[feat] Implement multi-level skiplist for havenask indexlib

aios/catalog/tools/TableSchemaConfig.cpp

@@ -57,6 +57,7 @@ std::vector<std::string> IndexConfig::intParams = {"term_payload_flag",
                                                    "position_payload_flag",
                                                    "term_frequency_flag",
                                                    "term_frequency_bitmap",
+                                                   "multi_level_skip_list",
                                                    "format_version_id"};
 std::vector<std::string> IndexConfig::boolParams = {"has_primary_key_attribute"};
 

aios/storage/indexlib/index/inverted_index/BUILD

@@ -427,15 +427,15 @@ strict_cc_library(
         '//aios/storage/indexlib/index/inverted_index/format:InMemPostingDecoder',
         '//aios/storage/indexlib/index/inverted_index/format:ShortListSegmentDecoder',
         '//aios/storage/indexlib/index/inverted_index/format:SkipListSegmentDecoder',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:TriValueSkipListReader'
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib'
     ]
 )
 strict_cc_library(
     name='InDocStateKeeper',
     deps=[
         '//aios/storage/indexlib/index/inverted_index/format:InMemPositionListDecoder',
         '//aios/storage/indexlib/index/inverted_index/format:TermMeta',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:PairValueSkipListReader'
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib'
     ]
 )
 strict_cc_library(
@@ -462,8 +462,7 @@ strict_cc_library(
         '//aios/storage/indexlib/index/inverted_index/format:ShortListSegmentDecoder',
         '//aios/storage/indexlib/index/inverted_index/format:SkipListSegmentDecoder',
         '//aios/storage/indexlib/index/inverted_index/format:TermMetaLoader',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:PairValueSkipListReader',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:TriValueSkipListReader',
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib',
         '//aios/storage/indexlib/util:simple_heap'
     ]
 )

aios/storage/indexlib/index/inverted_index/BufferedIndexDecoder.cpp

@@ -24,6 +24,7 @@
 #include "indexlib/index/inverted_index/format/TermMetaLoader.h"
 #include "indexlib/index/inverted_index/format/skiplist/PairValueSkipListReader.h"
 #include "indexlib/index/inverted_index/format/skiplist/TriValueSkipListReader.h"
+#include "indexlib/index/inverted_index/format/skiplist/MultiLevelSkipListReader.h"
 
 namespace indexlib::index {
 AUTIL_LOG_SETUP(indexlib.index, BufferedIndexDecoder);
@@ -257,12 +258,26 @@ BufferedSegmentIndexDecoder* BufferedIndexDecoder::CreateNormalSegmentDecoder(ui
                                             compressMode, enableShortListVbyteCompress);
     }
 
-    if (mCurSegPostingFormatOption.HasTfList()) {
-        return IE_POOL_COMPATIBLE_NEW_CLASS(_sessionPool, SkipListSegmentDecoder<TriValueSkipListReader>, _sessionPool,
-                                            &_docListReader, docListBeginPos, compressMode);
+    if (mCurSegPostingFormatOption.HasMultiLevelSkipList()) {
+        if (mCurSegPostingFormatOption.HasTfList()) {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<MultiLevelTriValueSkipListReader>,
+                _sessionPool, &_docListReader, docListBeginPos, compressMode);
+        } else {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<MultiLevelPairValueSkipListReader>,
+                _sessionPool, &_docListReader, docListBeginPos, compressMode);
+        }
     } else {
-        return IE_POOL_COMPATIBLE_NEW_CLASS(_sessionPool, SkipListSegmentDecoder<PairValueSkipListReader>, _sessionPool,
-                                            &_docListReader, docListBeginPos, compressMode);
+        if (mCurSegPostingFormatOption.HasTfList()) {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<TriValueSkipListReader>, _sessionPool,
+                &_docListReader, docListBeginPos, compressMode);
+        } else {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<PairValueSkipListReader>, _sessionPool,
+                &_docListReader, docListBeginPos, compressMode);
+        }
     }
 }
 } // namespace indexlib::index

aios/storage/indexlib/index/inverted_index/Constant.h

@@ -52,6 +52,8 @@ constexpr uint8_t MAX_DICT_INLINE_AVAILABLE_SIZE = 7;
 constexpr uint8_t COMPRESS_MODE_SIZE = 4;
 constexpr uint32_t MAX_UNCOMPRESSED_SKIP_LIST_SIZE = 10;
 constexpr uint8_t SKIP_LIST_BUFFER_SIZE = 32;
+constexpr uint8_t SKIP_MULTIPLIER = 8;
+constexpr uint8_t MAX_SKIP_LEVEL = 10;
 
 static constexpr const char* TRUNCATE_ATTRIBUTE_READER_CREATOR = "truncate_attribute_reader_creator";
 static constexpr const char* TRUNCATE_META_FILE_READER_CREATOR = "truncate_meta_file_reader_creator";

aios/storage/indexlib/index/inverted_index/InDocStateKeeper.h

@@ -21,7 +21,6 @@
 #include "indexlib/index/inverted_index/format/NormalInDocState.h"
 #include "indexlib/index/inverted_index/format/PositionListSegmentDecoder.h"
 #include "indexlib/index/inverted_index/format/TermMeta.h"
-#include "indexlib/index/inverted_index/format/skiplist/PairValueSkipListReader.h"
 
 namespace indexlib::index {
 

aios/storage/indexlib/index/inverted_index/RangeSegmentPostingsIterator.cpp

@@ -23,6 +23,7 @@
 #include "indexlib/index/inverted_index/format/TermMetaLoader.h"
 #include "indexlib/index/inverted_index/format/skiplist/PairValueSkipListReader.h"
 #include "indexlib/index/inverted_index/format/skiplist/TriValueSkipListReader.h"
+#include "indexlib/index/inverted_index/format/skiplist/MultiLevelSkipListReader.h"
 
 namespace indexlib::index {
 AUTIL_LOG_SETUP(indexlib.index, RangeSegmentPostingsIterator);
@@ -182,12 +183,26 @@ RangeSegmentPostingsIterator::CreateNormalSegmentDecoder(file_system::ByteSliceR
                                             compressMode, curSegPostingFormatOption.IsShortListVbyteCompress());
     }
 
-    if (curSegPostingFormatOption.HasTfList()) {
-        return IE_POOL_COMPATIBLE_NEW_CLASS(_sessionPool, SkipListSegmentDecoder<TriValueSkipListReader>, _sessionPool,
-                                            docListReader, docListBeginPos, compressMode);
+    if (curSegPostingFormatOption.HasMultiLevelSkipList()) {
+        if (curSegPostingFormatOption.HasTfList()) {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<MultiLevelTriValueSkipListReader>,
+                _sessionPool, docListReader, docListBeginPos, compressMode);
+        } else {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<MultiLevelPairValueSkipListReader>,
+                _sessionPool, docListReader, docListBeginPos, compressMode);
+        }
     } else {
-        return IE_POOL_COMPATIBLE_NEW_CLASS(_sessionPool, SkipListSegmentDecoder<PairValueSkipListReader>, _sessionPool,
-                                            docListReader, docListBeginPos, compressMode);
+        if (curSegPostingFormatOption.HasTfList()) {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<TriValueSkipListReader>, _sessionPool,
+                docListReader, docListBeginPos, compressMode);
+        } else {
+            return IE_POOL_COMPATIBLE_NEW_CLASS(
+                _sessionPool, SkipListSegmentDecoder<PairValueSkipListReader>, _sessionPool,
+                docListReader, docListBeginPos, compressMode);
+        }
     }
 }
 } // namespace indexlib::index

aios/storage/indexlib/index/inverted_index/config/InvertedIndexConfig.cpp

@@ -82,6 +82,7 @@ struct InvertedIndexConfig::Impl {
     bool isPatchCompressed = false;
     bool isVirtual = false;
     bool isShortListVbyteCompress = false;
+    bool isMultiLevelSkipList = false;
     bool hasTruncate = false;
     indexlib::config::PayloadConfig payloadConfig;
 
@@ -115,6 +116,7 @@ struct InvertedIndexConfig::Impl {
         , isPatchCompressed(other.isPatchCompressed)
         , isVirtual(other.isVirtual)
         , isShortListVbyteCompress(other.isShortListVbyteCompress)
+        , isMultiLevelSkipList(other.isMultiLevelSkipList)
         , hasTruncate(other.hasTruncate)
     {
     }
@@ -185,7 +187,9 @@ Status InvertedIndexConfig::CheckEqual(const InvertedIndexConfig& other) const
     CHECK_CONFIG_EQUAL(_impl->formatVersionId, other._impl->formatVersionId, "_impl->formatVersionId not equal");
     CHECK_CONFIG_EQUAL(_impl->isShortListVbyteCompress, other._impl->isShortListVbyteCompress,
                        "_impl->isShortListVbyteCompress not equal");
-
+    CHECK_CONFIG_EQUAL(_impl->isMultiLevelSkipList,
+                       other._impl->isMultiLevelSkipList,
+                       "_impl->isMultiLevelSkipList not equal");
     for (size_t i = 0; i < _impl->shardingIndexConfigs.size(); i++) {
         auto status = _impl->shardingIndexConfigs[i]->CheckEqual(*other._impl->shardingIndexConfigs[i]);
         RETURN_IF_STATUS_ERROR(status, "sharding index config [%s] not equal",
@@ -340,6 +344,16 @@ void InvertedIndexConfig::SetShortListVbyteCompress(bool isShortListVbyteCompres
     _impl->isShortListVbyteCompress = isShortListVbyteCompress;
 }
 
+bool InvertedIndexConfig::IsMultiLevelSkipList() const
+{
+    return _impl->isMultiLevelSkipList;
+}
+
+void InvertedIndexConfig::SetMultiLevelSkipList(bool isMultiLevelSkipList)
+{
+    _impl->isMultiLevelSkipList = isMultiLevelSkipList;
+}
+
 void InvertedIndexConfig::SetIsReferenceCompress(bool isReferenceCompress)
 {
     _impl->isReferenceCompress = isReferenceCompress;

aios/storage/indexlib/index/inverted_index/config/InvertedIndexConfig.h

@@ -178,6 +178,9 @@ class InvertedIndexConfig : public indexlibv2::config::IIndexConfig
     bool IsShortListVbyteCompress() const;
     void SetShortListVbyteCompress(bool isShortListVbyteCompress);
 
+    bool IsMultiLevelSkipList() const;
+    void SetMultiLevelSkipList(bool isMultiLevelSkipList);
+
     void SetIsReferenceCompress(bool isReferenceCompress);
     bool IsReferenceCompress() const;
 
@@ -272,6 +275,7 @@ class InvertedIndexConfig : public indexlibv2::config::IIndexConfig
     inline static const std::string DOC_PAYLOAD_FLAG = "doc_payload_flag";
     inline static const std::string POSITION_PAYLOAD_FLAG = "position_payload_flag";
     inline static const std::string POSITION_LIST_FLAG = "position_list_flag";
+    inline static const std::string MULTI_LEVEL_SKIP_LIST = "multi_level_skip_list";
     inline static const std::string TERM_FREQUENCY_FLAG = "term_frequency_flag";
     inline static const std::string TERM_FREQUENCY_BITMAP = "term_frequency_bitmap";
     inline static const std::string HIGH_FEQUENCY_DICTIONARY = "high_frequency_dictionary";

aios/storage/indexlib/index/inverted_index/config/InvertedIndexConfigSerializer.cpp

@@ -164,7 +164,9 @@ void InvertedIndexConfigSerializer::Serialize(const indexlibv2::config::Inverted
         std::string indexCompressMode = indexlibv2::config::InvertedIndexConfig::INDEX_COMPRESS_MODE_REFERENCE;
         json->Jsonize(indexlibv2::config::InvertedIndexConfig::INDEX_COMPRESS_MODE, indexCompressMode);
     }
-
+    if (indexConfig.IsMultiLevelSkipList()) {
+        json->Jsonize(indexlibv2::config::InvertedIndexConfig::MULTI_LEVEL_SKIP_LIST, enableValue);
+    }
     if (indexConfig.IsHashTypedDictionary()) {
         json->Jsonize(indexlibv2::config::InvertedIndexConfig::USE_HASH_DICTIONARY,
                       indexConfig.IsHashTypedDictionary());
@@ -209,6 +211,9 @@ void InvertedIndexConfigSerializer::DeserializeCommonFields(const autil::legacy:
                  indexlibv2::config::InvertedIndexConfig::INDEX_COMPRESS_MODE_PFOR_DELTA);
     indexConfig->SetIsReferenceCompress(
         (compressMode == indexlibv2::config::InvertedIndexConfig::INDEX_COMPRESS_MODE_REFERENCE) ? true : false);
+    int useMultiLevelSkipList = 0;
+    json.Jsonize(indexlibv2::config::InvertedIndexConfig::MULTI_LEVEL_SKIP_LIST, useMultiLevelSkipList, useMultiLevelSkipList);
+    indexConfig->SetMultiLevelSkipList(useMultiLevelSkipList == 1);
     bool isHashTypedDictionary = false;
     json.Jsonize(indexlibv2::config::InvertedIndexConfig::USE_HASH_DICTIONARY, isHashTypedDictionary,
                  isHashTypedDictionary);

aios/storage/indexlib/index/inverted_index/format/BUILD

@@ -104,7 +104,7 @@ strict_cc_library(
         '//aios/storage/indexlib/index/common/field_format/section_attribute:MultiSectionMeta',
         '//aios/storage/indexlib/index/common/numeric_compress:EncoderProvider',
         '//aios/storage/indexlib/index/inverted_index:InDocPositionState',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:PairValueSkipListReader',
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib',
         '//aios/storage/indexlib/index/inverted_index/section_attribute:SectionAttributeReaderImpl',
         '//aios/storage/indexlib/util:object_pool'
     ]
@@ -155,8 +155,7 @@ strict_cc_library(
         ':BufferedByteSlice', ':InMemPositionListDecoder',
         ':PositionListFormat', '//aios/autil:mem_pool_base',
         '//aios/storage/indexlib/file_system:byte_slice_rw',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:BufferedSkipListWriter',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:InMemPairValueSkipListReader'
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib'
     ]
 )
 strict_cc_library(
@@ -224,19 +223,15 @@ strict_cc_library(
     deps=[
         ':BufferedByteSlice', ':DocListFormat', ':InMemDocListDecoder',
         ':PositionBitmapWriter', '//aios/autil:mem_pool_base',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:BufferedSkipListWriter',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:InMemPairValueSkipListReader',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:InMemTriValueSkipListReader'
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib',
     ]
 )
 strict_cc_library(
     name='DocListEncoderImproved',
     deps=[
         ':BufferedByteSlice', ':DocListFormat', ':InMemDocListDecoder',
         ':PositionBitmapWriter', '//aios/autil:mem_pool_base',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:BufferedSkipListWriter',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:InMemPairValueSkipListReader',
-        '//aios/storage/indexlib/index/inverted_index/format/skiplist:InMemTriValueSkipListReader'
+        '//aios/storage/indexlib/index/inverted_index/format/skiplist:lib'
     ]
 )
 strict_cc_library(

aios/storage/indexlib/index/inverted_index/format/DocListEncoder.cpp

@@ -63,7 +63,7 @@ DocListEncoder::DocListEncoder(const DocListFormatOption& docListFormatOption, u
 DocListEncoder::~DocListEncoder()
 {
     if (_docSkipListWriter) {
-        _docSkipListWriter->~BufferedSkipListWriter();
+        _docSkipListWriter->~SkipListWriter();
         _docSkipListWriter = nullptr;
     }
     if (_ownDocListFormat) {
@@ -199,16 +199,21 @@ void DocListEncoder::FlushDocListBuffer(uint8_t compressMode)
 
 void DocListEncoder::CreateDocSkipListWriter()
 {
-    void* buffer = _byteSlicePool->allocate(sizeof(BufferedSkipListWriter));
-    autil::mem_pool::RecyclePool* bufferPool =
-        dynamic_cast<autil::mem_pool::RecyclePool*>(_docListBuffer.GetBufferPool());
-    assert(bufferPool);
-
-    BufferedSkipListWriter* docSkipListWriter =
-        new (buffer) BufferedSkipListWriter(_byteSlicePool, bufferPool, _compressMode);
-    docSkipListWriter->Init(_docListFormat->GetDocListSkipListFormat());
     // skip list writer should be atomic created;
-    _docSkipListWriter = docSkipListWriter;
+    if (_docListFormatOption.HasMultiLevelSkipList()) {
+        void *buffer = _byteSlicePool->allocate(sizeof(MultiLevelSkipListWriter));
+        MultiLevelSkipListWriter *docSkipListWriter =
+            new (buffer) MultiLevelSkipListWriter(_byteSlicePool);
+        _docSkipListWriter = docSkipListWriter;
+    } else {
+        void *buffer = _byteSlicePool->allocate(sizeof(BufferedSkipListWriter));
+        autil::mem_pool::RecyclePool *bufferPool = dynamic_cast<autil::mem_pool::RecyclePool *>(_docListBuffer.GetBufferPool());
+        assert(bufferPool);
+        BufferedSkipListWriter *docSkipListWriter =
+            new (buffer) BufferedSkipListWriter(_byteSlicePool, bufferPool, _compressMode);
+        docSkipListWriter->Init(_docListFormat->GetDocListSkipListFormat());
+        _docSkipListWriter = docSkipListWriter;
+    }
 }
 
 void DocListEncoder::AddSkipListItem(uint32_t itemSize)
@@ -234,17 +239,37 @@ InMemDocListDecoder* DocListEncoder::GetInMemDocListDecoder(autil::mem_pool::Poo
     if (_docSkipListWriter) {
         const DocListSkipListFormat* skipListFormat = _docListFormat->GetDocListSkipListFormat();
         assert(skipListFormat);
-
-        if (skipListFormat->HasTfList()) {
-            InMemTriValueSkipListReader* inMemSkipListReader =
-                IE_POOL_COMPATIBLE_NEW_CLASS(sessionPool, InMemTriValueSkipListReader, sessionPool);
-            inMemSkipListReader->Load(_docSkipListWriter);
-            skipListReader = inMemSkipListReader;
+        if (_docListFormatOption.HasMultiLevelSkipList()) {
+            MultiLevelSkipListWriter *skipwriter =
+                dynamic_cast<MultiLevelSkipListWriter *>(_docSkipListWriter);
+            assert(skipwriter != nullptr);
+            if (skipListFormat->HasTfList()) {
+                MultiLevelTriValueSkipListReader *inMemSkipListReader =
+                    IE_POOL_COMPATIBLE_NEW_CLASS(sessionPool, MultiLevelTriValueSkipListReader);
+                skipwriter->SnapShot(inMemSkipListReader, sessionPool);
+                skipListReader = inMemSkipListReader;
+            } else {
+                MultiLevelPairValueSkipListReader *inMemSkipListReader =
+                    IE_POOL_COMPATIBLE_NEW_CLASS(sessionPool, MultiLevelPairValueSkipListReader);
+                skipwriter->SnapShot(inMemSkipListReader, sessionPool);
+                skipListReader = inMemSkipListReader;
+            }
         } else {
-            InMemPairValueSkipListReader* inMemSkipListReader = IE_POOL_COMPATIBLE_NEW_CLASS(
-                sessionPool, InMemPairValueSkipListReader, sessionPool, _compressMode == REFERENCE_COMPRESS_MODE);
-            inMemSkipListReader->Load(_docSkipListWriter);
-            skipListReader = inMemSkipListReader;
+            BufferedSkipListWriter *bufferedWriter =
+                dynamic_cast<BufferedSkipListWriter *>(_docSkipListWriter);
+            assert(bufferedWriter != nullptr);
+            if (skipListFormat->HasTfList()) {
+                InMemTriValueSkipListReader *inMemSkipListReader = IE_POOL_COMPATIBLE_NEW_CLASS(
+                    sessionPool, InMemTriValueSkipListReader, sessionPool);
+                inMemSkipListReader->Load(bufferedWriter);
+                skipListReader = inMemSkipListReader;
+            } else {
+                InMemPairValueSkipListReader *inMemSkipListReader = IE_POOL_COMPATIBLE_NEW_CLASS(
+                    sessionPool, InMemPairValueSkipListReader, sessionPool,
+                    _compressMode == REFERENCE_COMPRESS_MODE);
+                inMemSkipListReader->Load(bufferedWriter);
+                skipListReader = inMemSkipListReader;
+            }
         }
     }
     BufferedByteSlice* docListBuffer =

aios/storage/indexlib/index/inverted_index/format/DocListEncoder.h

@@ -27,6 +27,8 @@
 #include "indexlib/index/inverted_index/format/InMemDocListDecoder.h"
 #include "indexlib/index/inverted_index/format/PositionBitmapWriter.h"
 #include "indexlib/index/inverted_index/format/skiplist/BufferedSkipListWriter.h"
+#include "indexlib/index/inverted_index/format/skiplist/SkipListWriter.h"
+#include "indexlib/index/inverted_index/format/skiplist/MultiLevelSkipListWriter.h"
 #include "indexlib/util/PoolUtil.h"
 #include "indexlib/util/SimplePool.h"
 
@@ -80,7 +82,7 @@ class DocListEncoder
 private:
     const DocListFormat* TEST_GetDocListFormat() const { return _docListFormat; }
     BufferedByteSlice* TEST_GetDocListBuffer() { return &_docListBuffer; }
-    void TEST_SetDocSkipListWriter(BufferedSkipListWriter* skipListBuffer) { _docSkipListWriter = skipListBuffer; }
+    void TEST_SetDocSkipListWriter(SkipListWriter* skipListBuffer) { _docSkipListWriter = skipListBuffer; }
 
     void TEST_SetPositionBitmapWriter(PositionBitmapWriter* tfBitmapWriter)
     {
@@ -104,7 +106,7 @@ class DocListEncoder
     index::CompressMode _compressMode; // 1byte
     // volatile for realtime
     PositionBitmapWriter* _tfBitmapWriter; // 8byte
-    BufferedSkipListWriter* volatile _docSkipListWriter;
+    SkipListWriter* volatile _docSkipListWriter;
     DocListFormat* _docListFormat;
     autil::mem_pool::Pool* _byteSlicePool;