Merge pull request #27562 from vespa-engine/havardpe/current-index-setup

add tool used to wire CurrentIndex dependencies in grouping

6 files changed, 191 insertions, 0 deletions

diff --git a/searchlib/CMakeLists.txt b/searchlib/CMakeLists.txt
index 07045684d6e..eb199a7ba44 100644
--- a/searchlib/CMakeLists.txt
+++ b/searchlib/CMakeLists.txt
@@ -126,6 +126,7 @@ vespa_define_module(
     src/tests/engine/proto_converter
     src/tests/engine/proto_rpc_adapter
     src/tests/expression/attributenode
+    src/tests/expression/current_index_setup
     src/tests/features
     src/tests/features/beta
     src/tests/features/bm25
diff --git a/searchlib/src/tests/expression/current_index_setup/CMakeLists.txt b/searchlib/src/tests/expression/current_index_setup/CMakeLists.txt
new file mode 100644
index 00000000000..dc50dea4ce3
--- /dev/null
+++ b/searchlib/src/tests/expression/current_index_setup/CMakeLists.txt
@@ -0,0 +1,9 @@
+# Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+vespa_add_executable(searchlib_current_index_setup_test_app TEST
+    SOURCES
+    current_index_setup_test.cpp
+    DEPENDS
+    searchlib
+    GTest::GTest
+)
+vespa_add_test(NAME searchlib_current_index_setup_test_app COMMAND searchlib_current_index_setup_test_app)
diff --git a/searchlib/src/tests/expression/current_index_setup/current_index_setup_test.cpp b/searchlib/src/tests/expression/current_index_setup/current_index_setup_test.cpp
new file mode 100644
index 00000000000..20818e576dd
--- /dev/null
+++ b/searchlib/src/tests/expression/current_index_setup/current_index_setup_test.cpp
@@ -0,0 +1,58 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include <vespa/searchlib/expression/current_index_setup.h>
+#include <vespa/vespalib/gtest/gtest.h>
+
+using search::expression::CurrentIndex;
+using search::expression::CurrentIndexSetup;
+
+TEST(CurrentIndexSetupTest, bound_structs_can_be_resolved) {
+    CurrentIndexSetup setup;
+    CurrentIndex foo_idx;
+    CurrentIndex bar_idx;
+    setup.bind("foo", foo_idx);
+    setup.bind("foo.bar", bar_idx);
+    EXPECT_EQ(setup.resolve("plain"), nullptr);
+    EXPECT_EQ(setup.resolve("foo.a"), &foo_idx);
+    EXPECT_EQ(setup.resolve("foo.b"), &foo_idx);
+    EXPECT_EQ(setup.resolve("foo.c"), &foo_idx);
+    EXPECT_EQ(setup.resolve("foo.bar.x"), &bar_idx);
+    EXPECT_EQ(setup.resolve("foo.bar.y"), &bar_idx);
+    EXPECT_EQ(setup.resolve("baz.f"), nullptr);
+    EXPECT_EQ(setup.resolve("foo.baz.f"), nullptr);
+}
+
+TEST(CurrentIndexSetupTest, unbound_struct_usage_can_be_captured) {
+    CurrentIndexSetup setup;
+    CurrentIndexSetup::Usage usage;
+    CurrentIndex foo_idx;
+    setup.bind("foo", foo_idx);
+    EXPECT_FALSE(usage.has_single_unbound_struct());
+    {
+        CurrentIndexSetup::Usage::Bind capture_guard(setup, usage);
+        EXPECT_EQ(setup.resolve("foo.a"), &foo_idx);
+        EXPECT_EQ(setup.resolve("bar.a"), nullptr);
+        EXPECT_EQ(setup.resolve("bar.b"), nullptr);
+        EXPECT_EQ(setup.resolve("plain"), nullptr);
+    }
+    EXPECT_EQ(setup.resolve("baz.a"), nullptr);
+    EXPECT_TRUE(usage.has_single_unbound_struct());
+    EXPECT_EQ(usage.get_unbound_struct_name(), "bar");
+}
+
+TEST(CurrentIndexSetupTest, multi_unbound_struct_conflict_can_be_captured) {
+    CurrentIndexSetup setup;
+    CurrentIndexSetup::Usage usage;
+    EXPECT_FALSE(usage.has_single_unbound_struct());
+    {
+        CurrentIndexSetup::Usage::Bind capture_guard(setup, usage);
+        EXPECT_FALSE(usage.has_single_unbound_struct());
+        EXPECT_EQ(setup.resolve("foo.a"), nullptr);
+        EXPECT_TRUE(usage.has_single_unbound_struct());
+        EXPECT_EQ(setup.resolve("bar.a"), nullptr);
+        EXPECT_FALSE(usage.has_single_unbound_struct());
+    }
+    EXPECT_FALSE(usage.has_single_unbound_struct());
+}
+
+GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchlib/src/vespa/searchlib/expression/CMakeLists.txt b/searchlib/src/vespa/searchlib/expression/CMakeLists.txt
index 9a3e377e5ef..6942a827fe9 100644
--- a/searchlib/src/vespa/searchlib/expression/CMakeLists.txt
+++ b/searchlib/src/vespa/searchlib/expression/CMakeLists.txt
@@ -4,6 +4,7 @@ vespa_add_library(searchlib_expression OBJECT
     attribute_map_lookup_node.cpp
     attributenode.cpp
     attributeresult.cpp
+    current_index_setup.cpp
     enumattributeresult.cpp
     perdocexpression.cpp
     expressiontree.cpp
diff --git a/searchlib/src/vespa/searchlib/expression/current_index_setup.cpp b/searchlib/src/vespa/searchlib/expression/current_index_setup.cpp
new file mode 100644
index 00000000000..6660728dea7
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/expression/current_index_setup.cpp
@@ -0,0 +1,74 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include "current_index_setup.h"
+#include <vespa/vespalib/stllike/hash_map.hpp>
+#include <cassert>
+
+namespace search::expression {
+
+void
+CurrentIndexSetup::Usage::notify_unbound_struct_usage(vespalib::stringref name)
+{
+    _unbound.insert(name);
+}
+
+CurrentIndexSetup::Usage::Usage()
+  : _unbound()
+{
+}
+
+CurrentIndexSetup::Usage::~Usage() = default;
+
+vespalib::stringref
+CurrentIndexSetup::Usage::get_unbound_struct_name() const
+{
+    assert(has_single_unbound_struct());
+    return *_unbound.begin();
+}
+
+CurrentIndexSetup::Usage::Bind::Bind(CurrentIndexSetup &setup, Usage &usage) noexcept
+  : _setup(setup)
+{
+    auto prev = setup.capture(std::addressof(usage));
+    assert(prev == nullptr); // no nesting
+}
+
+CurrentIndexSetup::Usage::Bind::~Bind()
+{
+    [[maybe_unused]] auto prev = _setup.capture(nullptr);
+}
+
+CurrentIndexSetup::CurrentIndexSetup()
+  : _bound(), _usage(nullptr)
+{
+}
+
+CurrentIndexSetup::~CurrentIndexSetup() = default;
+
+const CurrentIndex *
+CurrentIndexSetup::resolve(vespalib::stringref field_name) const
+{
+    size_t pos = field_name.rfind('.');
+    if (pos > field_name.size()) {
+        return nullptr;
+    }
+    auto struct_name = field_name.substr(0, pos);
+    auto entry = _bound.find(struct_name);
+    if (entry == _bound.end()) {
+        if (_usage != nullptr) {
+            _usage->notify_unbound_struct_usage(struct_name);
+        }
+        return nullptr;
+    }
+    return entry->second;
+}
+
+void
+CurrentIndexSetup::bind(vespalib::stringref struct_name, const CurrentIndex &index)
+{
+    auto res = _bound.insert(std::make_pair(vespalib::string(struct_name),
+                                            std::addressof(index)));
+    assert(res.second); // struct must be either bound or unbound
+}
+
+}
diff --git a/searchlib/src/vespa/searchlib/expression/current_index_setup.h b/searchlib/src/vespa/searchlib/expression/current_index_setup.h
new file mode 100644
index 00000000000..4f835d064bb
--- /dev/null
+++ b/searchlib/src/vespa/searchlib/expression/current_index_setup.h
@@ -0,0 +1,48 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include "currentindex.h"
+#include <vespa/vespalib/stllike/string.h>
+#include <vespa/vespalib/stllike/hash_map.h>
+#include <vespa/vespalib/stllike/hash_set.h>
+#include <utility>
+
+namespace search::expression {
+
+class CurrentIndexSetup {
+public:
+    class Usage {
+    private:
+        friend class CurrentIndexSetup;
+        vespalib::hash_set<vespalib::string> _unbound;
+        void notify_unbound_struct_usage(vespalib::stringref name);
+    public:
+        Usage();
+        ~Usage();
+        [[nodiscard]] bool has_single_unbound_struct() const noexcept {
+            return (_unbound.size() == 1);
+        }
+        vespalib::stringref get_unbound_struct_name() const;
+        class Bind {
+        private:
+            CurrentIndexSetup &_setup;
+        public:
+            Bind(CurrentIndexSetup &setup, Usage &usage) noexcept;
+            ~Bind();
+        };
+    };
+private:
+    vespalib::hash_map<vespalib::string, const CurrentIndex *> _bound;
+    Usage *_usage;
+    [[nodiscard]] Usage *capture(Usage *usage) noexcept {
+        return std::exchange(_usage, usage);
+    }
+public:
+    CurrentIndexSetup();
+    ~CurrentIndexSetup();
+    [[nodiscard]] const CurrentIndex *resolve(vespalib::stringref field_name) const;    
+    void bind(vespalib::stringref struct_name, const CurrentIndex &index);
+};
+
+}