diff options
Diffstat (limited to 'searchlib/src/vespa/searchlib')
11 files changed, 109 insertions, 34 deletions
diff --git a/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.cpp b/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.cpp index 3fcf983901d..8878c8c5cdc 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.cpp @@ -81,10 +81,10 @@ EquivQueryNode::unpack_match_data(uint32_t docid, const fef::ITermData& td, fef: unpack_match_data_helper(docid, td, match_data, hit_list, *this, is_filter(), index_env); } -EquivQueryNode* -EquivQueryNode::as_equiv_query_node() noexcept +bool +EquivQueryNode::multi_index_terms() const noexcept { - return this; + return true; } const EquivQueryNode* diff --git a/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.h b/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.h index a0485954675..997a861ee44 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.h +++ b/searchlib/src/vespa/searchlib/query/streaming/equiv_query_node.h @@ -17,7 +17,7 @@ public: bool evaluate() const override; const HitList & evaluateHits(HitList & hl) const override; void unpack_match_data(uint32_t docid, const fef::ITermData& td, fef::MatchData& match_data, const fef::IIndexEnvironment& index_env) override; - EquivQueryNode* as_equiv_query_node() noexcept override; + bool multi_index_terms() const noexcept override; const EquivQueryNode* as_equiv_query_node() const noexcept override; std::vector<std::unique_ptr<QueryTerm>> steal_terms(); }; diff --git a/searchlib/src/vespa/searchlib/query/streaming/multi_term.cpp b/searchlib/src/vespa/searchlib/query/streaming/multi_term.cpp index f5a09892551..2a96a967215 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/multi_term.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/multi_term.cpp @@ -51,4 +51,22 @@ MultiTerm::evaluate() const return false; } +MultiTerm* +MultiTerm::as_multi_term() noexcept +{ + return this; +} + +const MultiTerm* +MultiTerm::as_multi_term() const noexcept +{ + return this; +} + +bool +MultiTerm::multi_index_terms() const noexcept +{ + return false; +} + } diff --git a/searchlib/src/vespa/searchlib/query/streaming/multi_term.h b/searchlib/src/vespa/searchlib/query/streaming/multi_term.h index 9bf5f8de6b2..a2ae3d30d52 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/multi_term.h +++ b/searchlib/src/vespa/searchlib/query/streaming/multi_term.h @@ -29,7 +29,12 @@ public: std::unique_ptr<query::TermVector> terms, Normalizing normalizing); ~MultiTerm() override; void add_term(std::unique_ptr<QueryTerm> term); - MultiTerm* as_multi_term() noexcept override { return this; } + MultiTerm* as_multi_term() noexcept override; + const MultiTerm* as_multi_term() const noexcept override; + /* + * Terms below search in different indexes when multi_index_terms() returns true. + */ + virtual bool multi_index_terms() const noexcept; void reset() override; bool evaluate() const override; const std::vector<std::unique_ptr<QueryTerm>>& get_terms() const noexcept { return _terms; } diff --git a/searchlib/src/vespa/searchlib/query/streaming/query.cpp b/searchlib/src/vespa/searchlib/query/streaming/query.cpp index 94d9acd02cd..25fcb8c123e 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/query.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/query.cpp @@ -95,7 +95,6 @@ QueryConnector::create(ParseItem::ItemType type) case search::ParseItem::ITEM_OR: case search::ParseItem::ITEM_WEAK_AND: return std::make_unique<OrQueryNode>(); case search::ParseItem::ITEM_NOT: return std::make_unique<AndNotQueryNode>(); - case search::ParseItem::ITEM_SAME_ELEMENT: return std::make_unique<SameElementQueryNode>(); case search::ParseItem::ITEM_NEAR: return std::make_unique<NearQueryNode>(); case search::ParseItem::ITEM_ONEAR: return std::make_unique<ONearQueryNode>(); case search::ParseItem::ITEM_RANK: return std::make_unique<RankWithQueryNode>(); diff --git a/searchlib/src/vespa/searchlib/query/streaming/querynode.cpp b/searchlib/src/vespa/searchlib/query/streaming/querynode.cpp index 611e8d67d76..94a479fd2d3 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/querynode.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/querynode.cpp @@ -11,6 +11,7 @@ #include <vespa/searchlib/query/streaming/dot_product_term.h> #include <vespa/searchlib/query/streaming/equiv_query_node.h> #include <vespa/searchlib/query/streaming/in_term.h> +#include <vespa/searchlib/query/streaming/same_element_query_node.h> #include <vespa/searchlib/query/streaming/wand_term.h> #include <vespa/searchlib/query/streaming/weighted_set_term.h> #include <vespa/searchlib/query/tree/term_vector.h> @@ -26,9 +27,7 @@ namespace search::streaming { namespace { bool disableRewrite(const QueryNode * qn) { - return dynamic_cast<const NearQueryNode *> (qn) || - dynamic_cast<const PhraseQueryNode *> (qn) || - dynamic_cast<const SameElementQueryNode *>(qn); + return dynamic_cast<const NearQueryNode *> (qn); } bool possibleFloat(const QueryTerm & qt, const QueryTerm::string & term) { @@ -49,7 +48,6 @@ QueryNode::Build(const QueryNode * parent, const QueryNodeResultFactory & factor case ParseItem::ITEM_OR: case ParseItem::ITEM_WEAK_AND: case ParseItem::ITEM_NOT: - case ParseItem::ITEM_SAME_ELEMENT: case ParseItem::ITEM_NEAR: case ParseItem::ITEM_ONEAR: case ParseItem::ITEM_RANK: @@ -61,9 +59,7 @@ QueryNode::Build(const QueryNode * parent, const QueryNodeResultFactory & factor if (nqn) { nqn->distance(queryRep.getNearDistance()); } - if ((type == ParseItem::ITEM_WEAK_AND) || - (type == ParseItem::ITEM_SAME_ELEMENT)) - { + if (type == ParseItem::ITEM_WEAK_AND) { qn->setIndex(queryRep.getIndexName()); } for (size_t i=0; i < arity; i++) { @@ -197,6 +193,9 @@ QueryNode::Build(const QueryNode * parent, const QueryNodeResultFactory & factor case ParseItem::ITEM_EQUIV: qn = build_equiv_term(factory, queryRep, allowRewrite); break; + case ParseItem::ITEM_SAME_ELEMENT: + qn = build_same_element_term(factory, queryRep); + break; default: skip_unknown(queryRep); break; @@ -335,6 +334,25 @@ QueryNode::build_equiv_term(const QueryNodeResultFactory& factory, SimpleQuerySt return eqn; } +std::unique_ptr<QueryNode> +QueryNode::build_same_element_term(const QueryNodeResultFactory& factory, SimpleQueryStackDumpIterator& queryRep) +{ + auto sen = std::make_unique<SameElementQueryNode>(factory.create(), queryRep.getIndexName(), queryRep.getArity()); + auto arity = queryRep.getArity(); + sen->setWeight(queryRep.GetWeight()); + sen->setUniqueId(queryRep.getUniqueId()); + for (size_t i = 0; i < arity; ++i) { + queryRep.next(); + auto qn = Build(sen.get(), factory, queryRep, false); + auto qtp = dynamic_cast<QueryTerm*>(qn.get()); + assert(qtp != nullptr); + qn.release(); + std::unique_ptr<QueryTerm> qt(qtp); + sen->add_term(std::move(qt)); + } + return sen; +} + void QueryNode::skip_unknown(SimpleQueryStackDumpIterator& queryRep) { diff --git a/searchlib/src/vespa/searchlib/query/streaming/querynode.h b/searchlib/src/vespa/searchlib/query/streaming/querynode.h index fff3bb15d10..20e701e5c50 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/querynode.h +++ b/searchlib/src/vespa/searchlib/query/streaming/querynode.h @@ -35,6 +35,7 @@ class QueryNode static std::unique_ptr<QueryNode> build_weighted_set_term(const QueryNodeResultFactory& factory, SimpleQueryStackDumpIterator& queryRep); static std::unique_ptr<QueryNode> build_phrase_term(const QueryNodeResultFactory& factory, SimpleQueryStackDumpIterator& queryRep); static std::unique_ptr<QueryNode> build_equiv_term(const QueryNodeResultFactory& factory, SimpleQueryStackDumpIterator& queryRep, bool allow_rewrite); + static std::unique_ptr<QueryNode> build_same_element_term(const QueryNodeResultFactory& factory, SimpleQueryStackDumpIterator& queryRep); static void skip_unknown(SimpleQueryStackDumpIterator& queryRep); public: using UP = std::unique_ptr<QueryNode>; diff --git a/searchlib/src/vespa/searchlib/query/streaming/queryterm.cpp b/searchlib/src/vespa/searchlib/query/streaming/queryterm.cpp index 728b9a2ab76..07fc60d2243 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/queryterm.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/queryterm.cpp @@ -135,20 +135,20 @@ QueryTerm::as_multi_term() noexcept return nullptr; } -RegexpTerm* -QueryTerm::as_regexp_term() noexcept +const MultiTerm* +QueryTerm::as_multi_term() const noexcept { return nullptr; } -FuzzyTerm* -QueryTerm::as_fuzzy_term() noexcept +RegexpTerm* +QueryTerm::as_regexp_term() noexcept { return nullptr; } -EquivQueryNode* -QueryTerm::as_equiv_query_node() noexcept +FuzzyTerm* +QueryTerm::as_fuzzy_term() noexcept { return nullptr; } diff --git a/searchlib/src/vespa/searchlib/query/streaming/queryterm.h b/searchlib/src/vespa/searchlib/query/streaming/queryterm.h index e6b063231d6..78b0a1fea7d 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/queryterm.h +++ b/searchlib/src/vespa/searchlib/query/streaming/queryterm.h @@ -104,9 +104,9 @@ public: void setFuzzyPrefixLength(uint32_t fuzzyPrefixLength) { _fuzzyPrefixLength = fuzzyPrefixLength; } virtual NearestNeighborQueryNode* as_nearest_neighbor_query_node() noexcept; virtual MultiTerm* as_multi_term() noexcept; + virtual const MultiTerm* as_multi_term() const noexcept; virtual RegexpTerm* as_regexp_term() noexcept; virtual FuzzyTerm* as_fuzzy_term() noexcept; - virtual EquivQueryNode* as_equiv_query_node() noexcept; virtual const EquivQueryNode* as_equiv_query_node() const noexcept; virtual void unpack_match_data(uint32_t docid, const fef::ITermData& td, fef::MatchData& match_data, const fef::IIndexEnvironment& index_env); protected: diff --git a/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.cpp b/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.cpp index 49d5fb0f9fb..cd9c693ca1c 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.cpp +++ b/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.cpp @@ -1,30 +1,36 @@ // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root. #include "same_element_query_node.h" +#include <vespa/searchlib/fef/itermdata.h> +#include <vespa/searchlib/fef/matchdata.h> #include <cassert> namespace search::streaming { +SameElementQueryNode::SameElementQueryNode(std::unique_ptr<QueryNodeResultBase> result_base, const string& index, uint32_t num_terms) noexcept + : MultiTerm(std::move(result_base), index, num_terms) +{ +} + +SameElementQueryNode::~SameElementQueryNode() = default; + bool SameElementQueryNode::evaluate() const { HitList hl; return ! evaluateHits(hl).empty(); } -void -SameElementQueryNode::addChild(QueryNode::UP child) { - assert(dynamic_cast<const QueryTerm *>(child.get()) != nullptr); - AndQueryNode::addChild(std::move(child)); -} - const HitList & SameElementQueryNode::evaluateHits(HitList & hl) const { hl.clear(); - if ( !AndQueryNode::evaluate()) return hl; - + const auto & children = get_terms(); + for (auto& child : children) { + if ( ! child->evaluate() ) { + return hl; + } + } HitList tmpHL; - const auto & children = getChildren(); unsigned int numFields = children.size(); unsigned int currMatchCount = 0; std::vector<unsigned int> indexVector(numFields, 0); @@ -62,4 +68,31 @@ SameElementQueryNode::evaluateHits(HitList & hl) const return hl; } +void +SameElementQueryNode::unpack_match_data(uint32_t docid, const fef::ITermData& td, fef::MatchData& match_data, const fef::IIndexEnvironment&) +{ + HitList list; + const HitList & hit_list = evaluateHits(list); + if (!hit_list.empty()) { + auto num_fields = td.numFields(); + /* + * Currently reports hit for all fields for query node instead of + * just the fields where the related subfields had matches. + */ + for (size_t field_idx = 0; field_idx < num_fields; ++field_idx) { + auto& tfd = td.field(field_idx); + auto field_id = tfd.getFieldId(); + auto tmd = match_data.resolveTermField(tfd.getHandle()); + tmd->setFieldId(field_id); + tmd->reset(docid); + } + } +} + +bool +SameElementQueryNode::multi_index_terms() const noexcept +{ + return true; +} + } diff --git a/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.h b/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.h index 8e675feb569..37fb3dbba52 100644 --- a/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.h +++ b/searchlib/src/vespa/searchlib/query/streaming/same_element_query_node.h @@ -2,21 +2,22 @@ #pragma once -#include "query.h" +#include "multi_term.h" namespace search::streaming { /** N-ary Same element operator. All terms must be within the same element. */ -class SameElementQueryNode : public AndQueryNode +class SameElementQueryNode : public MultiTerm { public: - SameElementQueryNode() noexcept : AndQueryNode("SAME_ELEMENT") { } + SameElementQueryNode(std::unique_ptr<QueryNodeResultBase> result_base, const string& index, uint32_t num_terms) noexcept; + ~SameElementQueryNode() override; bool evaluate() const override; const HitList & evaluateHits(HitList & hl) const override; - bool isFlattenable(ParseItem::ItemType) const override { return false; } - void addChild(QueryNode::UP child) override; + void unpack_match_data(uint32_t docid, const fef::ITermData& td, fef::MatchData& match_data, const fef::IIndexEnvironment& index_env) override; + bool multi_index_terms() const noexcept override; }; } |