snappy 0.0.17 → 0.4.0

data/vendor/snappy/cmake/SnappyConfig.cmake.in

@@ -0,0 +1,33 @@
+# Copyright 2019 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.
+
+@PACKAGE_INIT@
+
+include("${CMAKE_CURRENT_LIST_DIR}/SnappyTargets.cmake")
+
+check_required_components(Snappy)

data/vendor/snappy/cmake/config.h.in

@@ -0,0 +1,66 @@
+#ifndef THIRD_PARTY_SNAPPY_OPENSOURCE_CMAKE_CONFIG_H_
+#define THIRD_PARTY_SNAPPY_OPENSOURCE_CMAKE_CONFIG_H_
+
+/* Define to 1 if the compiler supports __attribute__((always_inline)). */
+#cmakedefine01 HAVE_ATTRIBUTE_ALWAYS_INLINE
+
+/* Define to 1 if the compiler supports __builtin_ctz and friends. */
+#cmakedefine01 HAVE_BUILTIN_CTZ
+
+/* Define to 1 if the compiler supports __builtin_expect. */
+#cmakedefine01 HAVE_BUILTIN_EXPECT
+
+/* Define to 1 if you have a definition for mmap() in <sys/mman.h>. */
+#cmakedefine01 HAVE_FUNC_MMAP
+
+/* Define to 1 if you have a definition for sysconf() in <unistd.h>. */
+#cmakedefine01 HAVE_FUNC_SYSCONF
+
+/* Define to 1 if you have the `lzo2' library (-llzo2). */
+#cmakedefine01 HAVE_LIBLZO2
+
+/* Define to 1 if you have the `z' library (-lz). */
+#cmakedefine01 HAVE_LIBZ
+
+/* Define to 1 if you have the `lz4' library (-llz4). */
+#cmakedefine01 HAVE_LIBLZ4
+
+/* Define to 1 if you have the <sys/mman.h> header file. */
+#cmakedefine01 HAVE_SYS_MMAN_H
+
+/* Define to 1 if you have the <sys/resource.h> header file. */
+#cmakedefine01 HAVE_SYS_RESOURCE_H
+
+/* Define to 1 if you have the <sys/time.h> header file. */
+#cmakedefine01 HAVE_SYS_TIME_H
+
+/* Define to 1 if you have the <sys/uio.h> header file. */
+#cmakedefine01 HAVE_SYS_UIO_H
+
+/* Define to 1 if you have the <unistd.h> header file. */
+#cmakedefine01 HAVE_UNISTD_H
+
+/* Define to 1 if you have the <windows.h> header file. */
+#cmakedefine01 HAVE_WINDOWS_H
+
+/* Define to 1 if you target processors with SSSE3+ and have <tmmintrin.h>. */
+#cmakedefine01 SNAPPY_HAVE_SSSE3
+
+/* Define to 1 if you target processors with SSE4.2 and have <crc32intrin.h>. */
+#cmakedefine01 SNAPPY_HAVE_X86_CRC32
+
+/* Define to 1 if you target processors with BMI2+ and have <bmi2intrin.h>. */
+#cmakedefine01 SNAPPY_HAVE_BMI2
+
+/* Define to 1 if you target processors with NEON and have <arm_neon.h>. */
+#cmakedefine01 SNAPPY_HAVE_NEON
+
+/* Define to 1 if you have <arm_neon.h> and <arm_acle.h> and want to optimize
+   compression speed by using __crc32cw from <arm_acle.h>. */
+#cmakedefine01 SNAPPY_HAVE_NEON_CRC32
+
+/* Define to 1 if your processor stores words with the most significant byte
+   first (like Motorola and SPARC, unlike Intel and VAX). */
+#cmakedefine01 SNAPPY_IS_BIG_ENDIAN
+
+#endif  // THIRD_PARTY_SNAPPY_OPENSOURCE_CMAKE_CONFIG_H_

data/vendor/snappy/docs/README.md

@@ -0,0 +1,72 @@
+Snappy is a compression/decompression library. It does not aim for maximum
+compression, or compatibility with any other compression library; instead, it
+aims for very high speeds and reasonable compression. For instance, compared
+to the fastest mode of zlib, Snappy is an order of magnitude faster for most
+inputs, but the resulting compressed files are anywhere from 20% to 100%
+bigger. On a single core of a Core i7 processor in 64-bit mode, Snappy
+compresses at about 250 MB/sec or more and decompresses at about 500 MB/sec
+or more.
+
+Snappy is widely used inside Google, in everything from BigTable and MapReduce
+to our internal RPC systems. (Snappy has previously been referred to as "Zippy"
+in some presentations and the likes.)
+
+For more information, please see the [README](../README.md). Benchmarks against
+a few other compression libraries (zlib, LZO, LZF, FastLZ, and QuickLZ) are
+included in the source code distribution. The source code also contains a
+[formal format specification](../format_description.txt), as well
+as a specification for a [framing format](../framing_format.txt) useful for
+higher-level framing and encapsulation of Snappy data, e.g. for transporting
+Snappy-compressed data across HTTP in a streaming fashion. Note that the Snappy
+distribution currently has no code implementing the latter, but some of the
+ports do (see below).
+
+Snappy is written in C++, but C bindings are included, and several bindings to
+other languages are maintained by third parties:
+
+* C#: [Snappy for .NET](http://snappy4net.codeplex.com/) (P/Invoke wrapper),
+  [Snappy.NET](http://snappy.angeloflogic.com/) (P/Invoke wrapper),
+  [Snappy.Sharp](https://github.com/jeffesp/Snappy.Sharp) (native
+  reimplementation)
+* [C port](http://github.com/andikleen/snappy-c)
+* [C++ MSVC packaging](http://snappy.angeloflogic.com/) (plus Windows binaries,
+  NuGet packages and command-line tool)
+* Common Lisp: [Library bindings](http://flambard.github.com/thnappy/),
+  [native reimplementation](https://github.com/brown/snappy)
+* Erlang: [esnappy](https://github.com/thekvs/esnappy),
+  [snappy-erlang-nif](https://github.com/fdmanana/snappy-erlang-nif)
+* [Go](https://github.com/golang/snappy/)
+* [Haskell](http://hackage.haskell.org/package/snappy)
+* [Haxe](https://github.com/MaddinXx/hxsnappy) (C++/Neko)
+* [iOS packaging](https://github.com/ideawu/snappy-ios)
+* Java: [JNI wrapper](https://github.com/xerial/snappy-java) (including the
+  framing format), [native reimplementation](http://code.google.com/p/jsnappy/),
+  [other native reimplementation](https://github.com/dain/snappy) (including
+  the framing format)
+* [Lua](https://github.com/forhappy/lua-snappy)
+* [Node.js](https://github.com/kesla/node-snappy) (including the [framing
+  format](https://github.com/kesla/node-snappy-stream))
+* [Perl](http://search.cpan.org/dist/Compress-Snappy/)
+* [PHP](https://github.com/kjdev/php-ext-snappy)
+* [Python](http://pypi.python.org/pypi/python-snappy) (including a command-line
+  tool for the framing format)
+* [R](https://github.com/lulyon/R-snappy)
+* [Ruby](https://github.com/miyucy/snappy)
+* [Rust](https://github.com/BurntSushi/rust-snappy)
+* [Smalltalk](https://github.com/mumez/sqnappy) (including the framing format)
+
+Snappy is used or is available as an alternative in software such as
+
+* [MongoDB](https://www.mongodb.com/)
+* [Cassandra](http://cassandra.apache.org/)
+* [Couchbase](http://www.couchbase.com/)
+* [Hadoop](http://hadoop.apache.org/)
+* [LessFS](http://www.lessfs.com/wordpress/)
+* [LevelDB](https://github.com/google/leveldb) (which is in turn used by
+  [Google Chrome](http://chrome.google.com/))
+* [Lucene](http://lucene.apache.org/)
+* [VoltDB](http://voltdb.com/)
+
+If you know of more, do not hesitate to let us know. The easiest way to get in
+touch is via the
+[Snappy discussion mailing list](http://groups.google.com/group/snappy-compression).

data/vendor/snappy/snappy-internal.h

@@ -33,24 +33,108 @@
 
 #include "snappy-stubs-internal.h"
 
+#if SNAPPY_HAVE_SSSE3
+// Please do not replace with <x86intrin.h> or with headers that assume more
+// advanced SSE versions without checking with all the OWNERS.
+#include <emmintrin.h>
+#include <tmmintrin.h>
+#endif
+
+#if SNAPPY_HAVE_NEON
+#include <arm_neon.h>
+#endif
+
+#if SNAPPY_HAVE_SSSE3 || SNAPPY_HAVE_NEON
+#define SNAPPY_HAVE_VECTOR_BYTE_SHUFFLE 1
+#else
+#define SNAPPY_HAVE_VECTOR_BYTE_SHUFFLE 0
+#endif
+
 namespace snappy {
 namespace internal {
 
+#if SNAPPY_HAVE_VECTOR_BYTE_SHUFFLE
+#if SNAPPY_HAVE_SSSE3
+using V128 = __m128i;
+#elif SNAPPY_HAVE_NEON
+using V128 = uint8x16_t;
+#endif
+
+// Load 128 bits of integer data. `src` must be 16-byte aligned.
+inline V128 V128_Load(const V128* src);
+
+// Load 128 bits of integer data. `src` does not need to be aligned.
+inline V128 V128_LoadU(const V128* src);
+
+// Store 128 bits of integer data. `dst` does not need to be aligned.
+inline void V128_StoreU(V128* dst, V128 val);
+
+// Shuffle packed 8-bit integers using a shuffle mask.
+// Each packed integer in the shuffle mask must be in [0,16).
+inline V128 V128_Shuffle(V128 input, V128 shuffle_mask);
+
+// Constructs V128 with 16 chars |c|.
+inline V128 V128_DupChar(char c);
+
+#if SNAPPY_HAVE_SSSE3
+inline V128 V128_Load(const V128* src) { return _mm_load_si128(src); }
+
+inline V128 V128_LoadU(const V128* src) { return _mm_loadu_si128(src); }
+
+inline void V128_StoreU(V128* dst, V128 val) { _mm_storeu_si128(dst, val); }
+
+inline V128 V128_Shuffle(V128 input, V128 shuffle_mask) {
+  return _mm_shuffle_epi8(input, shuffle_mask);
+}
+
+inline V128 V128_DupChar(char c) { return _mm_set1_epi8(c); }
+
+#elif SNAPPY_HAVE_NEON
+inline V128 V128_Load(const V128* src) {
+  return vld1q_u8(reinterpret_cast<const uint8_t*>(src));
+}
+
+inline V128 V128_LoadU(const V128* src) {
+  return vld1q_u8(reinterpret_cast<const uint8_t*>(src));
+}
+
+inline void V128_StoreU(V128* dst, V128 val) {
+  vst1q_u8(reinterpret_cast<uint8_t*>(dst), val);
+}
+
+inline V128 V128_Shuffle(V128 input, V128 shuffle_mask) {
+  assert(vminvq_u8(shuffle_mask) >= 0 && vmaxvq_u8(shuffle_mask) <= 15);
+  return vqtbl1q_u8(input, shuffle_mask);
+}
+
+inline V128 V128_DupChar(char c) { return vdupq_n_u8(c); }
+#endif
+#endif  // SNAPPY_HAVE_VECTOR_BYTE_SHUFFLE
+
+// Working memory performs a single allocation to hold all scratch space
+// required for compression.
 class WorkingMemory {
  public:
-  WorkingMemory() : large_table_(NULL) { }
-  ~WorkingMemory() { delete[] large_table_; }
+  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.
-  uint16* GetHashTable(size_t input_size, int* table_size);
+  uint16_t* GetHashTable(size_t fragment_size, int* table_size) const;
+  char* GetScratchInput() const { return input_; }
+  char* GetScratchOutput() const { return output_; }
 
  private:
-  uint16 small_table_[1<<10];    // 2KB
-  uint16* large_table_;          // Allocated only when needed
+  char* mem_;        // the allocated memory, never nullptr
+  size_t size_;      // the size of the allocated memory, never 0
+  uint16_t* table_;  // the pointer to the hashtable
+  char* input_;      // the pointer to the input scratch buffer
+  char* output_;     // the pointer to the output scratch buffer
 
-  DISALLOW_COPY_AND_ASSIGN(WorkingMemory);
+  // No copying
+  WorkingMemory(const WorkingMemory&);
+  void operator=(const WorkingMemory&);
 };
 
 // Flat array compression that does not emit the "uncompressed length"
@@ -67,7 +151,7 @@ class WorkingMemory {
 char* CompressFragment(const char* input,
                        size_t input_length,
                        char* op,
-                       uint16* table,
+                       uint16_t* table,
                        const int table_size);
 
 // Find the largest n such that
@@ -80,12 +164,19 @@ char* CompressFragment(const char* input,
 // Does not read *(s1 + (s2_limit - s2)) or beyond.
 // Requires that s2_limit >= s2.
 //
-// Separate implementation for x86_64, for speed.  Uses the fact that
-// x86_64 is little endian.
-#if defined(ARCH_K8)
+// In addition populate *data with the next 5 bytes from the end of the match.
+// This is only done if 8 bytes are available (s2_limit - s2 >= 8). The point is
+// that on some arch's this can be done faster in this routine than subsequent
+// loading from s2 + n.
+//
+// Separate implementation for 64-bit, little-endian cpus.
+#if !SNAPPY_IS_BIG_ENDIAN && \
+    (defined(__x86_64__) || defined(_M_X64) || defined(ARCH_PPC) || \
+     defined(ARCH_ARM))
 static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
                                                       const char* s2,
-                                                      const char* s2_limit) {
+                                                      const char* s2_limit,
+                                                      uint64_t* data) {
   assert(s2_limit >= s2);
   size_t matched = 0;
 
@@ -94,12 +185,72 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
   // uncommon code paths that determine, without extra effort, whether the match
   // length is less than 8.  In short, we are hoping to avoid a conditional
   // branch, and perhaps get better code layout from the C++ compiler.
-  if (PREDICT_TRUE(s2 <= s2_limit - 8)) {
-    uint64 a1 = UNALIGNED_LOAD64(s1);
-    uint64 a2 = UNALIGNED_LOAD64(s2);
-    if (a1 != a2) {
-      return std::pair<size_t, bool>(Bits::FindLSBSetNonZero64(a1 ^ a2) >> 3,
-                                     true);
+  if (SNAPPY_PREDICT_TRUE(s2 <= s2_limit - 16)) {
+    uint64_t a1 = UNALIGNED_LOAD64(s1);
+    uint64_t a2 = UNALIGNED_LOAD64(s2);
+    if (SNAPPY_PREDICT_TRUE(a1 != a2)) {
+      // This code is critical for performance. The reason is that it determines
+      // how much to advance `ip` (s2). This obviously depends on both the loads
+      // from the `candidate` (s1) and `ip`. Furthermore the next `candidate`
+      // depends on the advanced `ip` calculated here through a load, hash and
+      // new candidate hash lookup (a lot of cycles). This makes s1 (ie.
+      // `candidate`) the variable that limits throughput. This is the reason we
+      // go through hoops to have this function update `data` for the next iter.
+      // The straightforward code would use *data, given by
+      //
+      // *data = UNALIGNED_LOAD64(s2 + matched_bytes) (Latency of 5 cycles),
+      //
+      // as input for the hash table lookup to find next candidate. However
+      // this forces the load on the data dependency chain of s1, because
+      // matched_bytes directly depends on s1. However matched_bytes is 0..7, so
+      // we can also calculate *data by
+      //
+      // *data = AlignRight(UNALIGNED_LOAD64(s2), UNALIGNED_LOAD64(s2 + 8),
+      //                    matched_bytes);
+      //
+      // The loads do not depend on s1 anymore and are thus off the bottleneck.
+      // The straightforward implementation on x86_64 would be to use
+      //
+      // shrd rax, rdx, cl  (cl being matched_bytes * 8)
+      //
+      // unfortunately shrd with a variable shift has a 4 cycle latency. So this
+      // only wins 1 cycle. The BMI2 shrx instruction is a 1 cycle variable
+      // shift instruction but can only shift 64 bits. If we focus on just
+      // obtaining the least significant 4 bytes, we can obtain this by
+      //
+      // *data = ConditionalMove(matched_bytes < 4, UNALIGNED_LOAD64(s2),
+      //     UNALIGNED_LOAD64(s2 + 4) >> ((matched_bytes & 3) * 8);
+      //
+      // Writen like above this is not a big win, the conditional move would be
+      // a cmp followed by a cmov (2 cycles) followed by a shift (1 cycle).
+      // However matched_bytes < 4 is equal to
+      // static_cast<uint32_t>(xorval) != 0. Writen that way, the conditional
+      // move (2 cycles) can execute in parallel with FindLSBSetNonZero64
+      // (tzcnt), which takes 3 cycles.
+      uint64_t xorval = a1 ^ a2;
+      int shift = Bits::FindLSBSetNonZero64(xorval);
+      size_t matched_bytes = shift >> 3;
+      uint64_t a3 = UNALIGNED_LOAD64(s2 + 4);
+#ifndef __x86_64__
+      a2 = static_cast<uint32_t>(xorval) == 0 ? a3 : a2;
+#else
+      // Ideally this would just be
+      //
+      // a2 = static_cast<uint32_t>(xorval) == 0 ? a3 : a2;
+      //
+      // However clang correctly infers that the above statement participates on
+      // a critical data dependency chain and thus, unfortunately, refuses to
+      // use a conditional move (it's tuned to cut data dependencies). In this
+      // case there is a longer parallel chain anyway AND this will be fairly
+      // unpredictable.
+      asm("testl %k2, %k2\n\t"
+          "cmovzq %1, %0\n\t"
+          : "+r"(a2)
+          : "r"(a3), "r"(xorval)
+          : "cc");
+#endif
+      *data = a2 >> (shift & (3 * 8));
+      return std::pair<size_t, bool>(matched_bytes, true);
     } else {
       matched = 8;
       s2 += 8;
@@ -110,23 +261,40 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
   // 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 (PREDICT_TRUE(s2 <= s2_limit - 8)) {
-    if (UNALIGNED_LOAD64(s2) == UNALIGNED_LOAD64(s1 + matched)) {
+  while (SNAPPY_PREDICT_TRUE(s2 <= s2_limit - 16)) {
+    uint64_t a1 = UNALIGNED_LOAD64(s1 + matched);
+    uint64_t a2 = UNALIGNED_LOAD64(s2);
+    if (a1 == a2) {
       s2 += 8;
       matched += 8;
     } else {
-      uint64 x = UNALIGNED_LOAD64(s2) ^ UNALIGNED_LOAD64(s1 + matched);
-      int matching_bits = Bits::FindLSBSetNonZero64(x);
-      matched += matching_bits >> 3;
+      uint64_t xorval = a1 ^ a2;
+      int shift = Bits::FindLSBSetNonZero64(xorval);
+      size_t matched_bytes = shift >> 3;
+      uint64_t a3 = UNALIGNED_LOAD64(s2 + 4);
+#ifndef __x86_64__
+      a2 = static_cast<uint32_t>(xorval) == 0 ? a3 : a2;
+#else
+      asm("testl %k2, %k2\n\t"
+          "cmovzq %1, %0\n\t"
+          : "+r"(a2)
+          : "r"(a3), "r"(xorval)
+          : "cc");
+#endif
+      *data = a2 >> (shift & (3 * 8));
+      matched += matched_bytes;
       assert(matched >= 8);
       return std::pair<size_t, bool>(matched, false);
     }
   }
-  while (PREDICT_TRUE(s2 < s2_limit)) {
+  while (SNAPPY_PREDICT_TRUE(s2 < s2_limit)) {
     if (s1[matched] == *s2) {
       ++s2;
       ++matched;
     } else {
+      if (s2 <= s2_limit - 8) {
+        *data = UNALIGNED_LOAD64(s2);
+      }
       return std::pair<size_t, bool>(matched, matched < 8);
     }
   }
@@ -135,7 +303,8 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
 #else
 static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
                                                       const char* s2,
-                                                      const char* s2_limit) {
+                                                      const char* s2_limit,
+                                                      uint64_t* data) {
   // Implementation based on the x86-64 version, above.
   assert(s2_limit >= s2);
   int matched = 0;
@@ -146,15 +315,17 @@ static inline std::pair<size_t, bool> FindMatchLength(const char* s1,
     matched += 4;
   }
   if (LittleEndian::IsLittleEndian() && s2 <= s2_limit - 4) {
-    uint32 x = UNALIGNED_LOAD32(s2) ^ UNALIGNED_LOAD32(s1 + matched);
+    uint32_t x = UNALIGNED_LOAD32(s2) ^ UNALIGNED_LOAD32(s1 + matched);
     int matching_bits = Bits::FindLSBSetNonZero(x);
     matched += matching_bits >> 3;
+    s2 += matching_bits >> 3;
   } else {
     while ((s2 < s2_limit) && (s1[matched] == *s2)) {
       ++s2;
       ++matched;
     }
   }
+  if (s2 <= s2_limit - 8) *data = LittleEndian::Load64(s2);
   return std::pair<size_t, bool>(matched, matched < 8);
 }
 #endif
@@ -170,11 +341,6 @@ enum {
 };
 static const int kMaximumTagLength = 5;  // COPY_4_BYTE_OFFSET plus the actual offset.
 
-// Mapping from i in range [0,4] to a mask to extract the bottom 8*i bits
-static const uint32 wordmask[] = {
-  0u, 0xffu, 0xffffu, 0xffffffu, 0xffffffffu
-};
-
 // Data stored per entry in lookup table:
 //      Range   Bits-used       Description
 //      ------------------------------------
@@ -186,7 +352,8 @@ static const uint32 wordmask[] = {
 // because of efficiency reasons:
 //      (1) Extracting a byte is faster than a bit-field
 //      (2) It properly aligns copy offset so we do not need a <<8
-static const uint16 char_table[256] = {
+static constexpr uint16_t char_table[256] = {
+    // clang-format off
   0x0001, 0x0804, 0x1001, 0x2001, 0x0002, 0x0805, 0x1002, 0x2002,
   0x0003, 0x0806, 0x1003, 0x2003, 0x0004, 0x0807, 0x1004, 0x2004,
   0x0005, 0x0808, 0x1005, 0x2005, 0x0006, 0x0809, 0x1006, 0x2006,
@@ -218,7 +385,8 @@ static const uint16 char_table[256] = {
   0x0039, 0x0f04, 0x1039, 0x2039, 0x003a, 0x0f05, 0x103a, 0x203a,
   0x003b, 0x0f06, 0x103b, 0x203b, 0x003c, 0x0f07, 0x103c, 0x203c,
   0x0801, 0x0f08, 0x103d, 0x203d, 0x1001, 0x0f09, 0x103e, 0x203e,
-  0x1801, 0x0f0a, 0x103f, 0x203f, 0x2001, 0x0f0b, 0x1040, 0x2040
+  0x1801, 0x0f0a, 0x103f, 0x203f, 0x2001, 0x0f0b, 0x1040, 0x2040,
+    // clang-format on
 };
 
 }  // end namespace internal

data/vendor/snappy/snappy-sinksource.cc

@@ -26,23 +26,31 @@
 // (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 <stddef.h>
+#include <cstring>
 
 #include "snappy-sinksource.h"
 
 namespace snappy {
 
-Source::~Source() { }
+Source::~Source() = default;
 
-Sink::~Sink() { }
+Sink::~Sink() = default;
 
 char* Sink::GetAppendBuffer(size_t length, char* scratch) {
+  // TODO: Switch to [[maybe_unused]] when we can assume C++17.
+  (void)length;
+
   return scratch;
 }
 
 char* Sink::GetAppendBufferVariable(
       size_t min_size, size_t desired_size_hint, char* scratch,
       size_t scratch_size, size_t* allocated_size) {
+  // TODO: Switch to [[maybe_unused]] when we can assume C++17.
+  (void)min_size;
+  (void)desired_size_hint;
+
   *allocated_size = scratch_size;
   return scratch;
 }
@@ -55,7 +63,7 @@ void Sink::AppendAndTakeOwnership(
   (*deleter)(deleter_arg, bytes, n);
 }
 
-ByteArraySource::~ByteArraySource() { }
+ByteArraySource::~ByteArraySource() = default;
 
 size_t ByteArraySource::Available() const { return left_; }
 
@@ -74,22 +82,26 @@ 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);
+    std::memcpy(dest_, data, n);
   }
   dest_ += n;
 }
 
 char* UncheckedByteArraySink::GetAppendBuffer(size_t len, char* scratch) {
+  // TODO: Switch to [[maybe_unused]] when we can assume C++17.
+  (void)len;
+  (void)scratch;
+
   return dest_;
 }
 
 void UncheckedByteArraySink::AppendAndTakeOwnership(
-    char* data, size_t n,
+    char* bytes, size_t n,
     void (*deleter)(void*, const char*, size_t),
     void *deleter_arg) {
-  if (data != dest_) {
-    memcpy(dest_, data, n);
-    (*deleter)(deleter_arg, data, n);
+  if (bytes != dest_) {
+    std::memcpy(dest_, bytes, n);
+    (*deleter)(deleter_arg, bytes, n);
   }
   dest_ += n;
 }
@@ -97,6 +109,11 @@ void UncheckedByteArraySink::AppendAndTakeOwnership(
 char* UncheckedByteArraySink::GetAppendBufferVariable(
       size_t min_size, size_t desired_size_hint, char* scratch,
       size_t scratch_size, size_t* allocated_size) {
+  // TODO: Switch to [[maybe_unused]] when we can assume C++17.
+  (void)min_size;
+  (void)scratch;
+  (void)scratch_size;
+
   *allocated_size = desired_size_hint;
   return dest_;
 }

data/vendor/snappy/snappy-sinksource.h

@@ -146,10 +146,10 @@ class Source {
 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);
+  ~ByteArraySource() override;
+  size_t Available() const override;
+  const char* Peek(size_t* len) override;
+  void Skip(size_t n) override;
  private:
   const char* ptr_;
   size_t left_;
@@ -159,15 +159,15 @@ class ByteArraySource : public Source {
 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(
+  ~UncheckedByteArraySink() override;
+  void Append(const char* data, size_t n) override;
+  char* GetAppendBuffer(size_t len, char* scratch) override;
+  char* GetAppendBufferVariable(
       size_t min_size, size_t desired_size_hint, char* scratch,
-      size_t scratch_size, size_t* allocated_size);
-  virtual void AppendAndTakeOwnership(
+      size_t scratch_size, size_t* allocated_size) override;
+  void AppendAndTakeOwnership(
       char* bytes, size_t n, void (*deleter)(void*, const char*, size_t),
-      void *deleter_arg);
+      void *deleter_arg) override;
 
   // Return the current output pointer so that a caller can see how
   // many bytes were produced.

data/vendor/snappy/snappy-stubs-internal.cc

@@ -33,7 +33,7 @@
 
 namespace snappy {
 
-void Varint::Append32(string* s, uint32 value) {
+void Varint::Append32(std::string* s, uint32_t value) {
   char buf[Varint::kMax32];
   const char* p = Varint::Encode32(buf, value);
   s->append(buf, p - buf);