Merge pull request #29706 from vespa-engine/geirst/prepare-for-in-term-used-for-ranking

Support TermFieldMatchData where doUnpack() sets docid.

3 files changed, 36 insertions, 10 deletions

diff --git a/searchlib/src/tests/attribute/multi_term_or_filter_search/multi_term_or_filter_search_test.cpp b/searchlib/src/tests/attribute/multi_term_or_filter_search/multi_term_or_filter_search_test.cpp
index dea2702ef0d..552a128c518 100644
--- a/searchlib/src/tests/attribute/multi_term_or_filter_search/multi_term_or_filter_search_test.cpp
+++ b/searchlib/src/tests/attribute/multi_term_or_filter_search/multi_term_or_filter_search_test.cpp
@@ -3,6 +3,7 @@
 #include <vespa/searchlib/attribute/i_direct_posting_store.h>
 #include <vespa/searchlib/attribute/multi_term_or_filter_search.h>
 #include <vespa/searchlib/common/bitvector.h>
+#include <vespa/searchlib/fef/termfieldmatchdata.h>
 #include <vespa/searchlib/queryeval/searchiterator.h>
 #include <vespa/vespalib/gtest/gtest.h>
 #define ENABLE_GTEST_MIGRATION
@@ -11,13 +12,16 @@
 using PostingList = search::attribute::PostingListTraits<int32_t>::PostingStoreBase;
 using Iterator = search::attribute::PostingListTraits<int32_t>::const_iterator;
 using KeyData = PostingList::KeyDataType;
+
 using search::BitVector;
 using search::attribute::MultiTermOrFilterSearch;
+using search::fef::TermFieldMatchData;
 using search::queryeval::SearchIterator;
 using vespalib::datastore::EntryRef;
 
 class MultiTermOrFilterSearchTest : public ::testing::Test {
     PostingList _postings;
+    mutable TermFieldMatchData _tfmd;
     vespalib::GenerationHandler _gens;
     std::vector<EntryRef> _trees;
     uint32_t _range_start;
@@ -62,7 +66,7 @@ public:
         for (size_t i = 0; i < num_trees(); ++i) {
             iterators.emplace_back(get_tree(i));
         }
-        auto result = MultiTermOrFilterSearch::create(std::move(iterators));
+        auto result = MultiTermOrFilterSearch::create(std::move(iterators), _tfmd);
         result->initRange(_range_start, _range_end);
         return result;
     };
@@ -73,6 +77,8 @@ public:
         while (doc_id < _range_end) {
             if (iterator.seek(doc_id)) {
                 result.emplace_back(doc_id);
+                iterator.unpack(doc_id);
+                EXPECT_EQ(doc_id, _tfmd.getDocId());
                 ++doc_id;
             } else {
                 doc_id = std::max(doc_id + 1, iterator.getDocId());
diff --git a/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.cpp b/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.cpp
index 19668522e17..214851e9681 100644
--- a/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.cpp
@@ -15,14 +15,19 @@ template<typename IteratorPack>
 class MultiTermOrFilterSearchImpl : public MultiTermOrFilterSearch
 {
     IteratorPack _children;
+    fef::TermFieldMatchData* _tfmd;
     void seek_all(uint32_t docId);
 public:
-    explicit MultiTermOrFilterSearchImpl(IteratorPack&& children);
+    explicit MultiTermOrFilterSearchImpl(IteratorPack&& children, fef::TermFieldMatchData* tfmd);
     ~MultiTermOrFilterSearchImpl() override;
 
     void doSeek(uint32_t docId) override;
     
-    void doUnpack(uint32_t) override { }
+    void doUnpack(uint32_t docid) override {
+        if (_tfmd != nullptr) {
+            _tfmd->resetOnlyDocId(docid);
+        }
+    }
 
     void initRange(uint32_t begin, uint32_t end) override {
         SearchIterator::initRange(begin, end);
@@ -46,9 +51,10 @@ public:
 };
 
 template<typename IteratorPack>
-MultiTermOrFilterSearchImpl<IteratorPack>::MultiTermOrFilterSearchImpl(IteratorPack&& children)
+MultiTermOrFilterSearchImpl<IteratorPack>::MultiTermOrFilterSearchImpl(IteratorPack&& children, fef::TermFieldMatchData* tfmd)
     : MultiTermOrFilterSearch(),
-      _children(std::move(children))
+      _children(std::move(children)),
+      _tfmd(tfmd)
 {
 }
 
@@ -89,7 +95,7 @@ namespace {
 
 template <typename IteratorType, typename IteratorPackType>
 std::unique_ptr<queryeval::SearchIterator>
-create_helper(std::vector<IteratorType>&& children)
+create_helper(std::vector<IteratorType>&& children, fef::TermFieldMatchData* tfmd)
 {
     if (children.empty()) {
         return std::make_unique<queryeval::EmptySearch>();
@@ -97,7 +103,7 @@ create_helper(std::vector<IteratorType>&& children)
         std::sort(children.begin(), children.end(),
                   [](const auto & a, const auto & b) { return a.size() > b.size(); });
         using OrFilter = MultiTermOrFilterSearchImpl<IteratorPackType>;
-        return std::make_unique<OrFilter>(IteratorPackType(std::move(children)));
+        return std::make_unique<OrFilter>(IteratorPackType(std::move(children)), tfmd);
     }
 }
 
@@ -106,13 +112,25 @@ create_helper(std::vector<IteratorType>&& children)
 std::unique_ptr<queryeval::SearchIterator>
 MultiTermOrFilterSearch::create(std::vector<DocidIterator>&& children)
 {
-    return create_helper<DocidIterator, DocidIteratorPack>(std::move(children));
+    return create_helper<DocidIterator, DocidIteratorPack>(std::move(children), nullptr);
+}
+
+std::unique_ptr<queryeval::SearchIterator>
+MultiTermOrFilterSearch::create(std::vector<DocidIterator>&& children, fef::TermFieldMatchData& tfmd)
+{
+    return create_helper<DocidIterator, DocidIteratorPack>(std::move(children), &tfmd);
 }
 
 std::unique_ptr<queryeval::SearchIterator>
 MultiTermOrFilterSearch::create(std::vector<DocidWithWeightIterator>&& children)
 {
-    return create_helper<DocidWithWeightIterator, DocidWithWeightIteratorPack>(std::move(children));
+    return create_helper<DocidWithWeightIterator, DocidWithWeightIteratorPack>(std::move(children), nullptr);
+}
+
+std::unique_ptr<queryeval::SearchIterator>
+MultiTermOrFilterSearch::create(std::vector<DocidWithWeightIterator>&& children, fef::TermFieldMatchData& tfmd)
+{
+    return create_helper<DocidWithWeightIterator, DocidWithWeightIteratorPack>(std::move(children), &tfmd);
 }
 
 std::unique_ptr<queryeval::SearchIterator>
@@ -123,7 +141,7 @@ MultiTermOrFilterSearch::create(const std::vector<SearchIterator *>& children,
         return std::make_unique<queryeval::EmptySearch>();
     } else {
         using OrFilter = MultiTermOrFilterSearchImpl<SearchIteratorPack>;
-        return std::make_unique<OrFilter>(SearchIteratorPack(children, std::move(md)));
+        return std::make_unique<OrFilter>(SearchIteratorPack(children, std::move(md)), nullptr);
     }
 }
 
diff --git a/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.h b/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.h
index 42eb33d2eed..1e8227c3007 100644
--- a/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.h
+++ b/searchlib/src/vespa/searchlib/attribute/multi_term_or_filter_search.h
@@ -18,7 +18,9 @@ protected:
     MultiTermOrFilterSearch() = default;
 public:
     static std::unique_ptr<SearchIterator> create(std::vector<DocidIterator>&& children);
+    static std::unique_ptr<SearchIterator> create(std::vector<DocidIterator>&& children, fef::TermFieldMatchData& tfmd);
     static std::unique_ptr<SearchIterator> create(std::vector<DocidWithWeightIterator>&& children);
+    static std::unique_ptr<SearchIterator> create(std::vector<DocidWithWeightIterator>&& children, fef::TermFieldMatchData& tfmd);
     static std::unique_ptr<SearchIterator> create(const std::vector<SearchIterator *>& children,
                                                   std::unique_ptr<fef::MatchData> md);
 };