Add unpack_match_data member function to search::streaming::QueryTerm.

4 files changed, 6 insertions, 76 deletions

diff --git a/streamingvisitors/src/tests/querywrapper/querywrapper_test.cpp b/streamingvisitors/src/tests/querywrapper/querywrapper_test.cpp
index dc3dbfca7ca..70b863e540b 100644
--- a/streamingvisitors/src/tests/querywrapper/querywrapper_test.cpp
+++ b/streamingvisitors/src/tests/querywrapper/querywrapper_test.cpp
@@ -84,12 +84,6 @@ QueryWrapperTest::testQueryWrapper()
         EXPECT_TRUE( tl[2].isPhraseTerm());
         EXPECT_TRUE( tl[3].isPhraseTerm());
         EXPECT_TRUE(!tl[4].isPhraseTerm());
-
-        EXPECT_EQUAL(tl[0].getPosAdjust(), 0u);
-        EXPECT_EQUAL(tl[1].getPosAdjust(), 2u);
-        EXPECT_EQUAL(tl[2].getPosAdjust(), 2u);
-        EXPECT_EQUAL(tl[3].getPosAdjust(), 2u);
-        EXPECT_EQUAL(tl[4].getPosAdjust(), 0u);
     }
 }
 
diff --git a/streamingvisitors/src/tests/rank_processor/rank_processor_test.cpp b/streamingvisitors/src/tests/rank_processor/rank_processor_test.cpp
index c9518b29884..0abff37d622 100644
--- a/streamingvisitors/src/tests/rank_processor/rank_processor_test.cpp
+++ b/streamingvisitors/src/tests/rank_processor/rank_processor_test.cpp
@@ -85,8 +85,8 @@ RankProcessorTest::test_unpack_match_data_for_term_node(bool interleaved_feature
     RankProcessor::unpack_match_data(1, *md, *_query_wrapper);
     EXPECT_EQ(invalid_id, tfmd->getDocId());
     node->add(field_id, 0, 1, 0);
+    node->add(field_id, 0, 1, 1);
     auto& field_info = node->getFieldInfo(field_id);
-    field_info.setHitCount(mock_num_occs);
     field_info.setFieldLength(mock_field_length);
     RankProcessor::unpack_match_data(2, *md, *_query_wrapper);
     EXPECT_EQ(2, tfmd->getDocId());
@@ -97,7 +97,7 @@ RankProcessorTest::test_unpack_match_data_for_term_node(bool interleaved_feature
         EXPECT_EQ(0, tfmd->getNumOccs());
         EXPECT_EQ(0, tfmd->getFieldLength());
     }
-    EXPECT_EQ(1, tfmd->size());
+    EXPECT_EQ(2, tfmd->size());
     node->reset();
     RankProcessor::unpack_match_data(3, *md, *_query_wrapper);
     EXPECT_EQ(2, tfmd->getDocId());
diff --git a/streamingvisitors/src/vespa/searchvisitor/querywrapper.h b/streamingvisitors/src/vespa/searchvisitor/querywrapper.h
index b24f695196e..420ff215833 100644
--- a/streamingvisitors/src/vespa/searchvisitor/querywrapper.h
+++ b/streamingvisitors/src/vespa/searchvisitor/querywrapper.h
@@ -47,7 +47,6 @@ public:
         size_t getIndex() const { return _index; }
         bool isPhraseTerm() const { return _parent != nullptr; }
         bool isFirstPhraseTerm() const { return isPhraseTerm() && getIndex() == 0; }
-        size_t getPosAdjust() const { return _parent != nullptr ? _parent->width() - 1 : 0; }
         bool isGeoPosTerm() const { return (_term != nullptr) && _term->isGeoLoc(); }
     };
 
diff --git a/streamingvisitors/src/vespa/searchvisitor/rankprocessor.cpp b/streamingvisitors/src/vespa/searchvisitor/rankprocessor.cpp
index 96e8ca89a04..09699f79427 100644
--- a/streamingvisitors/src/vespa/searchvisitor/rankprocessor.cpp
+++ b/streamingvisitors/src/vespa/searchvisitor/rankprocessor.cpp
@@ -53,11 +53,6 @@ getFeature(const RankProgram &rankProgram) {
     return resolver.resolve(0);
 }
 
-uint16_t
-cap_16_bits(uint32_t value) {
-    return std::min(value, static_cast<uint32_t>(std::numeric_limits<uint16_t>::max()));
-}
-
 }
 
 void
@@ -285,68 +280,10 @@ RankProcessor::unpack_match_data(uint32_t docid, MatchData &matchData, QueryWrap
             QueryTermData & qtd = static_cast<QueryTermData &>(term.getTerm()->getQueryItem());
             const ITermData &td = qtd.getTermData();
 
-            HitList list;
-            const HitList & hitList = isPhrase
-                                      ? term.getParent()->evaluateHits(list)
-                                      : term.getTerm()->evaluateHits(list);
-
-            if (hitList.size() > 0) { // only unpack if we have a hit
-                LOG(debug, "Unpack match data for query term '%s:%s' (%s)",
-                    term.getTerm()->index().c_str(), term.getTerm()->getTerm(), isPhrase ? "phrase" : "term");
-
-                uint32_t lastFieldId = -1;
-                TermFieldMatchData *tmd = nullptr;
-                uint32_t fieldLen = search::fef::FieldPositionsIterator::UNKNOWN_LENGTH;
-                uint32_t num_occs = 0;
-
-                // optimize for hitlist giving all hits for a single field in one chunk
-                for (const Hit & hit : hitList) {
-                    uint32_t fieldId = hit.field_id();
-                    if (fieldId != lastFieldId) {
-                        // reset to notfound/unknown values
-                        tmd = nullptr;
-                        fieldLen = search::fef::FieldPositionsIterator::UNKNOWN_LENGTH;
-                        num_occs = 0;
-
-                        // setup for new field that had a hit
-                        const ITermFieldData *tfd = td.lookupField(fieldId);
-                        if (tfd != nullptr) {
-                            tmd = matchData.resolveTermField(tfd->getHandle());
-                            tmd->setFieldId(fieldId);
-                            // reset field match data, but only once per docId
-                            if (tmd->getDocId() != docid) {
-                                tmd->reset(docid);
-                            }
-                        }
-                        // find fieldLen for new field
-                        if (isPhrase) {
-                            if (fieldId < term.getParent()->getFieldInfoSize()) {
-                                auto& field_info = term.getParent()->getFieldInfo(fieldId);
-                                fieldLen = field_info.getFieldLength();
-                                num_occs = field_info.getHitCount();
-                            }
-                        } else {
-                            if (fieldId < term.getTerm()->getFieldInfoSize()) {
-                                auto& field_info = term.getTerm()->getFieldInfo(fieldId);
-                                fieldLen = field_info.getFieldLength();
-                                num_occs = field_info.getHitCount();
-                            }
-                        }
-                        lastFieldId = fieldId;
-                    }
-                    if (tmd != nullptr) {
-                        // adjust so that the position for phrase terms equals the match for the first term
-                        TermFieldMatchDataPosition pos(hit.element_id(), hit.position() - term.getPosAdjust(),
-                                                       hit.element_weight(), hit.element_length());
-                        tmd->appendPosition(pos);
-                        LOG(debug, "Append elemId(%u),position(%u), weight(%d), tfmd.weight(%d)",
-                                   pos.getElementId(), pos.getPosition(), pos.getElementWeight(), tmd->getWeight());
-                        if (tmd->needs_interleaved_features()) {
-                            tmd->setFieldLength(cap_16_bits(fieldLen));
-                            tmd->setNumOccs(cap_16_bits(num_occs));
-                        }
-                    }
-                }
+            if (isPhrase) {
+                term.getParent()->unpack_match_data(docid, td, matchData);
+            } else {
+                term.getTerm()->unpack_match_data(docid, td, matchData);
             }
         }
     }