/*
- * Copyright 2015 Facebook, Inc.
+ * Copyright 2016 Facebook, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* "Quasi-succinct indices" (arxiv:1206.4300).
*/
-#ifndef FOLLY_EXPERIMENTAL_ELIAS_FANO_CODING_H
-#define FOLLY_EXPERIMENTAL_ELIAS_FANO_CODING_H
+#pragma once
+#include <algorithm>
#include <cstdlib>
#include <limits>
#include <type_traits>
#include <folly/experimental/Select64.h>
#include <glog/logging.h>
-#ifndef __GNUC__
-#error EliasFanoCoding.h requires GCC
-#endif
-
#if !FOLLY_X64
#error EliasFanoCoding.h requires x86_64
#endif
-#if __BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__
-#error EliasFanoCoding.h requires little endianness
-#endif
-
namespace folly { namespace compression {
+static_assert(kIsLittleEndian, "EliasFanoCoding.h requires little endianness");
+
template <class Pointer>
struct EliasFanoCompressedListBase {
EliasFanoCompressedListBase() = default;
template <class OtherPointer>
EliasFanoCompressedListBase(
- const EliasFanoCompressedListBase<OtherPointer>& other)
+ const EliasFanoCompressedListBase<OtherPointer>& other)
: size(other.size),
numLowerBits(other.numLowerBits),
data(other.data),
skipPointers(reinterpret_cast<Pointer>(other.skipPointers)),
forwardPointers(reinterpret_cast<Pointer>(other.forwardPointers)),
lower(reinterpret_cast<Pointer>(other.lower)),
- upper(reinterpret_cast<Pointer>(other.upper)) {
- }
+ upper(reinterpret_cast<Pointer>(other.upper)) { }
- void free() {
- ::free(const_cast<unsigned char*>(data.data()));
+ template <class T = Pointer>
+ auto free() -> decltype(::free(T(nullptr))) {
+ return ::free(data.data());
}
- size_t upperSize() const {
- return data.end() - upper;
- }
+ size_t upperSize() const { return data.end() - upper; }
size_t size = 0;
uint8_t numLowerBits = 0;
size_t kForwardQuantum = 0> // 0 = disabled
struct EliasFanoEncoderV2 {
static_assert(std::is_integral<Value>::value &&
- std::is_unsigned<Value>::value,
+ std::is_unsigned<Value>::value,
"Value should be unsigned integral");
typedef EliasFanoCompressedList CompressedList;
+ typedef MutableEliasFanoCompressedList MutableCompressedList;
typedef Value ValueType;
typedef SkipValue SkipValueType;
// EliasFanoCompressedList has no ownership of it, you need to call
// free() explicitly.
template <class RandomAccessIterator>
- static EliasFanoCompressedList encode(RandomAccessIterator begin,
- RandomAccessIterator end) {
+ static MutableCompressedList encode(RandomAccessIterator begin,
+ RandomAccessIterator end) {
if (begin == end) {
- return EliasFanoCompressedList();
+ return MutableCompressedList();
}
EliasFanoEncoderV2 encoder(end - begin, *(end - 1));
for (; begin != end; ++begin) {
return encoder.finish();
}
- explicit EliasFanoEncoderV2(const MutableEliasFanoCompressedList& result)
+ explicit EliasFanoEncoderV2(const MutableCompressedList& result)
: lower_(result.lower),
upper_(result.upper),
skipPointers_(reinterpret_cast<SkipValueType*>(
- result.skipPointers)),
+ result.skipPointers)),
forwardPointers_(reinterpret_cast<SkipValueType*>(
- result.forwardPointers)),
+ result.forwardPointers)),
result_(result) {
- memset(result.data.data(), 0, result.data.size());
+ std::fill(result.data.begin(), result.data.end(), 0);
}
EliasFanoEncoderV2(size_t size, ValueType upperBound)
: EliasFanoEncoderV2(
- Layout::fromUpperBoundAndSize(upperBound, size).allocList()) {
- }
+ Layout::fromUpperBoundAndSize(upperBound, size).allocList()) { }
void add(ValueType value) {
+ CHECK_LT(value, std::numeric_limits<ValueType>::max());
CHECK_GE(value, lastValue_);
const auto numLowerBits = result_.numLowerBits;
++size_;
}
- const EliasFanoCompressedList& finish() const {
+ const MutableCompressedList& finish() const {
CHECK_EQ(size_, result_.size);
return result_;
}
size_t size_ = 0;
size_t skipPointersSize_ = 0;
- EliasFanoCompressedList result_;
+ MutableCompressedList result_;
};
template <class Value,
// more serialization-friendly way (upperSizeBits doesn't need
// to be known by this function, unlike upper).
- // '?: 1' is a workaround for false 'division by zero'
- // compile-time error.
- size_t numSkipPointers = (8 * upper - size) / (skipQuantum ?: 1);
+ size_t numSkipPointers = (8 * upper - size) / skipQuantum;
layout.skipPointers = numSkipPointers * sizeof(SkipValueType);
}
// Store (1-indexed) position of every forwardQuantum-th
// 1-bit in upper bits sequence.
/* static */ if (forwardQuantum != 0) {
- size_t numForwardPointers = size / (forwardQuantum ?: 1);
+ size_t numForwardPointers = size / forwardQuantum;
layout.forwardPointers = numForwardPointers * sizeof(SkipValueType);
}
return lower + upper + skipPointers + forwardPointers;
}
- template <typename Range>
+ template <class Range>
EliasFanoCompressedListBase<typename Range::iterator>
openList(Range& buf) const {
EliasFanoCompressedListBase<typename Range::iterator> result;
return result;
}
- MutableEliasFanoCompressedList allocList() const {
+ MutableCompressedList allocList() const {
uint8_t* buf = nullptr;
// WARNING: Current read/write logic assumes that the 7 bytes
// following the last byte of lower and upper sequences are
public:
typedef typename Encoder::ValueType ValueType;
- explicit UpperBitsReader(const EliasFanoCompressedList& list)
- : forwardPointers_(list.forwardPointers),
- skipPointers_(list.skipPointers),
- start_(list.upper) {
+ explicit UpperBitsReader(const typename Encoder::CompressedList& list)
+ : forwardPointers_(list.forwardPointers),
+ skipPointers_(list.skipPointers),
+ start_(list.upper) {
reset();
}
// Use forward pointer.
if (Encoder::forwardQuantum > 0 && n > Encoder::forwardQuantum) {
- // Workaround to avoid 'division by zero' compile-time error.
- constexpr size_t q = Encoder::forwardQuantum ?: 1;
-
- const size_t steps = position_ / q;
+ const size_t steps = position_ / Encoder::forwardQuantum;
const size_t dest =
folly::loadUnaligned<SkipValueType>(
forwardPointers_ + (steps - 1) * sizeof(SkipValueType));
- reposition(dest + steps * q);
- n = position_ + 1 - steps * q; // n is > 0.
+ reposition(dest + steps * Encoder::forwardQuantum);
+ n = position_ + 1 - steps * Encoder::forwardQuantum; // n is > 0.
// Correct inner_ will be set at the end.
}
// Use skip pointer.
if (Encoder::skipQuantum > 0 && v >= value_ + Encoder::skipQuantum) {
- // Workaround to avoid 'division by zero' compile-time error.
- constexpr size_t q = Encoder::skipQuantum ?: 1;
-
- const size_t steps = v / q;
+ const size_t steps = v / Encoder::skipQuantum;
const size_t dest =
folly::loadUnaligned<SkipValueType>(
skipPointers_ + (steps - 1) * sizeof(SkipValueType));
- reposition(dest + q * steps);
+ reposition(dest + Encoder::skipQuantum * steps);
position_ = dest - 1;
// Correct inner_ and value_ will be set during the next()
typedef Encoder EncoderType;
typedef typename Encoder::ValueType ValueType;
- explicit EliasFanoReader(const EliasFanoCompressedList& list)
+ explicit EliasFanoReader(const typename Encoder::CompressedList& list)
: size_(list.size),
lower_(list.lower),
upper_(list),
void reset() {
upper_.reset();
- value_ = 0;
+ value_ = kInvalidValue;
}
bool next() {
}
bool skipTo(ValueType value) {
- DCHECK_GE(value, value_);
- if (value <= value_) {
- return true;
- } else if (!kUnchecked && value > lastValue_) {
+ // Also works when value_ == kInvalidValue.
+ if (value != kInvalidValue) { DCHECK_GE(value + 1, value_ + 1); }
+
+ if (!kUnchecked && value > lastValue_) {
return setDone();
+ } else if (value == value_) {
+ return true;
}
size_t upperValue = (value >> numLowerBits_);
}
bool jump(size_t n) {
- if (LIKELY(n - 1 < size_)) { // n > 0 && n <= size_
- value_ = readLowerPart(n - 1) | (upper_.jump(n) << numLowerBits_);
- return true;
- } else if (n == 0) {
- reset();
+ if (LIKELY(n < size_)) { // Also checks that n != -1.
+ value_ = readLowerPart(n) | (upper_.jump(n + 1) << numLowerBits_);
return true;
}
return setDone();
}
bool jumpTo(ValueType value) {
- if (value <= 0) {
- reset();
- return true;
- } else if (!kUnchecked && value > lastValue_) {
+ if (!kUnchecked && value > lastValue_) {
return setDone();
}
size_t size() const { return size_; }
+ bool valid() const {
+ return position() < size(); // Also checks that position() != -1.
+ }
+
size_t position() const { return upper_.position(); }
- ValueType value() const { return value_; }
+ ValueType value() const {
+ DCHECK(valid());
+ return value_;
+ }
private:
+ constexpr static ValueType kInvalidValue =
+ std::numeric_limits<ValueType>::max(); // Must hold kInvalidValue + 1 == 0.
+
bool setDone() {
- value_ = std::numeric_limits<ValueType>::max();
+ value_ = kInvalidValue;
upper_.setDone(size_);
return false;
}
const uint8_t* lower_;
detail::UpperBitsReader<Encoder, Instructions> upper_;
const ValueType lowerMask_;
- ValueType value_ = 0;
+ ValueType value_ = kInvalidValue;
ValueType lastValue_;
uint8_t numLowerBits_;
};
}} // namespaces
-
-#endif // FOLLY_EXPERIMENTAL_ELIAS_FANO_CODING_H
projects / folly.git / blobdiff