You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@couchdb.apache.org by dc...@apache.org on 2014/03/13 15:01:26 UTC
[2/3] snappy: import 1.1.1
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/c_src/snappy/snappy.cc
----------------------------------------------------------------------
diff --git a/src/snappy/c_src/snappy/snappy.cc b/src/snappy/c_src/snappy/snappy.cc
new file mode 100644
index 0000000..a591aba
--- /dev/null
+++ b/src/snappy/c_src/snappy/snappy.cc
@@ -0,0 +1,1021 @@
+// 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"
+
+#include <stdio.h>
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+
+namespace snappy {
+
+// 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;
+}
+
+enum {
+ LITERAL = 0,
+ COPY_1_BYTE_OFFSET = 1, // 3 bit length + 3 bits of offset in opcode
+ COPY_2_BYTE_OFFSET = 2,
+ COPY_4_BYTE_OFFSET = 3
+};
+
+// Copy "len" bytes from "src" to "op", one byte at a time. Used for
+// handling COPY operations where the input and output regions may
+// overlap. For example, suppose:
+// src == "ab"
+// op == src + 2
+// len == 20
+// After IncrementalCopy(src, op, len), the result will have
+// eleven copies of "ab"
+// ababababababababababab
+// Note that this does not match the semantics of either memcpy()
+// or memmove().
+static inline void IncrementalCopy(const char* src, char* op, int len) {
+ DCHECK_GT(len, 0);
+ do {
+ *op++ = *src++;
+ } while (--len > 0);
+}
+
+// Equivalent to IncrementalCopy except that it can write up to ten extra
+// bytes after the end of the copy, and that it is faster.
+//
+// The main part of this loop is a simple copy of eight bytes at a time until
+// we've copied (at least) the requested amount of bytes. However, if op and
+// src are less than eight bytes apart (indicating a repeating pattern of
+// length < 8), we first need to expand the pattern in order to get the correct
+// results. For instance, if the buffer looks like this, with the eight-byte
+// <src> and <op> patterns marked as intervals:
+//
+// abxxxxxxxxxxxx
+// [------] src
+// [------] op
+//
+// a single eight-byte copy from <src> to <op> will repeat the pattern once,
+// after which we can move <op> two bytes without moving <src>:
+//
+// ababxxxxxxxxxx
+// [------] src
+// [------] op
+//
+// and repeat the exercise until the two no longer overlap.
+//
+// This allows us to do very well in the special case of one single byte
+// repeated many times, without taking a big hit for more general cases.
+//
+// The worst case of extra writing past the end of the match occurs when
+// op - src == 1 and len == 1; the last copy will read from byte positions
+// [0..7] and write to [4..11], whereas it was only supposed to write to
+// position 1. Thus, ten excess bytes.
+
+namespace {
+
+const int kMaxIncrementCopyOverflow = 10;
+
+} // namespace
+
+static inline void IncrementalCopyFastPath(const char* src, char* op, int len) {
+ while (op - src < 8) {
+ UNALIGNED_STORE64(op, UNALIGNED_LOAD64(src));
+ len -= op - src;
+ op += op - src;
+ }
+ while (len > 0) {
+ UNALIGNED_STORE64(op, UNALIGNED_LOAD64(src));
+ src += 8;
+ op += 8;
+ len -= 8;
+ }
+}
+
+static inline char* EmitLiteral(char* op,
+ const char* literal,
+ int len,
+ bool allow_fast_path) {
+ int n = len - 1; // Zero-length literals are disallowed
+ if (n < 60) {
+ // Fits in tag byte
+ *op++ = LITERAL | (n << 2);
+
+ // The vast majority of copies are below 16 bytes, for which a
+ // call to memcpy is overkill. This fast path can sometimes
+ // copy up to 15 bytes too much, but that is okay in the
+ // main loop, since we have a bit to go on for both sides:
+ //
+ // - The input will always have kInputMarginBytes = 15 extra
+ // available bytes, as long as we're in the main loop, and
+ // if not, allow_fast_path = false.
+ // - The output will always have 32 spare bytes (see
+ // MaxCompressedLength).
+ if (allow_fast_path && len <= 16) {
+ UNALIGNED_STORE64(op, UNALIGNED_LOAD64(literal));
+ UNALIGNED_STORE64(op + 8, UNALIGNED_LOAD64(literal + 8));
+ return op + len;
+ }
+ } else {
+ // Encode in upcoming bytes
+ char* base = op;
+ int count = 0;
+ op++;
+ while (n > 0) {
+ *op++ = n & 0xff;
+ n >>= 8;
+ count++;
+ }
+ assert(count >= 1);
+ assert(count <= 4);
+ *base = LITERAL | ((59+count) << 2);
+ }
+ memcpy(op, literal, len);
+ return op + len;
+}
+
+static inline char* EmitCopyLessThan64(char* op, int offset, int len) {
+ DCHECK_LE(len, 64);
+ DCHECK_GE(len, 4);
+ DCHECK_LT(offset, 65536);
+
+ if ((len < 12) && (offset < 2048)) {
+ int len_minus_4 = len - 4;
+ assert(len_minus_4 < 8); // Must fit in 3 bits
+ *op++ = COPY_1_BYTE_OFFSET | ((len_minus_4) << 2) | ((offset >> 8) << 5);
+ *op++ = offset & 0xff;
+ } else {
+ *op++ = COPY_2_BYTE_OFFSET | ((len-1) << 2);
+ LittleEndian::Store16(op, offset);
+ op += 2;
+ }
+ return op;
+}
+
+static inline char* EmitCopy(char* op, int offset, int len) {
+ // Emit 64 byte copies but make sure to keep at least four bytes reserved
+ while (len >= 68) {
+ op = EmitCopyLessThan64(op, offset, 64);
+ len -= 64;
+ }
+
+ // Emit an extra 60 byte copy if have too much data to fit in one copy
+ if (len > 64) {
+ op = EmitCopyLessThan64(op, offset, 60);
+ len -= 60;
+ }
+
+ // Emit remainder
+ op = EmitCopyLessThan64(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;
+ }
+}
+
+namespace internal {
+uint16* WorkingMemory::GetHashTable(size_t input_size, int* table_size) {
+ // Use smaller hash table when input.size() is smaller, since we
+ // fill the table, incurring O(hash table size) overhead for
+ // compression, and if the input is short, we won't need that
+ // many hash table entries anyway.
+ assert(kMaxHashTableSize >= 256);
+ int htsize = 256;
+ while (htsize < kMaxHashTableSize && htsize < input_size) {
+ htsize <<= 1;
+ }
+ CHECK_EQ(0, htsize & (htsize - 1)) << ": must be power of two";
+ CHECK_LE(htsize, kMaxHashTableSize) << ": hash table too large";
+
+ uint16* table;
+ if (htsize <= ARRAYSIZE(small_table_)) {
+ table = small_table_;
+ } else {
+ if (large_table_ == NULL) {
+ large_table_ = new uint16[kMaxHashTableSize];
+ }
+ table = large_table_;
+ }
+
+ *table_size = htsize;
+ memset(table, 0, htsize * sizeof(*table));
+ return table;
+}
+} // end namespace internal
+
+// For 0 <= offset <= 4, GetUint32AtOffset(UNALIGNED_LOAD64(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.
+static inline uint32 GetUint32AtOffset(uint64 v, int offset) {
+ DCHECK(0 <= offset && offset <= 4) << offset;
+ return v >> (LittleEndian::IsLittleEndian() ? 8 * offset : 32 - 8 * offset);
+}
+
+// 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* const input,
+ const 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;
+ CHECK_LE(input_size, kBlockSize);
+ CHECK_EQ(table_size & (table_size - 1), 0) << ": table must be power of two";
+ const int shift = 32 - Bits::Log2Floor(table_size);
+ DCHECK_EQ(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 int kInputMarginBytes = 15;
+ if (PREDICT_TRUE(input_size >= kInputMarginBytes)) {
+ const char* ip_limit = input + input_size - kInputMarginBytes;
+
+ for (uint32 next_hash = Hash(++ip, shift); ; ) {
+ DCHECK_LT(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
+ // scanned, 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;
+ DCHECK_EQ(hash, Hash(ip, shift));
+ uint32 bytes_between_hash_lookups = skip++ >> 5;
+ next_ip = ip + bytes_between_hash_lookups;
+ if (PREDICT_FALSE(next_ip > ip_limit)) {
+ goto emit_remainder;
+ }
+ next_hash = Hash(next_ip, shift);
+ candidate = base_ip + table[hash];
+ DCHECK_GE(candidate, base_ip);
+ DCHECK_LT(candidate, ip);
+
+ table[hash] = ip - base_ip;
+ } while (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."
+ DCHECK_LE(next_emit + 16, ip_end);
+ op = EmitLiteral(op, next_emit, ip - next_emit, true);
+
+ // 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.
+ uint64 input_bytes = 0;
+ 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;
+ int matched = 4 + FindMatchLength(candidate + 4, ip + 4, ip_end);
+ ip += matched;
+ int offset = base - candidate;
+ DCHECK_EQ(0, memcmp(base, candidate, matched));
+ op = EmitCopy(op, offset, matched);
+ // We could immediately start working at ip now, but to improve
+ // compression we first update table[Hash(ip - 1, ...)].
+ const char* insert_tail = ip - 1;
+ next_emit = ip;
+ if (PREDICT_FALSE(ip >= ip_limit)) {
+ goto emit_remainder;
+ }
+ input_bytes = UNALIGNED_LOAD64(insert_tail);
+ 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(op, next_emit, ip_end - next_emit, false);
+ }
+
+ return op;
+}
+} // end namespace internal
+
+// 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, uint32 length, bool allow_fast_path);
+// bool AppendFromSelf(uint32 offset, uint32 length);
+// };
+//
+// "allow_fast_path" is a parameter that says if there is at least 16
+// readable bytes in "ip". It is currently only used by SnappyArrayWriter.
+
+// -----------------------------------------------------------------------
+// Lookup table for decompression code. Generated by ComputeTable() below.
+// -----------------------------------------------------------------------
+
+// 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
+// ------------------------------------
+// 1..64 0..7 Literal/copy length encoded in opcode byte
+// 0..7 8..10 Copy offset encoded in opcode byte / 256
+// 0..4 11..13 Extra bytes after opcode
+//
+// We use eight bits for the length even though 7 would have sufficed
+// 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] = {
+ 0x0001, 0x0804, 0x1001, 0x2001, 0x0002, 0x0805, 0x1002, 0x2002,
+ 0x0003, 0x0806, 0x1003, 0x2003, 0x0004, 0x0807, 0x1004, 0x2004,
+ 0x0005, 0x0808, 0x1005, 0x2005, 0x0006, 0x0809, 0x1006, 0x2006,
+ 0x0007, 0x080a, 0x1007, 0x2007, 0x0008, 0x080b, 0x1008, 0x2008,
+ 0x0009, 0x0904, 0x1009, 0x2009, 0x000a, 0x0905, 0x100a, 0x200a,
+ 0x000b, 0x0906, 0x100b, 0x200b, 0x000c, 0x0907, 0x100c, 0x200c,
+ 0x000d, 0x0908, 0x100d, 0x200d, 0x000e, 0x0909, 0x100e, 0x200e,
+ 0x000f, 0x090a, 0x100f, 0x200f, 0x0010, 0x090b, 0x1010, 0x2010,
+ 0x0011, 0x0a04, 0x1011, 0x2011, 0x0012, 0x0a05, 0x1012, 0x2012,
+ 0x0013, 0x0a06, 0x1013, 0x2013, 0x0014, 0x0a07, 0x1014, 0x2014,
+ 0x0015, 0x0a08, 0x1015, 0x2015, 0x0016, 0x0a09, 0x1016, 0x2016,
+ 0x0017, 0x0a0a, 0x1017, 0x2017, 0x0018, 0x0a0b, 0x1018, 0x2018,
+ 0x0019, 0x0b04, 0x1019, 0x2019, 0x001a, 0x0b05, 0x101a, 0x201a,
+ 0x001b, 0x0b06, 0x101b, 0x201b, 0x001c, 0x0b07, 0x101c, 0x201c,
+ 0x001d, 0x0b08, 0x101d, 0x201d, 0x001e, 0x0b09, 0x101e, 0x201e,
+ 0x001f, 0x0b0a, 0x101f, 0x201f, 0x0020, 0x0b0b, 0x1020, 0x2020,
+ 0x0021, 0x0c04, 0x1021, 0x2021, 0x0022, 0x0c05, 0x1022, 0x2022,
+ 0x0023, 0x0c06, 0x1023, 0x2023, 0x0024, 0x0c07, 0x1024, 0x2024,
+ 0x0025, 0x0c08, 0x1025, 0x2025, 0x0026, 0x0c09, 0x1026, 0x2026,
+ 0x0027, 0x0c0a, 0x1027, 0x2027, 0x0028, 0x0c0b, 0x1028, 0x2028,
+ 0x0029, 0x0d04, 0x1029, 0x2029, 0x002a, 0x0d05, 0x102a, 0x202a,
+ 0x002b, 0x0d06, 0x102b, 0x202b, 0x002c, 0x0d07, 0x102c, 0x202c,
+ 0x002d, 0x0d08, 0x102d, 0x202d, 0x002e, 0x0d09, 0x102e, 0x202e,
+ 0x002f, 0x0d0a, 0x102f, 0x202f, 0x0030, 0x0d0b, 0x1030, 0x2030,
+ 0x0031, 0x0e04, 0x1031, 0x2031, 0x0032, 0x0e05, 0x1032, 0x2032,
+ 0x0033, 0x0e06, 0x1033, 0x2033, 0x0034, 0x0e07, 0x1034, 0x2034,
+ 0x0035, 0x0e08, 0x1035, 0x2035, 0x0036, 0x0e09, 0x1036, 0x2036,
+ 0x0037, 0x0e0a, 0x1037, 0x2037, 0x0038, 0x0e0b, 0x1038, 0x2038,
+ 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
+};
+
+// In debug mode, allow optional computation of the table at startup.
+// Also, check that the decompression table is correct.
+#ifndef NDEBUG
+DEFINE_bool(snappy_dump_decompression_table, false,
+ "If true, we print the decompression table at startup.");
+
+static uint16 MakeEntry(unsigned int extra,
+ unsigned int len,
+ unsigned int copy_offset) {
+ // Check that all of the fields fit within the allocated space
+ DCHECK_EQ(extra, extra & 0x7); // At most 3 bits
+ DCHECK_EQ(copy_offset, copy_offset & 0x7); // At most 3 bits
+ DCHECK_EQ(len, len & 0x7f); // At most 7 bits
+ return len | (copy_offset << 8) | (extra << 11);
+}
+
+static void ComputeTable() {
+ uint16 dst[256];
+
+ // Place invalid entries in all places to detect missing initialization
+ int assigned = 0;
+ for (int i = 0; i < 256; i++) {
+ dst[i] = 0xffff;
+ }
+
+ // Small LITERAL entries. We store (len-1) in the top 6 bits.
+ for (unsigned int len = 1; len <= 60; len++) {
+ dst[LITERAL | ((len-1) << 2)] = MakeEntry(0, len, 0);
+ assigned++;
+ }
+
+ // Large LITERAL entries. We use 60..63 in the high 6 bits to
+ // encode the number of bytes of length info that follow the opcode.
+ for (unsigned int extra_bytes = 1; extra_bytes <= 4; extra_bytes++) {
+ // We set the length field in the lookup table to 1 because extra
+ // bytes encode len-1.
+ dst[LITERAL | ((extra_bytes+59) << 2)] = MakeEntry(extra_bytes, 1, 0);
+ assigned++;
+ }
+
+ // COPY_1_BYTE_OFFSET.
+ //
+ // The tag byte in the compressed data stores len-4 in 3 bits, and
+ // offset/256 in 5 bits. offset%256 is stored in the next byte.
+ //
+ // This format is used for length in range [4..11] and offset in
+ // range [0..2047]
+ for (unsigned int len = 4; len < 12; len++) {
+ for (unsigned int offset = 0; offset < 2048; offset += 256) {
+ dst[COPY_1_BYTE_OFFSET | ((len-4)<<2) | ((offset>>8)<<5)] =
+ MakeEntry(1, len, offset>>8);
+ assigned++;
+ }
+ }
+
+ // COPY_2_BYTE_OFFSET.
+ // Tag contains len-1 in top 6 bits, and offset in next two bytes.
+ for (unsigned int len = 1; len <= 64; len++) {
+ dst[COPY_2_BYTE_OFFSET | ((len-1)<<2)] = MakeEntry(2, len, 0);
+ assigned++;
+ }
+
+ // COPY_4_BYTE_OFFSET.
+ // Tag contents len-1 in top 6 bits, and offset in next four bytes.
+ for (unsigned int len = 1; len <= 64; len++) {
+ dst[COPY_4_BYTE_OFFSET | ((len-1)<<2)] = MakeEntry(4, len, 0);
+ assigned++;
+ }
+
+ // Check that each entry was initialized exactly once.
+ CHECK_EQ(assigned, 256);
+ for (int i = 0; i < 256; i++) {
+ CHECK_NE(dst[i], 0xffff);
+ }
+
+ if (FLAGS_snappy_dump_decompression_table) {
+ printf("static const uint16 char_table[256] = {\n ");
+ for (int i = 0; i < 256; i++) {
+ printf("0x%04x%s",
+ dst[i],
+ ((i == 255) ? "\n" : (((i%8) == 7) ? ",\n " : ", ")));
+ }
+ printf("};\n");
+ }
+
+ // Check that computed table matched recorded table
+ for (int i = 0; i < 256; i++) {
+ CHECK_EQ(dst[i], char_table[i]);
+ }
+}
+REGISTER_MODULE_INITIALIZER(snappy, ComputeTable());
+#endif /* !NDEBUG */
+
+// 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_[5]; // Temporary buffer for PeekFast() boundaries
+
+ // 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 succcess, stores the length in *result and returns true.
+ // On failure, returns false.
+ bool ReadUncompressedLength(uint32* result) {
+ DCHECK(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);
+ *result |= static_cast<uint32>(c & 0x7f) << 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>
+ void DecompressAllTags(Writer* writer) {
+ const char* ip = ip_;
+ for ( ;; ) {
+ if (ip_limit_ - ip < 5) {
+ ip_ = ip;
+ if (!RefillTag()) return;
+ ip = ip_;
+ }
+
+ const unsigned char c = *(reinterpret_cast<const unsigned char*>(ip++));
+ const uint32 entry = char_table[c];
+ const uint32 trailer = LittleEndian::Load32(ip) & wordmask[entry >> 11];
+ ip += entry >> 11;
+ const uint32 length = entry & 0xff;
+
+ if ((c & 0x3) == LITERAL) {
+ uint32 literal_length = length + trailer;
+ uint32 avail = ip_limit_ - ip;
+ while (avail < literal_length) {
+ bool allow_fast_path = (avail >= 16);
+ if (!writer->Append(ip, avail, allow_fast_path)) 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;
+ }
+ bool allow_fast_path = (avail >= 16);
+ if (!writer->Append(ip, literal_length, allow_fast_path)) {
+ return;
+ }
+ ip += literal_length;
+ } else {
+ // 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 uint32 copy_offset = entry & 0x700;
+ if (!writer->AppendFromSelf(copy_offset + trailer, length)) {
+ return;
+ }
+ }
+ }
+ }
+};
+
+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 (n == 0) {
+ eof_ = true;
+ return false;
+ }
+ ip_limit_ = ip + n;
+ }
+
+ // Read the tag character
+ DCHECK_LT(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'
+ DCHECK_LE(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;
+ uint32 to_add = min<uint32>(needed - nbuf, length);
+ memcpy(scratch_ + nbuf, src, to_add);
+ nbuf += to_add;
+ reader_->Skip(to_add);
+ }
+ DCHECK_EQ(nbuf, needed);
+ ip_ = scratch_;
+ ip_limit_ = scratch_ + needed;
+ } else if (nbuf < 5) {
+ // 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,
+ uint32 max_len) {
+ // 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;
+ // Protect against possible DoS attack
+ if (static_cast<uint64>(uncompressed_len) > max_len) {
+ return false;
+ }
+
+ writer->SetExpectedLength(uncompressed_len);
+
+ // Process the entire input
+ decompressor.DecompressAllTags(writer);
+ 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;
+ int N = reader->Available();
+ char ulength[Varint::kMax32];
+ char* p = Varint::Encode32(ulength, N);
+ writer->Append(ulength, p-ulength);
+ written += (p - ulength);
+
+ internal::WorkingMemory wmem;
+ char* scratch = NULL;
+ char* scratch_output = NULL;
+
+ while (N > 0) {
+ // Get next block to compress (without copying if possible)
+ size_t fragment_size;
+ const char* fragment = reader->Peek(&fragment_size);
+ DCHECK_NE(fragment_size, 0) << ": premature end of input";
+ const int num_to_read = min(N, kBlockSize);
+ size_t bytes_read = fragment_size;
+
+ int 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 {
+ // Read into scratch buffer
+ if (scratch == NULL) {
+ // If this is the last iteration, we want to allocate N bytes
+ // of space, otherwise the max possible kBlockSize space.
+ // num_to_read contains exactly the correct value
+ scratch = new char[num_to_read];
+ }
+ memcpy(scratch, fragment, bytes_read);
+ reader->Skip(bytes_read);
+
+ while (bytes_read < num_to_read) {
+ fragment = reader->Peek(&fragment_size);
+ size_t n = min<size_t>(fragment_size, num_to_read - bytes_read);
+ memcpy(scratch + bytes_read, fragment, n);
+ bytes_read += n;
+ reader->Skip(n);
+ }
+ DCHECK_EQ(bytes_read, num_to_read);
+ fragment = scratch;
+ fragment_size = num_to_read;
+ }
+ DCHECK_EQ(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.
+ if (scratch_output == NULL) {
+ scratch_output = new char[max_output];
+ } else {
+ // Since we encode kBlockSize regions followed by a region
+ // which is <= kBlockSize in length, a previously allocated
+ // scratch_output[] region is big enough for this iteration.
+ }
+ char* dest = writer->GetAppendBuffer(max_output, scratch_output);
+ 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);
+ }
+
+ delete[] scratch;
+ delete[] scratch_output;
+
+ return written;
+}
+
+// -----------------------------------------------------------------------
+// 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) {
+ }
+
+ inline void SetExpectedLength(size_t len) {
+ op_limit_ = op_ + len;
+ }
+
+ inline bool CheckLength() const {
+ return op_ == op_limit_;
+ }
+
+ inline bool Append(const char* ip, uint32 len, bool allow_fast_path) {
+ char* op = op_;
+ const int space_left = op_limit_ - op;
+ if (allow_fast_path && len <= 16 && space_left >= 16) {
+ // Fast path, used for the majority (about 90%) of dynamic invocations.
+ UNALIGNED_STORE64(op, UNALIGNED_LOAD64(ip));
+ UNALIGNED_STORE64(op + 8, UNALIGNED_LOAD64(ip + 8));
+ } else {
+ if (space_left < len) {
+ return false;
+ }
+ memcpy(op, ip, len);
+ }
+ op_ = op + len;
+ return true;
+ }
+
+ inline bool AppendFromSelf(uint32 offset, uint32 len) {
+ char* op = op_;
+ const int space_left = op_limit_ - op;
+
+ if (op - base_ <= offset - 1u) { // -1u catches offset==0
+ return false;
+ }
+ if (len <= 16 && offset >= 8 && space_left >= 16) {
+ // Fast path, used for the majority (70-80%) of dynamic invocations.
+ UNALIGNED_STORE64(op, UNALIGNED_LOAD64(op - offset));
+ UNALIGNED_STORE64(op + 8, UNALIGNED_LOAD64(op - offset + 8));
+ } else {
+ if (space_left >= len + kMaxIncrementCopyOverflow) {
+ IncrementalCopyFastPath(op - offset, op, len);
+ } else {
+ if (space_left < len) {
+ return false;
+ }
+ IncrementalCopy(op - offset, op, len);
+ }
+ }
+
+ op_ = op + len;
+ return true;
+ }
+};
+
+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, kuint32max);
+}
+
+bool Uncompress(const char* compressed, size_t n, string* uncompressed) {
+ size_t ulength;
+ if (!GetUncompressedLength(compressed, n, &ulength)) {
+ return false;
+ }
+ // Protect against possible DoS attack
+ if ((static_cast<uint64>(ulength) + uncompressed->size()) >
+ uncompressed->max_size()) {
+ return false;
+ }
+ STLStringResizeUninitialized(uncompressed, ulength);
+ return RawUncompress(compressed, n, string_as_array(uncompressed));
+}
+
+
+// A Writer that drops everything on the floor and just does validation
+class SnappyDecompressionValidator {
+ private:
+ size_t expected_;
+ size_t produced_;
+
+ public:
+ inline SnappyDecompressionValidator() : produced_(0) { }
+ inline void SetExpectedLength(size_t len) {
+ expected_ = len;
+ }
+ inline bool CheckLength() const {
+ return expected_ == produced_;
+ }
+ inline bool Append(const char* ip, uint32 len, bool allow_fast_path) {
+ produced_ += len;
+ return produced_ <= expected_;
+ }
+ inline bool AppendFromSelf(uint32 offset, uint32 len) {
+ if (produced_ <= offset - 1u) return false; // -1u catches offset==0
+ produced_ += len;
+ return produced_ <= expected_;
+ }
+};
+
+bool IsValidCompressedBuffer(const char* compressed, size_t n) {
+ ByteArraySource reader(compressed, n);
+ SnappyDecompressionValidator writer;
+ return InternalUncompress(&reader, &writer, kuint32max);
+}
+
+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, string* compressed) {
+ // Pre-grow the buffer to the max length of the compressed output
+ compressed->resize(MaxCompressedLength(input_length));
+
+ size_t compressed_length;
+ RawCompress(input, input_length, string_as_array(compressed),
+ &compressed_length);
+ compressed->resize(compressed_length);
+ return compressed_length;
+}
+
+
+} // end namespace snappy
+
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/c_src/snappy/snappy.h
----------------------------------------------------------------------
diff --git a/src/snappy/c_src/snappy/snappy.h b/src/snappy/c_src/snappy/snappy.h
new file mode 100644
index 0000000..8d6ef22
--- /dev/null
+++ b/src/snappy/c_src/snappy/snappy.h
@@ -0,0 +1,155 @@
+// 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 UTIL_SNAPPY_SNAPPY_H__
+#define UTIL_SNAPPY_SNAPPY_H__
+
+#include <stddef.h>
+#include <string>
+
+#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);
+
+ bool GetUncompressedLength(Source* source, uint32* result);
+
+ // ------------------------------------------------------------------------
+ // Higher-level string based routines (should be sufficient for most users)
+ // ------------------------------------------------------------------------
+
+ // Sets "*output" to the compressed version of "input[0,input_length-1]".
+ // Original contents of *output are lost.
+ //
+ // REQUIRES: "input[]" is not an alias of "*output".
+ size_t Compress(const char* input, size_t input_length, string* output);
+
+ // 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,
+ string* 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);
+
+ // 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);
+
+ // *** DO NOT CHANGE THE VALUE OF kBlockSize ***
+ //
+ // New Compression code chops up the input into blocks of at most
+ // the following size. This ensures that back-references in the
+ // output never cross kBlockSize block boundaries. This can be
+ // helpful in implementing blocked decompression. However the
+ // decompression code should not rely on this guarantee since older
+ // compression code may not obey it.
+ static const int kBlockLog = 15;
+ static const int kBlockSize = 1 << kBlockLog;
+
+ static const int kMaxHashTableBits = 14;
+ static const int kMaxHashTableSize = 1 << kMaxHashTableBits;
+
+} // end namespace snappy
+
+
+#endif // UTIL_SNAPPY_SNAPPY_H__
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/c_src/snappy_nif.cc
----------------------------------------------------------------------
diff --git a/src/snappy/c_src/snappy_nif.cc b/src/snappy/c_src/snappy_nif.cc
new file mode 100644
index 0000000..93c1859
--- /dev/null
+++ b/src/snappy/c_src/snappy_nif.cc
@@ -0,0 +1,265 @@
+/**
+ * Copyright 2011, Filipe David Manana <fd...@apache.org>
+ * Web: http://github.com/fdmanana/snappy-erlang-nif
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License"); you may not
+ * use this file except in compliance with the License. You may obtain a copy of
+ * the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+ * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+ * License for the specific language governing permissions and limitations under
+ * the License.
+ **/
+
+#include <iostream>
+#include <cstring>
+
+#include "erl_nif_compat.h"
+#include "snappy/snappy.h"
+#include "snappy/snappy-sinksource.h"
+
+#ifdef OTP_R13B03
+#error OTP R13B03 not supported. Upgrade to R13B04 or later.
+#endif
+
+#ifdef __cplusplus
+#define BEGIN_C extern "C" {
+#define END_C }
+#else
+#define BEGIN_C
+#define END_C
+#endif
+
+#define SC_PTR(c) reinterpret_cast<char *>(c)
+
+class SnappyNifSink : public snappy::Sink
+{
+ public:
+ SnappyNifSink(ErlNifEnv* e);
+ ~SnappyNifSink();
+
+ void Append(const char* data, size_t n);
+ char* GetAppendBuffer(size_t len, char* scratch);
+ ErlNifBinary& getBin();
+
+ private:
+ ErlNifEnv* env;
+ ErlNifBinary bin;
+ size_t length;
+};
+
+SnappyNifSink::SnappyNifSink(ErlNifEnv* e) : env(e), length(0)
+{
+ if(!enif_alloc_binary_compat(env, 0, &bin)) {
+ env = NULL;
+ throw std::bad_alloc();
+ }
+}
+
+SnappyNifSink::~SnappyNifSink()
+{
+ if(env != NULL) {
+ enif_release_binary_compat(env, &bin);
+ }
+}
+
+void
+SnappyNifSink::Append(const char *data, size_t n)
+{
+ if(data != (SC_PTR(bin.data) + length)) {
+ memcpy(bin.data + length, data, n);
+ }
+ length += n;
+}
+
+char*
+SnappyNifSink::GetAppendBuffer(size_t len, char* scratch)
+{
+ size_t sz;
+
+ if((length + len) > bin.size) {
+ sz = (len * 4) < 8192 ? 8192 : (len * 4);
+
+ if(!enif_realloc_binary_compat(env, &bin, bin.size + sz)) {
+ throw std::bad_alloc();
+ }
+ }
+
+ return SC_PTR(bin.data) + length;
+}
+
+ErlNifBinary&
+SnappyNifSink::getBin()
+{
+ if(bin.size > length) {
+ if(!enif_realloc_binary_compat(env, &bin, length)) {
+ throw std::bad_alloc();
+ }
+ }
+ return bin;
+}
+
+
+static inline ERL_NIF_TERM
+make_atom(ErlNifEnv* env, const char* name)
+{
+ ERL_NIF_TERM ret;
+ if(enif_make_existing_atom_compat(env, name, &ret, ERL_NIF_LATIN1)) {
+ return ret;
+ }
+ return enif_make_atom(env, name);
+}
+
+
+static inline ERL_NIF_TERM
+make_ok(ErlNifEnv* env, ERL_NIF_TERM mesg)
+{
+ ERL_NIF_TERM ok = make_atom(env, "ok");
+ return enif_make_tuple2(env, ok, mesg);
+}
+
+
+static inline ERL_NIF_TERM
+make_error(ErlNifEnv* env, const char* mesg)
+{
+ ERL_NIF_TERM error = make_atom(env, "error");
+ return enif_make_tuple2(env, error, make_atom(env, mesg));
+}
+
+
+BEGIN_C
+
+
+ERL_NIF_TERM
+snappy_compress(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+ ErlNifBinary input;
+
+ if(!enif_inspect_iolist_as_binary(env, argv[0], &input)) {
+ return enif_make_badarg(env);
+ }
+
+ try {
+ snappy::ByteArraySource source(SC_PTR(input.data), input.size);
+ SnappyNifSink sink(env);
+ snappy::Compress(&source, &sink);
+ return make_ok(env, enif_make_binary(env, &sink.getBin()));
+ } catch(std::bad_alloc e) {
+ return make_error(env, "insufficient_memory");
+ } catch(...) {
+ return make_error(env, "unknown");
+ }
+}
+
+
+ERL_NIF_TERM
+snappy_decompress(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+ ErlNifBinary bin;
+ ErlNifBinary ret;
+ size_t len;
+
+ if(!enif_inspect_iolist_as_binary(env, argv[0], &bin)) {
+ return enif_make_badarg(env);
+ }
+
+ try {
+ if(!snappy::GetUncompressedLength(SC_PTR(bin.data), bin.size, &len)) {
+ return make_error(env, "data_not_compressed");
+ }
+
+ if(!enif_alloc_binary_compat(env, len, &ret)) {
+ return make_error(env, "insufficient_memory");
+ }
+
+ if(!snappy::RawUncompress(SC_PTR(bin.data), bin.size,
+ SC_PTR(ret.data))) {
+ return make_error(env, "corrupted_data");
+ }
+
+ return make_ok(env, enif_make_binary(env, &ret));
+ } catch(...) {
+ return make_error(env, "unknown");
+ }
+}
+
+
+ERL_NIF_TERM
+snappy_uncompressed_length(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+ ErlNifBinary bin;
+ size_t len;
+
+ if(!enif_inspect_iolist_as_binary(env, argv[0], &bin)) {
+ return enif_make_badarg(env);
+ }
+
+ try {
+ if(!snappy::GetUncompressedLength(SC_PTR(bin.data), bin.size, &len)) {
+ return make_error(env, "data_not_compressed");
+ }
+ return make_ok(env, enif_make_ulong(env, len));
+ } catch(...) {
+ return make_error(env, "unknown");
+ }
+}
+
+
+ERL_NIF_TERM
+snappy_is_valid(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+ ErlNifBinary bin;
+
+ if (!enif_inspect_iolist_as_binary(env, argv[0], &bin)) {
+ return enif_make_badarg(env);
+ }
+
+ try {
+ if(snappy::IsValidCompressedBuffer(SC_PTR(bin.data), bin.size)) {
+ return make_atom(env, "true");
+ } else {
+ return make_atom(env, "false");
+ }
+ } catch(...) {
+ return make_error(env, "unknown");
+ }
+}
+
+
+int
+on_load(ErlNifEnv* env, void** priv, ERL_NIF_TERM info)
+{
+ return 0;
+}
+
+
+int
+on_reload(ErlNifEnv* env, void** priv, ERL_NIF_TERM info)
+{
+ return 0;
+}
+
+
+int
+on_upgrade(ErlNifEnv* env, void** priv, void** old_priv, ERL_NIF_TERM info)
+{
+ return 0;
+}
+
+
+static ErlNifFunc nif_functions[] = {
+ {"compress", 1, snappy_compress},
+ {"decompress", 1, snappy_decompress},
+ {"uncompressed_length", 1, snappy_uncompressed_length},
+ {"is_valid", 1, snappy_is_valid}
+};
+
+
+ERL_NIF_INIT(snappy, nif_functions, &on_load, &on_reload, &on_upgrade, NULL);
+
+
+END_C
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/erl_nif_compat.h
----------------------------------------------------------------------
diff --git a/src/snappy/erl_nif_compat.h b/src/snappy/erl_nif_compat.h
deleted file mode 100644
index b8eb9b0..0000000
--- a/src/snappy/erl_nif_compat.h
+++ /dev/null
@@ -1,129 +0,0 @@
-/* Copyright (c) 2010-2011 Basho Technologies, Inc.
- * With some minor modifications by Filipe David Manana
- * <fd...@apache.org>
- *
- * This file is provided to you under the Apache License,
- * Version 2.0 (the "License"); you may not use this file
- * except in compliance with the License. You may obtain
- * a copy of the License at
- *
- * http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing,
- * software distributed under the License is distributed on an
- * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
- * KIND, either express or implied. See the License for the
- * specific language governing permissions and limitations
- * under the License.
-*/
-
-#ifndef ERL_NIF_COMPAT_H_
-#define ERL_NIF_COMPAT_H_
-
-#ifdef __cplusplus
-extern "C" {
-#endif /* __cplusplus */
-
-#include "erl_nif.h"
-
-
-#if ERL_NIF_MAJOR_VERSION == 0 && ERL_NIF_MINOR_VERSION == 1
-#define OTP_R13B03
-#elif ERL_NIF_MAJOR_VERSION == 1 && ERL_NIF_MINOR_VERSION == 0
-#define OTP_R13B04
-#elif ERL_NIF_MAJOR_VERSION == 2 && ERL_NIF_MINOR_VERSION == 0
-#define OTP_R14A
-#define OTP_R14B
-#define OTP_R14B01
-#elif ERL_NIF_MAJOR_VERSION == 2 && ERL_NIF_MINOR_VERSION == 1
-#define OTP_R14B02
-#endif
-
-
-#ifdef OTP_R13B03
-
-#define enif_open_resource_type_compat enif_open_resource_type
-#define enif_alloc_resource_compat enif_alloc_resource
-#define enif_release_resource_compat enif_release_resource
-#define enif_alloc_binary_compat enif_alloc_binary
-#define enif_alloc_compat enif_alloc
-#define enif_release_binary_compat enif_release_binary
-#define enif_free_compat enif_free
-#define enif_get_atom_compat enif_get_atom
-#define enif_priv_data_compat enif_get_data
-#define enif_make_uint_compat enif_make_ulong
-
-#define enif_make_existing_atom_compat(E, N, R, Enc) \
- enif_make_existing_atom(E, N, R)
-
-#define enif_make_string_compat(E, B, Enc) \
- enif_make_string(E, B)
-
-#endif /* R13B03 */
-
-
-#ifdef OTP_R13B04
-
-#define enif_open_resource_type_compat enif_open_resource_type
-#define enif_alloc_resource_compat enif_alloc_resource
-#define enif_release_resource_compat enif_release_resource
-#define enif_alloc_binary_compat enif_alloc_binary
-#define enif_realloc_binary_compat enif_realloc_binary
-#define enif_release_binary_compat enif_release_binary
-#define enif_alloc_compat enif_alloc
-#define enif_free_compat enif_free
-#define enif_get_atom_compat enif_get_atom
-#define enif_priv_data_compat enif_priv_data
-#define enif_make_string_compat enif_make_string
-#define enif_make_uint_compat enif_make_uint
-
-#define enif_make_existing_atom_compat(E, N, R, Enc) \
- enif_make_existing_atom(E, N, R)
-
-
-#endif /* R13B04 */
-
-
-/* OTP R14A and future releases */
-#if !defined(OTP_R13B03) && !defined(OTP_R13B04)
-
-#define enif_open_resource_type_compat(E, N, D, F, T) \
- enif_open_resource_type(E, NULL, N, D, F, T)
-
-#define enif_alloc_resource_compat(E, T, S) \
- enif_alloc_resource(T, S)
-
-#define enif_release_resource_compat(E, H) \
- enif_release_resource(H)
-
-#define enif_alloc_binary_compat(E, S, B) \
- enif_alloc_binary(S, B)
-
-#define enif_realloc_binary_compat(E, S, B) \
- enif_realloc_binary(S, B)
-
-#define enif_release_binary_compat(E, B) \
- enif_release_binary(B)
-
-#define enif_alloc_compat(E, S) \
- enif_alloc(S)
-
-#define enif_free_compat(E, P) \
- enif_free(P)
-
-#define enif_get_atom_compat(E, T, B, S) \
- enif_get_atom(E, T, B, S, ERL_NIF_LATIN1)
-
-#define enif_priv_data_compat enif_priv_data
-#define enif_make_string_compat enif_make_string
-#define enif_make_existing_atom_compat enif_make_existing_atom
-#define enif_make_uint_compat enif_make_uint
-
-#endif /* R14 and future releases */
-
-
-#ifdef __cplusplus
-}
-#endif /* __cplusplus */
-
-#endif /* ERL_NIF_COMPAT_H_ */
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/AUTHORS
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/AUTHORS b/src/snappy/google-snappy/AUTHORS
deleted file mode 100644
index 4858b37..0000000
--- a/src/snappy/google-snappy/AUTHORS
+++ /dev/null
@@ -1 +0,0 @@
-opensource@google.com
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/COPYING
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/COPYING b/src/snappy/google-snappy/COPYING
deleted file mode 100644
index 8d6bd9f..0000000
--- a/src/snappy/google-snappy/COPYING
+++ /dev/null
@@ -1,28 +0,0 @@
-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.
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/config.h.in
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/config.h.in b/src/snappy/google-snappy/config.h.in
deleted file mode 100644
index 8f912f6..0000000
--- a/src/snappy/google-snappy/config.h.in
+++ /dev/null
@@ -1,128 +0,0 @@
-/* config.h.in. Generated from configure.ac by autoheader. */
-
-/* Define if building universal (internal helper macro) */
-#undef AC_APPLE_UNIVERSAL_BUILD
-
-/* Define to 1 if the compiler supports __builtin_ctz and friends. */
-#undef HAVE_BUILTIN_CTZ
-
-/* Define to 1 if the compiler supports __builtin_expect. */
-#undef HAVE_BUILTIN_EXPECT
-
-/* Define to 1 if you have the <byteswap.h> header file. */
-#undef HAVE_BYTESWAP_H
-
-/* Define to 1 if you have the <dlfcn.h> header file. */
-#undef HAVE_DLFCN_H
-
-/* Use the gflags package for command-line parsing. */
-#undef HAVE_GFLAGS
-
-/* Defined when Google Test is available. */
-#undef HAVE_GTEST
-
-/* Define to 1 if you have the <inttypes.h> header file. */
-#undef HAVE_INTTYPES_H
-
-/* Define to 1 if you have the `fastlz' library (-lfastlz). */
-#undef HAVE_LIBFASTLZ
-
-/* Define to 1 if you have the `lzf' library (-llzf). */
-#undef HAVE_LIBLZF
-
-/* Define to 1 if you have the `lzo2' library (-llzo2). */
-#undef HAVE_LIBLZO2
-
-/* Define to 1 if you have the `quicklz' library (-lquicklz). */
-#undef HAVE_LIBQUICKLZ
-
-/* Define to 1 if you have the `z' library (-lz). */
-#undef HAVE_LIBZ
-
-/* Define to 1 if you have the <memory.h> header file. */
-#undef HAVE_MEMORY_H
-
-/* Define to 1 if you have the <stddef.h> header file. */
-#undef HAVE_STDDEF_H
-
-/* Define to 1 if you have the <stdint.h> header file. */
-#undef HAVE_STDINT_H
-
-/* Define to 1 if you have the <stdlib.h> header file. */
-#undef HAVE_STDLIB_H
-
-/* Define to 1 if you have the <strings.h> header file. */
-#undef HAVE_STRINGS_H
-
-/* Define to 1 if you have the <string.h> header file. */
-#undef HAVE_STRING_H
-
-/* Define to 1 if you have the <sys/byteswap.h> header file. */
-#undef HAVE_SYS_BYTESWAP_H
-
-/* Define to 1 if you have the <sys/endian.h> header file. */
-#undef HAVE_SYS_ENDIAN_H
-
-/* Define to 1 if you have the <sys/mman.h> header file. */
-#undef HAVE_SYS_MMAN_H
-
-/* Define to 1 if you have the <sys/resource.h> header file. */
-#undef HAVE_SYS_RESOURCE_H
-
-/* Define to 1 if you have the <sys/stat.h> header file. */
-#undef HAVE_SYS_STAT_H
-
-/* Define to 1 if you have the <sys/time.h> header file. */
-#undef HAVE_SYS_TIME_H
-
-/* Define to 1 if you have the <sys/types.h> header file. */
-#undef HAVE_SYS_TYPES_H
-
-/* Define to 1 if you have the <unistd.h> header file. */
-#undef HAVE_UNISTD_H
-
-/* Define to 1 if you have the <windows.h> header file. */
-#undef HAVE_WINDOWS_H
-
-/* Define to the sub-directory in which libtool stores uninstalled libraries.
- */
-#undef LT_OBJDIR
-
-/* Name of package */
-#undef PACKAGE
-
-/* Define to the address where bug reports for this package should be sent. */
-#undef PACKAGE_BUGREPORT
-
-/* Define to the full name of this package. */
-#undef PACKAGE_NAME
-
-/* Define to the full name and version of this package. */
-#undef PACKAGE_STRING
-
-/* Define to the one symbol short name of this package. */
-#undef PACKAGE_TARNAME
-
-/* Define to the home page for this package. */
-#undef PACKAGE_URL
-
-/* Define to the version of this package. */
-#undef PACKAGE_VERSION
-
-/* Define to 1 if you have the ANSI C header files. */
-#undef STDC_HEADERS
-
-/* Version number of package */
-#undef VERSION
-
-/* Define WORDS_BIGENDIAN to 1 if your processor stores words with the most
- significant byte first (like Motorola and SPARC, unlike Intel). */
-#if defined AC_APPLE_UNIVERSAL_BUILD
-# if defined __BIG_ENDIAN__
-# define WORDS_BIGENDIAN 1
-# endif
-#else
-# ifndef WORDS_BIGENDIAN
-# undef WORDS_BIGENDIAN
-# endif
-#endif
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/snappy-internal.h
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/snappy-internal.h b/src/snappy/google-snappy/snappy-internal.h
deleted file mode 100644
index c99d331..0000000
--- a/src/snappy/google-snappy/snappy-internal.h
+++ /dev/null
@@ -1,150 +0,0 @@
-// 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 UTIL_SNAPPY_SNAPPY_INTERNAL_H_
-#define UTIL_SNAPPY_SNAPPY_INTERNAL_H_
-
-#include "snappy-stubs-internal.h"
-
-namespace snappy {
-namespace internal {
-
-class WorkingMemory {
- public:
- WorkingMemory() : large_table_(NULL) { }
- ~WorkingMemory() { delete[] large_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 input_size, int* table_size);
-
- private:
- uint16 small_table_[1<<10]; // 2KB
- uint16* large_table_; // Allocated only when needed
-
- DISALLOW_COPY_AND_ASSIGN(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);
-
-// Return the largest n such that
-//
-// s1[0,n-1] == s2[0,n-1]
-// and n <= (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 x86_64, for speed. Uses the fact that
-// x86_64 is little endian.
-#if defined(ARCH_K8)
-static inline int FindMatchLength(const char* s1,
- const char* s2,
- const char* s2_limit) {
- assert(s2_limit >= s2);
- int matched = 0;
-
- // 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 (PREDICT_TRUE(s2 <= s2_limit - 8)) {
- if (PREDICT_FALSE(UNALIGNED_LOAD64(s2) == UNALIGNED_LOAD64(s1 + matched))) {
- s2 += 8;
- matched += 8;
- } else {
- // On current (mid-2008) Opteron models there is a 3% more
- // efficient code sequence to find the first non-matching byte.
- // However, what follows is ~10% better on Intel Core 2 and newer,
- // and we expect AMD's bsf instruction to improve.
- uint64 x = UNALIGNED_LOAD64(s2) ^ UNALIGNED_LOAD64(s1 + matched);
- int matching_bits = Bits::FindLSBSetNonZero64(x);
- matched += matching_bits >> 3;
- return matched;
- }
- }
- while (PREDICT_TRUE(s2 < s2_limit)) {
- if (PREDICT_TRUE(s1[matched] == *s2)) {
- ++s2;
- ++matched;
- } else {
- return matched;
- }
- }
- return matched;
-}
-#else
-static inline int FindMatchLength(const char* s1,
- const char* s2,
- const char* s2_limit) {
- // 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;
- }
- }
- return matched;
-}
-#endif
-
-} // end namespace internal
-} // end namespace snappy
-
-#endif // UTIL_SNAPPY_SNAPPY_INTERNAL_H_
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/snappy-sinksource.cc
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/snappy-sinksource.cc b/src/snappy/google-snappy/snappy-sinksource.cc
deleted file mode 100644
index 5844552..0000000
--- a/src/snappy/google-snappy/snappy-sinksource.cc
+++ /dev/null
@@ -1,71 +0,0 @@
-// 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;
-}
-
-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_;
-}
-
-}
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/snappy-sinksource.h
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/snappy-sinksource.h b/src/snappy/google-snappy/snappy-sinksource.h
deleted file mode 100644
index faabfa1..0000000
--- a/src/snappy/google-snappy/snappy-sinksource.h
+++ /dev/null
@@ -1,137 +0,0 @@
-// 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 UTIL_SNAPPY_SNAPPY_SINKSOURCE_H_
-#define UTIL_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);
-
-
- 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);
-
- // 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_;
-};
-
-
-}
-
-#endif // UTIL_SNAPPY_SNAPPY_SINKSOURCE_H_
http://git-wip-us.apache.org/repos/asf/couchdb/blob/13c5dcfa/src/snappy/google-snappy/snappy-stubs-internal.cc
----------------------------------------------------------------------
diff --git a/src/snappy/google-snappy/snappy-stubs-internal.cc b/src/snappy/google-snappy/snappy-stubs-internal.cc
deleted file mode 100644
index 6ed3343..0000000
--- a/src/snappy/google-snappy/snappy-stubs-internal.cc
+++ /dev/null
@@ -1,42 +0,0 @@
-// 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 {
-
-void Varint::Append32(string* s, uint32 value) {
- char buf[Varint::kMax32];
- const char* p = Varint::Encode32(buf, value);
- s->append(buf, p - buf);
-}
-
-} // namespace snappy