Add DocumentIdDFW, used to print document id without using docsum blob.

11 files changed, 246 insertions, 0 deletions

diff --git a/searchsummary/CMakeLists.txt b/searchsummary/CMakeLists.txt
index 17b46499317..8c34f1b2016 100644
--- a/searchsummary/CMakeLists.txt
+++ b/searchsummary/CMakeLists.txt
@@ -19,6 +19,7 @@ vespa_define_module(
     src/tests/docsummary
     src/tests/docsummary/attribute_combiner
     src/tests/docsummary/attributedfw
+    src/tests/docsummary/document_id_dfw
     src/tests/docsummary/matched_elements_filter
     src/tests/docsummary/slime_summary
     src/tests/docsummary/summary_field_converter
diff --git a/searchsummary/src/tests/docsummary/document_id_dfw/CMakeLists.txt b/searchsummary/src/tests/docsummary/document_id_dfw/CMakeLists.txt
new file mode 100644
index 00000000000..3591a36c8de
--- /dev/null
+++ b/searchsummary/src/tests/docsummary/document_id_dfw/CMakeLists.txt
@@ -0,0 +1,10 @@
+# Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+vespa_add_executable(searchsummary_document_id_dfw_test_app TEST
+    SOURCES
+    document_id_dfw_test.cpp
+    DEPENDS
+    searchsummary
+    searchsummary_test
+    GTest::GTest
+)
+vespa_add_test(NAME searchsummary_document_id_dfw_test_app COMMAND searchsummary_document_id_dfw_test_app)
diff --git a/searchsummary/src/tests/docsummary/document_id_dfw/document_id_dfw_test.cpp b/searchsummary/src/tests/docsummary/document_id_dfw/document_id_dfw_test.cpp
new file mode 100644
index 00000000000..0ea78b722c0
--- /dev/null
+++ b/searchsummary/src/tests/docsummary/document_id_dfw/document_id_dfw_test.cpp
@@ -0,0 +1,153 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include <vespa/document/base/documentid.h>
+#include <vespa/document/datatype/documenttype.h>
+#include <vespa/document/fieldvalue/document.h>
+#include <vespa/document/repo/configbuilder.h>
+#include <vespa/document/repo/documenttyperepo.h>
+#include <vespa/searchsummary/docsummary/docsum_blob_entry_filter.h>
+#include <vespa/searchsummary/docsummary/docsum_store_document.h>
+#include <vespa/searchsummary/docsummary/document_id_dfw.h>
+#include <vespa/searchsummary/docsummary/general_result.h>
+#include <vespa/searchsummary/docsummary/resultconfig.h>
+#include <vespa/searchsummary/docsummary/resultpacker.h>
+#include <vespa/vespalib/data/slime/slime.h>
+#include <vespa/vespalib/gtest/gtest.h>
+#include <iostream>
+#include <memory>
+
+using document::Document;
+using document::DocumentId;
+using document::DocumentType;
+using document::DocumentTypeRepo;
+using document::config_builder::DocumenttypesConfigBuilderHelper;
+using document::config_builder::Struct;
+using search::docsummary::DocsumBlobEntryFilter;
+using search::docsummary::DocsumStoreDocument;
+using search::docsummary::DocsumStoreValue;
+using search::docsummary::DocumentIdDFW;
+using search::docsummary::IDocsumStoreDocument;
+using search::docsummary::ResultClass;
+using search::docsummary::ResultConfig;
+using search::docsummary::ResultPacker;
+using search::docsummary::GeneralResult;
+using vespalib::Slime;
+using vespalib::slime::Cursor;
+using vespalib::slime::ObjectInserter;
+using vespalib::slime::SlimeInserter;
+
+namespace {
+
+const int32_t          doc_type_id   = 787121340;
+const vespalib::string doc_type_name = "test";
+const vespalib::string header_name   = doc_type_name + ".header";
+const vespalib::string body_name     = doc_type_name + ".body";
+
+
+std::unique_ptr<const DocumentTypeRepo>
+make_doc_type_repo()
+{
+    DocumenttypesConfigBuilderHelper builder;
+    builder.document(doc_type_id, doc_type_name,
+                     Struct(header_name), Struct(body_name));
+    return std::unique_ptr<const DocumentTypeRepo>(new DocumentTypeRepo(builder.config()));
+}
+
+class DocumentIdDFWTest : public ::testing::Test
+{
+    vespalib::string                        _field_name;
+    vespalib::Memory                        _field_name_view;
+    DocsumBlobEntryFilter                   _docsum_blob_entry_filter;
+    std::unique_ptr<ResultConfig>           _result_config;
+    std::unique_ptr<ResultPacker>           _packer; // owns docsum blob
+    std::unique_ptr<const DocumentTypeRepo> _repo;
+    const DocumentType*                     _document_type;
+
+protected:
+    DocumentIdDFWTest();
+    ~DocumentIdDFWTest() override;
+
+    std::unique_ptr<DocsumStoreDocument> make_docsum_store_document(const vespalib::string &id);
+    std::unique_ptr<DocsumStoreValue> make_docsum_store_value(std::unique_ptr<DocsumStoreDocument> doc);
+    vespalib::Slime write(const DocsumStoreValue& value);
+    vespalib::Memory get_field_name_view() const noexcept { return _field_name_view; }
+};
+
+DocumentIdDFWTest::DocumentIdDFWTest()
+    : testing::Test(),
+      _field_name("documentid"),
+      _field_name_view(_field_name.data(), _field_name.size()),
+      _docsum_blob_entry_filter(DocsumBlobEntryFilter().add_skip(search::docsummary::RES_LONG_STRING)),
+      _result_config(std::make_unique<ResultConfig>(_docsum_blob_entry_filter)),
+      _packer(std::make_unique<ResultPacker>(_result_config.get())),
+      _repo(make_doc_type_repo()),
+      _document_type(_repo->getDocumentType(doc_type_name))
+{
+    auto* cfg = _result_config->AddResultClass("default", 0);
+    cfg->AddConfigEntry(_field_name.c_str(), search::docsummary::RES_LONG_STRING);
+    _result_config->CreateEnumMaps();
+}
+
+
+DocumentIdDFWTest::~DocumentIdDFWTest() = default;
+
+
+std::unique_ptr<DocsumStoreDocument>
+DocumentIdDFWTest::make_docsum_store_document(const vespalib::string& id)
+{
+    auto doc = std::make_unique<Document>(*_document_type, DocumentId(id));
+    doc->setRepo(*_repo);
+    return std::make_unique<DocsumStoreDocument>(std::move(doc));
+}
+
+std::unique_ptr<DocsumStoreValue>
+DocumentIdDFWTest::make_docsum_store_value(std::unique_ptr<DocsumStoreDocument> doc)
+{
+    EXPECT_TRUE(_packer->Init(0));
+    const char *ptr = nullptr;
+    uint32_t len = 0;
+    EXPECT_TRUE(_packer->GetDocsumBlob(&ptr, &len));
+    return std::make_unique<DocsumStoreValue>(ptr, len, std::move(doc));
+}
+
+vespalib::Slime
+DocumentIdDFWTest::write(const DocsumStoreValue& value)
+{
+    auto result = std::make_unique<GeneralResult>(_result_config->LookupResultClass(0));
+    EXPECT_TRUE(result->inplaceUnpack(value));
+    Slime slime;
+    SlimeInserter top_inserter(slime);
+    Cursor & docsum = top_inserter.insertObject();
+    ObjectInserter field_inserter(docsum, _field_name_view);
+    DocumentIdDFW writer;
+    writer.insertField(0, result.get(), nullptr, search::docsummary::RES_LONG_STRING, field_inserter);
+    return slime;
+}
+
+TEST_F(DocumentIdDFWTest, insert_document_id)
+{
+    vespalib::string id("id::test::0");
+    auto doc = make_docsum_store_document(id);
+    auto dsvalue = make_docsum_store_value(std::move(doc));
+    auto slime = write(*dsvalue);
+    EXPECT_TRUE(slime.get()[get_field_name_view()].valid());
+    EXPECT_EQ(id, slime.get()[get_field_name_view()].asString().make_string());
+}
+
+TEST_F(DocumentIdDFWTest, insert_document_id_no_document_doc)
+{
+    auto dsvalue = make_docsum_store_value(std::make_unique<DocsumStoreDocument>(std::unique_ptr<Document>()));
+    auto slime = write(*dsvalue);
+    EXPECT_FALSE(slime.get()[get_field_name_view()].valid());
+}
+
+TEST_F(DocumentIdDFWTest, insert_document_id_no_docsum_store_doc)
+{
+    auto dsvalue = make_docsum_store_value({});
+    auto slime = write(*dsvalue);
+    EXPECT_FALSE(slime.get()[get_field_name_view()].valid());
+}
+
+}
+
+GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt b/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
index e3272fb36de..e262482cddb 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
+++ b/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
@@ -13,6 +13,7 @@ vespa_add_library(searchsummary_docsummary OBJECT
     docsumstate.cpp
     docsumstorevalue.cpp
     docsumwriter.cpp
+    document_id_dfw.cpp
     dynamicteaserdfw.cpp
     empty_dfw.cpp
     general_result.cpp
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp
index e525989e972..e6c8d0b6ab8 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp
@@ -5,6 +5,7 @@
 #include "summaryfieldconverter.h"
 #include <vespa/document/datatype/datatype.h>
 #include <vespa/document/fieldvalue/document.h>
+#include <vespa/vespalib/data/slime/inserter.h>
 
 namespace search::docsummary {
 
@@ -39,4 +40,14 @@ DocsumStoreDocument::insert_summary_field(const vespalib::string& field_name, ve
     }
 }
 
+void
+DocsumStoreDocument::insert_document_id(vespalib::slime::Inserter& inserter) const
+{
+    if (_document) {
+        auto id = _document->getId().toString();
+        vespalib::Memory id_view(id.data(), id.size());
+        inserter.insertString(id_view);
+    }
+}
+
 }
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h
index 66a5a74fa8d..3b0bea6e721 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h
@@ -19,6 +19,7 @@ public:
     ~DocsumStoreDocument() override;
     std::unique_ptr<document::FieldValue> get_field_value(const vespalib::string& field_name) const override;
     void insert_summary_field(const vespalib::string& field_name, vespalib::slime::Inserter& inserter) const override;
+    void insert_document_id(vespalib::slime::Inserter& inserter) const override;
 };
 
 }
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsumconfig.cpp b/searchsummary/src/vespa/searchsummary/docsummary/docsumconfig.cpp
index 376d4f90204..fea11923858 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/docsumconfig.cpp
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsumconfig.cpp
@@ -4,6 +4,7 @@
 #include "attribute_combiner_dfw.h"
 #include "copy_dfw.h"
 #include "docsumwriter.h"
+#include "document_id_dfw.h"
 #include "empty_dfw.h"
 #include "geoposdfw.h"
 #include "idocsumenvironment.h"
@@ -113,6 +114,9 @@ DynamicDocsumConfig::createFieldWriter(const string & fieldName, const string &
                                                            *attr_ctx, matching_elems_fields);
             rc = static_cast<bool>(fieldWriter);
         }
+    } else if (overrideName == "documentid") {
+        fieldWriter = std::make_unique<DocumentIdDFW>();
+        rc = true;
     } else {
         throw IllegalArgumentException("unknown override operation '" + overrideName + "' for field '" + fieldName + "'.");
     }
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.cpp b/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.cpp
new file mode 100644
index 00000000000..8fcf7c95cdc
--- /dev/null
+++ b/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.cpp
@@ -0,0 +1,31 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include "document_id_dfw.h"
+#include "general_result.h"
+#include "i_docsum_store_document.h"
+
+namespace search::docsummary {
+
+DocumentIdDFW::DocumentIdDFW()
+{
+}
+
+DocumentIdDFW::~DocumentIdDFW() = default;
+
+bool
+DocumentIdDFW::IsGenerated() const
+{
+    return false;
+}
+
+void
+DocumentIdDFW::insertField(uint32_t, GeneralResult *gres, GetDocsumsState *, ResType,
+                           vespalib::slime::Inserter &target)
+{
+    const auto* document = gres->get_document();
+    if (document != nullptr) {
+        document->insert_document_id(target);
+    }
+}
+
+}
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.h b/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.h
new file mode 100644
index 00000000000..f6353912384
--- /dev/null
+++ b/searchsummary/src/vespa/searchsummary/docsummary/document_id_dfw.h
@@ -0,0 +1,22 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include "docsum_field_writer.h"
+
+namespace search::docsummary {
+
+/*
+ * Class for writing document id field.
+ */
+class DocumentIdDFW : public DocsumFieldWriter
+{
+private:
+public:
+    DocumentIdDFW();
+    ~DocumentIdDFW() override;
+    bool IsGenerated() const override;
+    void insertField(uint32_t docid, GeneralResult *gres, GetDocsumsState *state, ResType type, vespalib::slime::Inserter &target) override;
+};
+
+}
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h b/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h
index f81412eb34c..c177568c467 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h
+++ b/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h
@@ -22,6 +22,7 @@ public:
     virtual ~IDocsumStoreDocument() = default;
     virtual std::unique_ptr<document::FieldValue> get_field_value(const vespalib::string& field_name) const = 0;
     virtual void insert_summary_field(const vespalib::string& field_name, vespalib::slime::Inserter& inserter) const = 0;
+    virtual void insert_document_id(vespalib::slime::Inserter& inserter) const = 0;
 };
 
 }
diff --git a/streamingvisitors/src/vespa/vsm/vsm/docsumfilter.cpp b/streamingvisitors/src/vespa/vsm/vsm/docsumfilter.cpp
index 20083b9160e..5bcead79f97 100644
--- a/streamingvisitors/src/vespa/vsm/vsm/docsumfilter.cpp
+++ b/streamingvisitors/src/vespa/vsm/vsm/docsumfilter.cpp
@@ -135,6 +135,7 @@ public:
     ~DocsumStoreVsmDocument() override;
     std::unique_ptr<document::FieldValue> get_field_value(const vespalib::string& field_name) const override;
     void insert_summary_field(const vespalib::string& field_name, vespalib::slime::Inserter& inserter) const override;
+    void insert_document_id(vespalib::slime::Inserter& inserter) const override;
 };
 
 DocsumStoreVsmDocument::DocsumStoreVsmDocument(const document::Document* document)
@@ -168,6 +169,16 @@ DocsumStoreVsmDocument::insert_summary_field(const vespalib::string& field_name,
     }
 }
 
+void
+DocsumStoreVsmDocument::insert_document_id(vespalib::slime::Inserter& inserter) const
+{
+    if (_document) {
+        auto id = _document->getId().toString();
+        vespalib::Memory id_view(id.data(), id.size());
+        inserter.insertString(id_view);
+    }
+}
+
 }
 
 FieldPath