ClickHouse/src/Storages/MergeTree/MergeTreeIndexFullText.h

#pragma once

#include <Interpreters/BloomFilter.h>
#include <Storages/MergeTree/MergeTreeIndices.h>
#include <Storages/MergeTree/KeyCondition.h>

#include <memory>


namespace DB
{

/// Interface for string parsers.
struct ITokenExtractor
{
    virtual ~ITokenExtractor() = default;

    /// Fast inplace implementation for regular use.
    /// Gets string (data ptr and len) and start position for extracting next token (state of extractor).
    /// Returns false if parsing is finished, otherwise returns true.
    virtual bool nextInField(const char * data, size_t len, size_t * pos, size_t * token_start, size_t * token_len) const = 0;

    /// Optimized version that can assume at least 15 padding bytes after data + len (as our Columns provide).
    virtual bool nextInColumn(const char * data, size_t len, size_t * pos, size_t * token_start, size_t * token_len) const
    {
        return nextInField(data, len, pos, token_start, token_len);
    }

    /// Special implementation for creating bloom filter for LIKE function.
    /// It skips unescaped `%` and `_` and supports escaping symbols, but it is less lightweight.
    virtual bool nextLike(const String & str, size_t * pos, String & out) const = 0;

    virtual bool supportLike() const = 0;
};

using TokenExtractorPtr = const ITokenExtractor *;

struct MergeTreeIndexGranuleFullText final : public IMergeTreeIndexGranule
{
    explicit MergeTreeIndexGranuleFullText(
        const String & index_name_,
        size_t columns_number,
        const BloomFilterParameters & params_);

    ~MergeTreeIndexGranuleFullText() override = default;

    void serializeBinary(WriteBuffer & ostr) const override;
    void deserializeBinary(ReadBuffer & istr, MergeTreeIndexVersion version) override;

    bool empty() const override { return !has_elems; }

    String index_name;
    BloomFilterParameters params;

    std::vector<BloomFilter> bloom_filters;
    bool has_elems;
};

using MergeTreeIndexGranuleFullTextPtr = std::shared_ptr<MergeTreeIndexGranuleFullText>;

struct MergeTreeIndexAggregatorFullText final : IMergeTreeIndexAggregator
{
    explicit MergeTreeIndexAggregatorFullText(
        const Names & index_columns_,
        const String & index_name_,
        const BloomFilterParameters & params_,
        TokenExtractorPtr token_extractor_);

    ~MergeTreeIndexAggregatorFullText() override = default;

    bool empty() const override { return !granule || granule->empty(); }
    MergeTreeIndexGranulePtr getGranuleAndReset() override;

    void update(const Block & block, size_t * pos, size_t limit) override;

    Names index_columns;
    String index_name;
    BloomFilterParameters params;
    TokenExtractorPtr token_extractor;

    MergeTreeIndexGranuleFullTextPtr granule;
};


class MergeTreeConditionFullText final : public IMergeTreeIndexCondition
{
public:
    MergeTreeConditionFullText(
            const SelectQueryInfo & query_info,
            ContextPtr context,
            const Block & index_sample_block,
            const BloomFilterParameters & params_,
            TokenExtractorPtr token_extactor_);

    ~MergeTreeConditionFullText() override = default;

    bool alwaysUnknownOrTrue() const override;
    bool mayBeTrueOnGranule(MergeTreeIndexGranulePtr idx_granule) const override;
private:
    struct KeyTuplePositionMapping
    {
        KeyTuplePositionMapping(size_t tuple_index_, size_t key_index_) : tuple_index(tuple_index_), key_index(key_index_) {}

        size_t tuple_index;
        size_t key_index;
    };
    /// Uses RPN like KeyCondition
    struct RPNElement
    {
        enum Function
        {
            /// Atoms of a Boolean expression.
            FUNCTION_EQUALS,
            FUNCTION_NOT_EQUALS,
            FUNCTION_IN,
            FUNCTION_NOT_IN,
            FUNCTION_MULTI_SEARCH,
            FUNCTION_UNKNOWN, /// Can take any value.
            /// Operators of the logical expression.
            FUNCTION_NOT,
            FUNCTION_AND,
            FUNCTION_OR,
            /// Constants
            ALWAYS_FALSE,
            ALWAYS_TRUE,
        };

        RPNElement(
                Function function_ = FUNCTION_UNKNOWN, size_t key_column_ = 0, std::unique_ptr<BloomFilter> && const_bloom_filter_ = nullptr)
                : function(function_), key_column(key_column_), bloom_filter(std::move(const_bloom_filter_)) {}

        Function function = FUNCTION_UNKNOWN;
        /// For FUNCTION_EQUALS, FUNCTION_NOT_EQUALS and FUNCTION_MULTI_SEARCH
        size_t key_column;

        /// For FUNCTION_EQUALS, FUNCTION_NOT_EQUALS
        std::unique_ptr<BloomFilter> bloom_filter;

        /// For FUNCTION_IN, FUNCTION_NOT_IN and FUNCTION_MULTI_SEARCH
        std::vector<std::vector<BloomFilter>> set_bloom_filters;

        /// For FUNCTION_IN and FUNCTION_NOT_IN
        std::vector<size_t> set_key_position;
    };

    using RPN = std::vector<RPNElement>;

    bool atomFromAST(const ASTPtr & node, Block & block_with_constants, RPNElement & out);

    bool getKey(const ASTPtr & node, size_t & key_column_num);
    bool tryPrepareSetBloomFilter(const ASTs & args, RPNElement & out);

    static bool createFunctionEqualsCondition(
        RPNElement & out, const Field & value, const BloomFilterParameters & params, TokenExtractorPtr token_extractor);

    Names index_columns;
    DataTypes index_data_types;
    BloomFilterParameters params;
    TokenExtractorPtr token_extractor;
    RPN rpn;
    /// Sets from syntax analyzer.
    PreparedSets prepared_sets;
};


/// Parser extracting all ngrams from string.
struct NgramTokenExtractor final : public ITokenExtractor
{
    NgramTokenExtractor(size_t n_) : n(n_) {}

    static String getName() { return "ngrambf_v1"; }

    bool nextInField(const char * data, size_t len, size_t * pos, size_t * token_start, size_t * token_len) const override;
    bool nextLike(const String & str, size_t * pos, String & token) const override;

    bool supportLike() const override { return true; }

    size_t n;
};

/// Parser extracting tokens (sequences of numbers and ascii letters).
struct SplitTokenExtractor final : public ITokenExtractor
{
    static String getName() { return "tokenbf_v1"; }

    bool nextInField(const char * data, size_t len, size_t * pos, size_t * token_start, size_t * token_len) const override;
    bool nextInColumn(const char * data, size_t len, size_t * pos, size_t * token_start, size_t * token_len) const override;
    bool nextLike(const String & str, size_t * pos, String & token) const override;

    bool supportLike() const override { return true; }
};


class MergeTreeIndexFullText final : public IMergeTreeIndex
{
public:
    MergeTreeIndexFullText(
        const IndexDescription & index_,
        const BloomFilterParameters & params_,
        std::unique_ptr<ITokenExtractor> && token_extractor_)
        : IMergeTreeIndex(index_)
        , params(params_)
        , token_extractor(std::move(token_extractor_)) {}

    ~MergeTreeIndexFullText() override = default;

    MergeTreeIndexGranulePtr createIndexGranule() const override;
    MergeTreeIndexAggregatorPtr createIndexAggregator() const override;

    MergeTreeIndexConditionPtr createIndexCondition(
            const SelectQueryInfo & query, ContextPtr context) const override;

    bool mayBenefitFromIndexForIn(const ASTPtr & node) const override;

    BloomFilterParameters params;
    /// Function for selecting next token.
    std::unique_ptr<ITokenExtractor> token_extractor;
};

}