From 79e1d54e4c5da5c8aac758d3a1484800234dbd6a Mon Sep 17 00:00:00 2001
From: Jeff Becker <ampernand@gmail.com>
Date: Fri, 9 Dec 2016 09:08:03 -0500
Subject: [PATCH] implement simple bloom filter

---
 BloomFilter.cpp      | 69 ++++++++++++++++++++++++++++++++++++++++++++
 BloomFilter.h        | 31 ++++++++++++++++++++
 build/CMakeLists.txt |  1 +
 filelist.mk          |  2 +-
 4 files changed, 102 insertions(+), 1 deletion(-)
 create mode 100644 BloomFilter.cpp
 create mode 100644 BloomFilter.h

diff --git a/BloomFilter.cpp b/BloomFilter.cpp
new file mode 100644
index 00000000..eac931db
--- /dev/null
+++ b/BloomFilter.cpp
@@ -0,0 +1,69 @@
+#include "BloomFilter.h"
+#include "I2PEndian.h"
+#include <array>
+#include <openssl/sha.h>
+
+namespace i2p
+{
+namespace util
+{
+
+	/** @brief decaying bloom filter implementation */
+	class DecayingBloomFilter : public IBloomFilter
+	{
+	public:
+
+		DecayingBloomFilter(const std::size_t size)
+		{
+			m_Size = size;
+			m_Data = new uint8_t[size];
+		}
+
+		/** @brief implements IBloomFilter::~IBloomFilter */
+		~DecayingBloomFilter()
+		{
+			delete [] m_Data;
+		}
+
+		/** @brief implements IBloomFilter::Add */
+		bool Add(const uint8_t * data, std::size_t len)
+		{
+			std::size_t idx;
+			uint8_t mask;
+		  Get(data, len, idx, mask);
+			if(m_Data[idx] & mask) return false; // filter hit
+			m_Data[idx] |= mask;
+			return true;
+		}
+
+		/** @brief implements IBloomFilter::Decay */
+		void Decay()
+		{
+			// reset bloom filter buffer
+			memset(m_Data, 0, m_Size);
+		}
+
+	private:
+		/** @brief get bit index for for data */
+		void Get(const uint8_t * data, std::size_t len, std::size_t & idx, uint8_t & bm)
+		{
+			bm = 1;
+			uint8_t digest[32];
+			// TODO: use blake2 because it's faster
+			SHA256(data, len, digest);
+			uint64_t i = buf64toh(digest);
+			idx = i % m_Size;
+			bm <<= (i % 8);
+		}
+
+		uint8_t * m_Data;
+		std::size_t m_Size;
+	};
+
+
+	BloomFilterPtr BloomFilter(std::size_t capacity)
+	{
+		return std::make_shared<DecayingBloomFilter>(capacity);
+	}
+}
+}
diff --git a/BloomFilter.h b/BloomFilter.h
new file mode 100644
index 00000000..7d4b28fa
--- /dev/null
+++ b/BloomFilter.h
@@ -0,0 +1,31 @@
+#ifndef BLOOM_FILTER_H_
+#define BLOOM_FILTER_H_
+#include <memory>
+#include <cstdint>
+
+namespace i2p
+{
+namespace util
+{
+
+  /** @brief interface for bloom filter */
+  struct IBloomFilter
+  {
+
+    /** @brief destructor */
+    virtual ~IBloomFilter();
+    /** @brief add entry to bloom filter, return false if filter hit otherwise return true */
+    virtual bool Add(const uint8_t * data, std::size_t len) = 0;
+    /** @brief optionally decay old entries */
+    virtual void Decay();
+  };
+
+  typedef std::shared_ptr<IBloomFilter> BloomFilterPtr;
+
+  /** @brief create bloom filter */
+  BloomFilterPtr BloomFilter(std::size_t capacity = 1024 * 8);
+
+}
+}
+
+#endif
diff --git a/build/CMakeLists.txt b/build/CMakeLists.txt
index 8b42c063..a1fd19c2 100644
--- a/build/CMakeLists.txt
+++ b/build/CMakeLists.txt
@@ -26,6 +26,7 @@ set ( CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake_modules" )
 set ( CMAKE_SOURCE_DIR ".." )
 
 set (LIBI2PD_SRC
+  "${CMAKE_SOURCE_DIR}/BloomFilter.cpp"
   "${CMAKE_SOURCE_DIR}/Config.cpp"
   "${CMAKE_SOURCE_DIR}/Crypto.cpp"
   "${CMAKE_SOURCE_DIR}/Garlic.cpp"
diff --git a/filelist.mk b/filelist.mk
index 94ce2f22..76f58785 100644
--- a/filelist.mk
+++ b/filelist.mk
@@ -1,5 +1,5 @@
 LIB_SRC = \
-  Gzip.cpp Crypto.cpp Datagram.cpp Garlic.cpp I2NPProtocol.cpp LeaseSet.cpp \
+  BloomFilter.cpp Gzip.cpp Crypto.cpp Datagram.cpp Garlic.cpp I2NPProtocol.cpp LeaseSet.cpp \
   Log.cpp NTCPSession.cpp NetDb.cpp NetDbRequests.cpp Profiling.cpp \
   Reseed.cpp RouterContext.cpp RouterInfo.cpp Signature.cpp SSU.cpp \
   SSUSession.cpp SSUData.cpp Streaming.cpp Identity.cpp TransitTunnel.cpp \