Merge pull request #9749 from vespa-engine/toregge/extend-query-environment-to-get-average-field-length

Extend query environment with method to get average field length.

13 files changed, 48 insertions, 11 deletions

diff --git a/searchcore/src/vespa/searchcore/proton/matching/blueprintbuilder.cpp b/searchcore/src/vespa/searchcore/proton/matching/blueprintbuilder.cpp
index 268fe63ba4c..672e7f78784 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/blueprintbuilder.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/blueprintbuilder.cpp
@@ -4,6 +4,7 @@
 #include "blueprintbuilder.h"
 #include "termdatafromnode.h"
 #include "same_element_builder.h"
+#include <vespa/searchcorespi/index/indexsearchable.h>
 #include <vespa/searchlib/query/tree/customtypevisitor.h>
 #include <vespa/searchlib/queryeval/leaf_blueprints.h>
 #include <vespa/searchlib/queryeval/intermediate_blueprints.h>
diff --git a/searchcore/src/vespa/searchcore/proton/matching/fakesearchcontext.h b/searchcore/src/vespa/searchcore/proton/matching/fakesearchcontext.h
index 02aedf15d6e..fe9c20112f4 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/fakesearchcontext.h
+++ b/searchcore/src/vespa/searchcore/proton/matching/fakesearchcontext.h
@@ -52,7 +52,7 @@ public:
     search::queryeval::ISourceSelector &selector() { return *_selector; }
 
     // Implements ISearchContext
-    search::queryeval::Searchable &getIndexes() override {
+    IndexSearchable &getIndexes() override {
         return *_indexes;
     }
 
diff --git a/searchcore/src/vespa/searchcore/proton/matching/isearchcontext.h b/searchcore/src/vespa/searchcore/proton/matching/isearchcontext.h
index 2965e3796bf..dc840dc79ff 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/isearchcontext.h
+++ b/searchcore/src/vespa/searchcore/proton/matching/isearchcontext.h
@@ -6,6 +6,8 @@
 
 #include <memory>
 
+namespace searchcorespi { class IndexSearchable; }
+
 namespace proton::matching {
 
 /**
@@ -31,13 +33,14 @@ public:
     ISearchContext & operator = (const ISearchContext &) = delete;
 
     typedef search::queryeval::Searchable      Searchable;
+    using IndexSearchable = searchcorespi::IndexSearchable;
 
     /**
      * Obtain the index fields searchable.
      *
      * @return index fields searchable.
      **/
-    virtual Searchable &getIndexes() = 0;
+    virtual IndexSearchable &getIndexes() = 0;
 
     /**
      * Obtain the attribute fields searchable.
diff --git a/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp b/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
index 4a944dc3214..5d1e2212c83 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/match_tools.cpp
@@ -9,6 +9,7 @@
 #include <vespa/log/log.h>
 LOG_SETUP(".proton.matching.match_tools");
 #include <vespa/searchlib/query/tree/querytreecreator.h>
+#include <vespa/searchcorespi/index/indexsearchable.h>
 
 using search::attribute::IAttributeContext;
 using search::queryeval::IRequestContext;
@@ -158,7 +159,7 @@ MatchToolsFactory(QueryLimiter               & queryLimiter,
       _hardDoom(hardDoom),
       _query(),
       _match_limiter(),
-      _queryEnv(indexEnv, attributeContext, rankProperties),
+      _queryEnv(indexEnv, attributeContext, rankProperties, searchContext.getIndexes()),
       _mdl(),
       _rankSetup(rankSetup),
       _featureOverrides(featureOverrides),
diff --git a/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.cpp b/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.cpp
index d4320c87ab2..ec48ee7164b 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.cpp
@@ -1,6 +1,7 @@
 // Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 
 #include "queryenvironment.h"
+#include <vespa/searchlib/index/i_field_length_inspector.h>
 
 using search::attribute::IAttributeContext;
 using search::fef::IIndexEnvironment;
@@ -11,12 +12,14 @@ namespace proton::matching {
 
 QueryEnvironment::QueryEnvironment(const IIndexEnvironment &indexEnv,
                                    const IAttributeContext &attrContext,
-                                   const Properties &properties)
+                                   const Properties &properties,
+                                   const search::index::IFieldLengthInspector &field_length_inspector)
     : _indexEnv(indexEnv),
       _attrContext(attrContext),
       _properties(properties),
       _locations(1),
-      _terms()
+      _terms(),
+      _field_length_inspector(field_length_inspector)
 {
 }
 
@@ -53,6 +56,12 @@ QueryEnvironment::getAttributeContext() const
     return _attrContext;
 }
 
+double
+QueryEnvironment::get_average_field_length(const vespalib::string &field_name) const
+{
+    return _field_length_inspector.get_field_length_info(field_name).get_average_field_length();
+}
+
 const search::fef::IIndexEnvironment &
 QueryEnvironment::getIndexEnvironment() const
 {
diff --git a/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.h b/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.h
index d79ba1796f7..8f958870d52 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.h
+++ b/searchcore/src/vespa/searchcore/proton/matching/queryenvironment.h
@@ -6,6 +6,8 @@
 #include <vespa/searchlib/fef/properties.h>
 #include <vespa/searchlib/fef/location.h>
 
+namespace search::index { class IFieldLengthInspector; }
+
 namespace proton::matching {
 
 /**
@@ -19,6 +21,7 @@ private:
     search::fef::Properties                     _properties;
     std::vector<const search::fef::Location *>  _locations;
     std::vector<const search::fef::ITermData *> _terms;
+    const search::index::IFieldLengthInspector &_field_length_inspector;
 
     QueryEnvironment(const QueryEnvironment &);
     QueryEnvironment &operator=(const QueryEnvironment &);
@@ -33,7 +36,8 @@ public:
      **/
     QueryEnvironment(const search::fef::IIndexEnvironment &indexEnv,
                      const search::attribute::IAttributeContext &attrContext,
-                     const search::fef::Properties &properties);
+                     const search::fef::Properties &properties,
+                     const search::index::IFieldLengthInspector &field_length_inspector);
 
     /**
      * Used to edit the list of terms by the one setting up this query
@@ -71,6 +75,8 @@ public:
     // inherited from search::fef::IQueryEnvironment
     const search::attribute::IAttributeContext & getAttributeContext() const override;
 
+    double get_average_field_length(const vespalib::string &field_name) const override;
+
     // inherited from search::fef::IQueryEnvironment
     const search::fef::IIndexEnvironment & getIndexEnvironment() const override;
 
diff --git a/searchcore/src/vespa/searchcore/proton/matching/same_element_builder.cpp b/searchcore/src/vespa/searchcore/proton/matching/same_element_builder.cpp
index d3a0ec4726f..16c86e8a4f5 100644
--- a/searchcore/src/vespa/searchcore/proton/matching/same_element_builder.cpp
+++ b/searchcore/src/vespa/searchcore/proton/matching/same_element_builder.cpp
@@ -4,6 +4,7 @@
 #include "querynodes.h"
 #include <vespa/searchlib/query/tree/customtypevisitor.h>
 #include <vespa/searchlib/queryeval/leaf_blueprints.h>
+#include <vespa/searchcorespi/index/indexsearchable.h>
 
 using search::queryeval::Blueprint;
 using search::queryeval::EmptyBlueprint;
diff --git a/searchcore/src/vespa/searchcore/proton/server/searchcontext.cpp b/searchcore/src/vespa/searchcore/proton/server/searchcontext.cpp
index ea09c60bd52..d9207ef70e1 100644
--- a/searchcore/src/vespa/searchcore/proton/server/searchcontext.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/searchcontext.cpp
@@ -3,10 +3,11 @@
 #include "searchcontext.h"
 
 using search::queryeval::Searchable;
+using searchcorespi::IndexSearchable;
 
 namespace proton {
 
-Searchable &
+IndexSearchable &
 SearchContext::getIndexes()
 {
     return *_indexSearchable;
@@ -23,7 +24,7 @@ uint32_t SearchContext::getDocIdLimit()
     return _docIdLimit;
 }
 
-SearchContext::SearchContext(const Searchable::SP &indexSearchable, uint32_t docIdLimit)
+SearchContext::SearchContext(const std::shared_ptr<IndexSearchable> &indexSearchable, uint32_t docIdLimit)
     : _indexSearchable(indexSearchable),
       _attributeBlueprintFactory(),
       _docIdLimit(docIdLimit)
diff --git a/searchcore/src/vespa/searchcore/proton/server/searchcontext.h b/searchcore/src/vespa/searchcore/proton/server/searchcontext.h
index 71475c3decd..b9ea6b334b3 100644
--- a/searchcore/src/vespa/searchcore/proton/server/searchcontext.h
+++ b/searchcore/src/vespa/searchcore/proton/server/searchcontext.h
@@ -16,16 +16,16 @@ class SearchContext : public matching::ISearchContext
 {
 private:
     /// Snapshot of the indexes used.
-    Searchable::SP                    _indexSearchable;
+    std::shared_ptr<IndexSearchable>  _indexSearchable;
     search::AttributeBlueprintFactory _attributeBlueprintFactory;
     uint32_t                          _docIdLimit;
 
-    Searchable &getIndexes() override;
+    IndexSearchable &getIndexes() override;
     Searchable &getAttributes() override;
     uint32_t getDocIdLimit() override;
 
 public:
-    SearchContext(const Searchable::SP &indexSearchable, uint32_t docIdLimit);
+    SearchContext(const std::shared_ptr<IndexSearchable> &indexSearchable, uint32_t docIdLimit);
 };
 
 } // namespace proton
diff --git a/searchlib/src/vespa/searchlib/fef/iqueryenvironment.h b/searchlib/src/vespa/searchlib/fef/iqueryenvironment.h
index a7f268e5c6b..041e9ec67bc 100644
--- a/searchlib/src/vespa/searchlib/fef/iqueryenvironment.h
+++ b/searchlib/src/vespa/searchlib/fef/iqueryenvironment.h
@@ -70,6 +70,15 @@ public:
     virtual const search::attribute::IAttributeContext & getAttributeContext() const = 0;
 
     /**
+     * Returns the average field length for the given field.
+     *
+     * @param field_name field name
+     *
+     * @return average field length
+     **/
+    virtual double get_average_field_length(const vespalib::string &field_name) const = 0;
+
+    /**
      * Returns a const view of the index environment.
      *
      * @return index environment
diff --git a/searchlib/src/vespa/searchlib/fef/phrasesplitter.h b/searchlib/src/vespa/searchlib/fef/phrasesplitter.h
index 4c7ca4b67d7..4e46c9eaa7c 100644
--- a/searchlib/src/vespa/searchlib/fef/phrasesplitter.h
+++ b/searchlib/src/vespa/searchlib/fef/phrasesplitter.h
@@ -113,6 +113,7 @@ public:
     const Properties & getProperties() const override { return _queryEnv.getProperties(); }
     const Location & getLocation() const override { return _queryEnv.getLocation(); }
     const attribute::IAttributeContext & getAttributeContext() const override { return _queryEnv.getAttributeContext(); }
+    double get_average_field_length(const vespalib::string &field_name) const override { return _queryEnv.get_average_field_length(field_name); }
     const IIndexEnvironment & getIndexEnvironment() const override { return _queryEnv.getIndexEnvironment(); }
     void bind_match_data(const fef::MatchData &md) { _matchData = &md; }
 };
diff --git a/searchlib/src/vespa/searchlib/fef/test/queryenvironment.h b/searchlib/src/vespa/searchlib/fef/test/queryenvironment.h
index 0179b5020e6..4d7a92586e6 100644
--- a/searchlib/src/vespa/searchlib/fef/test/queryenvironment.h
+++ b/searchlib/src/vespa/searchlib/fef/test/queryenvironment.h
@@ -40,6 +40,9 @@ public:
     const ITermData *getTerm(uint32_t idx) const override { return idx < _terms.size() ? &_terms[idx] : NULL; }
     const Location & getLocation() const override { return _location; }
     const search::attribute::IAttributeContext &getAttributeContext() const override { return *_attrCtx; }
+    double get_average_field_length(const vespalib::string &) const override {
+        return 1.0;
+    }
     const IIndexEnvironment &getIndexEnvironment() const override { assert(_indexEnv != NULL); return *_indexEnv; }
 
     /** Returns a reference to the index environment of this. */
diff --git a/streamingvisitors/src/vespa/searchvisitor/queryenvironment.h b/streamingvisitors/src/vespa/searchvisitor/queryenvironment.h
index b9391ac838c..db0f95cb6bb 100644
--- a/streamingvisitors/src/vespa/searchvisitor/queryenvironment.h
+++ b/streamingvisitors/src/vespa/searchvisitor/queryenvironment.h
@@ -54,6 +54,8 @@ public:
     // inherit documentation
     virtual const search::attribute::IAttributeContext & getAttributeContext() const override { return *_attrCtx; }
 
+    double get_average_field_length(const vespalib::string &) const override { return 1.0; }
+
     // inherit documentation
     virtual const search::fef::IIndexEnvironment & getIndexEnvironment() const override { return _indexEnv; }