 0 | return poFSHandler->SupportsSequentialWrite(pszPath, bAllowLocalTempFile); |
3082 | 0 | } |
3083 | | |
3084 | | /************************************************************************/ |
3085 | | /* OpenGZipReadOnly() */ |
3086 | | /************************************************************************/ |
3087 | | |
3088 | | VSIGZipHandle * |
3089 | | VSIGZipFilesystemHandler::OpenGZipReadOnly(const char *pszFilename, |
3090 | | const char *pszAccess) |
3091 | 1.74k | { |
3092 | 1.74k | VSIFilesystemHandler *poFSHandler = |
3093 | 1.74k | VSIFileManager::GetHandler(pszFilename + strlen("/vsigzip/")); |
3094 | | |
3095 | 1.74k | std::unique_lock oLock(oMutex); |
3096 | | |
3097 | | #ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION |
3098 | | // Disable caching in fuzzing mode as the /vsigzip/ file is likely to |
3099 | | // change very often |
3100 | | // TODO: filename-based logic isn't enough. We should probably check |
3101 | | // timestamp and/or file size. |
3102 | | if (poHandleLastGZipFile != nullptr && |
3103 | | strcmp(pszFilename + strlen("/vsigzip/"), |
3104 | | poHandleLastGZipFile->GetBaseFileName()) == 0 && |
3105 | | EQUAL(pszAccess, "rb")) |
3106 | | { |
3107 | | VSIGZipHandle *poHandle = poHandleLastGZipFile->Duplicate(); |
3108 | | if (poHandle) |
3109 | | return poHandle; |
3110 | | } |
3111 | | #else |
3112 | 1.74k | CPL_IGNORE_RET_VAL(pszAccess); |
3113 | 1.74k | #endif |
3114 | | |
3115 | 1.74k | VSIVirtualHandleUniquePtr poVirtualHandle( |
3116 | 1.74k | poFSHandler->Open(pszFilename + strlen("/vsigzip/"), "rb")); |
3117 | | |
3118 | 1.74k | if (poVirtualHandle == nullptr) |
3119 | 1.09k | return nullptr; |
3120 | | |
3121 | 645 | unsigned char signature[2] = {'\0', '\0'}; |
3122 | 645 | if (poVirtualHandle->Read(signature, 1, 2) != 2 || |
3123 | 0 | signature[0] != gz_magic[0] || signature[1] != gz_magic[1]) |
3124 | 645 | { |
3125 | 645 | return nullptr; |
3126 | 645 | } |
3127 | | |
3128 | 0 | if (poHandleLastGZipFile) |
3129 | 0 | { |
3130 | 0 | poHandleLastGZipFile->UnsetCanSaveInfo(); |
3131 | 0 | poHandleLastGZipFile.reset(); |
3132 | 0 | } |
3133 | |
|
3134 | 0 | auto poHandle = std::make_unique<VSIGZipHandle>( |
3135 | 0 | std::move(poVirtualHandle), pszFilename + strlen("/vsigzip/")); |
3136 | 0 | if (!(poHandle->IsInitOK())) |
3137 | 0 | { |
3138 | 0 | return nullptr; |
3139 | 0 | } |
3140 | 0 | return poHandle.release(); |
3141 | 0 | } |
3142 | | |
3143 | | /************************************************************************/ |
3144 | | /* Stat() */ |
3145 | | /************************************************************************/ |
3146 | | |
3147 | | int VSIGZipFilesystemHandler::Stat(const char *pszFilename, |
3148 | | VSIStatBufL *pStatBuf, int nFlags) |
3149 | 4.23k | { |
3150 | 4.23k | if (!STARTS_WITH_CI(pszFilename, "/vsigzip/")) |
3151 | 105 | return -1; |
3152 | | |
3153 | 4.13k | std::unique_lock oLock(oMutex); |
3154 | | |
3155 | 4.13k | memset(pStatBuf, 0, sizeof(VSIStatBufL)); |
3156 | | |
3157 | 4.13k | if (poHandleLastGZipFile != nullptr && |
3158 | 0 | strcmp(pszFilename + strlen("/vsigzip/"), |
3159 | 0 | poHandleLastGZipFile->GetBaseFileName()) == 0) |
3160 | 0 | { |
3161 | 0 | if (poHandleLastGZipFile->GetUncompressedSize() != 0) |
3162 | 0 | { |
3163 | 0 | pStatBuf->st_mode = S_IFREG; |
3164 | 0 | pStatBuf->st_size = poHandleLastGZipFile->GetUncompressedSize(); |
3165 | 0 | return 0; |
3166 | 0 | } |
3167 | 0 | } |
3168 | | |
3169 | | // Begin by doing a stat on the real file. |
3170 | 4.13k | int ret = VSIStatExL(pszFilename + strlen("/vsigzip/"), pStatBuf, nFlags); |
3171 | | |
3172 | 4.13k | if (ret == 0 && (nFlags & VSI_STAT_SIZE_FLAG)) |
3173 | 222 | { |
3174 | 222 | CPLString osCacheFilename(pszFilename + strlen("/vsigzip/")); |
3175 | 222 | osCacheFilename += ".properties"; |
3176 | | |
3177 | | // Can we save a bit of seeking by using a .properties file? |
3178 | 222 | VSILFILE *fpCacheLength = VSIFOpenL(osCacheFilename.c_str(), "rb"); |
3179 | 222 | if (fpCacheLength) |
3180 | 113 | { |
3181 | 113 | const char *pszLine; |
3182 | 113 | GUIntBig nCompressedSize = 0; |
3183 | 113 | GUIntBig nUncompressedSize = 0; |
3184 | 113 | while ((pszLine = CPLReadLineL(fpCacheLength)) != nullptr) |
3185 | 0 | { |
3186 | 0 | if (STARTS_WITH_CI(pszLine, "compressed_size=")) |
3187 | 0 | { |
3188 | 0 | const char *pszBuffer = |
3189 | 0 | pszLine + strlen("compressed_size="); |
3190 | 0 | nCompressedSize = CPLScanUIntBig( |
3191 | 0 | pszBuffer, static_cast<int>(strlen(pszBuffer))); |
3192 | 0 | } |
3193 | 0 | else if (STARTS_WITH_CI(pszLine, "uncompressed_size=")) |
3194 | 0 | { |
3195 | 0 | const char *pszBuffer = |
3196 | 0 | pszLine + strlen("uncompressed_size="); |
3197 | 0 | nUncompressedSize = CPLScanUIntBig( |
3198 | 0 | pszBuffer, static_cast<int>(strlen(pszBuffer))); |
3199 | 0 | } |
3200 | 0 | } |
3201 | | |
3202 | 113 | CPL_IGNORE_RET_VAL(VSIFCloseL(fpCacheLength)); |
3203 | | |
3204 | 113 | if (nCompressedSize == static_cast<GUIntBig>(pStatBuf->st_size)) |
3205 | 113 | { |
3206 | | // Patch with the uncompressed size. |
3207 | 113 | pStatBuf->st_size = nUncompressedSize; |
3208 | | |
3209 | 113 | VSIGZipHandle *poHandle = |
3210 | 113 | VSIGZipFilesystemHandler::OpenGZipReadOnly(pszFilename, |
3211 | 113 | "rb"); |
3212 | 113 | if (poHandle) |
3213 | 0 | { |
3214 | 0 | poHandle->SetUncompressedSize(nUncompressedSize); |
3215 | 0 | SaveInfo_unlocked(poHandle); |
3216 | 0 | delete poHandle; |
3217 | 0 | } |
3218 | | |
3219 | 113 | return ret; |
3220 | 113 | } |
3221 | 113 | } |
3222 | | |
3223 | | // No, then seek at the end of the data (slow). |
3224 | 109 | VSIGZipHandle *poHandle = |
3225 | 109 | VSIGZipFilesystemHandler::OpenGZipReadOnly(pszFilename, "rb"); |
3226 | 109 | if (poHandle) |
3227 | 0 | { |
3228 | 0 | poHandle->Seek(0, SEEK_END); |
3229 | 0 | const GUIntBig uncompressed_size = |
3230 | 0 | static_cast<GUIntBig>(poHandle->Tell()); |
3231 | 0 | poHandle->Seek(0, SEEK_SET); |
3232 | | |
3233 | | // Patch with the uncompressed size. |
3234 | 0 | pStatBuf->st_size = uncompressed_size; |
3235 | |
|
3236 | 0 | delete poHandle; |
3237 | 0 | } |
3238 | 109 | else |
3239 | 109 | { |
3240 | 109 | ret = -1; |
3241 | 109 | } |
3242 | 109 | } |
3243 | | |
3244 | 4.02k | return ret; |
3245 | 4.13k | } |
3246 | | |
3247 | | /************************************************************************/ |
3248 | | /* ReadDirEx() */ |
3249 | | /************************************************************************/ |
3250 | | |
3251 | | char **VSIGZipFilesystemHandler::ReadDirEx(const char * /*pszDirname*/, |
3252 | | int /* nMaxFiles */) |
3253 | 0 | { |
3254 | 0 | return nullptr; |
3255 | 0 | } |
3256 | | |
3257 | | /************************************************************************/ |
3258 | | /* GetOptions() */ |
3259 | | /************************************************************************/ |
3260 | | |
3261 | | const char *VSIGZipFilesystemHandler::GetOptions() |
3262 | 0 | { |
3263 | 0 | return "<Options>" |
3264 | 0 | " <Option name='GDAL_NUM_THREADS' type='string' " |
3265 | 0 | "description='Number of threads for compression. Either a integer " |
3266 | 0 | "or ALL_CPUS'/>" |
3267 | 0 | " <Option name='CPL_VSIL_DEFLATE_CHUNK_SIZE' type='string' " |
3268 | 0 | "description='Chunk of uncompressed data for parallelization. " |
3269 | 0 | "Use K(ilobytes) or M(egabytes) suffix' default='1M'/>" |
3270 | 0 | "</Options>"; |
3271 | 0 | } |
3272 | | |
3273 | | //! @endcond |
3274 | | /************************************************************************/ |
3275 | | /* VSIInstallGZipFileHandler() */ |
3276 | | /************************************************************************/ |
3277 | | |
3278 | | /*! |
3279 | | \brief Install GZip file system handler. |
3280 | | |
3281 | | A special file handler is installed that allows reading on-the-fly and |
3282 | | writing in GZip (.gz) files. |
3283 | | |
3284 | | All portions of the file system underneath the base |
3285 | | path "/vsigzip/" will be handled by this driver. |
3286 | | |
3287 | | \verbatim embed:rst |
3288 | | See :ref:`/vsigzip/ documentation <vsigzip>` |
3289 | | \endverbatim |
3290 | | |
3291 | | */ |
3292 | | |
3293 | | void VSIInstallGZipFileHandler() |
3294 | 3 | { |
3295 | 3 | VSIFileManager::InstallHandler("/vsigzip/", new VSIGZipFilesystemHandler); |
3296 | 3 | } |
3297 | | |
3298 | | //! @cond Doxygen_Suppress |
3299 | | |
3300 | | /************************************************************************/ |
3301 | | /* ==================================================================== */ |
3302 | | /* VSIZipEntryFileOffset */ |
3303 | | /* ==================================================================== */ |
3304 | | /************************************************************************/ |
3305 | | |
3306 | | class VSIZipEntryFileOffset final : public VSIArchiveEntryFileOffset |
3307 | | { |
3308 | | public: |
3309 | | unz_file_pos m_file_pos; |
3310 | | |
3311 | 0 | explicit VSIZipEntryFileOffset(unz_file_pos file_pos) : m_file_pos() |
3312 | 0 | { |
3313 | 0 | m_file_pos.pos_in_zip_directory = file_pos.pos_in_zip_directory; |
3314 | 0 | m_file_pos.num_of_file = file_pos.num_of_file; |
3315 | 0 | } |
3316 | | |
3317 | | ~VSIZipEntryFileOffset() override; |
3318 | | }; |
3319 | | |
3320 | 0 | VSIZipEntryFileOffset::~VSIZipEntryFileOffset() = default; |
3321 | | |
3322 | | /************************************************************************/ |
3323 | | /* ==================================================================== */ |
3324 | | /* VSIZipReader */ |
3325 | | /* ==================================================================== */ |
3326 | | /************************************************************************/ |
3327 | | |
3328 | | class VSIZipReader final : public VSIArchiveReader |
3329 | | { |
3330 | | CPL_DISALLOW_COPY_ASSIGN(VSIZipReader) |
3331 | | |
3332 | | private: |
3333 | | unzFile unzF = nullptr; |
3334 | | unz_file_pos file_pos; |
3335 | | GUIntBig nNextFileSize = 0; |
3336 | | CPLString osNextFileName{}; |
3337 | | GIntBig nModifiedTime = 0; |
3338 | | |
3339 | | bool SetInfo(); |
3340 | | |
3341 | | public: |
3342 | | explicit VSIZipReader(const char *pszZipFileName); |
3343 | | ~VSIZipReader() override; |
3344 | | |
3345 | | int IsValid() |
3346 | 1.33k | { |
3347 | 1.33k | return unzF != nullptr; |
3348 | 1.33k | } |
3349 | | |
3350 | | unzFile GetUnzFileHandle() |
3351 | 0 | { |
3352 | 0 | return unzF; |
3353 | 0 | } |
3354 | | |
3355 | | int GotoFirstFile() override; |
3356 | | int GotoNextFile() override; |
3357 | | |
3358 | | VSIArchiveEntryFileOffset *GetFileOffset() override |
3359 | 0 | { |
3360 | 0 | return new VSIZipEntryFileOffset(file_pos); |
3361 | 0 | } |
3362 | | |
3363 | | GUIntBig GetFileSize() override |
3364 | 0 | { |
3365 | 0 | return nNextFileSize; |
3366 | 0 | } |
3367 | | |
3368 | | CPLString GetFileName() override |
3369 | 0 | { |
3370 | 0 | return osNextFileName; |
3371 | 0 | } |
3372 | | |
3373 | | GIntBig GetModifiedTime() override |
3374 | 0 | { |
3375 | 0 | return nModifiedTime; |
3376 | 0 | } |
3377 | | |
3378 | | int GotoFileOffset(VSIArchiveEntryFileOffset *pOffset) override; |
3379 | | }; |
3380 | | |
3381 | | /************************************************************************/ |
3382 | | /* VSIZipReader() */ |
3383 | | /************************************************************************/ |
3384 | | |
3385 | | VSIZipReader::VSIZipReader(const char *pszZipFileName) |
3386 | 1.33k | : unzF(cpl_unzOpen(pszZipFileName)), file_pos() |
3387 | 1.33k | { |
3388 | 1.33k | file_pos.pos_in_zip_directory = 0; |
3389 | 1.33k | file_pos.num_of_file = 0; |
3390 | 1.33k | } |
3391 | | |
3392 | | /************************************************************************/ |
3393 | | /* ~VSIZipReader() */ |
3394 | | /************************************************************************/ |
3395 | | |
3396 | | VSIZipReader::~VSIZipReader() |
3397 | 1.33k | { |
3398 | 1.33k | if (unzF) |
3399 | 0 | cpl_unzClose(unzF); |
3400 | 1.33k | } |
3401 | | |
3402 | | /************************************************************************/ |
3403 | | /* SetInfo() */ |
3404 | | /************************************************************************/ |
3405 | | |
3406 | | bool VSIZipReader::SetInfo() |
3407 | 0 | { |
3408 | 0 | char fileName[8193] = {}; |
3409 | 0 | unz_file_info file_info; |
3410 | 0 | if (UNZ_OK != cpl_unzGetCurrentFileInfo(unzF, &file_info, fileName, |
3411 | 0 | sizeof(fileName) - 1, nullptr, 0, |
3412 | 0 | nullptr, 0)) |
3413 | 0 | { |
3414 | 0 | CPLError(CE_Failure, CPLE_FileIO, "cpl_unzGetCurrentFileInfo failed"); |
3415 | 0 | cpl_unzGetFilePos(unzF, &file_pos); |
3416 | 0 | return false; |
3417 | 0 | } |
3418 | 0 | fileName[sizeof(fileName) - 1] = '\0'; |
3419 | 0 | osNextFileName = fileName; |
3420 | 0 | nNextFileSize = file_info.uncompressed_size; |
3421 | 0 | struct tm brokendowntime; |
3422 | 0 | brokendowntime.tm_sec = file_info.tmu_date.tm_sec; |
3423 | 0 | brokendowntime.tm_min = file_info.tmu_date.tm_min; |
3424 | 0 | brokendowntime.tm_hour = file_info.tmu_date.tm_hour; |
3425 | 0 | brokendowntime.tm_mday = file_info.tmu_date.tm_mday; |
3426 | 0 | brokendowntime.tm_mon = file_info.tmu_date.tm_mon; |
3427 | | // The minizip conventions differs from the Unix one. |
3428 | 0 | brokendowntime.tm_year = file_info.tmu_date.tm_year - 1900; |
3429 | 0 | nModifiedTime = CPLYMDHMSToUnixTime(&brokendowntime); |
3430 | |
|
3431 | 0 | cpl_unzGetFilePos(unzF, &file_pos); |
3432 | 0 | return true; |
3433 | 0 | } |
3434 | | |
3435 | | /************************************************************************/ |
3436 | | /* GotoNextFile() */ |
3437 | | /************************************************************************/ |
3438 | | |
3439 | | int VSIZipReader::GotoNextFile() |
3440 | 0 | { |
3441 | 0 | if (cpl_unzGoToNextFile(unzF) != UNZ_OK) |
3442 | 0 | return FALSE; |
3443 | | |
3444 | 0 | if (!SetInfo()) |
3445 | 0 | return FALSE; |
3446 | | |
3447 | 0 | return TRUE; |
3448 | 0 | } |
3449 | | |
3450 | | /************************************************************************/ |
3451 | | /* GotoFirstFile() */ |
3452 | | /************************************************************************/ |
3453 | | |
3454 | | int VSIZipReader::GotoFirstFile() |
3455 | 0 | { |
3456 | 0 | if (cpl_unzGoToFirstFile(unzF) != UNZ_OK) |
3457 | 0 | return FALSE; |
3458 | | |
3459 | 0 | if (!SetInfo()) |
3460 | 0 | return FALSE; |
3461 | | |
3462 | 0 | return TRUE; |
3463 | 0 | } |
3464 | | |
3465 | | /************************************************************************/ |
3466 | | /* GotoFileOffset() */ |
3467 | | /************************************************************************/ |
3468 | | |
3469 | | int VSIZipReader::GotoFileOffset(VSIArchiveEntryFileOffset *pOffset) |
3470 | 0 | { |
3471 | 0 | VSIZipEntryFileOffset *pZipEntryOffset = |
3472 | 0 | reinterpret_cast<VSIZipEntryFileOffset *>(pOffset); |
3473 | 0 | if (cpl_unzGoToFilePos(unzF, &(pZipEntryOffset->m_file_pos)) != UNZ_OK) |
3474 | 0 | { |
3475 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "GotoFileOffset failed"); |
3476 | 0 | return FALSE; |
3477 | 0 | } |
3478 | | |
3479 | 0 | if (!SetInfo()) |
3480 | 0 | return FALSE; |
3481 | | |
3482 | 0 | return TRUE; |
3483 | 0 | } |
3484 | | |
3485 | | /************************************************************************/ |
3486 | | /* ==================================================================== */ |
3487 | | /* VSIZipFilesystemHandler */ |
3488 | | /* ==================================================================== */ |
3489 | | /************************************************************************/ |
3490 | | |
3491 | | class VSIZipWriteHandle; |
3492 | | |
3493 | | class VSIZipFilesystemHandler final : public VSIArchiveFilesystemHandler |
3494 | | { |
3495 | | CPL_DISALLOW_COPY_ASSIGN(VSIZipFilesystemHandler) |
3496 | | |
3497 | | std::map<CPLString, VSIZipWriteHandle *> oMapZipWriteHandles{}; |
3498 | | VSIVirtualHandleUniquePtr OpenForWrite_unlocked(const char *pszFilename, |
3499 | | const char *pszAccess); |
3500 | | |
3501 | | struct VSIFileInZipInfo |
3502 | | { |
3503 | | VSIVirtualHandleUniquePtr poVirtualHandle{}; |
3504 | | std::map<std::string, std::string> oMapProperties{}; |
3505 | | int nCompressionMethod = 0; |
3506 | | uint64_t nUncompressedSize = 0; |
3507 | | uint64_t nCompressedSize = 0; |
3508 | | uint64_t nStartDataStream = 0; |
3509 | | uLong nCRC = 0; |
3510 | | bool bSOZipIndexFound = false; |
3511 | | bool bSOZipIndexValid = false; |
3512 | | uint32_t nSOZIPVersion = 0; |
3513 | | uint32_t nSOZIPToSkip = 0; |
3514 | | uint32_t nSOZIPChunkSize = 0; |
3515 | | uint32_t nSOZIPOffsetSize = 0; |
3516 | | uint64_t nSOZIPStartData = 0; |
3517 | | }; |
3518 | | |
3519 | | bool GetFileInfo(const char *pszFilename, VSIFileInZipInfo &info, |
3520 | | bool bSetError); |
3521 | | |
3522 | | public: |
3523 | 3 | VSIZipFilesystemHandler() = default; |
3524 | | ~VSIZipFilesystemHandler() override; |
3525 | | |
3526 | | const char *GetPrefix() const override |
3527 | 15.1k | { |
3528 | 15.1k | return "/vsizip"; |
3529 | 15.1k | } |
3530 | | |
3531 | | std::vector<CPLString> GetExtensions() const override; |
3532 | | std::unique_ptr<VSIArchiveReader> |
3533 | | CreateReader(const char *pszZipFileName) override; |
3534 | | |
3535 | | VSIVirtualHandleUniquePtr Open(const char *pszFilename, |
3536 | | const char *pszAccess, bool bSetError, |
3537 | | CSLConstList /* papszOptions */) override; |
3538 | | |
3539 | | char **GetFileMetadata(const char *pszFilename, const char *pszDomain, |
3540 | | CSLConstList papszOptions) override; |
3541 | | |
3542 | | VSIVirtualHandleUniquePtr OpenForWrite(const char *pszFilename, |
3543 | | const char *pszAccess); |
3544 | | |
3545 | | int CopyFile(const char *pszSource, const char *pszTarget, |
3546 | | VSILFILE *fpSource, vsi_l_offset nSourceSize, |
3547 | | const char *const *papszOptions, |
3548 | | GDALProgressFunc pProgressFunc, void *pProgressData) override; |
3549 | | |
3550 | | int Mkdir(const char *pszDirname, long nMode) override; |
3551 | | char **ReadDirEx(const char *pszDirname, int nMaxFiles) override; |
3552 | | int Stat(const char *pszFilename, VSIStatBufL *pStatBuf, |
3553 | | int nFlags) override; |
3554 | | |
3555 | | const char *GetOptions() override; |
3556 | | |
3557 | | void RemoveFromMap(VSIZipWriteHandle *poHandle); |
3558 | | }; |
3559 | | |
3560 | | /************************************************************************/ |
3561 | | /* ==================================================================== */ |
3562 | | /* VSIZipWriteHandle */ |
3563 | | /* ==================================================================== */ |
3564 | | /************************************************************************/ |
3565 | | |
3566 | | class VSIZipWriteHandle final : public VSIVirtualHandle |
3567 | | { |
3568 | | CPL_DISALLOW_COPY_ASSIGN(VSIZipWriteHandle) |
3569 | | |
3570 | | VSIZipFilesystemHandler *m_poFS = nullptr; |
3571 | | void *m_hZIP = nullptr; |
3572 | | VSIZipWriteHandle *poChildInWriting = nullptr; |
3573 | | VSIZipWriteHandle *m_poParent = nullptr; |
3574 | | bool bAutoDeleteParent = false; |
3575 | | vsi_l_offset nCurOffset = 0; |
3576 | | |
3577 | | public: |
3578 | | VSIZipWriteHandle(VSIZipFilesystemHandler *poFS, void *hZIP, |
3579 | | VSIZipWriteHandle *poParent); |
3580 | | |
3581 | | ~VSIZipWriteHandle() override; |
3582 | | |
3583 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
3584 | | vsi_l_offset Tell() override; |
3585 | | size_t Read(void *pBuffer, size_t nSize, size_t nMemb) override; |
3586 | | size_t Write(const void *pBuffer, size_t nSize, size_t nMemb) override; |
3587 | | |
3588 | | int Eof() override |
3589 | 0 | { |
3590 | 0 | return 0; |
3591 | 0 | } |
3592 | | |
3593 | | int Error() override |
3594 | 0 | { |
3595 | 0 | return 0; |
3596 | 0 | } |
3597 | | |
3598 | | void ClearErr() override |
3599 | 0 | { |
3600 | 0 | } |
3601 | | |
3602 | | int Flush() override; |
3603 | | int Close() override; |
3604 | | |
3605 | | void StartNewFile(VSIZipWriteHandle *poSubFile); |
3606 | | void StopCurrentFile(); |
3607 | | |
3608 | | void *GetHandle() |
3609 | 0 | { |
3610 | 0 | return m_hZIP; |
3611 | 0 | } |
3612 | | |
3613 | | VSIZipWriteHandle *GetChildInWriting() |
3614 | 0 | { |
3615 | 0 | return poChildInWriting; |
3616 | 0 | } |
3617 | | |
3618 | | void SetAutoDeleteParent() |
3619 | 0 | { |
3620 | 0 | bAutoDeleteParent = true; |
3621 | 0 | } |
3622 | | }; |
3623 | | |
3624 | | /************************************************************************/ |
3625 | | /* ~VSIZipFilesystemHandler() */ |
3626 | | /************************************************************************/ |
3627 | | |
3628 | | VSIZipFilesystemHandler::~VSIZipFilesystemHandler() |
3629 | 0 | { |
3630 | 0 | for (std::map<CPLString, VSIZipWriteHandle *>::const_iterator iter = |
3631 | 0 | oMapZipWriteHandles.begin(); |
3632 | 0 | iter != oMapZipWriteHandles.end(); ++iter) |
3633 | 0 | { |
3634 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "%s has not been closed", |
3635 | 0 | iter->first.c_str()); |
3636 | 0 | } |
3637 | 0 | } |
3638 | | |
3639 | | /************************************************************************/ |
3640 | | /* GetExtensions() */ |
3641 | | /************************************************************************/ |
3642 | | |
3643 | | std::vector<CPLString> VSIZipFilesystemHandler::GetExtensions() const |
3644 | 1.12k | { |
3645 | 1.12k | std::vector<CPLString> oList; |
3646 | 1.12k | oList.push_back(".zip"); |
3647 | 1.12k | oList.push_back(".kmz"); |
3648 | 1.12k | oList.push_back(".dwf"); |
3649 | 1.12k | oList.push_back(".ods"); |
3650 | 1.12k | oList.push_back(".xlsx"); |
3651 | 1.12k | oList.push_back(".xlsm"); |
3652 | | |
3653 | | // Add to zip FS handler extensions array additional extensions |
3654 | | // listed in CPL_VSIL_ZIP_ALLOWED_EXTENSIONS config option. |
3655 | | // The extensions are divided by commas. |
3656 | 1.12k | const char *pszAllowedExtensions = |
3657 | 1.12k | CPLGetConfigOption("CPL_VSIL_ZIP_ALLOWED_EXTENSIONS", nullptr); |
3658 | 1.12k | if (pszAllowedExtensions) |
3659 | 0 | { |
3660 | 0 | char **papszExtensions = |
3661 | 0 | CSLTokenizeString2(pszAllowedExtensions, ", ", 0); |
3662 | 0 | for (int i = 0; papszExtensions[i] != nullptr; i++) |
3663 | 0 | { |
3664 | 0 | oList.push_back(papszExtensions[i]); |
3665 | 0 | } |
3666 | 0 | CSLDestroy(papszExtensions); |
3667 | 0 | } |
3668 | | |
3669 | 1.12k | return oList; |
3670 | 1.12k | } |
3671 | | |
3672 | | /************************************************************************/ |
3673 | | /* CreateReader() */ |
3674 | | /************************************************************************/ |
3675 | | |
3676 | | std::unique_ptr<VSIArchiveReader> |
3677 | | VSIZipFilesystemHandler::CreateReader(const char *pszZipFileName) |
3678 | 1.33k | { |
3679 | 1.33k | auto poReader = std::make_unique<VSIZipReader>(pszZipFileName); |
3680 | | |
3681 | 1.33k | if (!poReader->IsValid() || !poReader->GotoFirstFile()) |
3682 | 1.33k | { |
3683 | 1.33k | return nullptr; |
3684 | 1.33k | } |
3685 | | |
3686 | 0 | return poReader; |
3687 | 1.33k | } |
3688 | | |
3689 | | /************************************************************************/ |
3690 | | /* VSISOZipHandle */ |
3691 | | /************************************************************************/ |
3692 | | |
3693 | | class VSISOZipHandle final : public VSIVirtualHandle |
3694 | | { |
3695 | | VSIVirtualHandleUniquePtr poBaseHandle_{}; |
3696 | | vsi_l_offset nPosCompressedStream_; |
3697 | | uint64_t compressed_size_; |
3698 | | uint64_t uncompressed_size_; |
3699 | | vsi_l_offset indexPos_; |
3700 | | uint32_t nToSkip_; |
3701 | | uint32_t nChunkSize_; |
3702 | | bool bEOF_ = false; |
3703 | | bool bError_ = false; |
3704 | | vsi_l_offset nCurPos_ = 0; |
3705 | | bool bOK_ = true; |
3706 | | #ifdef HAVE_LIBDEFLATE |
3707 | | struct libdeflate_decompressor *pDecompressor_ = nullptr; |
3708 | | #else |
3709 | | z_stream sStream_{}; |
3710 | | #endif |
3711 | | |
3712 | | VSISOZipHandle(const VSISOZipHandle &) = delete; |
3713 | | VSISOZipHandle &operator=(const VSISOZipHandle &) = delete; |
3714 | | |
3715 | | public: |
3716 | | VSISOZipHandle(VSIVirtualHandleUniquePtr poVirtualHandleIn, |
3717 | | vsi_l_offset nPosCompressedStream, uint64_t compressed_size, |
3718 | | uint64_t uncompressed_size, vsi_l_offset indexPos, |
3719 | | uint32_t nToSkip, uint32_t nChunkSize); |
3720 | | ~VSISOZipHandle() override; |
3721 | | |
3722 | | int Seek(vsi_l_offset nOffset, int nWhence) override; |
3723 | | |
3724 | | vsi_l_offset Tell() override |
3725 | 0 | { |
3726 | 0 | return nCurPos_; |
3727 | 0 | } |
3728 | | |
3729 | | size_t Read(void *pBuffer, size_t nSize, size_t nCount) override; |
3730 | | |
3731 | | size_t Write(const void *, size_t, size_t) override |
3732 | 0 | { |
3733 | 0 | return 0; |
3734 | 0 | } |
3735 | | |
3736 | | int Eof() override |
3737 | 0 | { |
3738 | 0 | return bEOF_; |
3739 | 0 | } |
3740 | | |
3741 | | int Error() override |
3742 | 0 | { |
3743 | 0 | return bError_; |
3744 | 0 | } |
3745 | | |
3746 | | void ClearErr() override |
3747 | 0 | { |
3748 | 0 | bEOF_ = false; |
3749 | 0 | bError_ = false; |
3750 | 0 | } |
3751 | | |
3752 | | int Close() override; |
3753 | | |
3754 | | bool IsOK() const |
3755 | 0 | { |
3756 | 0 | return bOK_; |
3757 | 0 | } |
3758 | | }; |
3759 | | |
3760 | | /************************************************************************/ |
3761 | | /* VSISOZipHandle() */ |
3762 | | /************************************************************************/ |
3763 | | |
3764 | | VSISOZipHandle::VSISOZipHandle(VSIVirtualHandleUniquePtr poVirtualHandleIn, |
3765 | | vsi_l_offset nPosCompressedStream, |
3766 | | uint64_t compressed_size, |
3767 | | uint64_t uncompressed_size, |
3768 | | vsi_l_offset indexPos, uint32_t nToSkip, |
3769 | | uint32_t nChunkSize) |
3770 | 0 | : poBaseHandle_(std::move(poVirtualHandleIn)), |
3771 | 0 | nPosCompressedStream_(nPosCompressedStream), |
3772 | 0 | compressed_size_(compressed_size), uncompressed_size_(uncompressed_size), |
3773 | 0 | indexPos_(indexPos), nToSkip_(nToSkip), nChunkSize_(nChunkSize) |
3774 | 0 | { |
3775 | | #ifdef HAVE_LIBDEFLATE |
3776 | | pDecompressor_ = libdeflate_alloc_decompressor(); |
3777 | | if (!pDecompressor_) |
3778 | | bOK_ = false; |
3779 | | #else |
3780 | 0 | memset(&sStream_, 0, sizeof(sStream_)); |
3781 | 0 | int err = inflateInit2(&sStream_, -MAX_WBITS); |
3782 | 0 | if (err != Z_OK) |
3783 | 0 | bOK_ = false; |
3784 | 0 | #endif |
3785 | 0 | } |
3786 | | |
3787 | | /************************************************************************/ |
3788 | | /* ~VSISOZipHandle() */ |
3789 | | /************************************************************************/ |
3790 | | |
3791 | | VSISOZipHandle::~VSISOZipHandle() |
3792 | 0 | { |
3793 | 0 | VSISOZipHandle::Close(); |
3794 | 0 | if (bOK_) |
3795 | 0 | { |
3796 | | #ifdef HAVE_LIBDEFLATE |
3797 | | libdeflate_free_decompressor(pDecompressor_); |
3798 | | #else |
3799 | 0 | inflateEnd(&sStream_); |
3800 | 0 | #endif |
3801 | 0 | } |
3802 | 0 | } |
3803 | | |
3804 | | /************************************************************************/ |
3805 | | /* Close() */ |
3806 | | /************************************************************************/ |
3807 | | |
3808 | | int VSISOZipHandle::Close() |
3809 | 0 | { |
3810 | 0 | int ret = 0; |
3811 | 0 | if (poBaseHandle_) |
3812 | 0 | { |
3813 | 0 | ret = poBaseHandle_->Close(); |
3814 | 0 | poBaseHandle_.reset(); |
3815 | 0 | } |
3816 | 0 | return ret; |
3817 | 0 | } |
3818 | | |
3819 | | /************************************************************************/ |
3820 | | /* Seek() */ |
3821 | | /************************************************************************/ |
3822 | | |
3823 | | int VSISOZipHandle::Seek(vsi_l_offset nOffset, int nWhence) |
3824 | 0 | { |
3825 | 0 | bEOF_ = false; |
3826 | 0 | if (nWhence == SEEK_SET) |
3827 | 0 | nCurPos_ = nOffset; |
3828 | 0 | else if (nWhence == SEEK_END) |
3829 | 0 | nCurPos_ = uncompressed_size_; |
3830 | 0 | else |
3831 | 0 | nCurPos_ += nOffset; |
3832 | 0 | return 0; |
3833 | 0 | } |
3834 | | |
3835 | | /************************************************************************/ |
3836 | | /* Read() */ |
3837 | | /************************************************************************/ |
3838 | | |
3839 | | size_t VSISOZipHandle::Read(void *pBuffer, size_t nSize, size_t nCount) |
3840 | 0 | { |
3841 | 0 | size_t nToRead = nSize * nCount; |
3842 | 0 | if (nCurPos_ >= uncompressed_size_ && nToRead > 0) |
3843 | 0 | { |
3844 | 0 | bEOF_ = true; |
3845 | 0 | return 0; |
3846 | 0 | } |
3847 | | |
3848 | 0 | if (nSize != 1) |
3849 | 0 | { |
3850 | 0 | bError_ = true; |
3851 | 0 | CPLError(CE_Failure, CPLE_NotSupported, "Unsupported nSize"); |
3852 | 0 | return 0; |
3853 | 0 | } |
3854 | 0 | if ((nCurPos_ % nChunkSize_) != 0) |
3855 | 0 | { |
3856 | 0 | bError_ = true; |
3857 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
3858 | 0 | "nCurPos is not a multiple of nChunkSize"); |
3859 | 0 | return 0; |
3860 | 0 | } |
3861 | 0 | if (nCurPos_ + nToRead > uncompressed_size_) |
3862 | 0 | { |
3863 | 0 | nToRead = static_cast<size_t>(uncompressed_size_ - nCurPos_); |
3864 | 0 | nCount = nToRead; |
3865 | 0 | } |
3866 | 0 | else if ((nToRead % nChunkSize_) != 0) |
3867 | 0 | { |
3868 | 0 | bError_ = true; |
3869 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
3870 | 0 | "nToRead is not a multiple of nChunkSize"); |
3871 | 0 | return 0; |
3872 | 0 | } |
3873 | | |
3874 | 0 | const auto ReadOffsetInCompressedStream = |
3875 | 0 | [this](uint64_t nChunkIdx) -> uint64_t |
3876 | 0 | { |
3877 | 0 | if (nChunkIdx == 0) |
3878 | 0 | return 0; |
3879 | 0 | if (nChunkIdx == 1 + (uncompressed_size_ - 1) / nChunkSize_) |
3880 | 0 | return compressed_size_; |
3881 | 0 | constexpr size_t nOffsetSize = 8; |
3882 | 0 | if (poBaseHandle_->Seek(indexPos_ + 32 + nToSkip_ + |
3883 | 0 | (nChunkIdx - 1) * nOffsetSize, |
3884 | 0 | SEEK_SET) != 0) |
3885 | 0 | return static_cast<uint64_t>(-1); |
3886 | | |
3887 | 0 | uint64_t nOffset; |
3888 | 0 | if (poBaseHandle_->Read(&nOffset, sizeof(nOffset), 1) != 1) |
3889 | 0 | return static_cast<uint64_t>(-1); |
3890 | 0 | CPL_LSBPTR64(&nOffset); |
3891 | 0 | return nOffset; |
3892 | 0 | }; |
3893 | |
|
3894 | 0 | size_t nOffsetInOutputBuffer = 0; |
3895 | 0 | while (true) |
3896 | 0 | { |
3897 | 0 | uint64_t nOffsetInCompressedStream = |
3898 | 0 | ReadOffsetInCompressedStream(nCurPos_ / nChunkSize_); |
3899 | 0 | if (nOffsetInCompressedStream == static_cast<uint64_t>(-1)) |
3900 | 0 | { |
3901 | 0 | bError_ = true; |
3902 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
3903 | 0 | "Cannot read nOffsetInCompressedStream"); |
3904 | 0 | return 0; |
3905 | 0 | } |
3906 | 0 | uint64_t nNextOffsetInCompressedStream = |
3907 | 0 | ReadOffsetInCompressedStream(1 + nCurPos_ / nChunkSize_); |
3908 | 0 | if (nNextOffsetInCompressedStream == static_cast<uint64_t>(-1)) |
3909 | 0 | { |
3910 | 0 | bError_ = true; |
3911 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
3912 | 0 | "Cannot read nNextOffsetInCompressedStream"); |
3913 | 0 | return 0; |
3914 | 0 | } |
3915 | | |
3916 | 0 | if (nNextOffsetInCompressedStream <= nOffsetInCompressedStream || |
3917 | 0 | nNextOffsetInCompressedStream - nOffsetInCompressedStream > |
3918 | 0 | 13 + 2 * nChunkSize_ || |
3919 | 0 | nNextOffsetInCompressedStream > compressed_size_) |
3920 | 0 | { |
3921 | 0 | bError_ = true; |
3922 | 0 | CPLError( |
3923 | 0 | CE_Failure, CPLE_AppDefined, |
3924 | 0 | "Invalid values for nOffsetInCompressedStream (" CPL_FRMT_GUIB |
3925 | 0 | ") / " |
3926 | 0 | "nNextOffsetInCompressedStream(" CPL_FRMT_GUIB ")", |
3927 | 0 | static_cast<GUIntBig>(nOffsetInCompressedStream), |
3928 | 0 | static_cast<GUIntBig>(nNextOffsetInCompressedStream)); |
3929 | 0 | return 0; |
3930 | 0 | } |
3931 | | |
3932 | | // CPLDebug("VSIZIP", "Seek to compressed data at offset " |
3933 | | // CPL_FRMT_GUIB, static_cast<GUIntBig>(nPosCompressedStream_ + |
3934 | | // nOffsetInCompressedStream)); |
3935 | 0 | if (poBaseHandle_->Seek( |
3936 | 0 | nPosCompressedStream_ + nOffsetInCompressedStream, SEEK_SET) != |
3937 | 0 | 0) |
3938 | 0 | { |
3939 | 0 | bError_ = true; |
3940 | 0 | return 0; |
3941 | 0 | } |
3942 | | |
3943 | 0 | const int nCompressedToRead = static_cast<int>( |
3944 | 0 | nNextOffsetInCompressedStream - nOffsetInCompressedStream); |
3945 | | // CPLDebug("VSIZIP", "nCompressedToRead = %d", nCompressedToRead); |
3946 | 0 | std::vector<GByte> abyCompressedData(nCompressedToRead); |
3947 | 0 | if (poBaseHandle_->Read(&abyCompressedData[0], nCompressedToRead, 1) != |
3948 | 0 | 1) |
3949 | 0 | { |
3950 | 0 | bError_ = true; |
3951 | 0 | return 0; |
3952 | 0 | } |
3953 | | |
3954 | 0 | size_t nToReadThisIter = |
3955 | 0 | std::min(nToRead, static_cast<size_t>(nChunkSize_)); |
3956 | |
|
3957 | 0 | if (nCompressedToRead >= 5 && |
3958 | 0 | abyCompressedData[nCompressedToRead - 5] == 0x00 && |
3959 | 0 | memcmp(&abyCompressedData[nCompressedToRead - 4], |
3960 | 0 | "\x00\x00\xFF\xFF", 4) == 0) |
3961 | 0 | { |
3962 | | // Tag this flush block as the last one. |
3963 | 0 | abyCompressedData[nCompressedToRead - 5] = 0x01; |
3964 | 0 | } |
3965 | |
|
3966 | | #ifdef HAVE_LIBDEFLATE |
3967 | | size_t nOut = 0; |
3968 | | if (libdeflate_deflate_decompress( |
3969 | | pDecompressor_, &abyCompressedData[0], nCompressedToRead, |
3970 | | static_cast<Bytef *>(pBuffer) + nOffsetInOutputBuffer, |
3971 | | nToReadThisIter, &nOut) != LIBDEFLATE_SUCCESS) |
3972 | | { |
3973 | | bError_ = true; |
3974 | | CPLError( |
3975 | | CE_Failure, CPLE_AppDefined, |
3976 | | "libdeflate_deflate_decompress() failed at pos " CPL_FRMT_GUIB, |
3977 | | static_cast<GUIntBig>(nCurPos_)); |
3978 | | return 0; |
3979 | | } |
3980 | | if (nOut != nToReadThisIter) |
3981 | | { |
3982 | | bError_ = true; |
3983 | | CPLError(CE_Failure, CPLE_AppDefined, |
3984 | | "Only %u bytes decompressed at pos " CPL_FRMT_GUIB |
3985 | | " whereas %u where expected", |
3986 | | static_cast<unsigned>(nOut), |
3987 | | static_cast<GUIntBig>(nCurPos_), |
3988 | | static_cast<unsigned>(nToReadThisIter)); |
3989 | | return 0; |
3990 | | } |
3991 | | #else |
3992 | 0 | sStream_.avail_in = nCompressedToRead; |
3993 | 0 | sStream_.next_in = &abyCompressedData[0]; |
3994 | 0 | sStream_.avail_out = static_cast<int>(nToReadThisIter); |
3995 | 0 | sStream_.next_out = |
3996 | 0 | static_cast<Bytef *>(pBuffer) + nOffsetInOutputBuffer; |
3997 | |
|
3998 | 0 | int err = inflate(&sStream_, Z_FINISH); |
3999 | 0 | if ((err != Z_OK && err != Z_STREAM_END)) |
4000 | 0 | { |
4001 | 0 | bError_ = true; |
4002 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4003 | 0 | "inflate() failed at pos " CPL_FRMT_GUIB, |
4004 | 0 | static_cast<GUIntBig>(nCurPos_)); |
4005 | 0 | inflateReset(&sStream_); |
4006 | 0 | return 0; |
4007 | 0 | } |
4008 | 0 | if (sStream_.avail_in != 0) |
4009 | 0 | CPLDebug("VSIZIP", "avail_in = %d", sStream_.avail_in); |
4010 | 0 | if (sStream_.avail_out != 0) |
4011 | 0 | { |
4012 | 0 | bError_ = true; |
4013 | 0 | CPLError( |
4014 | 0 | CE_Failure, CPLE_AppDefined, |
4015 | 0 | "Only %u bytes decompressed at pos " CPL_FRMT_GUIB |
4016 | 0 | " whereas %u where expected", |
4017 | 0 | static_cast<unsigned>(nToReadThisIter - sStream_.avail_out), |
4018 | 0 | static_cast<GUIntBig>(nCurPos_), |
4019 | 0 | static_cast<unsigned>(nToReadThisIter)); |
4020 | 0 | inflateReset(&sStream_); |
4021 | 0 | return 0; |
4022 | 0 | } |
4023 | 0 | inflateReset(&sStream_); |
4024 | 0 | #endif |
4025 | 0 | nOffsetInOutputBuffer += nToReadThisIter; |
4026 | 0 | nCurPos_ += nToReadThisIter; |
4027 | 0 | nToRead -= nToReadThisIter; |
4028 | 0 | if (nToRead == 0) |
4029 | 0 | break; |
4030 | 0 | } |
4031 | | |
4032 | 0 | return nCount; |
4033 | 0 | } |
4034 | | |
4035 | | /************************************************************************/ |
4036 | | /* GetFileInfo() */ |
4037 | | /************************************************************************/ |
4038 | | |
4039 | | bool VSIZipFilesystemHandler::GetFileInfo(const char *pszFilename, |
4040 | | VSIFileInZipInfo &info, |
4041 | | bool bSetError) |
4042 | 1.52k | { |
4043 | | |
4044 | 1.52k | CPLString osZipInFileName; |
4045 | 1.52k | std::unique_ptr<char, VSIFreeReleaser> zipFilename( |
4046 | 1.52k | SplitFilename(pszFilename, osZipInFileName, true, bSetError)); |
4047 | 1.52k | if (zipFilename == nullptr) |
4048 | 896 | return false; |
4049 | | |
4050 | 625 | { |
4051 | 625 | std::unique_lock oLock(oMutex); |
4052 | 625 | if (oMapZipWriteHandles.find(zipFilename.get()) != |
4053 | 625 | oMapZipWriteHandles.end()) |
4054 | 0 | { |
4055 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4056 | 0 | "Cannot read a zip file being written"); |
4057 | 0 | return false; |
4058 | 0 | } |
4059 | 625 | } |
4060 | | |
4061 | 625 | auto poReader = OpenArchiveFile(zipFilename.get(), osZipInFileName); |
4062 | 625 | if (poReader == nullptr) |
4063 | 625 | { |
4064 | 625 | return false; |
4065 | 625 | } |
4066 | | |
4067 | 0 | VSIFilesystemHandler *poFSHandler = |
4068 | 0 | VSIFileManager::GetHandler(zipFilename.get()); |
4069 | |
|
4070 | 0 | VSIVirtualHandleUniquePtr poVirtualHandle( |
4071 | 0 | poFSHandler->Open(zipFilename.get(), "rb")); |
4072 | |
|
4073 | 0 | if (poVirtualHandle == nullptr) |
4074 | 0 | { |
4075 | 0 | return false; |
4076 | 0 | } |
4077 | | |
4078 | 0 | unzFile unzF = |
4079 | 0 | cpl::down_cast<VSIZipReader *>(poReader.get())->GetUnzFileHandle(); |
4080 | |
|
4081 | 0 | if (cpl_unzOpenCurrentFile(unzF) != UNZ_OK) |
4082 | 0 | { |
4083 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4084 | 0 | "cpl_unzOpenCurrentFile() failed"); |
4085 | 0 | return false; |
4086 | 0 | } |
4087 | | |
4088 | 0 | info.nStartDataStream = cpl_unzGetCurrentFileZStreamPos(unzF); |
4089 | |
|
4090 | 0 | unz_file_info file_info; |
4091 | 0 | if (cpl_unzGetCurrentFileInfo(unzF, &file_info, nullptr, 0, nullptr, 0, |
4092 | 0 | nullptr, 0) != UNZ_OK) |
4093 | 0 | { |
4094 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4095 | 0 | "cpl_unzGetCurrentFileInfo() failed"); |
4096 | 0 | cpl_unzCloseCurrentFile(unzF); |
4097 | 0 | return false; |
4098 | 0 | } |
4099 | | |
4100 | 0 | if (file_info.size_file_extra) |
4101 | 0 | { |
4102 | 0 | std::vector<GByte> abyExtra(file_info.size_file_extra); |
4103 | 0 | poVirtualHandle->Seek(file_info.file_extra_abs_offset, SEEK_SET); |
4104 | 0 | if (poVirtualHandle->Read(&abyExtra[0], abyExtra.size(), 1) == 1) |
4105 | 0 | { |
4106 | 0 | size_t nPos = 0; |
4107 | 0 | while (nPos + 2 * sizeof(uint16_t) <= abyExtra.size()) |
4108 | 0 | { |
4109 | 0 | uint16_t nId; |
4110 | 0 | memcpy(&nId, &abyExtra[nPos], sizeof(uint16_t)); |
4111 | 0 | nPos += sizeof(uint16_t); |
4112 | 0 | CPL_LSBPTR16(&nId); |
4113 | 0 | uint16_t nSize; |
4114 | 0 | memcpy(&nSize, &abyExtra[nPos], sizeof(uint16_t)); |
4115 | 0 | nPos += sizeof(uint16_t); |
4116 | 0 | CPL_LSBPTR16(&nSize); |
4117 | 0 | if (nId == 0x564b && nPos + nSize <= abyExtra.size()) // "KV" |
4118 | 0 | { |
4119 | 0 | if (nSize >= strlen("KeyValuePairs") + 1 && |
4120 | 0 | memcmp(&abyExtra[nPos], "KeyValuePairs", |
4121 | 0 | strlen("KeyValuePairs")) == 0) |
4122 | 0 | { |
4123 | 0 | int nPos2 = static_cast<int>(strlen("KeyValuePairs")); |
4124 | 0 | int nKVPairs = abyExtra[nPos + nPos2]; |
4125 | 0 | nPos2++; |
4126 | 0 | for (int iKV = 0; iKV < nKVPairs; ++iKV) |
4127 | 0 | { |
4128 | 0 | if (nPos2 + sizeof(uint16_t) > nSize) |
4129 | 0 | break; |
4130 | 0 | uint16_t nKeyLen; |
4131 | 0 | memcpy(&nKeyLen, &abyExtra[nPos + nPos2], |
4132 | 0 | sizeof(uint16_t)); |
4133 | 0 | nPos2 += sizeof(uint16_t); |
4134 | 0 | CPL_LSBPTR16(&nKeyLen); |
4135 | 0 | if (nPos2 + nKeyLen > nSize) |
4136 | 0 | break; |
4137 | 0 | std::string osKey; |
4138 | 0 | osKey.resize(nKeyLen); |
4139 | 0 | memcpy(&osKey[0], &abyExtra[nPos + nPos2], nKeyLen); |
4140 | 0 | nPos2 += nKeyLen; |
4141 | |
|
4142 | 0 | if (nPos2 + sizeof(uint16_t) > nSize) |
4143 | 0 | break; |
4144 | 0 | uint16_t nValLen; |
4145 | 0 | memcpy(&nValLen, &abyExtra[nPos + nPos2], |
4146 | 0 | sizeof(uint16_t)); |
4147 | 0 | nPos2 += sizeof(uint16_t); |
4148 | 0 | CPL_LSBPTR16(&nValLen); |
4149 | 0 | if (nPos2 + nValLen > nSize) |
4150 | 0 | break; |
4151 | 0 | std::string osVal; |
4152 | 0 | osVal.resize(nValLen); |
4153 | 0 | memcpy(&osVal[0], &abyExtra[nPos + nPos2], nValLen); |
4154 | 0 | nPos2 += nValLen; |
4155 | |
|
4156 | 0 | info.oMapProperties[osKey] = std::move(osVal); |
4157 | 0 | } |
4158 | 0 | } |
4159 | 0 | } |
4160 | 0 | nPos += nSize; |
4161 | 0 | } |
4162 | 0 | } |
4163 | 0 | } |
4164 | |
|
4165 | 0 | info.nCRC = file_info.crc; |
4166 | 0 | info.nCompressionMethod = static_cast<int>(file_info.compression_method); |
4167 | 0 | info.nUncompressedSize = static_cast<uint64_t>(file_info.uncompressed_size); |
4168 | 0 | info.nCompressedSize = static_cast<uint64_t>(file_info.compressed_size); |
4169 | | |
4170 | | // Sanity checks |
4171 | 0 | if (info.nCompressedSize > |
4172 | 0 | std::numeric_limits<uint64_t>::max() - info.nStartDataStream) |
4173 | 0 | { |
4174 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4175 | 0 | "Invalid compressed size for file %s", pszFilename); |
4176 | 0 | return false; |
4177 | 0 | } |
4178 | 0 | const uLong64 afterFileOffset = |
4179 | 0 | info.nStartDataStream + info.nCompressedSize; |
4180 | | |
4181 | | // Cf https://stackoverflow.com/questions/16792189/gzip-compression-ratio-for-zeros/16794960 |
4182 | 0 | constexpr unsigned MAX_DEFLATE_COMPRESSION_RATIO = 1032; |
4183 | 0 | if (info.nCompressedSize == 0 && info.nUncompressedSize != 0) |
4184 | 0 | { |
4185 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4186 | 0 | "Invalid compressed size (=0) vs uncompressed size (!=0) for " |
4187 | 0 | "file %s", |
4188 | 0 | pszFilename); |
4189 | 0 | return false; |
4190 | 0 | } |
4191 | 0 | else if (info.nCompressedSize != 0 && |
4192 | 0 | info.nUncompressedSize / info.nCompressedSize > |
4193 | 0 | MAX_DEFLATE_COMPRESSION_RATIO) |
4194 | 0 | { |
4195 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4196 | 0 | "Invalid compression ratio for file %s: %" PRIu64, pszFilename, |
4197 | 0 | info.nUncompressedSize / info.nCompressedSize); |
4198 | 0 | return false; |
4199 | 0 | } |
4200 | | |
4201 | | // A bit arbitrary |
4202 | 0 | constexpr unsigned THRESHOLD_FOR_BIG_ALLOCS = 1024 * 1024 * 1024; |
4203 | 0 | if (info.nUncompressedSize > THRESHOLD_FOR_BIG_ALLOCS) |
4204 | 0 | { |
4205 | | // Check that the compressed file size is consistent with the ZIP file size |
4206 | 0 | poVirtualHandle->Seek(0, SEEK_END); |
4207 | 0 | if (afterFileOffset > poVirtualHandle->Tell()) |
4208 | 0 | { |
4209 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4210 | 0 | "Invalid compressed size for file %s: %" PRIu64, |
4211 | 0 | pszFilename, info.nCompressedSize); |
4212 | 0 | return false; |
4213 | 0 | } |
4214 | 0 | } |
4215 | | |
4216 | | // Try to locate .sozip.idx file |
4217 | 0 | unz_file_info file_info2; |
4218 | 0 | std::string osAuxName; |
4219 | 0 | osAuxName.resize(1024); |
4220 | 0 | uLong64 indexPos; |
4221 | 0 | if (file_info.compression_method == 8 && |
4222 | 0 | cpl_unzCurrentFileInfoFromLocalHeader( |
4223 | 0 | unzF, afterFileOffset, &file_info2, &osAuxName[0], osAuxName.size(), |
4224 | 0 | &indexPos) == UNZ_OK) |
4225 | 0 | { |
4226 | 0 | osAuxName.resize(strlen(osAuxName.c_str())); |
4227 | 0 | if (osAuxName.find(".sozip.idx") != std::string::npos) |
4228 | 0 | { |
4229 | 0 | info.bSOZipIndexFound = true; |
4230 | 0 | info.nSOZIPStartData = indexPos; |
4231 | 0 | poVirtualHandle->Seek(indexPos, SEEK_SET); |
4232 | 0 | uint32_t nVersion = 0; |
4233 | 0 | poVirtualHandle->Read(&nVersion, sizeof(nVersion), 1); |
4234 | 0 | CPL_LSBPTR32(&nVersion); |
4235 | 0 | uint32_t nToSkip = 0; |
4236 | 0 | poVirtualHandle->Read(&nToSkip, sizeof(nToSkip), 1); |
4237 | 0 | CPL_LSBPTR32(&nToSkip); |
4238 | 0 | uint32_t nChunkSize = 0; |
4239 | 0 | poVirtualHandle->Read(&nChunkSize, sizeof(nChunkSize), 1); |
4240 | 0 | CPL_LSBPTR32(&nChunkSize); |
4241 | 0 | uint32_t nOffsetSize = 0; |
4242 | 0 | poVirtualHandle->Read(&nOffsetSize, sizeof(nOffsetSize), 1); |
4243 | 0 | CPL_LSBPTR32(&nOffsetSize); |
4244 | 0 | uint64_t nUncompressedSize = 0; |
4245 | 0 | poVirtualHandle->Read(&nUncompressedSize, sizeof(nUncompressedSize), |
4246 | 0 | 1); |
4247 | 0 | CPL_LSBPTR64(&nUncompressedSize); |
4248 | 0 | uint64_t nCompressedSize = 0; |
4249 | 0 | poVirtualHandle->Read(&nCompressedSize, sizeof(nCompressedSize), 1); |
4250 | 0 | CPL_LSBPTR64(&nCompressedSize); |
4251 | |
|
4252 | 0 | info.nSOZIPVersion = nVersion; |
4253 | 0 | info.nSOZIPToSkip = nToSkip; |
4254 | 0 | info.nSOZIPChunkSize = nChunkSize; |
4255 | 0 | info.nSOZIPOffsetSize = nOffsetSize; |
4256 | |
|
4257 | 0 | bool bValid = true; |
4258 | 0 | if (nVersion != 1) |
4259 | 0 | { |
4260 | 0 | CPLDebug("SOZIP", "version = %u, expected 1", nVersion); |
4261 | 0 | bValid = false; |
4262 | 0 | } |
4263 | 0 | if (nCompressedSize != file_info.compressed_size) |
4264 | 0 | { |
4265 | 0 | CPLDebug("SOZIP", |
4266 | 0 | "compressedSize field inconsistent with file"); |
4267 | 0 | bValid = false; |
4268 | 0 | } |
4269 | 0 | if (nUncompressedSize != file_info.uncompressed_size) |
4270 | 0 | { |
4271 | 0 | CPLDebug("SOZIP", |
4272 | 0 | "uncompressedSize field inconsistent with file"); |
4273 | 0 | bValid = false; |
4274 | 0 | } |
4275 | 0 | if (!(nChunkSize > 0 && nChunkSize < 100 * 1024 * 1024)) |
4276 | 0 | { |
4277 | 0 | CPLDebug("SOZIP", "invalid chunkSize = %u", nChunkSize); |
4278 | 0 | bValid = false; |
4279 | 0 | } |
4280 | 0 | if (nOffsetSize != 8) |
4281 | 0 | { |
4282 | 0 | CPLDebug("SOZIP", "invalid offsetSize = %u", nOffsetSize); |
4283 | 0 | bValid = false; |
4284 | 0 | } |
4285 | 0 | if (file_info2.compression_method != 0) |
4286 | 0 | { |
4287 | 0 | CPLDebug("SOZIP", "unexpected compression_method = %u", |
4288 | 0 | static_cast<unsigned>(file_info2.compression_method)); |
4289 | 0 | bValid = false; |
4290 | 0 | } |
4291 | 0 | if (bValid) |
4292 | 0 | { |
4293 | 0 | const auto nExpectedIndexSize = |
4294 | 0 | 32 + static_cast<uint64_t>(nToSkip) + |
4295 | 0 | ((nUncompressedSize - 1) / nChunkSize) * nOffsetSize; |
4296 | 0 | if (nExpectedIndexSize != file_info2.uncompressed_size) |
4297 | 0 | { |
4298 | 0 | CPLDebug("SOZIP", "invalid file size for index"); |
4299 | 0 | bValid = false; |
4300 | 0 | } |
4301 | 0 | } |
4302 | 0 | if (bValid) |
4303 | 0 | { |
4304 | 0 | info.bSOZipIndexValid = true; |
4305 | 0 | CPLDebug("SOZIP", "Found valid SOZIP index: %s", |
4306 | 0 | osAuxName.c_str()); |
4307 | 0 | } |
4308 | 0 | else |
4309 | 0 | { |
4310 | 0 | CPLDebug("SOZIP", "Found *invalid* SOZIP index: %s", |
4311 | 0 | osAuxName.c_str()); |
4312 | 0 | } |
4313 | 0 | } |
4314 | 0 | } |
4315 | |
|
4316 | 0 | cpl_unzCloseCurrentFile(unzF); |
4317 | |
|
4318 | 0 | info.poVirtualHandle = std::move(poVirtualHandle); |
4319 | |
|
4320 | 0 | return true; |
4321 | 0 | } |
4322 | | |
4323 | | /************************************************************************/ |
4324 | | /* Open() */ |
4325 | | /************************************************************************/ |
4326 | | |
4327 | | VSIVirtualHandleUniquePtr |
4328 | | VSIZipFilesystemHandler::Open(const char *pszFilename, const char *pszAccess, |
4329 | | bool bSetError, CSLConstList /* papszOptions */) |
4330 | 1.52k | { |
4331 | | |
4332 | 1.52k | if (strchr(pszAccess, 'w') != nullptr) |
4333 | 0 | { |
4334 | 0 | return OpenForWrite(pszFilename, pszAccess); |
4335 | 0 | } |
4336 | | |
4337 | 1.52k | if (strchr(pszAccess, '+') != nullptr) |
4338 | 0 | { |
4339 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4340 | 0 | "Read-write random access not supported for /vsizip"); |
4341 | 0 | return nullptr; |
4342 | 0 | } |
4343 | | |
4344 | 1.52k | VSIFileInZipInfo info; |
4345 | 1.52k | if (!GetFileInfo(pszFilename, info, bSetError)) |
4346 | 1.52k | return nullptr; |
4347 | | |
4348 | 0 | #ifdef ENABLE_DEFLATE64 |
4349 | 0 | if (info.nCompressionMethod == 9) |
4350 | 0 | { |
4351 | 0 | auto poGZIPHandle = std::make_unique<VSIDeflate64Handle>( |
4352 | 0 | std::move(info.poVirtualHandle), nullptr, info.nStartDataStream, |
4353 | 0 | info.nCompressedSize, info.nUncompressedSize, info.nCRC); |
4354 | 0 | if (!(poGZIPHandle->IsInitOK())) |
4355 | 0 | { |
4356 | 0 | return nullptr; |
4357 | 0 | } |
4358 | | |
4359 | | // Wrap the VSIGZipHandle inside a buffered reader that will |
4360 | | // improve dramatically performance when doing small backward |
4361 | | // seeks. |
4362 | 0 | return VSIVirtualHandleUniquePtr( |
4363 | 0 | VSICreateBufferedReaderHandle(poGZIPHandle.release())); |
4364 | 0 | } |
4365 | 0 | else |
4366 | 0 | #endif |
4367 | 0 | { |
4368 | 0 | if (info.bSOZipIndexValid) |
4369 | 0 | { |
4370 | 0 | auto poSOZIPHandle = std::make_unique<VSISOZipHandle>( |
4371 | 0 | std::move(info.poVirtualHandle), info.nStartDataStream, |
4372 | 0 | info.nCompressedSize, info.nUncompressedSize, |
4373 | 0 | info.nSOZIPStartData, info.nSOZIPToSkip, info.nSOZIPChunkSize); |
4374 | 0 | if (!poSOZIPHandle->IsOK()) |
4375 | 0 | { |
4376 | 0 | return nullptr; |
4377 | 0 | } |
4378 | 0 | return VSIVirtualHandleUniquePtr(VSICreateCachedFile( |
4379 | 0 | poSOZIPHandle.release(), info.nSOZIPChunkSize, 0)); |
4380 | 0 | } |
4381 | | |
4382 | 0 | auto poGZIPHandle = std::make_unique<VSIGZipHandle>( |
4383 | 0 | std::move(info.poVirtualHandle), nullptr, info.nStartDataStream, |
4384 | 0 | info.nCompressedSize, info.nUncompressedSize, info.nCRC, |
4385 | 0 | info.nCompressionMethod == 0); |
4386 | 0 | if (!(poGZIPHandle->IsInitOK())) |
4387 | 0 | { |
4388 | 0 | return nullptr; |
4389 | 0 | } |
4390 | | |
4391 | | // Wrap the VSIGZipHandle inside a buffered reader that will |
4392 | | // improve dramatically performance when doing small backward |
4393 | | // seeks. |
4394 | 0 | return VSIVirtualHandleUniquePtr( |
4395 | 0 | VSICreateBufferedReaderHandle(poGZIPHandle.release())); |
4396 | 0 | } |
4397 | 0 | } |
4398 | | |
4399 | | /************************************************************************/ |
4400 | | /* GetFileMetadata() */ |
4401 | | /************************************************************************/ |
4402 | | |
4403 | | char **VSIZipFilesystemHandler::GetFileMetadata(const char *pszFilename, |
4404 | | const char *pszDomain, |
4405 | | CSLConstList /*papszOptions*/) |
4406 | 0 | { |
4407 | 0 | VSIFileInZipInfo info; |
4408 | 0 | if (!GetFileInfo(pszFilename, info, true)) |
4409 | 0 | return nullptr; |
4410 | | |
4411 | 0 | if (!pszDomain) |
4412 | 0 | { |
4413 | 0 | CPLStringList aosMetadata; |
4414 | 0 | for (const auto &kv : info.oMapProperties) |
4415 | 0 | { |
4416 | 0 | aosMetadata.AddNameValue(kv.first.c_str(), kv.second.c_str()); |
4417 | 0 | } |
4418 | 0 | return aosMetadata.StealList(); |
4419 | 0 | } |
4420 | 0 | else if (EQUAL(pszDomain, "ZIP")) |
4421 | 0 | { |
4422 | 0 | CPLStringList aosMetadata; |
4423 | 0 | aosMetadata.SetNameValue( |
4424 | 0 | "START_DATA_OFFSET", |
4425 | 0 | CPLSPrintf(CPL_FRMT_GUIB, |
4426 | 0 | static_cast<GUIntBig>(info.nStartDataStream))); |
4427 | |
|
4428 | 0 | if (info.nCompressionMethod == 0) |
4429 | 0 | aosMetadata.SetNameValue("COMPRESSION_METHOD", "0 (STORED)"); |
4430 | 0 | else if (info.nCompressionMethod == 8) |
4431 | 0 | aosMetadata.SetNameValue("COMPRESSION_METHOD", "8 (DEFLATE)"); |
4432 | 0 | else |
4433 | 0 | { |
4434 | 0 | aosMetadata.SetNameValue("COMPRESSION_METHOD", |
4435 | 0 | CPLSPrintf("%d", info.nCompressionMethod)); |
4436 | 0 | } |
4437 | 0 | aosMetadata.SetNameValue( |
4438 | 0 | "COMPRESSED_SIZE", |
4439 | 0 | CPLSPrintf(CPL_FRMT_GUIB, |
4440 | 0 | static_cast<GUIntBig>(info.nCompressedSize))); |
4441 | 0 | aosMetadata.SetNameValue( |
4442 | 0 | "UNCOMPRESSED_SIZE", |
4443 | 0 | CPLSPrintf(CPL_FRMT_GUIB, |
4444 | 0 | static_cast<GUIntBig>(info.nUncompressedSize))); |
4445 | |
|
4446 | 0 | if (info.bSOZipIndexFound) |
4447 | 0 | { |
4448 | 0 | aosMetadata.SetNameValue("SOZIP_FOUND", "YES"); |
4449 | |
|
4450 | 0 | aosMetadata.SetNameValue("SOZIP_VERSION", |
4451 | 0 | CPLSPrintf("%u", info.nSOZIPVersion)); |
4452 | |
|
4453 | 0 | aosMetadata.SetNameValue("SOZIP_OFFSET_SIZE", |
4454 | 0 | CPLSPrintf("%u", info.nSOZIPOffsetSize)); |
4455 | |
|
4456 | 0 | aosMetadata.SetNameValue("SOZIP_CHUNK_SIZE", |
4457 | 0 | CPLSPrintf("%u", info.nSOZIPChunkSize)); |
4458 | |
|
4459 | 0 | aosMetadata.SetNameValue( |
4460 | 0 | "SOZIP_START_DATA_OFFSET", |
4461 | 0 | CPLSPrintf(CPL_FRMT_GUIB, |
4462 | 0 | static_cast<GUIntBig>(info.nSOZIPStartData))); |
4463 | |
|
4464 | 0 | if (info.bSOZipIndexValid) |
4465 | 0 | { |
4466 | 0 | aosMetadata.SetNameValue("SOZIP_VALID", "YES"); |
4467 | 0 | } |
4468 | 0 | } |
4469 | |
|
4470 | 0 | return aosMetadata.StealList(); |
4471 | 0 | } |
4472 | 0 | return nullptr; |
4473 | 0 | } |
4474 | | |
4475 | | /************************************************************************/ |
4476 | | /* Mkdir() */ |
4477 | | /************************************************************************/ |
4478 | | |
4479 | | int VSIZipFilesystemHandler::Mkdir(const char *pszDirname, long /* nMode */) |
4480 | 0 | { |
4481 | 0 | CPLString osDirname = pszDirname; |
4482 | 0 | if (!osDirname.empty() && osDirname.back() != '/') |
4483 | 0 | osDirname += "/"; |
4484 | 0 | return OpenForWrite(osDirname, "wb") != nullptr ? 0 : -1; |
4485 | 0 | } |
4486 | | |
4487 | | /************************************************************************/ |
4488 | | /* ReadDirEx() */ |
4489 | | /************************************************************************/ |
4490 | | |
4491 | | char **VSIZipFilesystemHandler::ReadDirEx(const char *pszDirname, int nMaxFiles) |
4492 | 0 | { |
4493 | 0 | CPLString osInArchiveSubDir; |
4494 | 0 | char *zipFilename = |
4495 | 0 | SplitFilename(pszDirname, osInArchiveSubDir, true, true); |
4496 | 0 | if (zipFilename == nullptr) |
4497 | 0 | return nullptr; |
4498 | | |
4499 | 0 | { |
4500 | 0 | std::unique_lock oLock(oMutex); |
4501 | |
|
4502 | 0 | if (oMapZipWriteHandles.find(zipFilename) != oMapZipWriteHandles.end()) |
4503 | 0 | { |
4504 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4505 | 0 | "Cannot read a zip file being written"); |
4506 | 0 | CPLFree(zipFilename); |
4507 | 0 | return nullptr; |
4508 | 0 | } |
4509 | 0 | } |
4510 | 0 | CPLFree(zipFilename); |
4511 | |
|
4512 | 0 | return VSIArchiveFilesystemHandler::ReadDirEx(pszDirname, nMaxFiles); |
4513 | 0 | } |
4514 | | |
4515 | | /************************************************************************/ |
4516 | | /* Stat() */ |
4517 | | /************************************************************************/ |
4518 | | |
4519 | | int VSIZipFilesystemHandler::Stat(const char *pszFilename, |
4520 | | VSIStatBufL *pStatBuf, int nFlags) |
4521 | 2.34k | { |
4522 | 2.34k | CPLString osInArchiveSubDir; |
4523 | | |
4524 | 2.34k | memset(pStatBuf, 0, sizeof(VSIStatBufL)); |
4525 | | |
4526 | 2.34k | char *zipFilename = SplitFilename(pszFilename, osInArchiveSubDir, true, |
4527 | 2.34k | (nFlags & VSI_STAT_SET_ERROR_FLAG) != 0); |
4528 | 2.34k | if (zipFilename == nullptr) |
4529 | 1.56k | return -1; |
4530 | | |
4531 | 780 | { |
4532 | 780 | std::unique_lock oLock(oMutex); |
4533 | | |
4534 | 780 | if (oMapZipWriteHandles.find(zipFilename) != oMapZipWriteHandles.end()) |
4535 | 0 | { |
4536 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4537 | 0 | "Cannot read a zip file being written"); |
4538 | 0 | CPLFree(zipFilename); |
4539 | 0 | return -1; |
4540 | 0 | } |
4541 | 780 | } |
4542 | 780 | CPLFree(zipFilename); |
4543 | | |
4544 | 780 | return VSIArchiveFilesystemHandler::Stat(pszFilename, pStatBuf, nFlags); |
4545 | 780 | } |
4546 | | |
4547 | | /************************************************************************/ |
4548 | | /* RemoveFromMap() */ |
4549 | | /************************************************************************/ |
4550 | | |
4551 | | void VSIZipFilesystemHandler::RemoveFromMap(VSIZipWriteHandle *poHandle) |
4552 | 0 | { |
4553 | 0 | std::unique_lock oLock(oMutex); |
4554 | |
|
4555 | 0 | for (std::map<CPLString, VSIZipWriteHandle *>::iterator iter = |
4556 | 0 | oMapZipWriteHandles.begin(); |
4557 | 0 | iter != oMapZipWriteHandles.end(); ++iter) |
4558 | 0 | { |
4559 | 0 | if (iter->second == poHandle) |
4560 | 0 | { |
4561 | 0 | oMapZipWriteHandles.erase(iter); |
4562 | 0 | break; |
4563 | 0 | } |
4564 | 0 | } |
4565 | 0 | } |
4566 | | |
4567 | | /************************************************************************/ |
4568 | | /* OpenForWrite() */ |
4569 | | /************************************************************************/ |
4570 | | |
4571 | | VSIVirtualHandleUniquePtr |
4572 | | VSIZipFilesystemHandler::OpenForWrite(const char *pszFilename, |
4573 | | const char *pszAccess) |
4574 | 0 | { |
4575 | 0 | std::unique_lock oLock(oMutex); |
4576 | 0 | return OpenForWrite_unlocked(pszFilename, pszAccess); |
4577 | 0 | } |
4578 | | |
4579 | | VSIVirtualHandleUniquePtr |
4580 | | VSIZipFilesystemHandler::OpenForWrite_unlocked(const char *pszFilename, |
4581 | | const char *pszAccess) |
4582 | 0 | { |
4583 | 0 | CPLString osZipInFileName; |
4584 | |
|
4585 | 0 | char *zipFilename = |
4586 | 0 | SplitFilename(pszFilename, osZipInFileName, false, false); |
4587 | 0 | if (zipFilename == nullptr) |
4588 | 0 | return nullptr; |
4589 | 0 | CPLString osZipFilename = zipFilename; |
4590 | 0 | CPLFree(zipFilename); |
4591 | 0 | zipFilename = nullptr; |
4592 | | |
4593 | | // Invalidate cached file list. |
4594 | 0 | auto iter = oFileList.find(osZipFilename); |
4595 | 0 | if (iter != oFileList.end()) |
4596 | 0 | { |
4597 | 0 | oFileList.erase(iter); |
4598 | 0 | } |
4599 | |
|
4600 | 0 | auto oIter = oMapZipWriteHandles.find(osZipFilename); |
4601 | 0 | if (oIter != oMapZipWriteHandles.end()) |
4602 | 0 | { |
4603 | 0 | if (strchr(pszAccess, '+') != nullptr) |
4604 | 0 | { |
4605 | 0 | CPLError( |
4606 | 0 | CE_Failure, CPLE_AppDefined, |
4607 | 0 | "Random access not supported for writable file in /vsizip"); |
4608 | 0 | return nullptr; |
4609 | 0 | } |
4610 | | |
4611 | 0 | VSIZipWriteHandle *poZIPHandle = oIter->second; |
4612 | |
|
4613 | 0 | if (poZIPHandle->GetChildInWriting() != nullptr) |
4614 | 0 | { |
4615 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4616 | 0 | "Cannot create %s while another file is being " |
4617 | 0 | "written in the .zip", |
4618 | 0 | osZipInFileName.c_str()); |
4619 | 0 | return nullptr; |
4620 | 0 | } |
4621 | | |
4622 | 0 | poZIPHandle->StopCurrentFile(); |
4623 | | |
4624 | | // Re-add path separator when creating directories. |
4625 | 0 | char chLastChar = pszFilename[strlen(pszFilename) - 1]; |
4626 | 0 | if (chLastChar == '/' || chLastChar == '\\') |
4627 | 0 | osZipInFileName += chLastChar; |
4628 | |
|
4629 | 0 | if (CPLCreateFileInZip(poZIPHandle->GetHandle(), osZipInFileName, |
4630 | 0 | nullptr) != CE_None) |
4631 | 0 | return nullptr; |
4632 | | |
4633 | 0 | auto poChildHandle = |
4634 | 0 | std::make_unique<VSIZipWriteHandle>(this, nullptr, poZIPHandle); |
4635 | |
|
4636 | 0 | poZIPHandle->StartNewFile(poChildHandle.get()); |
4637 | |
|
4638 | 0 | return VSIVirtualHandleUniquePtr(poChildHandle.release()); |
4639 | 0 | } |
4640 | 0 | else |
4641 | 0 | { |
4642 | 0 | char **papszOptions = nullptr; |
4643 | 0 | if ((strchr(pszAccess, '+') && osZipInFileName.empty()) || |
4644 | 0 | !osZipInFileName.empty()) |
4645 | 0 | { |
4646 | 0 | VSIStatBufL sBuf; |
4647 | 0 | if (VSIStatExL(osZipFilename, &sBuf, VSI_STAT_EXISTS_FLAG) == 0) |
4648 | 0 | papszOptions = CSLAddNameValue(papszOptions, "APPEND", "TRUE"); |
4649 | 0 | } |
4650 | |
|
4651 | 0 | void *hZIP = CPLCreateZip(osZipFilename, papszOptions); |
4652 | 0 | CSLDestroy(papszOptions); |
4653 | |
|
4654 | 0 | if (hZIP == nullptr) |
4655 | 0 | return nullptr; |
4656 | | |
4657 | 0 | auto poHandle = new VSIZipWriteHandle(this, hZIP, nullptr); |
4658 | 0 | oMapZipWriteHandles[osZipFilename] = poHandle; |
4659 | |
|
4660 | 0 | if (!osZipInFileName.empty()) |
4661 | 0 | { |
4662 | 0 | auto poRes = std::unique_ptr<VSIZipWriteHandle>( |
4663 | 0 | cpl::down_cast<VSIZipWriteHandle *>( |
4664 | 0 | OpenForWrite_unlocked(pszFilename, pszAccess).release())); |
4665 | 0 | if (poRes == nullptr) |
4666 | 0 | { |
4667 | 0 | delete poHandle; |
4668 | 0 | oMapZipWriteHandles.erase(osZipFilename); |
4669 | 0 | return nullptr; |
4670 | 0 | } |
4671 | | |
4672 | 0 | poRes->SetAutoDeleteParent(); |
4673 | |
|
4674 | 0 | return VSIVirtualHandleUniquePtr(poRes.release()); |
4675 | 0 | } |
4676 | | |
4677 | 0 | return VSIVirtualHandleUniquePtr(poHandle); |
4678 | 0 | } |
4679 | 0 | } |
4680 | | |
4681 | | /************************************************************************/ |
4682 | | /* GetOptions() */ |
4683 | | /************************************************************************/ |
4684 | | |
4685 | | const char *VSIZipFilesystemHandler::GetOptions() |
4686 | 0 | { |
4687 | 0 | return "<Options>" |
4688 | 0 | " <Option name='GDAL_NUM_THREADS' type='string' " |
4689 | 0 | "description='Number of threads for compression. Either a integer " |
4690 | 0 | "or ALL_CPUS'/>" |
4691 | 0 | " <Option name='CPL_VSIL_DEFLATE_CHUNK_SIZE' type='string' " |
4692 | 0 | "description='Chunk of uncompressed data for parallelization. " |
4693 | 0 | "Use K(ilobytes) or M(egabytes) suffix' default='1M'/>" |
4694 | 0 | "</Options>"; |
4695 | 0 | } |
4696 | | |
4697 | | /************************************************************************/ |
4698 | | /* CopyFile() */ |
4699 | | /************************************************************************/ |
4700 | | |
4701 | | int VSIZipFilesystemHandler::CopyFile(const char *pszSource, |
4702 | | const char *pszTarget, VSILFILE *fpSource, |
4703 | | vsi_l_offset /* nSourceSize */, |
4704 | | CSLConstList papszOptions, |
4705 | | GDALProgressFunc pProgressFunc, |
4706 | | void *pProgressData) |
4707 | 0 | { |
4708 | 0 | CPLString osZipInFileName; |
4709 | |
|
4710 | 0 | char *zipFilename = SplitFilename(pszTarget, osZipInFileName, false, false); |
4711 | 0 | if (zipFilename == nullptr) |
4712 | 0 | return -1; |
4713 | 0 | CPLString osZipFilename = zipFilename; |
4714 | 0 | CPLFree(zipFilename); |
4715 | 0 | zipFilename = nullptr; |
4716 | 0 | if (osZipInFileName.empty()) |
4717 | 0 | { |
4718 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4719 | 0 | "Target filename should be of the form " |
4720 | 0 | "/vsizip/path_to.zip/filename_within_zip"); |
4721 | 0 | return -1; |
4722 | 0 | } |
4723 | | |
4724 | | // Invalidate cached file list. |
4725 | 0 | auto oIterFileList = oFileList.find(osZipFilename); |
4726 | 0 | if (oIterFileList != oFileList.end()) |
4727 | 0 | { |
4728 | 0 | oFileList.erase(oIterFileList); |
4729 | 0 | } |
4730 | |
|
4731 | 0 | const auto oIter = oMapZipWriteHandles.find(osZipFilename); |
4732 | 0 | if (oIter != oMapZipWriteHandles.end()) |
4733 | 0 | { |
4734 | 0 | VSIZipWriteHandle *poZIPHandle = oIter->second; |
4735 | |
|
4736 | 0 | if (poZIPHandle->GetChildInWriting() != nullptr) |
4737 | 0 | { |
4738 | 0 | CPLError(CE_Failure, CPLE_AppDefined, |
4739 | 0 | "Cannot create %s while another file is being " |
4740 | 0 | "written in the .zip", |
4741 | 0 | osZipInFileName.c_str()); |
4742 | 0 | return -1; |
4743 | 0 | } |
4744 | | |
4745 | 0 | if (CPLAddFileInZip(poZIPHandle->GetHandle(), osZipInFileName.c_str(), |
4746 | 0 | pszSource, fpSource, papszOptions, pProgressFunc, |
4747 | 0 | pProgressData) != CE_None) |
4748 | 0 | { |
4749 | 0 | return -1; |
4750 | 0 | } |
4751 | 0 | return 0; |
4752 | 0 | } |
4753 | 0 | else |
4754 | 0 | { |
4755 | 0 | CPLStringList aosOptionsCreateZip; |
4756 | 0 | VSIStatBufL sBuf; |
4757 | 0 | if (VSIStatExL(osZipFilename, &sBuf, VSI_STAT_EXISTS_FLAG) == 0) |
4758 | 0 | aosOptionsCreateZip.SetNameValue("APPEND", "TRUE"); |
4759 | |
|
4760 | 0 | void *hZIP = CPLCreateZip(osZipFilename, aosOptionsCreateZip.List()); |
4761 | |
|
4762 | 0 | if (hZIP == nullptr) |
4763 | 0 | return -1; |
4764 | | |
4765 | 0 | if (CPLAddFileInZip(hZIP, osZipInFileName.c_str(), pszSource, fpSource, |
4766 | 0 | papszOptions, pProgressFunc, |
4767 | 0 | pProgressData) != CE_None) |
4768 | 0 | { |
4769 | 0 | CPLCloseZip(hZIP); |
4770 | 0 | return -1; |
4771 | 0 | } |
4772 | 0 | CPLCloseZip(hZIP); |
4773 | 0 | return 0; |
4774 | 0 | } |
4775 | 0 | } |
4776 | | |
4777 | | /************************************************************************/ |
4778 | | /* VSIZipWriteHandle() */ |
4779 | | /************************************************************************/ |
4780 | | |
4781 | | VSIZipWriteHandle::VSIZipWriteHandle(VSIZipFilesystemHandler *poFS, void *hZIP, |
4782 | | VSIZipWriteHandle *poParent) |
4783 | 0 | : m_poFS(poFS), m_hZIP(hZIP), m_poParent(poParent) |
4784 | 0 | { |
4785 | 0 | } |
4786 | | |
4787 | | /************************************************************************/ |
4788 | | /* ~VSIZipWriteHandle() */ |
4789 | | /************************************************************************/ |
4790 | | |
4791 | | VSIZipWriteHandle::~VSIZipWriteHandle() |
4792 | 0 | { |
4793 | 0 | VSIZipWriteHandle::Close(); |
4794 | 0 | } |
4795 | | |
4796 | | /************************************************************************/ |
4797 | | /* Seek() */ |
4798 | | /************************************************************************/ |
4799 | | |
4800 | | int VSIZipWriteHandle::Seek(vsi_l_offset nOffset, int nWhence) |
4801 | 0 | { |
4802 | 0 | if (nOffset == 0 && (nWhence == SEEK_END || nWhence == SEEK_CUR)) |
4803 | 0 | return 0; |
4804 | 0 | if (nOffset == nCurOffset && nWhence == SEEK_SET) |
4805 | 0 | return 0; |
4806 | | |
4807 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
4808 | 0 | "VSIFSeekL() is not supported on writable Zip files"); |
4809 | 0 | return -1; |
4810 | 0 | } |
4811 | | |
4812 | | /************************************************************************/ |
4813 | | /* Tell() */ |
4814 | | /************************************************************************/ |
4815 | | |
4816 | | vsi_l_offset VSIZipWriteHandle::Tell() |
4817 | 0 | { |
4818 | 0 | return nCurOffset; |
4819 | 0 | } |
4820 | | |
4821 | | /************************************************************************/ |
4822 | | /* Read() */ |
4823 | | /************************************************************************/ |
4824 | | |
4825 | | size_t VSIZipWriteHandle::Read(void * /* pBuffer */, size_t /* nSize */, |
4826 | | size_t /* nMemb */) |
4827 | 0 | { |
4828 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
4829 | 0 | "VSIFReadL() is not supported on writable Zip files"); |
4830 | 0 | return 0; |
4831 | 0 | } |
4832 | | |
4833 | | /************************************************************************/ |
4834 | | /* Write() */ |
4835 | | /************************************************************************/ |
4836 | | |
4837 | | size_t VSIZipWriteHandle::Write(const void *pBuffer, size_t nSize, size_t nMemb) |
4838 | 0 | { |
4839 | 0 | if (m_poParent == nullptr) |
4840 | 0 | { |
4841 | 0 | CPLError(CE_Failure, CPLE_NotSupported, |
4842 | 0 | "VSIFWriteL() is not supported on " |
4843 | 0 | "main Zip file or closed subfiles"); |
4844 | 0 | return 0; |
4845 | 0 | } |
4846 | | |
4847 | 0 | const GByte *pabyBuffer = static_cast<const GByte *>(pBuffer); |
4848 | 0 | size_t nBytesToWrite = nSize * nMemb; |
4849 | 0 | size_t nWritten = 0; |
4850 | 0 | while (nWritten < nBytesToWrite) |
4851 | 0 | { |
4852 | 0 | int nToWrite = static_cast<int>( |
4853 | 0 | std::min(static_cast<size_t>(INT_MAX), nBytesToWrite)); |
4854 | 0 | if (CPLWriteFileInZip(m_poParent->m_hZIP, pabyBuffer, nToWrite) != |
4855 | 0 | CE_None) |
4856 | 0 | return 0; |
4857 | 0 | nWritten += nToWrite; |
4858 | 0 | pabyBuffer += nToWrite; |
4859 | 0 | } |
4860 | | |
4861 | 0 | nCurOffset += nSize * nMemb; |
4862 | |
|
4863 | 0 | return nMemb; |
4864 | 0 | } |
4865 | | |
4866 | | /************************************************************************/ |
4867 | | /* Flush() */ |
4868 | | /************************************************************************/ |
4869 | | |
4870 | | int VSIZipWriteHandle::Flush() |
4871 | 0 | { |
4872 | | /*CPLError(CE_Failure, CPLE_NotSupported, |
4873 | | "VSIFFlushL() is not supported on writable Zip files");*/ |
4874 | 0 | return 0; |
4875 | 0 | } |
4876 | | |
4877 | | /************************************************************************/ |
4878 | | /* Close() */ |
4879 | | /************************************************************************/ |
4880 | | |
4881 | | int VSIZipWriteHandle::Close() |
4882 | 0 | { |
4883 | 0 | int nRet = 0; |
4884 | 0 | if (m_poParent) |
4885 | 0 | { |
4886 | 0 | CPLCloseFileInZip(m_poParent->m_hZIP); |
4887 | 0 | m_poParent->poChildInWriting = nullptr; |
4888 | 0 | if (bAutoDeleteParent) |
4889 | 0 | { |
4890 | 0 | if (m_poParent->Close() != 0) |
4891 | 0 | nRet = -1; |
4892 | 0 | delete m_poParent; |
4893 | 0 | } |
4894 | 0 | m_poParent = nullptr; |
4895 | 0 | } |
4896 | 0 | if (poChildInWriting) |
4897 | 0 | { |
4898 | 0 | if (poChildInWriting->Close() != 0) |
4899 | 0 | nRet = -1; |
4900 | 0 | poChildInWriting = nullptr; |
4901 | 0 | } |
4902 | 0 | if (m_hZIP) |
4903 | 0 | { |
4904 | 0 | if (CPLCloseZip(m_hZIP) != CE_None) |
4905 | 0 | nRet = -1; |
4906 | 0 | m_hZIP = nullptr; |
4907 | |
|
4908 | 0 | m_poFS->RemoveFromMap(this); |
4909 | 0 | } |
4910 | |
|
4911 | 0 | return nRet; |
4912 | 0 | } |
4913 | | |
4914 | | /************************************************************************/ |
4915 | | /* StopCurrentFile() */ |
4916 | | /************************************************************************/ |
4917 | | |
4918 | | void VSIZipWriteHandle::StopCurrentFile() |
4919 | 0 | { |
4920 | 0 | if (poChildInWriting) |
4921 | 0 | poChildInWriting->Close(); |
4922 | 0 | poChildInWriting = nullptr; |
4923 | 0 | } |
4924 | | |
4925 | | /************************************************************************/ |
4926 | | /* StartNewFile() */ |
4927 | | /************************************************************************/ |
4928 | | |
4929 | | void VSIZipWriteHandle::StartNewFile(VSIZipWriteHandle *poSubFile) |
4930 | 0 | { |
4931 | 0 | poChildInWriting = poSubFile; |
4932 | 0 | } |
4933 | | |
4934 | | //! @endcond |
4935 | | |
4936 | | /************************************************************************/ |
4937 | | /* VSIInstallZipFileHandler() */ |
4938 | | /************************************************************************/ |
4939 | | |
4940 | | /*! |
4941 | | \brief Install ZIP file system handler. |
4942 | | |
4943 | | A special file handler is installed that allows reading on-the-fly in ZIP |
4944 | | (.zip) archives. |
4945 | | |
4946 | | All portions of the file system underneath the base path "/vsizip/" will be |
4947 | | handled by this driver. |
4948 | | |
4949 | | \verbatim embed:rst |
4950 | | See :ref:`/vsizip/ documentation <vsizip>` |
4951 | | \endverbatim |
4952 | | |
4953 | | */ |
4954 | | |
4955 | | void VSIInstallZipFileHandler() |
4956 | 3 | { |
4957 | 3 | VSIFileManager::InstallHandler("/vsizip/", new VSIZipFilesystemHandler()); |
4958 | 3 | } |
4959 | | |
4960 | | /************************************************************************/ |
4961 | | /* CPLZLibDeflate() */ |
4962 | | /************************************************************************/ |
4963 | | |
4964 | | /** |
4965 | | * \brief Compress a buffer with ZLib compression. |
4966 | | * |
4967 | | * @param ptr input buffer. |
4968 | | * @param nBytes size of input buffer in bytes. |
4969 | | * @param nLevel ZLib compression level (-1 for default). |
4970 | | * @param outptr output buffer, or NULL to let the function allocate it. |
4971 | | * @param nOutAvailableBytes size of output buffer if provided, or ignored. |
4972 | | * @param pnOutBytes pointer to a size_t, where to store the size of the |
4973 | | * output buffer. |
4974 | | * |
4975 | | * @return the output buffer (to be freed with VSIFree() if not provided) |
4976 | | * or NULL in case of error. |
4977 | | * |
4978 | | */ |
4979 | | |
4980 | | void *CPLZLibDeflate(const void *ptr, size_t nBytes, int nLevel, void *outptr, |
4981 | | size_t nOutAvailableBytes, size_t *pnOutBytes) |
4982 | 0 | { |
4983 | 0 | if (pnOutBytes != nullptr) |
4984 | 0 | *pnOutBytes = 0; |
4985 | |
|
4986 | 0 | size_t nTmpSize = 0; |
4987 | 0 | void *pTmp; |
4988 | | #ifdef HAVE_LIBDEFLATE |
4989 | | struct libdeflate_compressor *enc = |
4990 | | libdeflate_alloc_compressor(nLevel < 0 ? 7 : nLevel); |
4991 | | if (enc == nullptr) |
4992 | | { |
4993 | | return nullptr; |
4994 | | } |
4995 | | #endif |
4996 | 0 | if (outptr == nullptr) |
4997 | 0 | { |
4998 | | #ifdef HAVE_LIBDEFLATE |
4999 | | nTmpSize = libdeflate_zlib_compress_bound(enc, nBytes); |
5000 | | #else |
5001 | 0 | nTmpSize = 32 + nBytes * 2; |
5002 | 0 | #endif |
5003 | 0 | pTmp = VSIMalloc(nTmpSize); |
5004 | 0 | if (pTmp == nullptr) |
5005 | 0 | { |
5006 | | #ifdef HAVE_LIBDEFLATE |
5007 | | libdeflate_free_compressor(enc); |
5008 | | #endif |
5009 | 0 | return nullptr; |
5010 | 0 | } |
5011 | 0 | } |
5012 | 0 | else |
5013 | 0 | { |
5014 | 0 | pTmp = outptr; |
5015 | 0 | nTmpSize = nOutAvailableBytes; |
5016 | 0 | } |
5017 | | |
5018 | | #ifdef HAVE_LIBDEFLATE |
5019 | | size_t nCompressedBytes = |
5020 | | libdeflate_zlib_compress(enc, ptr, nBytes, pTmp, nTmpSize); |
5021 | | libdeflate_free_compressor(enc); |
5022 | | if (nCompressedBytes == 0) |
5023 | | { |
5024 | | if (pTmp != outptr) |
5025 | | VSIFree(pTmp); |
5026 | | return nullptr; |
5027 | | } |
5028 | | if (pnOutBytes != nullptr) |
5029 | | *pnOutBytes = nCompressedBytes; |
5030 | | #else |
5031 | 0 | z_stream strm; |
5032 | 0 | strm.zalloc = nullptr; |
5033 | 0 | strm.zfree = nullptr; |
5034 | 0 | strm.opaque = nullptr; |
5035 | 0 | int ret = deflateInit(&strm, nLevel < 0 ? Z_DEFAULT_COMPRESSION : nLevel); |
5036 | 0 | if (ret != Z_OK) |
5037 | 0 | { |
5038 | 0 | if (pTmp != outptr) |
5039 | 0 | VSIFree(pTmp); |
5040 | 0 | return nullptr; |
5041 | 0 | } |
5042 | | |
5043 | 0 | strm.avail_in = static_cast<uInt>(nBytes); |
5044 | 0 | strm.next_in = reinterpret_cast<Bytef *>(const_cast<void *>(ptr)); |
5045 | 0 | strm.avail_out = static_cast<uInt>(nTmpSize); |
5046 | 0 | strm.next_out = reinterpret_cast<Bytef *>(pTmp); |
5047 | 0 | ret = deflate(&strm, Z_FINISH); |
5048 | 0 | if (ret != Z_STREAM_END) |
5049 | 0 | { |
5050 | 0 | if (pTmp != outptr) |
5051 | 0 | VSIFree(pTmp); |
5052 | 0 | return nullptr; |
5053 | 0 | } |
5054 | 0 | if (pnOutBytes != nullptr) |
5055 | 0 | *pnOutBytes = nTmpSize - strm.avail_out; |
5056 | 0 | deflateEnd(&strm); |
5057 | 0 | #endif |
5058 | |
|
5059 | 0 | return pTmp; |
5060 | 0 | } |
5061 | | |
5062 | | /************************************************************************/ |
5063 | | /* CPLZLibInflate() */ |
5064 | | /************************************************************************/ |
5065 | | |
5066 | | /** |
5067 | | * \brief Uncompress a buffer compressed with ZLib compression. |
5068 | | * |
5069 | | * @param ptr input buffer. |
5070 | | * @param nBytes size of input buffer in bytes. |
5071 | | * @param outptr output buffer, or NULL to let the function allocate it. |
5072 | | * @param nOutAvailableBytes size of output buffer if provided, or ignored. |
5073 | | * @param pnOutBytes pointer to a size_t, where to store the size of the |
5074 | | * output buffer. |
5075 | | * |
5076 | | * @return the output buffer (to be freed with VSIFree() if not provided) |
5077 | | * or NULL in case of error. |
5078 | | * |
5079 | | */ |
5080 | | |
5081 | | void *CPLZLibInflate(const void *ptr, size_t nBytes, void *outptr, |
5082 | | size_t nOutAvailableBytes, size_t *pnOutBytes) |
5083 | 0 | { |
5084 | 0 | return CPLZLibInflateEx(ptr, nBytes, outptr, nOutAvailableBytes, false, |
5085 | 0 | pnOutBytes); |
5086 | 0 | } |
5087 | | |
5088 | | /************************************************************************/ |
5089 | | /* CPLZLibInflateEx() */ |
5090 | | /************************************************************************/ |
5091 | | |
5092 | | /** |
5093 | | * \brief Uncompress a buffer compressed with ZLib compression. |
5094 | | * |
5095 | | * @param ptr input buffer. |
5096 | | * @param nBytes size of input buffer in bytes. |
5097 | | * @param outptr output buffer, or NULL to let the function allocate it. |
5098 | | * @param nOutAvailableBytes size of output buffer if provided, or ignored. |
5099 | | * @param bAllowResizeOutptr whether the function is allowed to grow outptr |
5100 | | * (using VSIRealloc) if its initial capacity |
5101 | | * provided by nOutAvailableBytes is not |
5102 | | * large enough. Ignored if outptr is NULL. |
5103 | | * @param pnOutBytes pointer to a size_t, where to store the size of the |
5104 | | * output buffer. |
5105 | | * |
5106 | | * @return the output buffer (to be freed with VSIFree() if not provided) |
5107 | | * or NULL in case of error. If bAllowResizeOutptr is set to true, |
5108 | | * only the returned pointer should be freed by the caller, as outptr |
5109 | | * might have been reallocated or freed. |
5110 | | * |
5111 | | * @since GDAL 3.9.0 |
5112 | | */ |
5113 | | |
5114 | | void *CPLZLibInflateEx(const void *ptr, size_t nBytes, void *outptr, |
5115 | | size_t nOutAvailableBytes, bool bAllowResizeOutptr, |
5116 | | size_t *pnOutBytes) |
5117 | 0 | { |
5118 | 0 | if (pnOutBytes != nullptr) |
5119 | 0 | *pnOutBytes = 0; |
5120 | 0 | char *pszReallocatableBuf = nullptr; |
5121 | |
|
5122 | | #ifdef HAVE_LIBDEFLATE |
5123 | | if (outptr) |
5124 | | { |
5125 | | struct libdeflate_decompressor *dec = libdeflate_alloc_decompressor(); |
5126 | | if (dec == nullptr) |
5127 | | { |
5128 | | if (bAllowResizeOutptr) |
5129 | | VSIFree(outptr); |
5130 | | return nullptr; |
5131 | | } |
5132 | | enum libdeflate_result res; |
5133 | | size_t nOutBytes = 0; |
5134 | | if (nBytes > 2 && static_cast<const GByte *>(ptr)[0] == 0x1F && |
5135 | | static_cast<const GByte *>(ptr)[1] == 0x8B) |
5136 | | { |
5137 | | res = libdeflate_gzip_decompress(dec, ptr, nBytes, outptr, |
5138 | | nOutAvailableBytes, &nOutBytes); |
5139 | | } |
5140 | | else |
5141 | | { |
5142 | | res = libdeflate_zlib_decompress(dec, ptr, nBytes, outptr, |
5143 | | nOutAvailableBytes, &nOutBytes); |
5144 | | } |
5145 | | if (pnOutBytes) |
5146 | | *pnOutBytes = nOutBytes; |
5147 | | libdeflate_free_decompressor(dec); |
5148 | | if (res == LIBDEFLATE_INSUFFICIENT_SPACE && bAllowResizeOutptr) |
5149 | | { |
5150 | | if (nOutAvailableBytes > |
5151 | | (std::numeric_limits<size_t>::max() - 1) / 2) |
5152 | | { |
5153 | | VSIFree(outptr); |
5154 | | return nullptr; |
5155 | | } |
5156 | | size_t nOutBufSize = nOutAvailableBytes * 2; |
5157 | | pszReallocatableBuf = static_cast<char *>( |
5158 | | VSI_REALLOC_VERBOSE(outptr, nOutBufSize + 1)); |
5159 | | if (!pszReallocatableBuf) |
5160 | | { |
5161 | | VSIFree(outptr); |
5162 | | return nullptr; |
5163 | | } |
5164 | | outptr = nullptr; |
5165 | | nOutAvailableBytes = nOutBufSize; |
5166 | | } |
5167 | | else if (res != LIBDEFLATE_SUCCESS) |
5168 | | { |
5169 | | if (bAllowResizeOutptr) |
5170 | | VSIFree(outptr); |
5171 | | return nullptr; |
5172 | | } |
5173 | | else |
5174 | | { |
5175 | | // Nul-terminate if possible. |
5176 | | if (nOutBytes < nOutAvailableBytes) |
5177 | | { |
5178 | | static_cast<char *>(outptr)[nOutBytes] = '\0'; |
5179 | | } |
5180 | | return outptr; |
5181 | | } |
5182 | | } |
5183 | | #endif |
5184 | |
|
5185 | 0 | z_stream strm; |
5186 | 0 | memset(&strm, 0, sizeof(strm)); |
5187 | 0 | strm.zalloc = nullptr; |
5188 | 0 | strm.zfree = nullptr; |
5189 | 0 | strm.opaque = nullptr; |
5190 | 0 | int ret; |
5191 | | // MAX_WBITS + 32 mode which detects automatically gzip vs zlib |
5192 | | // encapsulation seems to be broken with |
5193 | | // /opt/intel/oneapi/intelpython/latest/lib/libz.so.1 from |
5194 | | // intel/oneapi-basekit Docker image |
5195 | 0 | if (nBytes > 2 && static_cast<const GByte *>(ptr)[0] == 0x1F && |
5196 | 0 | static_cast<const GByte *>(ptr)[1] == 0x8B) |
5197 | 0 | { |
5198 | 0 | ret = inflateInit2(&strm, MAX_WBITS + 16); // gzip |
5199 | 0 | } |
5200 | 0 | else |
5201 | 0 | { |
5202 | 0 | ret = inflateInit2(&strm, MAX_WBITS); // zlib |
5203 | 0 | } |
5204 | 0 | if (ret != Z_OK) |
5205 | 0 | { |
5206 | 0 | if (bAllowResizeOutptr) |
5207 | 0 | VSIFree(outptr); |
5208 | 0 | VSIFree(pszReallocatableBuf); |
5209 | 0 | return nullptr; |
5210 | 0 | } |
5211 | | |
5212 | 0 | size_t nOutBufSize = 0; |
5213 | 0 | char *pszOutBuf = nullptr; |
5214 | |
|
5215 | | #ifdef HAVE_LIBDEFLATE |
5216 | | if (pszReallocatableBuf) |
5217 | | { |
5218 | | pszOutBuf = pszReallocatableBuf; |
5219 | | nOutBufSize = nOutAvailableBytes; |
5220 | | } |
5221 | | else |
5222 | | #endif |
5223 | 0 | if (!outptr) |
5224 | 0 | { |
5225 | 0 | if (nBytes > (std::numeric_limits<size_t>::max() - 1) / 2) |
5226 | 0 | { |
5227 | 0 | inflateEnd(&strm); |
5228 | 0 | return nullptr; |
5229 | 0 | } |
5230 | 0 | nOutBufSize = 2 * nBytes + 1; |
5231 | 0 | pszOutBuf = static_cast<char *>(VSI_MALLOC_VERBOSE(nOutBufSize)); |
5232 | 0 | if (pszOutBuf == nullptr) |
5233 | 0 | { |
5234 | 0 | inflateEnd(&strm); |
5235 | 0 | return nullptr; |
5236 | 0 | } |
5237 | 0 | pszReallocatableBuf = pszOutBuf; |
5238 | 0 | bAllowResizeOutptr = true; |
5239 | 0 | } |
5240 | 0 | #ifndef HAVE_LIBDEFLATE |
5241 | 0 | else |
5242 | 0 | { |
5243 | 0 | pszOutBuf = static_cast<char *>(outptr); |
5244 | 0 | nOutBufSize = nOutAvailableBytes; |
5245 | 0 | if (bAllowResizeOutptr) |
5246 | 0 | pszReallocatableBuf = pszOutBuf; |
5247 | 0 | } |
5248 | 0 | #endif |
5249 | | |
5250 | 0 | strm.next_in = static_cast<Bytef *>(const_cast<void *>(ptr)); |
5251 | 0 | strm.next_out = reinterpret_cast<Bytef *>(pszOutBuf); |
5252 | 0 | size_t nInBytesRemaining = nBytes; |
5253 | 0 | size_t nOutBytesRemaining = nOutBufSize; |
5254 | |
|
5255 | 0 | while (true) |
5256 | 0 | { |
5257 | 0 | strm.avail_in = static_cast<uInt>(std::min<size_t>( |
5258 | 0 | nInBytesRemaining, std::numeric_limits<uInt>::max())); |
5259 | 0 | const auto avail_in_before = strm.avail_in; |
5260 | 0 | strm.avail_out = static_cast<uInt>(std::min<size_t>( |
5261 | 0 | nOutBytesRemaining, std::numeric_limits<uInt>::max())); |
5262 | 0 | const auto avail_out_before = strm.avail_out; |
5263 | 0 | ret = inflate(&strm, Z_FINISH); |
5264 | 0 | nInBytesRemaining -= (avail_in_before - strm.avail_in); |
5265 | 0 | nOutBytesRemaining -= (avail_out_before - strm.avail_out); |
5266 | |
|
5267 | 0 | if (ret == Z_BUF_ERROR && strm.avail_out == 0) |
5268 | 0 | { |
5269 | | #ifdef HAVE_LIBDEFLATE |
5270 | | CPLAssert(bAllowResizeOutptr); |
5271 | | #else |
5272 | 0 | if (!bAllowResizeOutptr) |
5273 | 0 | { |
5274 | 0 | VSIFree(pszReallocatableBuf); |
5275 | 0 | inflateEnd(&strm); |
5276 | 0 | return nullptr; |
5277 | 0 | } |
5278 | 0 | #endif |
5279 | | |
5280 | 0 | const size_t nAlreadyWritten = nOutBufSize - nOutBytesRemaining; |
5281 | 0 | if (nOutBufSize > (std::numeric_limits<size_t>::max() - 1) / 2) |
5282 | 0 | { |
5283 | 0 | VSIFree(pszReallocatableBuf); |
5284 | 0 | inflateEnd(&strm); |
5285 | 0 | return nullptr; |
5286 | 0 | } |
5287 | 0 | nOutBufSize = nOutBufSize * 2 + 1; |
5288 | 0 | char *pszNew = static_cast<char *>( |
5289 | 0 | VSI_REALLOC_VERBOSE(pszReallocatableBuf, nOutBufSize)); |
5290 | 0 | if (!pszNew) |
5291 | 0 | { |
5292 | 0 | VSIFree(pszReallocatableBuf); |
5293 | 0 | inflateEnd(&strm); |
5294 | 0 | return nullptr; |
5295 | 0 | } |
5296 | 0 | pszOutBuf = pszNew; |
5297 | 0 | pszReallocatableBuf = pszOutBuf; |
5298 | 0 | nOutBytesRemaining = nOutBufSize - nAlreadyWritten; |
5299 | 0 | strm.next_out = |
5300 | 0 | reinterpret_cast<Bytef *>(pszOutBuf + nAlreadyWritten); |
5301 | 0 | } |
5302 | 0 | else if (ret != Z_OK || nInBytesRemaining == 0) |
5303 | 0 | break; |
5304 | 0 | } |
5305 | | |
5306 | 0 | if (ret == Z_OK || ret == Z_STREAM_END) |
5307 | 0 | { |
5308 | 0 | size_t nOutBytes = nOutBufSize - nOutBytesRemaining; |
5309 | | // Nul-terminate if possible. |
5310 | 0 | if (nOutBytes < nOutBufSize) |
5311 | 0 | { |
5312 | 0 | pszOutBuf[nOutBytes] = '\0'; |
5313 | 0 | } |
5314 | 0 | inflateEnd(&strm); |
5315 | 0 | if (pnOutBytes != nullptr) |
5316 | 0 | *pnOutBytes = nOutBytes; |
5317 | 0 | return pszOutBuf; |
5318 | 0 | } |
5319 | 0 | else |
5320 | 0 | { |
5321 | 0 | VSIFree(pszReallocatableBuf); |
5322 | 0 | inflateEnd(&strm); |
5323 | 0 | return nullptr; |
5324 | 0 | } |
5325 | 0 | } |