diff options
author | Anton Samokhvalov <pg83@yandex.ru> | 2022-02-10 16:45:15 +0300 |
---|---|---|
committer | Daniil Cherednik <dcherednik@yandex-team.ru> | 2022-02-10 16:45:15 +0300 |
commit | 72cb13b4aff9bc9cf22e49251bc8fd143f82538f (patch) | |
tree | da2c34829458c7d4e74bdfbdf85dff449e9e7fb8 /contrib/libs/snappy | |
parent | 778e51ba091dc39e7b7fcab2b9cf4dbedfb6f2b5 (diff) | |
download | ydb-72cb13b4aff9bc9cf22e49251bc8fd143f82538f.tar.gz |
Restoring authorship annotation for Anton Samokhvalov <pg83@yandex.ru>. Commit 1 of 2.
Diffstat (limited to 'contrib/libs/snappy')
-rw-r--r-- | contrib/libs/snappy/snappy-c.cc | 180 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-c.h | 268 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-internal.h | 232 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-sinksource.cc | 142 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-sinksource.h | 262 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-stubs-internal.cc | 78 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-stubs-internal.h | 720 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy-stubs-public.h | 92 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy.cc | 1374 | ||||
-rw-r--r-- | contrib/libs/snappy/snappy.h | 266 | ||||
-rw-r--r-- | contrib/libs/snappy/ya.make | 18 |
11 files changed, 1816 insertions, 1816 deletions
diff --git a/contrib/libs/snappy/snappy-c.cc b/contrib/libs/snappy/snappy-c.cc index 473a0b0978..0cb59c7296 100644 --- a/contrib/libs/snappy/snappy-c.cc +++ b/contrib/libs/snappy/snappy-c.cc @@ -1,90 +1,90 @@ -// Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - -#include "snappy.h" -#include "snappy-c.h" - -extern "C" { - -snappy_status snappy_compress(const char* input, - size_t input_length, - char* compressed, - size_t *compressed_length) { - if (*compressed_length < snappy_max_compressed_length(input_length)) { - return SNAPPY_BUFFER_TOO_SMALL; - } - snappy::RawCompress(input, input_length, compressed, compressed_length); - return SNAPPY_OK; -} - -snappy_status snappy_uncompress(const char* compressed, - size_t compressed_length, - char* uncompressed, - size_t* uncompressed_length) { - size_t real_uncompressed_length; - if (!snappy::GetUncompressedLength(compressed, - compressed_length, - &real_uncompressed_length)) { - return SNAPPY_INVALID_INPUT; - } - if (*uncompressed_length < real_uncompressed_length) { - return SNAPPY_BUFFER_TOO_SMALL; - } - if (!snappy::RawUncompress(compressed, compressed_length, uncompressed)) { - return SNAPPY_INVALID_INPUT; - } - *uncompressed_length = real_uncompressed_length; - return SNAPPY_OK; -} - -size_t snappy_max_compressed_length(size_t source_length) { - return snappy::MaxCompressedLength(source_length); -} - -snappy_status snappy_uncompressed_length(const char *compressed, - size_t compressed_length, - size_t *result) { - if (snappy::GetUncompressedLength(compressed, - compressed_length, - result)) { - return SNAPPY_OK; - } else { - return SNAPPY_INVALID_INPUT; - } -} - -snappy_status snappy_validate_compressed_buffer(const char *compressed, - size_t compressed_length) { - if (snappy::IsValidCompressedBuffer(compressed, compressed_length)) { - return SNAPPY_OK; - } else { - return SNAPPY_INVALID_INPUT; - } -} - -} // extern "C" +// Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include "snappy.h" +#include "snappy-c.h" + +extern "C" { + +snappy_status snappy_compress(const char* input, + size_t input_length, + char* compressed, + size_t *compressed_length) { + if (*compressed_length < snappy_max_compressed_length(input_length)) { + return SNAPPY_BUFFER_TOO_SMALL; + } + snappy::RawCompress(input, input_length, compressed, compressed_length); + return SNAPPY_OK; +} + +snappy_status snappy_uncompress(const char* compressed, + size_t compressed_length, + char* uncompressed, + size_t* uncompressed_length) { + size_t real_uncompressed_length; + if (!snappy::GetUncompressedLength(compressed, + compressed_length, + &real_uncompressed_length)) { + return SNAPPY_INVALID_INPUT; + } + if (*uncompressed_length < real_uncompressed_length) { + return SNAPPY_BUFFER_TOO_SMALL; + } + if (!snappy::RawUncompress(compressed, compressed_length, uncompressed)) { + return SNAPPY_INVALID_INPUT; + } + *uncompressed_length = real_uncompressed_length; + return SNAPPY_OK; +} + +size_t snappy_max_compressed_length(size_t source_length) { + return snappy::MaxCompressedLength(source_length); +} + +snappy_status snappy_uncompressed_length(const char *compressed, + size_t compressed_length, + size_t *result) { + if (snappy::GetUncompressedLength(compressed, + compressed_length, + result)) { + return SNAPPY_OK; + } else { + return SNAPPY_INVALID_INPUT; + } +} + +snappy_status snappy_validate_compressed_buffer(const char *compressed, + size_t compressed_length) { + if (snappy::IsValidCompressedBuffer(compressed, compressed_length)) { + return SNAPPY_OK; + } else { + return SNAPPY_INVALID_INPUT; + } +} + +} // extern "C" diff --git a/contrib/libs/snappy/snappy-c.h b/contrib/libs/snappy/snappy-c.h index 32aa0c6b8b..826bccfded 100644 --- a/contrib/libs/snappy/snappy-c.h +++ b/contrib/libs/snappy/snappy-c.h @@ -1,138 +1,138 @@ -/* - * Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions are - * met: - * - * * Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * * Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following disclaimer - * in the documentation and/or other materials provided with the - * distribution. - * * Neither the name of Google Inc. nor the names of its - * contributors may be used to endorse or promote products derived from - * this software without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS - * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT - * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR - * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT - * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, - * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT - * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, - * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY - * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT - * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE - * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - * - * Plain C interface (a wrapper around the C++ implementation). - */ - +/* + * Copyright 2011 Martin Gieseking <martin.gieseking@uos.de>. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are + * met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following disclaimer + * in the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Google Inc. nor the names of its + * contributors may be used to endorse or promote products derived from + * this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * Plain C interface (a wrapper around the C++ implementation). + */ + #ifndef THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_C_H_ #define THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_C_H_ - -#ifdef __cplusplus -extern "C" { -#endif - -#include <stddef.h> - -/* - * Return values; see the documentation for each function to know - * what each can return. - */ -typedef enum { - SNAPPY_OK = 0, - SNAPPY_INVALID_INPUT = 1, + +#ifdef __cplusplus +extern "C" { +#endif + +#include <stddef.h> + +/* + * Return values; see the documentation for each function to know + * what each can return. + */ +typedef enum { + SNAPPY_OK = 0, + SNAPPY_INVALID_INPUT = 1, SNAPPY_BUFFER_TOO_SMALL = 2 -} snappy_status; - -/* - * Takes the data stored in "input[0..input_length-1]" and stores - * it in the array pointed to by "compressed". - * - * <compressed_length> signals the space available in "compressed". - * If it is not at least equal to "snappy_max_compressed_length(input_length)", - * SNAPPY_BUFFER_TOO_SMALL is returned. After successful compression, - * <compressed_length> contains the true length of the compressed output, - * and SNAPPY_OK is returned. - * - * Example: - * size_t output_length = snappy_max_compressed_length(input_length); - * char* output = (char*)malloc(output_length); - * if (snappy_compress(input, input_length, output, &output_length) - * == SNAPPY_OK) { - * ... Process(output, output_length) ... - * } - * free(output); - */ -snappy_status snappy_compress(const char* input, - size_t input_length, - char* compressed, - size_t* compressed_length); - -/* - * Given data in "compressed[0..compressed_length-1]" generated by - * calling the snappy_compress routine, this routine stores - * the uncompressed data to - * uncompressed[0..uncompressed_length-1]. - * Returns failure (a value not equal to SNAPPY_OK) if the message - * is corrupted and could not be decrypted. - * - * <uncompressed_length> signals the space available in "uncompressed". - * If it is not at least equal to the value returned by - * snappy_uncompressed_length for this stream, SNAPPY_BUFFER_TOO_SMALL - * is returned. After successful decompression, <uncompressed_length> - * contains the true length of the decompressed output. - * - * Example: - * size_t output_length; - * if (snappy_uncompressed_length(input, input_length, &output_length) - * != SNAPPY_OK) { - * ... fail ... - * } - * char* output = (char*)malloc(output_length); - * if (snappy_uncompress(input, input_length, output, &output_length) - * == SNAPPY_OK) { - * ... Process(output, output_length) ... - * } - * free(output); - */ -snappy_status snappy_uncompress(const char* compressed, - size_t compressed_length, - char* uncompressed, - size_t* uncompressed_length); - -/* - * Returns the maximal size of the compressed representation of - * input data that is "source_length" bytes in length. - */ -size_t snappy_max_compressed_length(size_t source_length); - -/* - * REQUIRES: "compressed[]" was produced by snappy_compress() - * Returns SNAPPY_OK and stores the length of the uncompressed data in - * *result normally. Returns SNAPPY_INVALID_INPUT on parsing error. - * This operation takes O(1) time. - */ -snappy_status snappy_uncompressed_length(const char* compressed, - size_t compressed_length, - size_t* result); - -/* - * Check if the contents of "compressed[]" can be uncompressed successfully. - * Does not return the uncompressed data; if so, returns SNAPPY_OK, - * or if not, returns SNAPPY_INVALID_INPUT. - * Takes time proportional to compressed_length, but is usually at least a - * factor of four faster than actual decompression. - */ -snappy_status snappy_validate_compressed_buffer(const char* compressed, - size_t compressed_length); - -#ifdef __cplusplus -} // extern "C" -#endif - +} snappy_status; + +/* + * Takes the data stored in "input[0..input_length-1]" and stores + * it in the array pointed to by "compressed". + * + * <compressed_length> signals the space available in "compressed". + * If it is not at least equal to "snappy_max_compressed_length(input_length)", + * SNAPPY_BUFFER_TOO_SMALL is returned. After successful compression, + * <compressed_length> contains the true length of the compressed output, + * and SNAPPY_OK is returned. + * + * Example: + * size_t output_length = snappy_max_compressed_length(input_length); + * char* output = (char*)malloc(output_length); + * if (snappy_compress(input, input_length, output, &output_length) + * == SNAPPY_OK) { + * ... Process(output, output_length) ... + * } + * free(output); + */ +snappy_status snappy_compress(const char* input, + size_t input_length, + char* compressed, + size_t* compressed_length); + +/* + * Given data in "compressed[0..compressed_length-1]" generated by + * calling the snappy_compress routine, this routine stores + * the uncompressed data to + * uncompressed[0..uncompressed_length-1]. + * Returns failure (a value not equal to SNAPPY_OK) if the message + * is corrupted and could not be decrypted. + * + * <uncompressed_length> signals the space available in "uncompressed". + * If it is not at least equal to the value returned by + * snappy_uncompressed_length for this stream, SNAPPY_BUFFER_TOO_SMALL + * is returned. After successful decompression, <uncompressed_length> + * contains the true length of the decompressed output. + * + * Example: + * size_t output_length; + * if (snappy_uncompressed_length(input, input_length, &output_length) + * != SNAPPY_OK) { + * ... fail ... + * } + * char* output = (char*)malloc(output_length); + * if (snappy_uncompress(input, input_length, output, &output_length) + * == SNAPPY_OK) { + * ... Process(output, output_length) ... + * } + * free(output); + */ +snappy_status snappy_uncompress(const char* compressed, + size_t compressed_length, + char* uncompressed, + size_t* uncompressed_length); + +/* + * Returns the maximal size of the compressed representation of + * input data that is "source_length" bytes in length. + */ +size_t snappy_max_compressed_length(size_t source_length); + +/* + * REQUIRES: "compressed[]" was produced by snappy_compress() + * Returns SNAPPY_OK and stores the length of the uncompressed data in + * *result normally. Returns SNAPPY_INVALID_INPUT on parsing error. + * This operation takes O(1) time. + */ +snappy_status snappy_uncompressed_length(const char* compressed, + size_t compressed_length, + size_t* result); + +/* + * Check if the contents of "compressed[]" can be uncompressed successfully. + * Does not return the uncompressed data; if so, returns SNAPPY_OK, + * or if not, returns SNAPPY_INVALID_INPUT. + * Takes time proportional to compressed_length, but is usually at least a + * factor of four faster than actual decompression. + */ +snappy_status snappy_validate_compressed_buffer(const char* compressed, + size_t compressed_length); + +#ifdef __cplusplus +} // extern "C" +#endif + #endif /* THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_C_H_ */ diff --git a/contrib/libs/snappy/snappy-internal.h b/contrib/libs/snappy/snappy-internal.h index 1e1c307fef..28c179ee46 100644 --- a/contrib/libs/snappy/snappy-internal.h +++ b/contrib/libs/snappy/snappy-internal.h @@ -1,94 +1,94 @@ -// Copyright 2008 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -// -// Internals shared between the Snappy implementation and its unittest. - +// Copyright 2008 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Internals shared between the Snappy implementation and its unittest. + #ifndef THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_ #define THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_ - -#include "snappy-stubs-internal.h" - -namespace snappy { -namespace internal { - + +#include "snappy-stubs-internal.h" + +namespace snappy { +namespace internal { + // Working memory performs a single allocation to hold all scratch space // required for compression. -class WorkingMemory { - public: +class WorkingMemory { + public: explicit WorkingMemory(size_t input_size); ~WorkingMemory(); - - // Allocates and clears a hash table using memory in "*this", - // stores the number of buckets in "*table_size" and returns a pointer to - // the base of the hash table. + + // Allocates and clears a hash table using memory in "*this", + // stores the number of buckets in "*table_size" and returns a pointer to + // the base of the hash table. uint16* GetHashTable(size_t fragment_size, int* table_size) const; char* GetScratchInput() const { return input_; } char* GetScratchOutput() const { return output_; } - - private: + + private: char* mem_; // the allocated memory, never nullptr size_t size_; // the size of the allocated memory, never 0 uint16* table_; // the pointer to the hashtable char* input_; // the pointer to the input scratch buffer char* output_; // the pointer to the output scratch buffer - + // No copying WorkingMemory(const WorkingMemory&); void operator=(const WorkingMemory&); -}; - -// Flat array compression that does not emit the "uncompressed length" -// prefix. Compresses "input" string to the "*op" buffer. -// -// REQUIRES: "input_length <= kBlockSize" -// REQUIRES: "op" points to an array of memory that is at least -// "MaxCompressedLength(input_length)" in size. -// REQUIRES: All elements in "table[0..table_size-1]" are initialized to zero. -// REQUIRES: "table_size" is a power of two -// -// Returns an "end" pointer into "op" buffer. -// "end - op" is the compressed size of "input". -char* CompressFragment(const char* input, - size_t input_length, - char* op, - uint16* table, - const int table_size); - +}; + +// Flat array compression that does not emit the "uncompressed length" +// prefix. Compresses "input" string to the "*op" buffer. +// +// REQUIRES: "input_length <= kBlockSize" +// REQUIRES: "op" points to an array of memory that is at least +// "MaxCompressedLength(input_length)" in size. +// REQUIRES: All elements in "table[0..table_size-1]" are initialized to zero. +// REQUIRES: "table_size" is a power of two +// +// Returns an "end" pointer into "op" buffer. +// "end - op" is the compressed size of "input". +char* CompressFragment(const char* input, + size_t input_length, + char* op, + uint16* table, + const int table_size); + // Find the largest n such that -// -// s1[0,n-1] == s2[0,n-1] -// and n <= (s2_limit - s2). -// +// +// s1[0,n-1] == s2[0,n-1] +// and n <= (s2_limit - s2). +// // Return make_pair(n, n < 8). -// Does not read *s2_limit or beyond. -// Does not read *(s1 + (s2_limit - s2)) or beyond. -// Requires that s2_limit >= s2. -// +// Does not read *s2_limit or beyond. +// Does not read *(s1 + (s2_limit - s2)) or beyond. +// Requires that s2_limit >= s2. +// // Separate implementation for 64-bit, little-endian cpus. #if !defined(SNAPPY_IS_BIG_ENDIAN) && \ (defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM)) @@ -97,7 +97,7 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1, const char* s2_limit) { assert(s2_limit >= s2); size_t matched = 0; - + // This block isn't necessary for correctness; we could just start looping // immediately. As an optimization though, it is useful. It creates some not // uncommon code paths that determine, without extra effort, whether the match @@ -115,59 +115,59 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1, } } - // Find out how long the match is. We loop over the data 64 bits at a - // time until we find a 64-bit block that doesn't match; then we find - // the first non-matching bit and use that to calculate the total - // length of the match. + // Find out how long the match is. We loop over the data 64 bits at a + // time until we find a 64-bit block that doesn't match; then we find + // the first non-matching bit and use that to calculate the total + // length of the match. while (SNAPPY_PREDICT_TRUE(s2 <= s2_limit - 8)) { if (UNALIGNED_LOAD64(s2) == UNALIGNED_LOAD64(s1 + matched)) { - s2 += 8; - matched += 8; - } else { - uint64 x = UNALIGNED_LOAD64(s2) ^ UNALIGNED_LOAD64(s1 + matched); - int matching_bits = Bits::FindLSBSetNonZero64(x); - matched += matching_bits >> 3; + s2 += 8; + matched += 8; + } else { + uint64 x = UNALIGNED_LOAD64(s2) ^ UNALIGNED_LOAD64(s1 + matched); + int matching_bits = Bits::FindLSBSetNonZero64(x); + matched += matching_bits >> 3; assert(matched >= 8); return std::pair<size_t, bool>(matched, false); - } - } + } + } while (SNAPPY_PREDICT_TRUE(s2 < s2_limit)) { if (s1[matched] == *s2) { - ++s2; - ++matched; - } else { + ++s2; + ++matched; + } else { return std::pair<size_t, bool>(matched, matched < 8); - } - } + } + } return std::pair<size_t, bool>(matched, matched < 8); -} -#else +} +#else static inline std::pair<size_t, bool> FindMatchLength(const char* s1, const char* s2, const char* s2_limit) { - // Implementation based on the x86-64 version, above. + // Implementation based on the x86-64 version, above. assert(s2_limit >= s2); - int matched = 0; - - while (s2 <= s2_limit - 4 && - UNALIGNED_LOAD32(s2) == UNALIGNED_LOAD32(s1 + matched)) { - s2 += 4; - matched += 4; - } - if (LittleEndian::IsLittleEndian() && s2 <= s2_limit - 4) { - uint32 x = UNALIGNED_LOAD32(s2) ^ UNALIGNED_LOAD32(s1 + matched); - int matching_bits = Bits::FindLSBSetNonZero(x); - matched += matching_bits >> 3; - } else { - while ((s2 < s2_limit) && (s1[matched] == *s2)) { - ++s2; - ++matched; - } - } + int matched = 0; + + while (s2 <= s2_limit - 4 && + UNALIGNED_LOAD32(s2) == UNALIGNED_LOAD32(s1 + matched)) { + s2 += 4; + matched += 4; + } + if (LittleEndian::IsLittleEndian() && s2 <= s2_limit - 4) { + uint32 x = UNALIGNED_LOAD32(s2) ^ UNALIGNED_LOAD32(s1 + matched); + int matching_bits = Bits::FindLSBSetNonZero(x); + matched += matching_bits >> 3; + } else { + while ((s2 < s2_limit) && (s1[matched] == *s2)) { + ++s2; + ++matched; + } + } return std::pair<size_t, bool>(matched, matched < 8); -} -#endif - +} +#endif + // Lookup tables for decompression code. Give --snappy_dump_decompression_table // to the unit test to recompute char_table. @@ -225,7 +225,7 @@ static const uint16 char_table[256] = { 0x1801, 0x0f0a, 0x103f, 0x203f, 0x2001, 0x0f0b, 0x1040, 0x2040 }; -} // end namespace internal -} // end namespace snappy - +} // end namespace internal +} // end namespace snappy + #endif // THIRD_PARTY_SNAPPY_SNAPPY_INTERNAL_H_ diff --git a/contrib/libs/snappy/snappy-sinksource.cc b/contrib/libs/snappy/snappy-sinksource.cc index 369a13215b..42651664bf 100644 --- a/contrib/libs/snappy/snappy-sinksource.cc +++ b/contrib/libs/snappy/snappy-sinksource.cc @@ -1,45 +1,45 @@ -// Copyright 2011 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - -#include <string.h> - -#include "snappy-sinksource.h" - -namespace snappy { - -Source::~Source() { } - -Sink::~Sink() { } - -char* Sink::GetAppendBuffer(size_t length, char* scratch) { - return scratch; -} - +// Copyright 2011 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include <string.h> + +#include "snappy-sinksource.h" + +namespace snappy { + +Source::~Source() { } + +Sink::~Sink() { } + +char* Sink::GetAppendBuffer(size_t length, char* scratch) { + return scratch; +} + char* Sink::GetAppendBufferVariable( size_t min_size, size_t desired_size_hint, char* scratch, size_t scratch_size, size_t* allocated_size) { @@ -55,34 +55,34 @@ void Sink::AppendAndTakeOwnership( (*deleter)(deleter_arg, bytes, n); } -ByteArraySource::~ByteArraySource() { } - -size_t ByteArraySource::Available() const { return left_; } - -const char* ByteArraySource::Peek(size_t* len) { - *len = left_; - return ptr_; -} - -void ByteArraySource::Skip(size_t n) { - left_ -= n; - ptr_ += n; -} - -UncheckedByteArraySink::~UncheckedByteArraySink() { } - -void UncheckedByteArraySink::Append(const char* data, size_t n) { - // Do no copying if the caller filled in the result of GetAppendBuffer() - if (data != dest_) { - memcpy(dest_, data, n); - } - dest_ += n; -} - -char* UncheckedByteArraySink::GetAppendBuffer(size_t len, char* scratch) { - return dest_; -} - +ByteArraySource::~ByteArraySource() { } + +size_t ByteArraySource::Available() const { return left_; } + +const char* ByteArraySource::Peek(size_t* len) { + *len = left_; + return ptr_; +} + +void ByteArraySource::Skip(size_t n) { + left_ -= n; + ptr_ += n; +} + +UncheckedByteArraySink::~UncheckedByteArraySink() { } + +void UncheckedByteArraySink::Append(const char* data, size_t n) { + // Do no copying if the caller filled in the result of GetAppendBuffer() + if (data != dest_) { + memcpy(dest_, data, n); + } + dest_ += n; +} + +char* UncheckedByteArraySink::GetAppendBuffer(size_t len, char* scratch) { + return dest_; +} + void UncheckedByteArraySink::AppendAndTakeOwnership( char* data, size_t n, void (*deleter)(void*, const char*, size_t), @@ -92,7 +92,7 @@ void UncheckedByteArraySink::AppendAndTakeOwnership( (*deleter)(deleter_arg, data, n); } dest_ += n; -} +} char* UncheckedByteArraySink::GetAppendBufferVariable( size_t min_size, size_t desired_size_hint, char* scratch, diff --git a/contrib/libs/snappy/snappy-sinksource.h b/contrib/libs/snappy/snappy-sinksource.h index 8afcdaaa2c..75aa872653 100644 --- a/contrib/libs/snappy/snappy-sinksource.h +++ b/contrib/libs/snappy/snappy-sinksource.h @@ -1,69 +1,69 @@ -// Copyright 2011 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - +// Copyright 2011 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + #ifndef THIRD_PARTY_SNAPPY_SNAPPY_SINKSOURCE_H_ #define THIRD_PARTY_SNAPPY_SNAPPY_SINKSOURCE_H_ - -#include <stddef.h> - -namespace snappy { - -// A Sink is an interface that consumes a sequence of bytes. -class Sink { - public: - Sink() { } - virtual ~Sink(); - - // Append "bytes[0,n-1]" to this. - virtual void Append(const char* bytes, size_t n) = 0; - - // Returns a writable buffer of the specified length for appending. - // May return a pointer to the caller-owned scratch buffer which - // must have at least the indicated length. The returned buffer is - // only valid until the next operation on this Sink. - // - // After writing at most "length" bytes, call Append() with the - // pointer returned from this function and the number of bytes - // written. Many Append() implementations will avoid copying - // bytes if this function returned an internal buffer. - // - // If a non-scratch buffer is returned, the caller may only pass a - // prefix of it to Append(). That is, it is not correct to pass an - // interior pointer of the returned array to Append(). - // - // The default implementation always returns the scratch buffer. - virtual char* GetAppendBuffer(size_t length, char* scratch); - + +#include <stddef.h> + +namespace snappy { + +// A Sink is an interface that consumes a sequence of bytes. +class Sink { + public: + Sink() { } + virtual ~Sink(); + + // Append "bytes[0,n-1]" to this. + virtual void Append(const char* bytes, size_t n) = 0; + + // Returns a writable buffer of the specified length for appending. + // May return a pointer to the caller-owned scratch buffer which + // must have at least the indicated length. The returned buffer is + // only valid until the next operation on this Sink. + // + // After writing at most "length" bytes, call Append() with the + // pointer returned from this function and the number of bytes + // written. Many Append() implementations will avoid copying + // bytes if this function returned an internal buffer. + // + // If a non-scratch buffer is returned, the caller may only pass a + // prefix of it to Append(). That is, it is not correct to pass an + // interior pointer of the returned array to Append(). + // + // The default implementation always returns the scratch buffer. + virtual char* GetAppendBuffer(size_t length, char* scratch); + // For higher performance, Sink implementations can provide custom // AppendAndTakeOwnership() and GetAppendBufferVariable() methods. // These methods can reduce the number of copies done during // compression/decompression. - + // Append "bytes[0,n-1] to the sink. Takes ownership of "bytes" // and calls the deleter function as (*deleter)(deleter_arg, bytes, n) // to free the buffer. deleter function must be non NULL. @@ -101,82 +101,82 @@ class Sink { size_t min_size, size_t desired_size_hint, char* scratch, size_t scratch_size, size_t* allocated_size); - private: - // No copying - Sink(const Sink&); - void operator=(const Sink&); -}; - -// A Source is an interface that yields a sequence of bytes -class Source { - public: - Source() { } - virtual ~Source(); - - // Return the number of bytes left to read from the source - virtual size_t Available() const = 0; - - // Peek at the next flat region of the source. Does not reposition - // the source. The returned region is empty iff Available()==0. - // - // Returns a pointer to the beginning of the region and store its - // length in *len. - // - // The returned region is valid until the next call to Skip() or - // until this object is destroyed, whichever occurs first. - // - // The returned region may be larger than Available() (for example - // if this ByteSource is a view on a substring of a larger source). - // The caller is responsible for ensuring that it only reads the - // Available() bytes. - virtual const char* Peek(size_t* len) = 0; - - // Skip the next n bytes. Invalidates any buffer returned by - // a previous call to Peek(). - // REQUIRES: Available() >= n - virtual void Skip(size_t n) = 0; - - private: - // No copying - Source(const Source&); - void operator=(const Source&); -}; - -// A Source implementation that yields the contents of a flat array -class ByteArraySource : public Source { - public: - ByteArraySource(const char* p, size_t n) : ptr_(p), left_(n) { } - virtual ~ByteArraySource(); - virtual size_t Available() const; - virtual const char* Peek(size_t* len); - virtual void Skip(size_t n); - private: - const char* ptr_; - size_t left_; -}; - -// A Sink implementation that writes to a flat array without any bound checks. -class UncheckedByteArraySink : public Sink { - public: - explicit UncheckedByteArraySink(char* dest) : dest_(dest) { } - virtual ~UncheckedByteArraySink(); - virtual void Append(const char* data, size_t n); - virtual char* GetAppendBuffer(size_t len, char* scratch); + private: + // No copying + Sink(const Sink&); + void operator=(const Sink&); +}; + +// A Source is an interface that yields a sequence of bytes +class Source { + public: + Source() { } + virtual ~Source(); + + // Return the number of bytes left to read from the source + virtual size_t Available() const = 0; + + // Peek at the next flat region of the source. Does not reposition + // the source. The returned region is empty iff Available()==0. + // + // Returns a pointer to the beginning of the region and store its + // length in *len. + // + // The returned region is valid until the next call to Skip() or + // until this object is destroyed, whichever occurs first. + // + // The returned region may be larger than Available() (for example + // if this ByteSource is a view on a substring of a larger source). + // The caller is responsible for ensuring that it only reads the + // Available() bytes. + virtual const char* Peek(size_t* len) = 0; + + // Skip the next n bytes. Invalidates any buffer returned by + // a previous call to Peek(). + // REQUIRES: Available() >= n + virtual void Skip(size_t n) = 0; + + private: + // No copying + Source(const Source&); + void operator=(const Source&); +}; + +// A Source implementation that yields the contents of a flat array +class ByteArraySource : public Source { + public: + ByteArraySource(const char* p, size_t n) : ptr_(p), left_(n) { } + virtual ~ByteArraySource(); + virtual size_t Available() const; + virtual const char* Peek(size_t* len); + virtual void Skip(size_t n); + private: + const char* ptr_; + size_t left_; +}; + +// A Sink implementation that writes to a flat array without any bound checks. +class UncheckedByteArraySink : public Sink { + public: + explicit UncheckedByteArraySink(char* dest) : dest_(dest) { } + virtual ~UncheckedByteArraySink(); + virtual void Append(const char* data, size_t n); + virtual char* GetAppendBuffer(size_t len, char* scratch); virtual char* GetAppendBufferVariable( size_t min_size, size_t desired_size_hint, char* scratch, size_t scratch_size, size_t* allocated_size); virtual void AppendAndTakeOwnership( char* bytes, size_t n, void (*deleter)(void*, const char*, size_t), void *deleter_arg); - - // Return the current output pointer so that a caller can see how - // many bytes were produced. - // Note: this is not a Sink method. - char* CurrentDestination() const { return dest_; } - private: - char* dest_; -}; - + + // Return the current output pointer so that a caller can see how + // many bytes were produced. + // Note: this is not a Sink method. + char* CurrentDestination() const { return dest_; } + private: + char* dest_; +}; + } // namespace snappy - + #endif // THIRD_PARTY_SNAPPY_SNAPPY_SINKSOURCE_H_ diff --git a/contrib/libs/snappy/snappy-stubs-internal.cc b/contrib/libs/snappy/snappy-stubs-internal.cc index 66ed2e9039..3e43a2124b 100644 --- a/contrib/libs/snappy/snappy-stubs-internal.cc +++ b/contrib/libs/snappy/snappy-stubs-internal.cc @@ -1,42 +1,42 @@ -// Copyright 2011 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - -#include <algorithm> +// Copyright 2011 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include <algorithm> #include <string> - -#include "snappy-stubs-internal.h" - -namespace snappy { - + +#include "snappy-stubs-internal.h" + +namespace snappy { + void Varint::Append32(std::string* s, uint32 value) { - char buf[Varint::kMax32]; - const char* p = Varint::Encode32(buf, value); + char buf[Varint::kMax32]; + const char* p = Varint::Encode32(buf, value); s->append(buf, p - buf); -} - -} // namespace snappy +} + +} // namespace snappy diff --git a/contrib/libs/snappy/snappy-stubs-internal.h b/contrib/libs/snappy/snappy-stubs-internal.h index 4854689d17..48b40bac98 100644 --- a/contrib/libs/snappy/snappy-stubs-internal.h +++ b/contrib/libs/snappy/snappy-stubs-internal.h @@ -1,46 +1,46 @@ -// Copyright 2011 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -// -// Various stubs for the open-source version of Snappy. - +// Copyright 2011 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Various stubs for the open-source version of Snappy. + #ifndef THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_ #define THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_ - + #ifdef HAVE_CONFIG_H #include "config.h" -#endif - +#endif + #include <string> - -#include <assert.h> -#include <stdlib.h> -#include <string.h> - + +#include <assert.h> +#include <stdlib.h> +#include <string.h> + #ifdef HAVE_SYS_MMAN_H #include <sys/mman.h> #endif @@ -65,13 +65,13 @@ #define SNAPPY_ANNOTATE_MEMORY_IS_INITIALIZED(address, size) /* empty */ #endif // __has_feature(memory_sanitizer) -#include "snappy-stubs-public.h" - -#if defined(__x86_64__) - -// Enable 64-bit optimized versions of some routines. -#define ARCH_K8 1 - +#include "snappy-stubs-public.h" + +#if defined(__x86_64__) + +// Enable 64-bit optimized versions of some routines. +#define ARCH_K8 1 + #elif defined(__ppc64__) #define ARCH_PPC 1 @@ -80,21 +80,21 @@ #define ARCH_ARM 1 -#endif - -// Needed by OS X, among others. -#ifndef MAP_ANONYMOUS -#define MAP_ANONYMOUS MAP_ANON -#endif - -// The size of an array, if known at compile-time. -// Will give unexpected results if used on a pointer. -// We undefine it first, since some compilers already have a definition. -#ifdef ARRAYSIZE -#undef ARRAYSIZE -#endif -#define ARRAYSIZE(a) (sizeof(a) / sizeof(*(a))) - +#endif + +// Needed by OS X, among others. +#ifndef MAP_ANONYMOUS +#define MAP_ANONYMOUS MAP_ANON +#endif + +// The size of an array, if known at compile-time. +// Will give unexpected results if used on a pointer. +// We undefine it first, since some compilers already have a definition. +#ifdef ARRAYSIZE +#undef ARRAYSIZE +#endif +#define ARRAYSIZE(a) (sizeof(a) / sizeof(*(a))) + // Static prediction hints. #ifdef HAVE_BUILTIN_EXPECT #define SNAPPY_PREDICT_FALSE(x) (__builtin_expect(x, 0)) @@ -103,44 +103,44 @@ #define SNAPPY_PREDICT_FALSE(x) x #define SNAPPY_PREDICT_TRUE(x) x #endif - -// This is only used for recomputing the tag byte table used during -// decompression; for simplicity we just remove it from the open-source -// version (anyone who wants to regenerate it can just do the call -// themselves within main()). -#define DEFINE_bool(flag_name, default_value, description) \ - bool FLAGS_ ## flag_name = default_value -#define DECLARE_bool(flag_name) \ - extern bool FLAGS_ ## flag_name - -namespace snappy { - -static const uint32 kuint32max = static_cast<uint32>(0xFFFFFFFF); -static const int64 kint64max = static_cast<int64>(0x7FFFFFFFFFFFFFFFLL); - -// Potentially unaligned loads and stores. - + +// This is only used for recomputing the tag byte table used during +// decompression; for simplicity we just remove it from the open-source +// version (anyone who wants to regenerate it can just do the call +// themselves within main()). +#define DEFINE_bool(flag_name, default_value, description) \ + bool FLAGS_ ## flag_name = default_value +#define DECLARE_bool(flag_name) \ + extern bool FLAGS_ ## flag_name + +namespace snappy { + +static const uint32 kuint32max = static_cast<uint32>(0xFFFFFFFF); +static const int64 kint64max = static_cast<int64>(0x7FFFFFFFFFFFFFFFLL); + +// Potentially unaligned loads and stores. + // x86, PowerPC, and ARM64 can simply do these loads and stores native. - + #if defined(__i386__) || defined(__x86_64__) || defined(__powerpc__) || \ defined(__aarch64__) - -#define UNALIGNED_LOAD16(_p) (*reinterpret_cast<const uint16 *>(_p)) -#define UNALIGNED_LOAD32(_p) (*reinterpret_cast<const uint32 *>(_p)) -#define UNALIGNED_LOAD64(_p) (*reinterpret_cast<const uint64 *>(_p)) - -#define UNALIGNED_STORE16(_p, _val) (*reinterpret_cast<uint16 *>(_p) = (_val)) -#define UNALIGNED_STORE32(_p, _val) (*reinterpret_cast<uint32 *>(_p) = (_val)) -#define UNALIGNED_STORE64(_p, _val) (*reinterpret_cast<uint64 *>(_p) = (_val)) - -// ARMv7 and newer support native unaligned accesses, but only of 16-bit -// and 32-bit values (not 64-bit); older versions either raise a fatal signal, -// do an unaligned read and rotate the words around a bit, or do the reads very -// slowly (trip through kernel mode). There's no simple #define that says just -// “ARMv7 or higher”, so we have to filter away all ARMv5 and ARMv6 -// sub-architectures. -// -// This is a mess, but there's not much we can do about it. + +#define UNALIGNED_LOAD16(_p) (*reinterpret_cast<const uint16 *>(_p)) +#define UNALIGNED_LOAD32(_p) (*reinterpret_cast<const uint32 *>(_p)) +#define UNALIGNED_LOAD64(_p) (*reinterpret_cast<const uint64 *>(_p)) + +#define UNALIGNED_STORE16(_p, _val) (*reinterpret_cast<uint16 *>(_p) = (_val)) +#define UNALIGNED_STORE32(_p, _val) (*reinterpret_cast<uint32 *>(_p) = (_val)) +#define UNALIGNED_STORE64(_p, _val) (*reinterpret_cast<uint64 *>(_p) = (_val)) + +// ARMv7 and newer support native unaligned accesses, but only of 16-bit +// and 32-bit values (not 64-bit); older versions either raise a fatal signal, +// do an unaligned read and rotate the words around a bit, or do the reads very +// slowly (trip through kernel mode). There's no simple #define that says just +// “ARMv7 or higher”, so we have to filter away all ARMv5 and ARMv6 +// sub-architectures. +// +// This is a mess, but there's not much we can do about it. // // To further complicate matters, only LDR instructions (single reads) are // allowed to be unaligned, not LDRD (two reads) or LDM (many reads). Unless we @@ -150,30 +150,30 @@ static const int64 kint64max = static_cast<int64>(0x7FFFFFFFFFFFFFFFLL); // (it ignores __attribute__((packed)) on individual variables). However, // we can tell it that a _struct_ is unaligned, which has the same effect, // so we do that. - -#elif defined(__arm__) && \ + +#elif defined(__arm__) && \ !defined(__ARM_ARCH_4__) && \ !defined(__ARM_ARCH_4T__) && \ - !defined(__ARM_ARCH_5__) && \ - !defined(__ARM_ARCH_5T__) && \ - !defined(__ARM_ARCH_5TE__) && \ - !defined(__ARM_ARCH_5TEJ__) && \ - !defined(__ARM_ARCH_6__) && \ - !defined(__ARM_ARCH_6J__) && \ - !defined(__ARM_ARCH_6K__) && \ - !defined(__ARM_ARCH_6Z__) && \ - !defined(__ARM_ARCH_6ZK__) && \ - !defined(__ARM_ARCH_6T2__) - + !defined(__ARM_ARCH_5__) && \ + !defined(__ARM_ARCH_5T__) && \ + !defined(__ARM_ARCH_5TE__) && \ + !defined(__ARM_ARCH_5TEJ__) && \ + !defined(__ARM_ARCH_6__) && \ + !defined(__ARM_ARCH_6J__) && \ + !defined(__ARM_ARCH_6K__) && \ + !defined(__ARM_ARCH_6Z__) && \ + !defined(__ARM_ARCH_6ZK__) && \ + !defined(__ARM_ARCH_6T2__) + #if __GNUC__ #define ATTRIBUTE_PACKED __attribute__((__packed__)) #else #define ATTRIBUTE_PACKED #endif - + namespace base { namespace internal { - + struct Unaligned16Struct { uint16 value; uint8 dummy; // To make the size non-power-of-two. @@ -200,59 +200,59 @@ struct Unaligned32Struct { (_val)) // TODO: NEON supports unaligned 64-bit loads and stores. -// See if that would be more efficient on platforms supporting it, -// at least for copies. - -inline uint64 UNALIGNED_LOAD64(const void *p) { - uint64 t; - memcpy(&t, p, sizeof t); - return t; -} - -inline void UNALIGNED_STORE64(void *p, uint64 v) { - memcpy(p, &v, sizeof v); -} - -#else - -// These functions are provided for architectures that don't support -// unaligned loads and stores. - -inline uint16 UNALIGNED_LOAD16(const void *p) { - uint16 t; - memcpy(&t, p, sizeof t); - return t; -} - -inline uint32 UNALIGNED_LOAD32(const void *p) { - uint32 t; - memcpy(&t, p, sizeof t); - return t; -} - -inline uint64 UNALIGNED_LOAD64(const void *p) { - uint64 t; - memcpy(&t, p, sizeof t); - return t; -} - -inline void UNALIGNED_STORE16(void *p, uint16 v) { - memcpy(p, &v, sizeof v); -} - -inline void UNALIGNED_STORE32(void *p, uint32 v) { - memcpy(p, &v, sizeof v); -} - -inline void UNALIGNED_STORE64(void *p, uint64 v) { - memcpy(p, &v, sizeof v); -} - -#endif - +// See if that would be more efficient on platforms supporting it, +// at least for copies. + +inline uint64 UNALIGNED_LOAD64(const void *p) { + uint64 t; + memcpy(&t, p, sizeof t); + return t; +} + +inline void UNALIGNED_STORE64(void *p, uint64 v) { + memcpy(p, &v, sizeof v); +} + +#else + +// These functions are provided for architectures that don't support +// unaligned loads and stores. + +inline uint16 UNALIGNED_LOAD16(const void *p) { + uint16 t; + memcpy(&t, p, sizeof t); + return t; +} + +inline uint32 UNALIGNED_LOAD32(const void *p) { + uint32 t; + memcpy(&t, p, sizeof t); + return t; +} + +inline uint64 UNALIGNED_LOAD64(const void *p) { + uint64 t; + memcpy(&t, p, sizeof t); + return t; +} + +inline void UNALIGNED_STORE16(void *p, uint16 v) { + memcpy(p, &v, sizeof v); +} + +inline void UNALIGNED_STORE32(void *p, uint32 v) { + memcpy(p, &v, sizeof v); +} + +inline void UNALIGNED_STORE64(void *p, uint64 v) { + memcpy(p, &v, sizeof v); +} + +#endif + // The following guarantees declaration of the byte swap functions. #if defined(SNAPPY_IS_BIG_ENDIAN) - + #ifdef HAVE_SYS_BYTEORDER_H #include <sys/byteorder.h> #endif @@ -293,8 +293,8 @@ inline void UNALIGNED_STORE64(void *p, uint64 v) { inline uint16 bswap_16(uint16 x) { return (x << 8) | (x >> 8); -} - +} + inline uint32 bswap_32(uint32 x) { x = ((x & 0xff00ff00UL) >> 8) | ((x & 0x00ff00ffUL) << 8); return (x >> 16) | (x << 16); @@ -310,28 +310,28 @@ inline uint64 bswap_64(uint64 x) { #endif // defined(SNAPPY_IS_BIG_ENDIAN) -// Convert to little-endian storage, opposite of network format. -// Convert x from host to little endian: x = LittleEndian.FromHost(x); -// convert x from little endian to host: x = LittleEndian.ToHost(x); -// -// Store values into unaligned memory converting to little endian order: -// LittleEndian.Store16(p, x); -// -// Load unaligned values stored in little endian converting to host order: -// x = LittleEndian.Load16(p); -class LittleEndian { - public: - // Conversion functions. +// Convert to little-endian storage, opposite of network format. +// Convert x from host to little endian: x = LittleEndian.FromHost(x); +// convert x from little endian to host: x = LittleEndian.ToHost(x); +// +// Store values into unaligned memory converting to little endian order: +// LittleEndian.Store16(p, x); +// +// Load unaligned values stored in little endian converting to host order: +// x = LittleEndian.Load16(p); +class LittleEndian { + public: + // Conversion functions. #if defined(SNAPPY_IS_BIG_ENDIAN) - + static uint16 FromHost16(uint16 x) { return bswap_16(x); } static uint16 ToHost16(uint16 x) { return bswap_16(x); } - + static uint32 FromHost32(uint32 x) { return bswap_32(x); } static uint32 ToHost32(uint32 x) { return bswap_32(x); } - static bool IsLittleEndian() { return false; } - + static bool IsLittleEndian() { return false; } + #else // !defined(SNAPPY_IS_BIG_ENDIAN) static uint16 FromHost16(uint16 x) { return x; } @@ -344,50 +344,50 @@ class LittleEndian { #endif // !defined(SNAPPY_IS_BIG_ENDIAN) - // Functions to do unaligned loads and stores in little-endian order. - static uint16 Load16(const void *p) { - return ToHost16(UNALIGNED_LOAD16(p)); - } - - static void Store16(void *p, uint16 v) { - UNALIGNED_STORE16(p, FromHost16(v)); - } - - static uint32 Load32(const void *p) { - return ToHost32(UNALIGNED_LOAD32(p)); - } - - static void Store32(void *p, uint32 v) { - UNALIGNED_STORE32(p, FromHost32(v)); - } -}; - -// Some bit-manipulation functions. -class Bits { - public: + // Functions to do unaligned loads and stores in little-endian order. + static uint16 Load16(const void *p) { + return ToHost16(UNALIGNED_LOAD16(p)); + } + + static void Store16(void *p, uint16 v) { + UNALIGNED_STORE16(p, FromHost16(v)); + } + + static uint32 Load32(const void *p) { + return ToHost32(UNALIGNED_LOAD32(p)); + } + + static void Store32(void *p, uint32 v) { + UNALIGNED_STORE32(p, FromHost32(v)); + } +}; + +// Some bit-manipulation functions. +class Bits { + public: // Return floor(log2(n)) for positive integer n. static int Log2FloorNonZero(uint32 n); - // Return floor(log2(n)) for positive integer n. Returns -1 iff n == 0. - static int Log2Floor(uint32 n); - - // Return the first set least / most significant bit, 0-indexed. Returns an - // undefined value if n == 0. FindLSBSetNonZero() is similar to ffs() except - // that it's 0-indexed. - static int FindLSBSetNonZero(uint32 n); + // Return floor(log2(n)) for positive integer n. Returns -1 iff n == 0. + static int Log2Floor(uint32 n); + + // Return the first set least / most significant bit, 0-indexed. Returns an + // undefined value if n == 0. FindLSBSetNonZero() is similar to ffs() except + // that it's 0-indexed. + static int FindLSBSetNonZero(uint32 n); #if defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) - static int FindLSBSetNonZero64(uint64 n); + static int FindLSBSetNonZero64(uint64 n); #endif // defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) - - private: + + private: // No copying Bits(const Bits&); void operator=(const Bits&); -}; - -#ifdef HAVE_BUILTIN_CTZ - +}; + +#ifdef HAVE_BUILTIN_CTZ + inline int Bits::Log2FloorNonZero(uint32 n) { assert(n != 0); // (31 ^ x) is equivalent to (31 - x) for x in [0, 31]. An easy proof @@ -399,22 +399,22 @@ inline int Bits::Log2FloorNonZero(uint32 n) { return 31 ^ __builtin_clz(n); } -inline int Bits::Log2Floor(uint32 n) { +inline int Bits::Log2Floor(uint32 n) { return (n == 0) ? -1 : Bits::Log2FloorNonZero(n); -} - -inline int Bits::FindLSBSetNonZero(uint32 n) { +} + +inline int Bits::FindLSBSetNonZero(uint32 n) { assert(n != 0); - return __builtin_ctz(n); -} - + return __builtin_ctz(n); +} + #if defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) -inline int Bits::FindLSBSetNonZero64(uint64 n) { +inline int Bits::FindLSBSetNonZero64(uint64 n) { assert(n != 0); - return __builtin_ctzll(n); -} + return __builtin_ctzll(n); +} #endif // defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) - + #elif defined(_MSC_VER) inline int Bits::Log2FloorNonZero(uint32 n) { @@ -449,158 +449,158 @@ inline int Bits::FindLSBSetNonZero64(uint64 n) { } #endif // defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) -#else // Portable versions. - +#else // Portable versions. + inline int Bits::Log2FloorNonZero(uint32 n) { assert(n != 0); - int log = 0; - uint32 value = n; - for (int i = 4; i >= 0; --i) { - int shift = (1 << i); - uint32 x = value >> shift; - if (x != 0) { - value = x; - log += shift; - } - } - assert(value == 1); - return log; -} - + int log = 0; + uint32 value = n; + for (int i = 4; i >= 0; --i) { + int shift = (1 << i); + uint32 x = value >> shift; + if (x != 0) { + value = x; + log += shift; + } + } + assert(value == 1); + return log; +} + inline int Bits::Log2Floor(uint32 n) { return (n == 0) ? -1 : Bits::Log2FloorNonZero(n); } -inline int Bits::FindLSBSetNonZero(uint32 n) { +inline int Bits::FindLSBSetNonZero(uint32 n) { assert(n != 0); - int rc = 31; - for (int i = 4, shift = 1 << 4; i >= 0; --i) { - const uint32 x = n << shift; - if (x != 0) { - n = x; - rc -= shift; - } - shift >>= 1; - } - return rc; -} - + int rc = 31; + for (int i = 4, shift = 1 << 4; i >= 0; --i) { + const uint32 x = n << shift; + if (x != 0) { + n = x; + rc -= shift; + } + shift >>= 1; + } + return rc; +} + #if defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) -// FindLSBSetNonZero64() is defined in terms of FindLSBSetNonZero(). -inline int Bits::FindLSBSetNonZero64(uint64 n) { +// FindLSBSetNonZero64() is defined in terms of FindLSBSetNonZero(). +inline int Bits::FindLSBSetNonZero64(uint64 n) { assert(n != 0); - const uint32 bottombits = static_cast<uint32>(n); - if (bottombits == 0) { - // Bottom bits are zero, so scan in top bits - return 32 + FindLSBSetNonZero(static_cast<uint32>(n >> 32)); - } else { - return FindLSBSetNonZero(bottombits); - } -} + const uint32 bottombits = static_cast<uint32>(n); + if (bottombits == 0) { + // Bottom bits are zero, so scan in top bits + return 32 + FindLSBSetNonZero(static_cast<uint32>(n >> 32)); + } else { + return FindLSBSetNonZero(bottombits); + } +} #endif // defined(ARCH_K8) || defined(ARCH_PPC) || defined(ARCH_ARM) - -#endif // End portable versions. - -// Variable-length integer encoding. -class Varint { - public: - // Maximum lengths of varint encoding of uint32. - static const int kMax32 = 5; - - // Attempts to parse a varint32 from a prefix of the bytes in [ptr,limit-1]. - // Never reads a character at or beyond limit. If a valid/terminated varint32 - // was found in the range, stores it in *OUTPUT and returns a pointer just - // past the last byte of the varint32. Else returns NULL. On success, - // "result <= limit". - static const char* Parse32WithLimit(const char* ptr, const char* limit, - uint32* OUTPUT); - - // REQUIRES "ptr" points to a buffer of length sufficient to hold "v". - // EFFECTS Encodes "v" into "ptr" and returns a pointer to the - // byte just past the last encoded byte. - static char* Encode32(char* ptr, uint32 v); - - // EFFECTS Appends the varint representation of "value" to "*s". + +#endif // End portable versions. + +// Variable-length integer encoding. +class Varint { + public: + // Maximum lengths of varint encoding of uint32. + static const int kMax32 = 5; + + // Attempts to parse a varint32 from a prefix of the bytes in [ptr,limit-1]. + // Never reads a character at or beyond limit. If a valid/terminated varint32 + // was found in the range, stores it in *OUTPUT and returns a pointer just + // past the last byte of the varint32. Else returns NULL. On success, + // "result <= limit". + static const char* Parse32WithLimit(const char* ptr, const char* limit, + uint32* OUTPUT); + + // REQUIRES "ptr" points to a buffer of length sufficient to hold "v". + // EFFECTS Encodes "v" into "ptr" and returns a pointer to the + // byte just past the last encoded byte. + static char* Encode32(char* ptr, uint32 v); + + // EFFECTS Appends the varint representation of "value" to "*s". static void Append32(std::string* s, uint32 value); -}; - -inline const char* Varint::Parse32WithLimit(const char* p, - const char* l, - uint32* OUTPUT) { - const unsigned char* ptr = reinterpret_cast<const unsigned char*>(p); - const unsigned char* limit = reinterpret_cast<const unsigned char*>(l); - uint32 b, result; - if (ptr >= limit) return NULL; - b = *(ptr++); result = b & 127; if (b < 128) goto done; - if (ptr >= limit) return NULL; - b = *(ptr++); result |= (b & 127) << 7; if (b < 128) goto done; - if (ptr >= limit) return NULL; - b = *(ptr++); result |= (b & 127) << 14; if (b < 128) goto done; - if (ptr >= limit) return NULL; - b = *(ptr++); result |= (b & 127) << 21; if (b < 128) goto done; - if (ptr >= limit) return NULL; - b = *(ptr++); result |= (b & 127) << 28; if (b < 16) goto done; - return NULL; // Value is too long to be a varint32 - done: - *OUTPUT = result; - return reinterpret_cast<const char*>(ptr); -} - -inline char* Varint::Encode32(char* sptr, uint32 v) { - // Operate on characters as unsigneds - unsigned char* ptr = reinterpret_cast<unsigned char*>(sptr); - static const int B = 128; - if (v < (1<<7)) { - *(ptr++) = v; - } else if (v < (1<<14)) { - *(ptr++) = v | B; - *(ptr++) = v>>7; - } else if (v < (1<<21)) { - *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = v>>14; - } else if (v < (1<<28)) { - *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = (v>>14) | B; - *(ptr++) = v>>21; - } else { - *(ptr++) = v | B; - *(ptr++) = (v>>7) | B; - *(ptr++) = (v>>14) | B; - *(ptr++) = (v>>21) | B; - *(ptr++) = v>>28; - } - return reinterpret_cast<char*>(ptr); -} - +}; + +inline const char* Varint::Parse32WithLimit(const char* p, + const char* l, + uint32* OUTPUT) { + const unsigned char* ptr = reinterpret_cast<const unsigned char*>(p); + const unsigned char* limit = reinterpret_cast<const unsigned char*>(l); + uint32 b, result; + if (ptr >= limit) return NULL; + b = *(ptr++); result = b & 127; if (b < 128) goto done; + if (ptr >= limit) return NULL; + b = *(ptr++); result |= (b & 127) << 7; if (b < 128) goto done; + if (ptr >= limit) return NULL; + b = *(ptr++); result |= (b & 127) << 14; if (b < 128) goto done; + if (ptr >= limit) return NULL; + b = *(ptr++); result |= (b & 127) << 21; if (b < 128) goto done; + if (ptr >= limit) return NULL; + b = *(ptr++); result |= (b & 127) << 28; if (b < 16) goto done; + return NULL; // Value is too long to be a varint32 + done: + *OUTPUT = result; + return reinterpret_cast<const char*>(ptr); +} + +inline char* Varint::Encode32(char* sptr, uint32 v) { + // Operate on characters as unsigneds + unsigned char* ptr = reinterpret_cast<unsigned char*>(sptr); + static const int B = 128; + if (v < (1<<7)) { + *(ptr++) = v; + } else if (v < (1<<14)) { + *(ptr++) = v | B; + *(ptr++) = v>>7; + } else if (v < (1<<21)) { + *(ptr++) = v | B; + *(ptr++) = (v>>7) | B; + *(ptr++) = v>>14; + } else if (v < (1<<28)) { + *(ptr++) = v | B; + *(ptr++) = (v>>7) | B; + *(ptr++) = (v>>14) | B; + *(ptr++) = v>>21; + } else { + *(ptr++) = v | B; + *(ptr++) = (v>>7) | B; + *(ptr++) = (v>>14) | B; + *(ptr++) = (v>>21) | B; + *(ptr++) = v>>28; + } + return reinterpret_cast<char*>(ptr); +} + // If you know the internal layout of the std::string in use, you can -// replace this function with one that resizes the string without -// filling the new space with zeros (if applicable) -- -// it will be non-portable but faster. +// replace this function with one that resizes the string without +// filling the new space with zeros (if applicable) -- +// it will be non-portable but faster. inline void STLStringResizeUninitialized(std::string* s, size_t new_size) { s->resize(new_size); -} - -// Return a mutable char* pointing to a string's internal buffer, -// which may not be null-terminated. Writing through this pointer will -// modify the string. -// -// string_as_array(&str)[i] is valid for 0 <= i < str.size() until the -// next call to a string method that invalidates iterators. -// -// As of 2006-04, there is no standard-blessed way of getting a -// mutable reference to a string's internal buffer. However, issue 530 -// (http://www.open-std.org/JTC1/SC22/WG21/docs/lwg-defects.html#530) -// proposes this as the method. It will officially be part of the standard -// for C++0x. This should already work on all current implementations. +} + +// Return a mutable char* pointing to a string's internal buffer, +// which may not be null-terminated. Writing through this pointer will +// modify the string. +// +// string_as_array(&str)[i] is valid for 0 <= i < str.size() until the +// next call to a string method that invalidates iterators. +// +// As of 2006-04, there is no standard-blessed way of getting a +// mutable reference to a string's internal buffer. However, issue 530 +// (http://www.open-std.org/JTC1/SC22/WG21/docs/lwg-defects.html#530) +// proposes this as the method. It will officially be part of the standard +// for C++0x. This should already work on all current implementations. inline char* string_as_array(std::string* str) { return str->empty() ? NULL : &*str->begin(); -} - -} // namespace snappy - +} + +} // namespace snappy + #endif // THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_ diff --git a/contrib/libs/snappy/snappy-stubs-public.h b/contrib/libs/snappy/snappy-stubs-public.h index 357c4b2e4b..2a2931c4a4 100644 --- a/contrib/libs/snappy/snappy-stubs-public.h +++ b/contrib/libs/snappy/snappy-stubs-public.h @@ -1,58 +1,58 @@ -// Copyright 2011 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -// -// Various type stubs for the open-source version of Snappy. -// -// This file cannot include config.h, as it is included from snappy.h, -// which is a public header. Instead, snappy-stubs-public.h is generated by -// from snappy-stubs-public.h.in at configure time. - +// Copyright 2011 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// Various type stubs for the open-source version of Snappy. +// +// This file cannot include config.h, as it is included from snappy.h, +// which is a public header. Instead, snappy-stubs-public.h is generated by +// from snappy-stubs-public.h.in at configure time. + #ifndef THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_PUBLIC_H_ #define THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_PUBLIC_H_ - + #include <cstddef> #include <cstdint> #include <string> - + #include "config.h" #if defined(HAVE_SYS_UIO_H) #include <sys/uio.h> #endif // HAVE_SYS_UIO_H -#define SNAPPY_MAJOR 1 +#define SNAPPY_MAJOR 1 #define SNAPPY_MINOR 1 #define SNAPPY_PATCHLEVEL 8 -#define SNAPPY_VERSION \ - ((SNAPPY_MAJOR << 16) | (SNAPPY_MINOR << 8) | SNAPPY_PATCHLEVEL) - -namespace snappy { - +#define SNAPPY_VERSION \ + ((SNAPPY_MAJOR << 16) | (SNAPPY_MINOR << 8) | SNAPPY_PATCHLEVEL) + +namespace snappy { + using int8 = std::int8_t; using uint8 = std::uint8_t; using int16 = std::int16_t; @@ -61,7 +61,7 @@ using int32 = std::int32_t; using uint32 = std::uint32_t; using int64 = std::int64_t; using uint64 = std::uint64_t; - + #if !defined(HAVE_SYS_UIO_H) // Windows does not have an iovec type, yet the concept is universally useful. // It is simple to define it ourselves, so we put it inside our own namespace. @@ -70,7 +70,7 @@ struct iovec { size_t iov_len; }; #endif // !HAVE_SYS_UIO_H - -} // namespace snappy - + +} // namespace snappy + #endif // THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_PUBLIC_H_ diff --git a/contrib/libs/snappy/snappy.cc b/contrib/libs/snappy/snappy.cc index 9351b0f21e..27e491c043 100644 --- a/contrib/libs/snappy/snappy.cc +++ b/contrib/libs/snappy/snappy.cc @@ -1,35 +1,35 @@ -// Copyright 2005 Google Inc. All Rights Reserved. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - -#include "snappy.h" -#include "snappy-internal.h" -#include "snappy-sinksource.h" - +// Copyright 2005 Google Inc. All Rights Reserved. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +#include "snappy.h" +#include "snappy-internal.h" +#include "snappy-sinksource.h" + #if !defined(SNAPPY_HAVE_SSSE3) // __SSSE3__ is defined by GCC and Clang. Visual Studio doesn't target SIMD // support between SSE2 and AVX (so SSSE3 instructions require AVX support), and @@ -68,60 +68,60 @@ #include <immintrin.h> #endif -#include <stdio.h> - -#include <algorithm> +#include <stdio.h> + +#include <algorithm> #include <string> #include <vector> #include <util/generic/string.h> - -namespace snappy { - + +namespace snappy { + using internal::COPY_1_BYTE_OFFSET; using internal::COPY_2_BYTE_OFFSET; using internal::LITERAL; using internal::char_table; using internal::kMaximumTagLength; -// Any hash function will produce a valid compressed bitstream, but a good -// hash function reduces the number of collisions and thus yields better -// compression for compressible input, and more speed for incompressible -// input. Of course, it doesn't hurt if the hash function is reasonably fast -// either, as it gets called a lot. -static inline uint32 HashBytes(uint32 bytes, int shift) { - uint32 kMul = 0x1e35a7bd; - return (bytes * kMul) >> shift; -} -static inline uint32 Hash(const char* p, int shift) { - return HashBytes(UNALIGNED_LOAD32(p), shift); -} - -size_t MaxCompressedLength(size_t source_len) { - // Compressed data can be defined as: - // compressed := item* literal* - // item := literal* copy - // - // The trailing literal sequence has a space blowup of at most 62/60 - // since a literal of length 60 needs one tag byte + one extra byte - // for length information. - // - // Item blowup is trickier to measure. Suppose the "copy" op copies - // 4 bytes of data. Because of a special check in the encoding code, - // we produce a 4-byte copy only if the offset is < 65536. Therefore - // the copy op takes 3 bytes to encode, and this type of item leads - // to at most the 62/60 blowup for representing literals. - // - // Suppose the "copy" op copies 5 bytes of data. If the offset is big - // enough, it will take 5 bytes to encode the copy op. Therefore the - // worst case here is a one-byte literal followed by a five-byte copy. - // I.e., 6 bytes of input turn into 7 bytes of "compressed" data. - // - // This last factor dominates the blowup, so the final estimate is: - return 32 + source_len + source_len/6; -} - +// Any hash function will produce a valid compressed bitstream, but a good +// hash function reduces the number of collisions and thus yields better +// compression for compressible input, and more speed for incompressible +// input. Of course, it doesn't hurt if the hash function is reasonably fast +// either, as it gets called a lot. +static inline uint32 HashBytes(uint32 bytes, int shift) { + uint32 kMul = 0x1e35a7bd; + return (bytes * kMul) >> shift; +} +static inline uint32 Hash(const char* p, int shift) { + return HashBytes(UNALIGNED_LOAD32(p), shift); +} + +size_t MaxCompressedLength(size_t source_len) { + // Compressed data can be defined as: + // compressed := item* literal* + // item := literal* copy + // + // The trailing literal sequence has a space blowup of at most 62/60 + // since a literal of length 60 needs one tag byte + one extra byte + // for length information. + // + // Item blowup is trickier to measure. Suppose the "copy" op copies + // 4 bytes of data. Because of a special check in the encoding code, + // we produce a 4-byte copy only if the offset is < 65536. Therefore + // the copy op takes 3 bytes to encode, and this type of item leads + // to at most the 62/60 blowup for representing literals. + // + // Suppose the "copy" op copies 5 bytes of data. If the offset is big + // enough, it will take 5 bytes to encode the copy op. Therefore the + // worst case here is a one-byte literal followed by a five-byte copy. + // I.e., 6 bytes of input turn into 7 bytes of "compressed" data. + // + // This last factor dominates the blowup, so the final estimate is: + return 32 + source_len + source_len/6; +} + namespace { - + void UnalignedCopy64(const void* src, void* dst) { char tmp[8]; memcpy(tmp, src, 8); @@ -145,7 +145,7 @@ void UnalignedCopy128(const void* src, void* dst) { // op_limit == op + 20 // After IncrementalCopySlow(src, op, op_limit), the result will have eleven // copies of "ab" -// ababababababababababab +// ababababababababababab // Note that this does not match the semantics of either memcpy() or memmove(). inline char* IncrementalCopySlow(const char* src, char* op, char* const op_limit) { @@ -156,13 +156,13 @@ inline char* IncrementalCopySlow(const char* src, char* op, #pragma clang loop unroll(disable) #endif while (op < op_limit) { - *op++ = *src++; + *op++ = *src++; } return op_limit; -} - +} + #if SNAPPY_HAVE_SSSE3 - + // This is a table of shuffle control masks that can be used as the source // operand for PSHUFB to permute the contents of the destination XMM register // into a repeating byte pattern. @@ -175,9 +175,9 @@ alignas(16) const char pshufb_fill_patterns[7][16] = { {0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3}, {0, 1, 2, 3, 4, 5, 6, 0, 1, 2, 3, 4, 5, 6, 0, 1}, }; - + #endif // SNAPPY_HAVE_SSSE3 - + // Copy [src, src+(op_limit-op)) to [op, (op_limit-op)) but faster than // IncrementalCopySlow. buf_limit is the address past the end of the writable // region of the buffer. @@ -194,7 +194,7 @@ inline char* IncrementalCopy(const char* src, char* op, char* const op_limit, // NOTE: The compressor always emits 4 <= len <= 64. It is ok to assume that // to optimize this function but we have to also handle other cases in case // the input does not satisfy these conditions. - + size_t pattern_size = op - src; // The cases are split into different branches to allow the branch predictor, // FDO, and static prediction hints to work better. For each input we list the @@ -286,7 +286,7 @@ inline char* IncrementalCopy(const char* src, char* op, char* const op_limit, // conditionals instead of a loop allows FDO to layout the code with respect // to the actual probabilities of each length. // TODO: Replace with loop with trip count hint. - UnalignedCopy64(src, op); + UnalignedCopy64(src, op); UnalignedCopy64(src + 8, op + 8); if (op + 16 < op_limit) { @@ -302,7 +302,7 @@ inline char* IncrementalCopy(const char* src, char* op, char* const op_limit, UnalignedCopy64(src + 56, op + 56); } return op_limit; - } + } // Fall back to doing as much as we can with the available slop in the // buffer. This code path is relatively cold however so we save code size by @@ -314,7 +314,7 @@ inline char* IncrementalCopy(const char* src, char* op, char* const op_limit, #pragma clang loop unroll(disable) #endif for (char *op_end = buf_limit - 16; op < op_end; op += 16, src += 16) { - UnalignedCopy64(src, op); + UnalignedCopy64(src, op); UnalignedCopy64(src + 8, op + 8); } if (op >= op_limit) @@ -324,17 +324,17 @@ inline char* IncrementalCopy(const char* src, char* op, char* const op_limit, // single 8 byte copy. if (SNAPPY_PREDICT_FALSE(op <= buf_limit - 8)) { UnalignedCopy64(src, op); - src += 8; - op += 8; - } + src += 8; + op += 8; + } return IncrementalCopySlow(src, op, op_limit); -} - +} + } // namespace template <bool allow_fast_path> -static inline char* EmitLiteral(char* op, - const char* literal, +static inline char* EmitLiteral(char* op, + const char* literal, int len) { // The vast majority of copies are below 16 bytes, for which a // call to memcpy is overkill. This fast path can sometimes @@ -356,13 +356,13 @@ static inline char* EmitLiteral(char* op, return op + len; } - if (n < 60) { - // Fits in tag byte - *op++ = LITERAL | (n << 2); - } else { + if (n < 60) { + // Fits in tag byte + *op++ = LITERAL | (n << 2); + } else { int count = (Bits::Log2Floor(n) >> 3) + 1; - assert(count >= 1); - assert(count <= 4); + assert(count >= 1); + assert(count <= 4); *op++ = LITERAL | ((59 + count) << 2); // Encode in upcoming bytes. // Write 4 bytes, though we may care about only 1 of them. The output buffer @@ -370,33 +370,33 @@ static inline char* EmitLiteral(char* op, // here and there is a memcpy of size 'len' below. LittleEndian::Store32(op, n); op += count; - } - memcpy(op, literal, len); - return op + len; -} - + } + memcpy(op, literal, len); + return op + len; +} + template <bool len_less_than_12> static inline char* EmitCopyAtMost64(char* op, size_t offset, size_t len) { assert(len <= 64); assert(len >= 4); assert(offset < 65536); assert(len_less_than_12 == (len < 12)); - + if (len_less_than_12 && SNAPPY_PREDICT_TRUE(offset < 2048)) { // offset fits in 11 bits. The 3 highest go in the top of the first byte, // and the rest go in the second byte. *op++ = COPY_1_BYTE_OFFSET + ((len - 4) << 2) + ((offset >> 3) & 0xe0); - *op++ = offset & 0xff; - } else { + *op++ = offset & 0xff; + } else { // Write 4 bytes, though we only care about 3 of them. The output buffer // is required to have some slack, so the extra byte won't overrun it. uint32 u = COPY_2_BYTE_OFFSET + ((len - 1) << 2) + (offset << 8); LittleEndian::Store32(op, u); op += 3; - } - return op; -} - + } + return op; +} + template <bool len_less_than_12> static inline char* EmitCopy(char* op, size_t offset, size_t len) { assert(len_less_than_12 == (len < 12)); @@ -405,7 +405,7 @@ static inline char* EmitCopy(char* op, size_t offset, size_t len) { } else { // A special case for len <= 64 might help, but so far measurements suggest // it's in the noise. - + // Emit 64 byte copies but make sure to keep at least four bytes reserved. while (SNAPPY_PREDICT_FALSE(len >= 68)) { op = EmitCopyAtMost64</*len_less_than_12=*/false>(op, offset, 64); @@ -425,20 +425,20 @@ static inline char* EmitCopy(char* op, size_t offset, size_t len) { op = EmitCopyAtMost64</*len_less_than_12=*/false>(op, offset, len); } return op; - } -} - -bool GetUncompressedLength(const char* start, size_t n, size_t* result) { - uint32 v = 0; - const char* limit = start + n; - if (Varint::Parse32WithLimit(start, limit, &v) != NULL) { - *result = v; - return true; - } else { - return false; - } -} - + } +} + +bool GetUncompressedLength(const char* start, size_t n, size_t* result) { + uint32 v = 0; + const char* limit = start + n; + if (Varint::Parse32WithLimit(start, limit, &v) != NULL) { + *result = v; + return true; + } else { + return false; + } +} + namespace { uint32 CalculateTableSize(uint32 input_size) { static_assert( @@ -446,16 +446,16 @@ uint32 CalculateTableSize(uint32 input_size) { "kMaxHashTableSize should be greater or equal to kMinHashTableSize."); if (input_size > kMaxHashTableSize) { return kMaxHashTableSize; - } + } if (input_size < kMinHashTableSize) { return kMinHashTableSize; - } + } // This is equivalent to Log2Ceiling(input_size), assuming input_size > 1. // 2 << Log2Floor(x - 1) is equivalent to 1 << (1 + Log2Floor(x - 1)). return 2u << Bits::Log2Floor(input_size - 1); } } // namespace - + namespace internal { WorkingMemory::WorkingMemory(size_t input_size) { const size_t max_fragment_size = std::min(input_size, kBlockSize); @@ -476,225 +476,225 @@ uint16* WorkingMemory::GetHashTable(size_t fragment_size, int* table_size) const { const size_t htsize = CalculateTableSize(fragment_size); memset(table_, 0, htsize * sizeof(*table_)); - *table_size = htsize; + *table_size = htsize; return table_; -} -} // end namespace internal - -// For 0 <= offset <= 4, GetUint32AtOffset(GetEightBytesAt(p), offset) will -// equal UNALIGNED_LOAD32(p + offset). Motivation: On x86-64 hardware we have -// empirically found that overlapping loads such as -// UNALIGNED_LOAD32(p) ... UNALIGNED_LOAD32(p+1) ... UNALIGNED_LOAD32(p+2) -// are slower than UNALIGNED_LOAD64(p) followed by shifts and casts to uint32. -// -// We have different versions for 64- and 32-bit; ideally we would avoid the -// two functions and just inline the UNALIGNED_LOAD64 call into -// GetUint32AtOffset, but GCC (at least not as of 4.6) is seemingly not clever -// enough to avoid loading the value multiple times then. For 64-bit, the load -// is done when GetEightBytesAt() is called, whereas for 32-bit, the load is -// done at GetUint32AtOffset() time. - -#ifdef ARCH_K8 - -typedef uint64 EightBytesReference; - -static inline EightBytesReference GetEightBytesAt(const char* ptr) { - return UNALIGNED_LOAD64(ptr); -} - -static inline uint32 GetUint32AtOffset(uint64 v, int offset) { +} +} // end namespace internal + +// For 0 <= offset <= 4, GetUint32AtOffset(GetEightBytesAt(p), offset) will +// equal UNALIGNED_LOAD32(p + offset). Motivation: On x86-64 hardware we have +// empirically found that overlapping loads such as +// UNALIGNED_LOAD32(p) ... UNALIGNED_LOAD32(p+1) ... UNALIGNED_LOAD32(p+2) +// are slower than UNALIGNED_LOAD64(p) followed by shifts and casts to uint32. +// +// We have different versions for 64- and 32-bit; ideally we would avoid the +// two functions and just inline the UNALIGNED_LOAD64 call into +// GetUint32AtOffset, but GCC (at least not as of 4.6) is seemingly not clever +// enough to avoid loading the value multiple times then. For 64-bit, the load +// is done when GetEightBytesAt() is called, whereas for 32-bit, the load is +// done at GetUint32AtOffset() time. + +#ifdef ARCH_K8 + +typedef uint64 EightBytesReference; + +static inline EightBytesReference GetEightBytesAt(const char* ptr) { + return UNALIGNED_LOAD64(ptr); +} + +static inline uint32 GetUint32AtOffset(uint64 v, int offset) { assert(offset >= 0); assert(offset <= 4); - return v >> (LittleEndian::IsLittleEndian() ? 8 * offset : 32 - 8 * offset); -} - -#else - -typedef const char* EightBytesReference; - -static inline EightBytesReference GetEightBytesAt(const char* ptr) { - return ptr; -} - -static inline uint32 GetUint32AtOffset(const char* v, int offset) { + return v >> (LittleEndian::IsLittleEndian() ? 8 * offset : 32 - 8 * offset); +} + +#else + +typedef const char* EightBytesReference; + +static inline EightBytesReference GetEightBytesAt(const char* ptr) { + return ptr; +} + +static inline uint32 GetUint32AtOffset(const char* v, int offset) { assert(offset >= 0); assert(offset <= 4); - return UNALIGNED_LOAD32(v + offset); -} - -#endif - -// Flat array compression that does not emit the "uncompressed length" -// prefix. Compresses "input" string to the "*op" buffer. -// -// REQUIRES: "input" is at most "kBlockSize" bytes long. -// REQUIRES: "op" points to an array of memory that is at least -// "MaxCompressedLength(input.size())" in size. -// REQUIRES: All elements in "table[0..table_size-1]" are initialized to zero. -// REQUIRES: "table_size" is a power of two -// -// Returns an "end" pointer into "op" buffer. -// "end - op" is the compressed size of "input". -namespace internal { -char* CompressFragment(const char* input, - size_t input_size, - char* op, - uint16* table, - const int table_size) { - // "ip" is the input pointer, and "op" is the output pointer. - const char* ip = input; + return UNALIGNED_LOAD32(v + offset); +} + +#endif + +// Flat array compression that does not emit the "uncompressed length" +// prefix. Compresses "input" string to the "*op" buffer. +// +// REQUIRES: "input" is at most "kBlockSize" bytes long. +// REQUIRES: "op" points to an array of memory that is at least +// "MaxCompressedLength(input.size())" in size. +// REQUIRES: All elements in "table[0..table_size-1]" are initialized to zero. +// REQUIRES: "table_size" is a power of two +// +// Returns an "end" pointer into "op" buffer. +// "end - op" is the compressed size of "input". +namespace internal { +char* CompressFragment(const char* input, + size_t input_size, + char* op, + uint16* table, + const int table_size) { + // "ip" is the input pointer, and "op" is the output pointer. + const char* ip = input; assert(input_size <= kBlockSize); assert((table_size & (table_size - 1)) == 0); // table must be power of two - const int shift = 32 - Bits::Log2Floor(table_size); + const int shift = 32 - Bits::Log2Floor(table_size); assert(static_cast<int>(kuint32max >> shift) == table_size - 1); - const char* ip_end = input + input_size; - const char* base_ip = ip; - // Bytes in [next_emit, ip) will be emitted as literal bytes. Or - // [next_emit, ip_end) after the main loop. - const char* next_emit = ip; - - const size_t kInputMarginBytes = 15; + const char* ip_end = input + input_size; + const char* base_ip = ip; + // Bytes in [next_emit, ip) will be emitted as literal bytes. Or + // [next_emit, ip_end) after the main loop. + const char* next_emit = ip; + + const size_t kInputMarginBytes = 15; if (SNAPPY_PREDICT_TRUE(input_size >= kInputMarginBytes)) { - const char* ip_limit = input + input_size - kInputMarginBytes; - - for (uint32 next_hash = Hash(++ip, shift); ; ) { + const char* ip_limit = input + input_size - kInputMarginBytes; + + for (uint32 next_hash = Hash(++ip, shift); ; ) { assert(next_emit < ip); - // The body of this loop calls EmitLiteral once and then EmitCopy one or - // more times. (The exception is that when we're close to exhausting - // the input we goto emit_remainder.) - // - // In the first iteration of this loop we're just starting, so - // there's nothing to copy, so calling EmitLiteral once is - // necessary. And we only start a new iteration when the - // current iteration has determined that a call to EmitLiteral will - // precede the next call to EmitCopy (if any). - // - // Step 1: Scan forward in the input looking for a 4-byte-long match. - // If we get close to exhausting the input then goto emit_remainder. - // - // Heuristic match skipping: If 32 bytes are scanned with no matches - // found, start looking only at every other byte. If 32 more bytes are + // The body of this loop calls EmitLiteral once and then EmitCopy one or + // more times. (The exception is that when we're close to exhausting + // the input we goto emit_remainder.) + // + // In the first iteration of this loop we're just starting, so + // there's nothing to copy, so calling EmitLiteral once is + // necessary. And we only start a new iteration when the + // current iteration has determined that a call to EmitLiteral will + // precede the next call to EmitCopy (if any). + // + // Step 1: Scan forward in the input looking for a 4-byte-long match. + // If we get close to exhausting the input then goto emit_remainder. + // + // Heuristic match skipping: If 32 bytes are scanned with no matches + // found, start looking only at every other byte. If 32 more bytes are // scanned (or skipped), look at every third byte, etc.. When a match is // found, immediately go back to looking at every byte. This is a small // loss (~5% performance, ~0.1% density) for compressible data due to more - // bookkeeping, but for non-compressible data (such as JPEG) it's a huge - // win since the compressor quickly "realizes" the data is incompressible - // and doesn't bother looking for matches everywhere. - // - // The "skip" variable keeps track of how many bytes there are since the - // last match; dividing it by 32 (ie. right-shifting by five) gives the - // number of bytes to move ahead for each iteration. - uint32 skip = 32; - - const char* next_ip = ip; - const char* candidate; - do { - ip = next_ip; - uint32 hash = next_hash; + // bookkeeping, but for non-compressible data (such as JPEG) it's a huge + // win since the compressor quickly "realizes" the data is incompressible + // and doesn't bother looking for matches everywhere. + // + // The "skip" variable keeps track of how many bytes there are since the + // last match; dividing it by 32 (ie. right-shifting by five) gives the + // number of bytes to move ahead for each iteration. + uint32 skip = 32; + + const char* next_ip = ip; + const char* candidate; + do { + ip = next_ip; + uint32 hash = next_hash; assert(hash == Hash(ip, shift)); uint32 bytes_between_hash_lookups = skip >> 5; skip += bytes_between_hash_lookups; - next_ip = ip + bytes_between_hash_lookups; + next_ip = ip + bytes_between_hash_lookups; if (SNAPPY_PREDICT_FALSE(next_ip > ip_limit)) { - goto emit_remainder; - } - next_hash = Hash(next_ip, shift); - candidate = base_ip + table[hash]; + goto emit_remainder; + } + next_hash = Hash(next_ip, shift); + candidate = base_ip + table[hash]; assert(candidate >= base_ip); assert(candidate < ip); - - table[hash] = ip - base_ip; + + table[hash] = ip - base_ip; } while (SNAPPY_PREDICT_TRUE(UNALIGNED_LOAD32(ip) != UNALIGNED_LOAD32(candidate))); - - // Step 2: A 4-byte match has been found. We'll later see if more - // than 4 bytes match. But, prior to the match, input - // bytes [next_emit, ip) are unmatched. Emit them as "literal bytes." + + // Step 2: A 4-byte match has been found. We'll later see if more + // than 4 bytes match. But, prior to the match, input + // bytes [next_emit, ip) are unmatched. Emit them as "literal bytes." assert(next_emit + 16 <= ip_end); op = EmitLiteral</*allow_fast_path=*/true>(op, next_emit, ip - next_emit); - - // Step 3: Call EmitCopy, and then see if another EmitCopy could - // be our next move. Repeat until we find no match for the - // input immediately after what was consumed by the last EmitCopy call. - // - // If we exit this loop normally then we need to call EmitLiteral next, - // though we don't yet know how big the literal will be. We handle that - // by proceeding to the next iteration of the main loop. We also can exit - // this loop via goto if we get close to exhausting the input. - EightBytesReference input_bytes; - uint32 candidate_bytes = 0; - - do { - // We have a 4-byte match at ip, and no need to emit any - // "literal bytes" prior to ip. - const char* base = ip; + + // Step 3: Call EmitCopy, and then see if another EmitCopy could + // be our next move. Repeat until we find no match for the + // input immediately after what was consumed by the last EmitCopy call. + // + // If we exit this loop normally then we need to call EmitLiteral next, + // though we don't yet know how big the literal will be. We handle that + // by proceeding to the next iteration of the main loop. We also can exit + // this loop via goto if we get close to exhausting the input. + EightBytesReference input_bytes; + uint32 candidate_bytes = 0; + + do { + // We have a 4-byte match at ip, and no need to emit any + // "literal bytes" prior to ip. + const char* base = ip; std::pair<size_t, bool> p = FindMatchLength(candidate + 4, ip + 4, ip_end); size_t matched = 4 + p.first; - ip += matched; - size_t offset = base - candidate; + ip += matched; + size_t offset = base - candidate; assert(0 == memcmp(base, candidate, matched)); if (p.second) { op = EmitCopy</*len_less_than_12=*/true>(op, offset, matched); } else { op = EmitCopy</*len_less_than_12=*/false>(op, offset, matched); } - next_emit = ip; + next_emit = ip; if (SNAPPY_PREDICT_FALSE(ip >= ip_limit)) { - goto emit_remainder; - } + goto emit_remainder; + } // We are now looking for a 4-byte match again. We read // table[Hash(ip, shift)] for that. To improve compression, // we also update table[Hash(ip - 1, shift)] and table[Hash(ip, shift)]. input_bytes = GetEightBytesAt(ip - 1); - uint32 prev_hash = HashBytes(GetUint32AtOffset(input_bytes, 0), shift); - table[prev_hash] = ip - base_ip - 1; - uint32 cur_hash = HashBytes(GetUint32AtOffset(input_bytes, 1), shift); - candidate = base_ip + table[cur_hash]; - candidate_bytes = UNALIGNED_LOAD32(candidate); - table[cur_hash] = ip - base_ip; - } while (GetUint32AtOffset(input_bytes, 1) == candidate_bytes); - - next_hash = HashBytes(GetUint32AtOffset(input_bytes, 2), shift); - ++ip; - } - } - - emit_remainder: - // Emit the remaining bytes as a literal - if (next_emit < ip_end) { + uint32 prev_hash = HashBytes(GetUint32AtOffset(input_bytes, 0), shift); + table[prev_hash] = ip - base_ip - 1; + uint32 cur_hash = HashBytes(GetUint32AtOffset(input_bytes, 1), shift); + candidate = base_ip + table[cur_hash]; + candidate_bytes = UNALIGNED_LOAD32(candidate); + table[cur_hash] = ip - base_ip; + } while (GetUint32AtOffset(input_bytes, 1) == candidate_bytes); + + next_hash = HashBytes(GetUint32AtOffset(input_bytes, 2), shift); + ++ip; + } + } + + emit_remainder: + // Emit the remaining bytes as a literal + if (next_emit < ip_end) { op = EmitLiteral</*allow_fast_path=*/false>(op, next_emit, ip_end - next_emit); - } - - return op; -} -} // end namespace internal - + } + + return op; +} +} // end namespace internal + // Called back at avery compression call to trace parameters and sizes. static inline void Report(const char *algorithm, size_t compressed_size, size_t uncompressed_size) {} -// Signature of output types needed by decompression code. -// The decompression code is templatized on a type that obeys this -// signature so that we do not pay virtual function call overhead in -// the middle of a tight decompression loop. -// -// class DecompressionWriter { -// public: -// // Called before decompression -// void SetExpectedLength(size_t length); -// -// // Called after decompression -// bool CheckLength() const; -// -// // Called repeatedly during decompression -// bool Append(const char* ip, size_t length); -// bool AppendFromSelf(uint32 offset, size_t length); -// +// Signature of output types needed by decompression code. +// The decompression code is templatized on a type that obeys this +// signature so that we do not pay virtual function call overhead in +// the middle of a tight decompression loop. +// +// class DecompressionWriter { +// public: +// // Called before decompression +// void SetExpectedLength(size_t length); +// +// // Called after decompression +// bool CheckLength() const; +// +// // Called repeatedly during decompression +// bool Append(const char* ip, size_t length); +// bool AppendFromSelf(uint32 offset, size_t length); +// // // The rules for how TryFastAppend differs from Append are somewhat // // convoluted: -// // +// // // // - TryFastAppend is allowed to decline (return false) at any // // time, for any reason -- just "return false" would be // // a perfectly legal implementation of TryFastAppend. @@ -711,10 +711,10 @@ static inline void Report(const char *algorithm, size_t compressed_size, // // decoded fully. In practice, this should not be a big problem, // // as it is unlikely that one would implement a fast path accepting // // this much data. -// // -// bool TryFastAppend(const char* ip, size_t available, size_t length); -// }; - +// // +// bool TryFastAppend(const char* ip, size_t available, size_t length); +// }; + static inline uint32 ExtractLowBytes(uint32 v, int n) { assert(n >= 0); assert(n <= 4); @@ -726,8 +726,8 @@ static inline uint32 ExtractLowBytes(uint32 v, int n) { uint64 mask = 0xffffffff; return v & ~(mask << (8 * n)); #endif -} - +} + static inline bool LeftShiftOverflows(uint8 value, uint32 shift) { assert(shift < 32); static const uint8 masks[] = { @@ -736,77 +736,77 @@ static inline bool LeftShiftOverflows(uint8 value, uint32 shift) { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // 0x00, 0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe}; return (value & masks[shift]) != 0; -} - -// Helper class for decompression -class SnappyDecompressor { - private: - Source* reader_; // Underlying source of bytes to decompress - const char* ip_; // Points to next buffered byte - const char* ip_limit_; // Points just past buffered bytes - uint32 peeked_; // Bytes peeked from reader (need to skip) - bool eof_; // Hit end of input without an error? +} + +// Helper class for decompression +class SnappyDecompressor { + private: + Source* reader_; // Underlying source of bytes to decompress + const char* ip_; // Points to next buffered byte + const char* ip_limit_; // Points just past buffered bytes + uint32 peeked_; // Bytes peeked from reader (need to skip) + bool eof_; // Hit end of input without an error? char scratch_[kMaximumTagLength]; // See RefillTag(). - - // Ensure that all of the tag metadata for the next tag is available - // in [ip_..ip_limit_-1]. Also ensures that [ip,ip+4] is readable even - // if (ip_limit_ - ip_ < 5). - // - // Returns true on success, false on error or end of input. - bool RefillTag(); - - public: - explicit SnappyDecompressor(Source* reader) - : reader_(reader), - ip_(NULL), - ip_limit_(NULL), - peeked_(0), - eof_(false) { - } - - ~SnappyDecompressor() { - // Advance past any bytes we peeked at from the reader - reader_->Skip(peeked_); - } - - // Returns true iff we have hit the end of the input without an error. - bool eof() const { - return eof_; - } - - // Read the uncompressed length stored at the start of the compressed data. + + // Ensure that all of the tag metadata for the next tag is available + // in [ip_..ip_limit_-1]. Also ensures that [ip,ip+4] is readable even + // if (ip_limit_ - ip_ < 5). + // + // Returns true on success, false on error or end of input. + bool RefillTag(); + + public: + explicit SnappyDecompressor(Source* reader) + : reader_(reader), + ip_(NULL), + ip_limit_(NULL), + peeked_(0), + eof_(false) { + } + + ~SnappyDecompressor() { + // Advance past any bytes we peeked at from the reader + reader_->Skip(peeked_); + } + + // Returns true iff we have hit the end of the input without an error. + bool eof() const { + return eof_; + } + + // Read the uncompressed length stored at the start of the compressed data. // On success, stores the length in *result and returns true. - // On failure, returns false. - bool ReadUncompressedLength(uint32* result) { + // On failure, returns false. + bool ReadUncompressedLength(uint32* result) { assert(ip_ == NULL); // Must not have read anything yet - // Length is encoded in 1..5 bytes - *result = 0; - uint32 shift = 0; - while (true) { - if (shift >= 32) return false; - size_t n; - const char* ip = reader_->Peek(&n); - if (n == 0) return false; - const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip)); - reader_->Skip(1); + // Length is encoded in 1..5 bytes + *result = 0; + uint32 shift = 0; + while (true) { + if (shift >= 32) return false; + size_t n; + const char* ip = reader_->Peek(&n); + if (n == 0) return false; + const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip)); + reader_->Skip(1); uint32 val = c & 0x7f; if (LeftShiftOverflows(static_cast<uint8>(val), shift)) return false; *result |= val << shift; - if (c < 128) { - break; - } - shift += 7; - } - return true; - } - - // Process the next item found in the input. - // Returns true if successful, false on error or end of input. - template <class Writer> + if (c < 128) { + break; + } + shift += 7; + } + return true; + } + + // Process the next item found in the input. + // Returns true if successful, false on error or end of input. + template <class Writer> #if defined(__GNUC__) && defined(__x86_64__) __attribute__((aligned(32))) #endif - void DecompressAllTags(Writer* writer) { + void DecompressAllTags(Writer* writer) { // In x86, pad the function body to start 16 bytes later. This function has // a couple of hotspots that are highly sensitive to alignment: we have // observed regressions by more than 20% in some metrics just by moving the @@ -823,22 +823,22 @@ class SnappyDecompressor { asm(".byte 0x0f, 0x1f, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00"); #endif - const char* ip = ip_; - // We could have put this refill fragment only at the beginning of the loop. - // However, duplicating it at the end of each branch gives the compiler more - // scope to optimize the <ip_limit_ - ip> expression based on the local - // context, which overall increases speed. - #define MAYBE_REFILL() \ + const char* ip = ip_; + // We could have put this refill fragment only at the beginning of the loop. + // However, duplicating it at the end of each branch gives the compiler more + // scope to optimize the <ip_limit_ - ip> expression based on the local + // context, which overall increases speed. + #define MAYBE_REFILL() \ if (ip_limit_ - ip < kMaximumTagLength) { \ - ip_ = ip; \ - if (!RefillTag()) return; \ - ip = ip_; \ - } - - MAYBE_REFILL(); - for ( ;; ) { - const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip++)); - + ip_ = ip; \ + if (!RefillTag()) return; \ + ip = ip_; \ + } + + MAYBE_REFILL(); + for ( ;; ) { + const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip++)); + // Ratio of iterations that have LITERAL vs non-LITERAL for different // inputs. // @@ -852,202 +852,202 @@ class SnappyDecompressor { // pb 24% 76% // bin 24% 76% if (SNAPPY_PREDICT_FALSE((c & 0x3) == LITERAL)) { - size_t literal_length = (c >> 2) + 1u; - if (writer->TryFastAppend(ip, ip_limit_ - ip, literal_length)) { + size_t literal_length = (c >> 2) + 1u; + if (writer->TryFastAppend(ip, ip_limit_ - ip, literal_length)) { assert(literal_length < 61); - ip += literal_length; + ip += literal_length; // NOTE: There is no MAYBE_REFILL() here, as TryFastAppend() // will not return true unless there's already at least five spare // bytes in addition to the literal. - continue; - } + continue; + } if (SNAPPY_PREDICT_FALSE(literal_length >= 61)) { - // Long literal. - const size_t literal_length_length = literal_length - 60; - literal_length = + // Long literal. + const size_t literal_length_length = literal_length - 60; + literal_length = ExtractLowBytes(LittleEndian::Load32(ip), literal_length_length) + 1; - ip += literal_length_length; - } - - size_t avail = ip_limit_ - ip; - while (avail < literal_length) { - if (!writer->Append(ip, avail)) return; - literal_length -= avail; - reader_->Skip(peeked_); - size_t n; - ip = reader_->Peek(&n); - avail = n; - peeked_ = avail; - if (avail == 0) return; // Premature end of input - ip_limit_ = ip + avail; - } - if (!writer->Append(ip, literal_length)) { - return; - } - ip += literal_length; - MAYBE_REFILL(); - } else { + ip += literal_length_length; + } + + size_t avail = ip_limit_ - ip; + while (avail < literal_length) { + if (!writer->Append(ip, avail)) return; + literal_length -= avail; + reader_->Skip(peeked_); + size_t n; + ip = reader_->Peek(&n); + avail = n; + peeked_ = avail; + if (avail == 0) return; // Premature end of input + ip_limit_ = ip + avail; + } + if (!writer->Append(ip, literal_length)) { + return; + } + ip += literal_length; + MAYBE_REFILL(); + } else { const size_t entry = char_table[c]; const size_t trailer = ExtractLowBytes(LittleEndian::Load32(ip), entry >> 11); const size_t length = entry & 0xff; - ip += entry >> 11; - - // copy_offset/256 is encoded in bits 8..10. By just fetching - // those bits, we get copy_offset (since the bit-field starts at - // bit 8). + ip += entry >> 11; + + // copy_offset/256 is encoded in bits 8..10. By just fetching + // those bits, we get copy_offset (since the bit-field starts at + // bit 8). const size_t copy_offset = entry & 0x700; - if (!writer->AppendFromSelf(copy_offset + trailer, length)) { - return; - } - MAYBE_REFILL(); - } - } - -#undef MAYBE_REFILL - } -}; - -bool SnappyDecompressor::RefillTag() { - const char* ip = ip_; - if (ip == ip_limit_) { - // Fetch a new fragment from the reader - reader_->Skip(peeked_); // All peeked bytes are used up - size_t n; - ip = reader_->Peek(&n); - peeked_ = n; + if (!writer->AppendFromSelf(copy_offset + trailer, length)) { + return; + } + MAYBE_REFILL(); + } + } + +#undef MAYBE_REFILL + } +}; + +bool SnappyDecompressor::RefillTag() { + const char* ip = ip_; + if (ip == ip_limit_) { + // Fetch a new fragment from the reader + reader_->Skip(peeked_); // All peeked bytes are used up + size_t n; + ip = reader_->Peek(&n); + peeked_ = n; eof_ = (n == 0); if (eof_) return false; - ip_limit_ = ip + n; - } - - // Read the tag character + ip_limit_ = ip + n; + } + + // Read the tag character assert(ip < ip_limit_); - const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip)); - const uint32 entry = char_table[c]; - const uint32 needed = (entry >> 11) + 1; // +1 byte for 'c' + const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip)); + const uint32 entry = char_table[c]; + const uint32 needed = (entry >> 11) + 1; // +1 byte for 'c' assert(needed <= sizeof(scratch_)); - - // Read more bytes from reader if needed - uint32 nbuf = ip_limit_ - ip; - if (nbuf < needed) { - // Stitch together bytes from ip and reader to form the word - // contents. We store the needed bytes in "scratch_". They - // will be consumed immediately by the caller since we do not - // read more than we need. - memmove(scratch_, ip, nbuf); - reader_->Skip(peeked_); // All peeked bytes are used up - peeked_ = 0; - while (nbuf < needed) { - size_t length; - const char* src = reader_->Peek(&length); - if (length == 0) return false; + + // Read more bytes from reader if needed + uint32 nbuf = ip_limit_ - ip; + if (nbuf < needed) { + // Stitch together bytes from ip and reader to form the word + // contents. We store the needed bytes in "scratch_". They + // will be consumed immediately by the caller since we do not + // read more than we need. + memmove(scratch_, ip, nbuf); + reader_->Skip(peeked_); // All peeked bytes are used up + peeked_ = 0; + while (nbuf < needed) { + size_t length; + const char* src = reader_->Peek(&length); + if (length == 0) return false; uint32 to_add = std::min<uint32>(needed - nbuf, length); - memcpy(scratch_ + nbuf, src, to_add); - nbuf += to_add; - reader_->Skip(to_add); - } + memcpy(scratch_ + nbuf, src, to_add); + nbuf += to_add; + reader_->Skip(to_add); + } assert(nbuf == needed); - ip_ = scratch_; - ip_limit_ = scratch_ + needed; + ip_ = scratch_; + ip_limit_ = scratch_ + needed; } else if (nbuf < kMaximumTagLength) { - // Have enough bytes, but move into scratch_ so that we do not - // read past end of input - memmove(scratch_, ip, nbuf); - reader_->Skip(peeked_); // All peeked bytes are used up - peeked_ = 0; - ip_ = scratch_; - ip_limit_ = scratch_ + nbuf; - } else { - // Pass pointer to buffer returned by reader_. - ip_ = ip; - } - return true; -} - -template <typename Writer> + // Have enough bytes, but move into scratch_ so that we do not + // read past end of input + memmove(scratch_, ip, nbuf); + reader_->Skip(peeked_); // All peeked bytes are used up + peeked_ = 0; + ip_ = scratch_; + ip_limit_ = scratch_ + nbuf; + } else { + // Pass pointer to buffer returned by reader_. + ip_ = ip; + } + return true; +} + +template <typename Writer> static bool InternalUncompress(Source* r, Writer* writer) { - // Read the uncompressed length from the front of the compressed input - SnappyDecompressor decompressor(r); - uint32 uncompressed_len = 0; - if (!decompressor.ReadUncompressedLength(&uncompressed_len)) return false; + // Read the uncompressed length from the front of the compressed input + SnappyDecompressor decompressor(r); + uint32 uncompressed_len = 0; + if (!decompressor.ReadUncompressedLength(&uncompressed_len)) return false; return InternalUncompressAllTags(&decompressor, writer, r->Available(), uncompressed_len); -} - -template <typename Writer> -static bool InternalUncompressAllTags(SnappyDecompressor* decompressor, - Writer* writer, +} + +template <typename Writer> +static bool InternalUncompressAllTags(SnappyDecompressor* decompressor, + Writer* writer, uint32 compressed_len, uint32 uncompressed_len) { Report("snappy_uncompress", compressed_len, uncompressed_len); - - writer->SetExpectedLength(uncompressed_len); - - // Process the entire input - decompressor->DecompressAllTags(writer); + + writer->SetExpectedLength(uncompressed_len); + + // Process the entire input + decompressor->DecompressAllTags(writer); writer->Flush(); - return (decompressor->eof() && writer->CheckLength()); -} - -bool GetUncompressedLength(Source* source, uint32* result) { - SnappyDecompressor decompressor(source); - return decompressor.ReadUncompressedLength(result); -} - -size_t Compress(Source* reader, Sink* writer) { - size_t written = 0; - size_t N = reader->Available(); + return (decompressor->eof() && writer->CheckLength()); +} + +bool GetUncompressedLength(Source* source, uint32* result) { + SnappyDecompressor decompressor(source); + return decompressor.ReadUncompressedLength(result); +} + +size_t Compress(Source* reader, Sink* writer) { + size_t written = 0; + size_t N = reader->Available(); const size_t uncompressed_size = N; - char ulength[Varint::kMax32]; - char* p = Varint::Encode32(ulength, N); - writer->Append(ulength, p-ulength); - written += (p - ulength); - + char ulength[Varint::kMax32]; + char* p = Varint::Encode32(ulength, N); + writer->Append(ulength, p-ulength); + written += (p - ulength); + internal::WorkingMemory wmem(N); - - while (N > 0) { - // Get next block to compress (without copying if possible) - size_t fragment_size; - const char* fragment = reader->Peek(&fragment_size); + + while (N > 0) { + // Get next block to compress (without copying if possible) + size_t fragment_size; + const char* fragment = reader->Peek(&fragment_size); assert(fragment_size != 0); // premature end of input const size_t num_to_read = std::min(N, kBlockSize); - size_t bytes_read = fragment_size; - - size_t pending_advance = 0; - if (bytes_read >= num_to_read) { - // Buffer returned by reader is large enough - pending_advance = num_to_read; - fragment_size = num_to_read; - } else { + size_t bytes_read = fragment_size; + + size_t pending_advance = 0; + if (bytes_read >= num_to_read) { + // Buffer returned by reader is large enough + pending_advance = num_to_read; + fragment_size = num_to_read; + } else { char* scratch = wmem.GetScratchInput(); - memcpy(scratch, fragment, bytes_read); - reader->Skip(bytes_read); - - while (bytes_read < num_to_read) { - fragment = reader->Peek(&fragment_size); + memcpy(scratch, fragment, bytes_read); + reader->Skip(bytes_read); + + while (bytes_read < num_to_read) { + fragment = reader->Peek(&fragment_size); size_t n = std::min<size_t>(fragment_size, num_to_read - bytes_read); - memcpy(scratch + bytes_read, fragment, n); - bytes_read += n; - reader->Skip(n); - } + memcpy(scratch + bytes_read, fragment, n); + bytes_read += n; + reader->Skip(n); + } assert(bytes_read == num_to_read); - fragment = scratch; - fragment_size = num_to_read; - } + fragment = scratch; + fragment_size = num_to_read; + } assert(fragment_size == num_to_read); - - // Get encoding table for compression - int table_size; - uint16* table = wmem.GetHashTable(num_to_read, &table_size); - - // Compress input_fragment and append to dest - const int max_output = MaxCompressedLength(num_to_read); - - // Need a scratch buffer for the output, in case the byte sink doesn't - // have room for us directly. + + // Get encoding table for compression + int table_size; + uint16* table = wmem.GetHashTable(num_to_read, &table_size); + + // Compress input_fragment and append to dest + const int max_output = MaxCompressedLength(num_to_read); + + // Need a scratch buffer for the output, in case the byte sink doesn't + // have room for us directly. // Since we encode kBlockSize regions followed by a region // which is <= kBlockSize in length, a previously allocated @@ -1055,19 +1055,19 @@ size_t Compress(Source* reader, Sink* writer) { char* dest = writer->GetAppendBuffer(max_output, wmem.GetScratchOutput()); char* end = internal::CompressFragment(fragment, fragment_size, dest, table, table_size); - writer->Append(dest, end - dest); - written += (end - dest); - - N -= num_to_read; - reader->Skip(pending_advance); - } - + writer->Append(dest, end - dest); + written += (end - dest); + + N -= num_to_read; + reader->Skip(pending_advance); + } + Report("snappy_compress", written, uncompressed_size); - - return written; -} - -// ----------------------------------------------------------------------- + + return written; +} + +// ----------------------------------------------------------------------- // IOVec interfaces // ----------------------------------------------------------------------- @@ -1260,60 +1260,60 @@ bool RawUncompressToIOVec(Source* compressed, const struct iovec* iov, } // ----------------------------------------------------------------------- -// Flat array interfaces -// ----------------------------------------------------------------------- - -// A type that writes to a flat array. -// Note that this is not a "ByteSink", but a type that matches the -// Writer template argument to SnappyDecompressor::DecompressAllTags(). -class SnappyArrayWriter { - private: - char* base_; - char* op_; - char* op_limit_; - - public: - inline explicit SnappyArrayWriter(char* dst) - : base_(dst), +// Flat array interfaces +// ----------------------------------------------------------------------- + +// A type that writes to a flat array. +// Note that this is not a "ByteSink", but a type that matches the +// Writer template argument to SnappyDecompressor::DecompressAllTags(). +class SnappyArrayWriter { + private: + char* base_; + char* op_; + char* op_limit_; + + public: + inline explicit SnappyArrayWriter(char* dst) + : base_(dst), op_(dst), op_limit_(dst) { - } - - inline void SetExpectedLength(size_t len) { - op_limit_ = op_ + len; - } - - inline bool CheckLength() const { - return op_ == op_limit_; - } - - inline bool Append(const char* ip, size_t len) { - char* op = op_; - const size_t space_left = op_limit_ - op; - if (space_left < len) { - return false; - } - memcpy(op, ip, len); - op_ = op + len; - return true; - } - - inline bool TryFastAppend(const char* ip, size_t available, size_t len) { - char* op = op_; - const size_t space_left = op_limit_ - op; + } + + inline void SetExpectedLength(size_t len) { + op_limit_ = op_ + len; + } + + inline bool CheckLength() const { + return op_ == op_limit_; + } + + inline bool Append(const char* ip, size_t len) { + char* op = op_; + const size_t space_left = op_limit_ - op; + if (space_left < len) { + return false; + } + memcpy(op, ip, len); + op_ = op + len; + return true; + } + + inline bool TryFastAppend(const char* ip, size_t available, size_t len) { + char* op = op_; + const size_t space_left = op_limit_ - op; if (len <= 16 && available >= 16 + kMaximumTagLength && space_left >= 16) { - // Fast path, used for the majority (about 95%) of invocations. + // Fast path, used for the majority (about 95%) of invocations. UnalignedCopy128(ip, op); - op_ = op + len; - return true; - } else { - return false; - } - } - - inline bool AppendFromSelf(size_t offset, size_t len) { + op_ = op + len; + return true; + } else { + return false; + } + } + + inline bool AppendFromSelf(size_t offset, size_t len) { char* const op_end = op_ + len; - + // Check if we try to append from before the start of the buffer. // Normally this would just be a check for "produced < offset", // but "produced <= offset - 1u" is equivalent for every case @@ -1323,40 +1323,40 @@ class SnappyArrayWriter { // into an infinite loop. if (Produced() <= offset - 1u || op_end > op_limit_) return false; op_ = IncrementalCopy(op_ - offset, op_, op_end, op_limit_); - - return true; - } + + return true; + } inline size_t Produced() const { assert(op_ >= base_); return op_ - base_; } inline void Flush() {} -}; - -bool RawUncompress(const char* compressed, size_t n, char* uncompressed) { - ByteArraySource reader(compressed, n); - return RawUncompress(&reader, uncompressed); -} - -bool RawUncompress(Source* compressed, char* uncompressed) { - SnappyArrayWriter output(uncompressed); +}; + +bool RawUncompress(const char* compressed, size_t n, char* uncompressed) { + ByteArraySource reader(compressed, n); + return RawUncompress(&reader, uncompressed); +} + +bool RawUncompress(Source* compressed, char* uncompressed) { + SnappyArrayWriter output(uncompressed); return InternalUncompress(compressed, &output); -} - +} + bool Uncompress(const char* compressed, size_t n, std::string* uncompressed) { - size_t ulength; - if (!GetUncompressedLength(compressed, n, &ulength)) { - return false; - } + size_t ulength; + if (!GetUncompressedLength(compressed, n, &ulength)) { + return false; + } // On 32-bit builds: max_size() < kuint32max. Check for that instead // of crashing (e.g., consider externally specified compressed data). if (ulength > uncompressed->max_size()) { - return false; - } - STLStringResizeUninitialized(uncompressed, ulength); - return RawUncompress(compressed, n, string_as_array(uncompressed)); -} - + return false; + } + STLStringResizeUninitialized(uncompressed, ulength); + return RawUncompress(compressed, n, string_as_array(uncompressed)); +} + bool Uncompress(const char* compressed, size_t n, TString* uncompressed) { size_t ulength; if (!GetUncompressedLength(compressed, n, &ulength)) { @@ -1371,72 +1371,72 @@ bool Uncompress(const char* compressed, size_t n, TString* uncompressed) { return RawUncompress(compressed, n, uncompressed->begin()); } -// A Writer that drops everything on the floor and just does validation -class SnappyDecompressionValidator { - private: - size_t expected_; - size_t produced_; - - public: +// A Writer that drops everything on the floor and just does validation +class SnappyDecompressionValidator { + private: + size_t expected_; + size_t produced_; + + public: inline SnappyDecompressionValidator() : expected_(0), produced_(0) { } - inline void SetExpectedLength(size_t len) { - expected_ = len; - } - inline bool CheckLength() const { - return expected_ == produced_; - } - inline bool Append(const char* ip, size_t len) { - produced_ += len; - return produced_ <= expected_; - } - inline bool TryFastAppend(const char* ip, size_t available, size_t length) { - return false; - } - inline bool AppendFromSelf(size_t offset, size_t len) { + inline void SetExpectedLength(size_t len) { + expected_ = len; + } + inline bool CheckLength() const { + return expected_ == produced_; + } + inline bool Append(const char* ip, size_t len) { + produced_ += len; + return produced_ <= expected_; + } + inline bool TryFastAppend(const char* ip, size_t available, size_t length) { + return false; + } + inline bool AppendFromSelf(size_t offset, size_t len) { // See SnappyArrayWriter::AppendFromSelf for an explanation of // the "offset - 1u" trick. if (produced_ <= offset - 1u) return false; - produced_ += len; - return produced_ <= expected_; - } + produced_ += len; + return produced_ <= expected_; + } inline void Flush() {} -}; - -bool IsValidCompressedBuffer(const char* compressed, size_t n) { - ByteArraySource reader(compressed, n); - SnappyDecompressionValidator writer; +}; + +bool IsValidCompressedBuffer(const char* compressed, size_t n) { + ByteArraySource reader(compressed, n); + SnappyDecompressionValidator writer; return InternalUncompress(&reader, &writer); -} - +} + bool IsValidCompressed(Source* compressed) { SnappyDecompressionValidator writer; return InternalUncompress(compressed, &writer); } -void RawCompress(const char* input, - size_t input_length, - char* compressed, - size_t* compressed_length) { - ByteArraySource reader(input, input_length); - UncheckedByteArraySink writer(compressed); - Compress(&reader, &writer); - - // Compute how many bytes were added - *compressed_length = (writer.CurrentDestination() - compressed); -} - +void RawCompress(const char* input, + size_t input_length, + char* compressed, + size_t* compressed_length) { + ByteArraySource reader(input, input_length); + UncheckedByteArraySink writer(compressed); + Compress(&reader, &writer); + + // Compute how many bytes were added + *compressed_length = (writer.CurrentDestination() - compressed); +} + size_t Compress(const char* input, size_t input_length, std::string* compressed) { - // Pre-grow the buffer to the max length of the compressed output + // Pre-grow the buffer to the max length of the compressed output STLStringResizeUninitialized(compressed, MaxCompressedLength(input_length)); - - size_t compressed_length; - RawCompress(input, input_length, string_as_array(compressed), - &compressed_length); - compressed->resize(compressed_length); - return compressed_length; -} - + + size_t compressed_length; + RawCompress(input, input_length, string_as_array(compressed), + &compressed_length); + compressed->resize(compressed_length); + return compressed_length; +} + size_t Compress(const char* input, size_t input_length, TString* compressed) { // Pre-grow the buffer to the max length of the compressed output @@ -1452,14 +1452,14 @@ size_t Compress(const char* input, size_t input_length, // ----------------------------------------------------------------------- // Sink interface // ----------------------------------------------------------------------- - + // A type that decompresses into a Sink. The template parameter // Allocator must export one method "char* Allocate(int size);", which // allocates a buffer of "size" and appends that to the destination. template <typename Allocator> class SnappyScatteredWriter { Allocator allocator_; - + // We need random access into the data generated so far. Therefore // we keep track of all of the generated data as an array of blocks. // All of the blocks except the last have length kBlockSize. diff --git a/contrib/libs/snappy/snappy.h b/contrib/libs/snappy/snappy.h index 9a3bc3fa64..a91ef2b4d5 100644 --- a/contrib/libs/snappy/snappy.h +++ b/contrib/libs/snappy/snappy.h @@ -1,63 +1,63 @@ -// Copyright 2005 and onwards Google Inc. -// -// Redistribution and use in source and binary forms, with or without -// modification, are permitted provided that the following conditions are -// met: -// -// * Redistributions of source code must retain the above copyright -// notice, this list of conditions and the following disclaimer. -// * Redistributions in binary form must reproduce the above -// copyright notice, this list of conditions and the following disclaimer -// in the documentation and/or other materials provided with the -// distribution. -// * Neither the name of Google Inc. nor the names of its -// contributors may be used to endorse or promote products derived from -// this software without specific prior written permission. -// -// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS -// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT -// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR -// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT -// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, -// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT -// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, -// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY -// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE -// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -// -// A light-weight compression algorithm. It is designed for speed of -// compression and decompression, rather than for the utmost in space -// savings. -// -// For getting better compression ratios when you are compressing data -// with long repeated sequences or compressing data that is similar to -// other data, while still compressing fast, you might look at first -// using BMDiff and then compressing the output of BMDiff with -// Snappy. - +// Copyright 2005 and onwards Google Inc. +// +// Redistribution and use in source and binary forms, with or without +// modification, are permitted provided that the following conditions are +// met: +// +// * Redistributions of source code must retain the above copyright +// notice, this list of conditions and the following disclaimer. +// * Redistributions in binary form must reproduce the above +// copyright notice, this list of conditions and the following disclaimer +// in the documentation and/or other materials provided with the +// distribution. +// * Neither the name of Google Inc. nor the names of its +// contributors may be used to endorse or promote products derived from +// this software without specific prior written permission. +// +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +// +// A light-weight compression algorithm. It is designed for speed of +// compression and decompression, rather than for the utmost in space +// savings. +// +// For getting better compression ratios when you are compressing data +// with long repeated sequences or compressing data that is similar to +// other data, while still compressing fast, you might look at first +// using BMDiff and then compressing the output of BMDiff with +// Snappy. + #ifndef THIRD_PARTY_SNAPPY_SNAPPY_H__ #define THIRD_PARTY_SNAPPY_SNAPPY_H__ - + #include <cstddef> #include <string> - + #include <util/generic/fwd.h> -#include "snappy-stubs-public.h" - -namespace snappy { - class Source; - class Sink; - - // ------------------------------------------------------------------------ - // Generic compression/decompression routines. - // ------------------------------------------------------------------------ - - // Compress the bytes read from "*source" and append to "*sink". Return the - // number of bytes written. - size_t Compress(Source* source, Sink* sink); - +#include "snappy-stubs-public.h" + +namespace snappy { + class Source; + class Sink; + + // ------------------------------------------------------------------------ + // Generic compression/decompression routines. + // ------------------------------------------------------------------------ + + // Compress the bytes read from "*source" and append to "*sink". Return the + // number of bytes written. + size_t Compress(Source* source, Sink* sink); + // Find the uncompressed length of the given stream, as given by the header. // Note that the true length could deviate from this; the stream could e.g. // be truncated. @@ -65,37 +65,37 @@ namespace snappy { // Also note that this leaves "*source" in a state that is unsuitable for // further operations, such as RawUncompress(). You will need to rewind // or recreate the source yourself before attempting any further calls. - bool GetUncompressedLength(Source* source, uint32* result); - - // ------------------------------------------------------------------------ - // Higher-level string based routines (should be sufficient for most users) - // ------------------------------------------------------------------------ - + bool GetUncompressedLength(Source* source, uint32* result); + + // ------------------------------------------------------------------------ + // Higher-level string based routines (should be sufficient for most users) + // ------------------------------------------------------------------------ + // Sets "*compressed" to the compressed version of "input[0,input_length-1]". // Original contents of *compressed are lost. - // + // // REQUIRES: "input[]" is not an alias of "*compressed". size_t Compress(const char* input, size_t input_length, std::string* compressed); size_t Compress(const char* input, size_t input_length, TString* compressed); - - // Decompresses "compressed[0,compressed_length-1]" to "*uncompressed". - // Original contents of "*uncompressed" are lost. - // - // REQUIRES: "compressed[]" is not an alias of "*uncompressed". - // - // returns false if the message is corrupted and could not be decompressed - bool Uncompress(const char* compressed, size_t compressed_length, + + // Decompresses "compressed[0,compressed_length-1]" to "*uncompressed". + // Original contents of "*uncompressed" are lost. + // + // REQUIRES: "compressed[]" is not an alias of "*uncompressed". + // + // returns false if the message is corrupted and could not be decompressed + bool Uncompress(const char* compressed, size_t compressed_length, std::string* uncompressed); bool Uncompress(const char* compressed, size_t compressed_length, TString* uncompressed); - + // Decompresses "compressed" to "*uncompressed". // // returns false if the message is corrupted and could not be decompressed bool Uncompress(Source* compressed, Sink* uncompressed); - + // This routine uncompresses as much of the "compressed" as possible // into sink. It returns the number of valid bytes added to sink // (extra invalid bytes may have been added due to errors; the caller @@ -104,45 +104,45 @@ namespace snappy { // encountered. size_t UncompressAsMuchAsPossible(Source* compressed, Sink* uncompressed); - // ------------------------------------------------------------------------ - // Lower-level character array based routines. May be useful for - // efficiency reasons in certain circumstances. - // ------------------------------------------------------------------------ - - // REQUIRES: "compressed" must point to an area of memory that is at - // least "MaxCompressedLength(input_length)" bytes in length. - // - // Takes the data stored in "input[0..input_length]" and stores - // it in the array pointed to by "compressed". - // - // "*compressed_length" is set to the length of the compressed output. - // - // Example: - // char* output = new char[snappy::MaxCompressedLength(input_length)]; - // size_t output_length; - // RawCompress(input, input_length, output, &output_length); - // ... Process(output, output_length) ... - // delete [] output; - void RawCompress(const char* input, - size_t input_length, - char* compressed, - size_t* compressed_length); - - // Given data in "compressed[0..compressed_length-1]" generated by - // calling the Snappy::Compress routine, this routine - // stores the uncompressed data to - // uncompressed[0..GetUncompressedLength(compressed)-1] - // returns false if the message is corrupted and could not be decrypted - bool RawUncompress(const char* compressed, size_t compressed_length, - char* uncompressed); - - // Given data from the byte source 'compressed' generated by calling - // the Snappy::Compress routine, this routine stores the uncompressed - // data to - // uncompressed[0..GetUncompressedLength(compressed,compressed_length)-1] - // returns false if the message is corrupted and could not be decrypted - bool RawUncompress(Source* compressed, char* uncompressed); - + // ------------------------------------------------------------------------ + // Lower-level character array based routines. May be useful for + // efficiency reasons in certain circumstances. + // ------------------------------------------------------------------------ + + // REQUIRES: "compressed" must point to an area of memory that is at + // least "MaxCompressedLength(input_length)" bytes in length. + // + // Takes the data stored in "input[0..input_length]" and stores + // it in the array pointed to by "compressed". + // + // "*compressed_length" is set to the length of the compressed output. + // + // Example: + // char* output = new char[snappy::MaxCompressedLength(input_length)]; + // size_t output_length; + // RawCompress(input, input_length, output, &output_length); + // ... Process(output, output_length) ... + // delete [] output; + void RawCompress(const char* input, + size_t input_length, + char* compressed, + size_t* compressed_length); + + // Given data in "compressed[0..compressed_length-1]" generated by + // calling the Snappy::Compress routine, this routine + // stores the uncompressed data to + // uncompressed[0..GetUncompressedLength(compressed)-1] + // returns false if the message is corrupted and could not be decrypted + bool RawUncompress(const char* compressed, size_t compressed_length, + char* uncompressed); + + // Given data from the byte source 'compressed' generated by calling + // the Snappy::Compress routine, this routine stores the uncompressed + // data to + // uncompressed[0..GetUncompressedLength(compressed,compressed_length)-1] + // returns false if the message is corrupted and could not be decrypted + bool RawUncompress(Source* compressed, char* uncompressed); + // Given data in "compressed[0..compressed_length-1]" generated by // calling the Snappy::Compress routine, this routine // stores the uncompressed data to the iovec "iov". The number of physical @@ -165,24 +165,24 @@ namespace snappy { bool RawUncompressToIOVec(Source* compressed, const struct iovec* iov, size_t iov_cnt); - // Returns the maximal size of the compressed representation of - // input data that is "source_bytes" bytes in length; - size_t MaxCompressedLength(size_t source_bytes); - - // REQUIRES: "compressed[]" was produced by RawCompress() or Compress() - // Returns true and stores the length of the uncompressed data in - // *result normally. Returns false on parsing error. - // This operation takes O(1) time. - bool GetUncompressedLength(const char* compressed, size_t compressed_length, - size_t* result); - - // Returns true iff the contents of "compressed[]" can be uncompressed - // successfully. Does not return the uncompressed data. Takes - // time proportional to compressed_length, but is usually at least - // a factor of four faster than actual decompression. - bool IsValidCompressedBuffer(const char* compressed, - size_t compressed_length); - + // Returns the maximal size of the compressed representation of + // input data that is "source_bytes" bytes in length; + size_t MaxCompressedLength(size_t source_bytes); + + // REQUIRES: "compressed[]" was produced by RawCompress() or Compress() + // Returns true and stores the length of the uncompressed data in + // *result normally. Returns false on parsing error. + // This operation takes O(1) time. + bool GetUncompressedLength(const char* compressed, size_t compressed_length, + size_t* result); + + // Returns true iff the contents of "compressed[]" can be uncompressed + // successfully. Does not return the uncompressed data. Takes + // time proportional to compressed_length, but is usually at least + // a factor of four faster than actual decompression. + bool IsValidCompressedBuffer(const char* compressed, + size_t compressed_length); + // Returns true iff the contents of "compressed" can be uncompressed // successfully. Does not return the uncompressed data. Takes // time proportional to *compressed length, but is usually at least @@ -196,18 +196,18 @@ namespace snappy { // can only store 16-bit offsets, and EmitCopy() also assumes the offset // is 65535 bytes or less. Note also that if you change this, it will // affect the framing format (see framing_format.txt). - // + // // Note that there might be older data around that is compressed with larger // block sizes, so the decompression code should not rely on the // non-existence of long backreferences. static constexpr int kBlockLog = 16; static constexpr size_t kBlockSize = 1 << kBlockLog; - + static constexpr int kMinHashTableBits = 8; static constexpr size_t kMinHashTableSize = 1 << kMinHashTableBits; - + static constexpr int kMaxHashTableBits = 14; static constexpr size_t kMaxHashTableSize = 1 << kMaxHashTableBits; -} // end namespace snappy - +} // end namespace snappy + #endif // THIRD_PARTY_SNAPPY_SNAPPY_H__ diff --git a/contrib/libs/snappy/ya.make b/contrib/libs/snappy/ya.make index 472daa0c80..4fb0aef69c 100644 --- a/contrib/libs/snappy/ya.make +++ b/contrib/libs/snappy/ya.make @@ -1,6 +1,6 @@ # Generated by devtools/yamaker from nixpkgs 92c884dfd7140a6c3e6c717cf8990f7a78524331. -LIBRARY() +LIBRARY() OWNER(g:cpp-contrib) @@ -11,7 +11,7 @@ ORIGINAL_SOURCE(https://github.com/google/snappy/archive/1.1.8.tar.gz) LICENSE(BSD-3-Clause) LICENSE_TEXTS(.yandex_meta/licenses.list.txt) - + ADDINCL( GLOBAL contrib/libs/snappy/include ) @@ -21,12 +21,12 @@ NO_COMPILER_WARNINGS() CFLAGS( -DHAVE_CONFIG_H ) - -SRCS( - snappy-c.cc + +SRCS( + snappy-c.cc snappy-sinksource.cc - snappy-stubs-internal.cc + snappy-stubs-internal.cc snappy.cc -) - -END() +) + +END() |