Merge pull request #30018 from vespa-engine/toregge/add-hit-iterator-pack

Add hit iterator pack and use it for phrase search in streaming mode.

3 files changed, 185 insertions, 0 deletions

diff --git a/searchlib/src/tests/query/streaming/CMakeLists.txt b/searchlib/src/tests/query/streaming/CMakeLists.txt
new file mode 100644
index 00000000000..d40250fbc3f
--- /dev/null
+++ b/searchlib/src/tests/query/streaming/CMakeLists.txt
@@ -0,0 +1,19 @@
+# Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+vespa_add_executable(searchlib_query_streaming_hit_iterator_test_app TEST
+    SOURCES
+    hit_iterator_test.cpp
+    DEPENDS
+    searchlib
+    GTest::gtest
+)
+vespa_add_test(NAME searchlib_query_streaming_hit_iterator_test_app COMMAND searchlib_query_streaming_hit_iterator_test_app)
+
+vespa_add_executable(searchlib_query_streaming_hit_iterator_pack_test_app TEST
+    SOURCES
+    hit_iterator_pack_test.cpp
+    DEPENDS
+    searchlib
+    GTest::gtest
+)
+vespa_add_test(NAME searchlib_query_streaming_hit_iterator_pack_test_app COMMAND searchlib_query_streaming_hit_iterator_pack_test_app)
diff --git a/searchlib/src/tests/query/streaming/hit_iterator_pack_test.cpp b/searchlib/src/tests/query/streaming/hit_iterator_pack_test.cpp
new file mode 100644
index 00000000000..7d7d8307920
--- /dev/null
+++ b/searchlib/src/tests/query/streaming/hit_iterator_pack_test.cpp
@@ -0,0 +1,44 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include <vespa/searchlib/query/streaming/hit_iterator_pack.h>
+#include <vespa/vespalib/gtest/gtest.h>
+
+using search::streaming::HitIterator;
+using search::streaming::HitIteratorPack;
+using search::streaming::QueryNodeList;
+using search::streaming::QueryTerm;
+using search::streaming::QueryNodeResultBase;
+
+using FieldElement = HitIterator::FieldElement;
+
+TEST(HitIteratorPackTest, seek_to_matching_field_element)
+{
+    QueryNodeList qnl;
+    auto qt = std::make_unique<QueryTerm>(std::unique_ptr<QueryNodeResultBase>(), "7", "", QueryTerm::Type::WORD);
+    qt->add(11, 0, 10, 0);
+    qt->add(11, 0, 10, 5);
+    qt->add(11, 1, 12, 0);
+    qt->add(11, 1, 12, 0);
+    qt->add(12, 1, 13, 0);
+    qt->add(12, 1, 13, 0);
+    qnl.emplace_back(std::move(qt));
+    qt = std::make_unique<QueryTerm>(std::unique_ptr<QueryNodeResultBase>(), "8", "", QueryTerm::Type::WORD);
+    qt->add(2, 0, 4, 0);
+    qt->add(11, 0, 10, 0);
+    qt->add(12, 1, 13, 0);
+    qt->add(12, 2, 14, 0);
+    qnl.emplace_back(std::move(qt));
+    HitIteratorPack itr_pack(qnl);
+    EXPECT_TRUE(itr_pack.all_valid());
+    EXPECT_TRUE(itr_pack.seek_to_matching_field_element());
+    EXPECT_EQ(FieldElement(11, 0), itr_pack.get_field_element_ref());
+    EXPECT_TRUE(itr_pack.seek_to_matching_field_element());
+    EXPECT_EQ(FieldElement(11, 0), itr_pack.get_field_element_ref());
+    ++itr_pack.get_field_element_ref().second;
+    EXPECT_TRUE(itr_pack.seek_to_matching_field_element());
+    EXPECT_EQ(FieldElement(12, 1), itr_pack.get_field_element_ref());
+    ++itr_pack.get_field_element_ref().second;
+    EXPECT_FALSE(itr_pack.seek_to_matching_field_element());
+}
+
+GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchlib/src/tests/query/streaming/hit_iterator_test.cpp b/searchlib/src/tests/query/streaming/hit_iterator_test.cpp
new file mode 100644
index 00000000000..a9588ea3d6c
--- /dev/null
+++ b/searchlib/src/tests/query/streaming/hit_iterator_test.cpp
@@ -0,0 +1,122 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include <vespa/searchlib/query/streaming/hit_iterator.h>
+#include <vespa/vespalib/gtest/gtest.h>
+
+using search::streaming::Hit;
+using search::streaming::HitList;
+using search::streaming::HitIterator;
+
+using FieldElement = HitIterator::FieldElement;
+
+namespace {
+
+HitList
+make_hit_list()
+{
+    HitList hl;
+    hl.emplace_back(11, 0, 10, 0);
+    hl.emplace_back(11, 0, 10, 5);
+    hl.emplace_back(11, 1, 12, 0);
+    hl.emplace_back(11, 1, 12, 7);
+    hl.emplace_back(12, 1, 13, 0);
+    hl.emplace_back(12, 1, 13, 9);
+    return hl;
+}
+
+void
+check_seek_to_field_elem(HitIterator& it, const FieldElement& field_element, const Hit* exp_ptr, const vespalib::string& label)
+{
+    SCOPED_TRACE(label);
+    EXPECT_TRUE(it.seek_to_field_element(field_element));
+    EXPECT_TRUE(it.valid());
+    EXPECT_EQ(exp_ptr, &*it);
+}
+
+void
+check_seek_to_field_elem_failure(HitIterator& it, const FieldElement& field_element, const vespalib::string& label)
+{
+    SCOPED_TRACE(label);
+    EXPECT_FALSE(it.seek_to_field_element(field_element));
+    EXPECT_FALSE(it.valid());
+}
+
+void
+check_step_in_field_element(HitIterator& it, FieldElement& field_element, bool exp_success, const Hit* exp_ptr, const vespalib::string& label)
+{
+    SCOPED_TRACE(label);
+    EXPECT_EQ(exp_success, it.step_in_field_element(field_element));
+    if (exp_ptr) {
+        EXPECT_TRUE(it.valid());
+        EXPECT_EQ(it.get_field_element(), field_element);
+        EXPECT_EQ(exp_ptr, &*it);
+    } else {
+        EXPECT_FALSE(it.valid());
+    }
+}
+
+void
+check_seek_in_field_element(HitIterator& it, uint32_t position, FieldElement& field_element, bool exp_success, const Hit* exp_ptr, const vespalib::string& label)
+{
+    SCOPED_TRACE(label);
+    EXPECT_EQ(exp_success, it.seek_in_field_element(position, field_element));
+    if (exp_ptr) {
+        EXPECT_TRUE(it.valid());
+        EXPECT_EQ(it.get_field_element(), field_element);
+        EXPECT_EQ(exp_ptr, &*it);
+    } else {
+        EXPECT_FALSE(it.valid());
+    }
+}
+
+}
+
+TEST(HitITeratorTest, seek_to_field_element)
+{
+    auto hl = make_hit_list();
+    HitIterator it(hl);
+    EXPECT_TRUE(it.valid());
+    EXPECT_EQ(&hl[0], &*it);
+    check_seek_to_field_elem(it, FieldElement(0, 0), &hl[0], "(0, 0)");
+    check_seek_to_field_elem(it, FieldElement(11, 0), &hl[0], "(11, 0)");
+    check_seek_to_field_elem(it, FieldElement(11, 1), &hl[2], "(11, 1)");
+    check_seek_to_field_elem(it, FieldElement(11, 2), &hl[4], "(11, 2)");
+    check_seek_to_field_elem(it, FieldElement(12, 0), &hl[4], "(12, 0)");
+    check_seek_to_field_elem(it, FieldElement(12, 1), &hl[4], "(12, 1)");
+    check_seek_to_field_elem_failure(it, FieldElement(12, 2), "(12, 2)");
+    check_seek_to_field_elem_failure(it, FieldElement(13, 0), "(13, 0)");
+}
+
+TEST(HitIteratorTest, step_in_field_element)
+{
+    auto hl = make_hit_list();
+    HitIterator it(hl);
+    auto field_element = it.get_field_element();
+    check_step_in_field_element(it, field_element, true, &hl[1], "1");
+    check_step_in_field_element(it, field_element, false, &hl[2], "2");
+    check_step_in_field_element(it, field_element, true, &hl[3], "3");
+    check_step_in_field_element(it, field_element, false, &hl[4], "4");
+    check_step_in_field_element(it, field_element, true, &hl[5], "5");
+    check_step_in_field_element(it, field_element, false, nullptr, "end");
+}
+
+TEST(hitIteratorTest, seek_in_field_elem)
+{
+    auto hl = make_hit_list();
+    HitIterator it(hl);
+    auto field_element = it.get_field_element();
+    check_seek_in_field_element(it, 0, field_element, true, &hl[0], "0a");
+    check_seek_in_field_element(it, 2, field_element, true, &hl[1], "2");
+    check_seek_in_field_element(it, 5, field_element, true, &hl[1], "5");
+    check_seek_in_field_element(it, 6, field_element, false, &hl[2], "6");
+    check_seek_in_field_element(it, 0, field_element, true, &hl[2], "0b");
+    check_seek_in_field_element(it, 1, field_element, true, &hl[3], "1");
+    check_seek_in_field_element(it, 7, field_element, true, &hl[3], "7");
+    check_seek_in_field_element(it, 8, field_element, false, &hl[4], "8");
+    check_seek_in_field_element(it, 0, field_element, true, &hl[4], "0c");
+    check_seek_in_field_element(it, 3, field_element, true, &hl[5], "3");
+    check_seek_in_field_element(it, 9, field_element, true, &hl[5], "9");
+    check_seek_in_field_element(it, 10, field_element, false, nullptr, "end");
+}
+
+GTEST_MAIN_RUN_ALL_TESTS()