Branch data Line data Source code
# 1 : : // Copyright (c) 2012-2020 The Bitcoin Core developers
# 2 : : // Distributed under the MIT software license, see the accompanying
# 3 : : // file COPYING or http://www.opensource.org/licenses/mit-license.php.
# 4 : :
# 5 : : #ifndef BITCOIN_BLOOM_H
# 6 : : #define BITCOIN_BLOOM_H
# 7 : :
# 8 : : #include <serialize.h>
# 9 : :
# 10 : : #include <vector>
# 11 : :
# 12 : : class COutPoint;
# 13 : : class CTransaction;
# 14 : : class uint256;
# 15 : :
# 16 : : //! 20,000 items with fp rate < 0.1% or 10,000 items and <0.0001%
# 17 : : static const unsigned int MAX_BLOOM_FILTER_SIZE = 36000; // bytes
# 18 : : static const unsigned int MAX_HASH_FUNCS = 50;
# 19 : :
# 20 : : /**
# 21 : : * First two bits of nFlags control how much IsRelevantAndUpdate actually updates
# 22 : : * The remaining bits are reserved
# 23 : : */
# 24 : : enum bloomflags
# 25 : : {
# 26 : : BLOOM_UPDATE_NONE = 0,
# 27 : : BLOOM_UPDATE_ALL = 1,
# 28 : : // Only adds outpoints to the filter if the output is a pay-to-pubkey/pay-to-multisig script
# 29 : : BLOOM_UPDATE_P2PUBKEY_ONLY = 2,
# 30 : : BLOOM_UPDATE_MASK = 3,
# 31 : : };
# 32 : :
# 33 : : /**
# 34 : : * BloomFilter is a probabilistic filter which SPV clients provide
# 35 : : * so that we can filter the transactions we send them.
# 36 : : *
# 37 : : * This allows for significantly more efficient transaction and block downloads.
# 38 : : *
# 39 : : * Because bloom filters are probabilistic, a SPV node can increase the false-
# 40 : : * positive rate, making us send it transactions which aren't actually its,
# 41 : : * allowing clients to trade more bandwidth for more privacy by obfuscating which
# 42 : : * keys are controlled by them.
# 43 : : */
# 44 : : class CBloomFilter
# 45 : : {
# 46 : : private:
# 47 : : std::vector<unsigned char> vData;
# 48 : : unsigned int nHashFuncs;
# 49 : : unsigned int nTweak;
# 50 : : unsigned char nFlags;
# 51 : :
# 52 : : unsigned int Hash(unsigned int nHashNum, const std::vector<unsigned char>& vDataToHash) const;
# 53 : :
# 54 : : public:
# 55 : : /**
# 56 : : * Creates a new bloom filter which will provide the given fp rate when filled with the given number of elements
# 57 : : * Note that if the given parameters will result in a filter outside the bounds of the protocol limits,
# 58 : : * the filter created will be as close to the given parameters as possible within the protocol limits.
# 59 : : * This will apply if nFPRate is very low or nElements is unreasonably high.
# 60 : : * nTweak is a constant which is added to the seed value passed to the hash function
# 61 : : * It should generally always be a random value (and is largely only exposed for unit testing)
# 62 : : * nFlags should be one of the BLOOM_UPDATE_* enums (not _MASK)
# 63 : : */
# 64 : : CBloomFilter(const unsigned int nElements, const double nFPRate, const unsigned int nTweak, unsigned char nFlagsIn);
# 65 : 9 : CBloomFilter() : nHashFuncs(0), nTweak(0), nFlags(0) {}
# 66 : :
# 67 : 15 : SERIALIZE_METHODS(CBloomFilter, obj) { READWRITE(obj.vData, obj.nHashFuncs, obj.nTweak, obj.nFlags); }
# 68 : :
# 69 : : void insert(const std::vector<unsigned char>& vKey);
# 70 : : void insert(const COutPoint& outpoint);
# 71 : : void insert(const uint256& hash);
# 72 : :
# 73 : : bool contains(const std::vector<unsigned char>& vKey) const;
# 74 : : bool contains(const COutPoint& outpoint) const;
# 75 : : bool contains(const uint256& hash) const;
# 76 : :
# 77 : : //! True if the size is <= MAX_BLOOM_FILTER_SIZE and the number of hash functions is <= MAX_HASH_FUNCS
# 78 : : //! (catch a filter which was just deserialized which was too big)
# 79 : : bool IsWithinSizeConstraints() const;
# 80 : :
# 81 : : //! Also adds any outputs which match the filter to the filter (to match their spending txes)
# 82 : : bool IsRelevantAndUpdate(const CTransaction& tx);
# 83 : : };
# 84 : :
# 85 : : /**
# 86 : : * RollingBloomFilter is a probabilistic "keep track of most recently inserted" set.
# 87 : : * Construct it with the number of items to keep track of, and a false-positive
# 88 : : * rate. Unlike CBloomFilter, by default nTweak is set to a cryptographically
# 89 : : * secure random value for you. Similarly rather than clear() the method
# 90 : : * reset() is provided, which also changes nTweak to decrease the impact of
# 91 : : * false-positives.
# 92 : : *
# 93 : : * contains(item) will always return true if item was one of the last N to 1.5*N
# 94 : : * insert()'ed ... but may also return true for items that were not inserted.
# 95 : : *
# 96 : : * It needs around 1.8 bytes per element per factor 0.1 of false positive rate.
# 97 : : * For example, if we want 1000 elements, we'd need:
# 98 : : * - ~1800 bytes for a false positive rate of 0.1
# 99 : : * - ~3600 bytes for a false positive rate of 0.01
# 100 : : * - ~5400 bytes for a false positive rate of 0.001
# 101 : : *
# 102 : : * If we make these simplifying assumptions:
# 103 : : * - logFpRate / log(0.5) doesn't get rounded or clamped in the nHashFuncs calculation
# 104 : : * - nElements is even, so that nEntriesPerGeneration == nElements / 2
# 105 : : *
# 106 : : * Then we get a more accurate estimate for filter bytes:
# 107 : : *
# 108 : : * 3/(log(256)*log(2)) * log(1/fpRate) * nElements
# 109 : : */
# 110 : : class CRollingBloomFilter
# 111 : : {
# 112 : : public:
# 113 : : CRollingBloomFilter(const unsigned int nElements, const double nFPRate);
# 114 : :
# 115 : : void insert(const std::vector<unsigned char>& vKey);
# 116 : : void insert(const uint256& hash);
# 117 : : bool contains(const std::vector<unsigned char>& vKey) const;
# 118 : : bool contains(const uint256& hash) const;
# 119 : :
# 120 : : void reset();
# 121 : :
# 122 : : private:
# 123 : : int nEntriesPerGeneration;
# 124 : : int nEntriesThisGeneration;
# 125 : : int nGeneration;
# 126 : : std::vector<uint64_t> data;
# 127 : : unsigned int nTweak;
# 128 : : int nHashFuncs;
# 129 : : };
# 130 : :
# 131 : : #endif // BITCOIN_BLOOM_H
|