Add IDocsumStoreDocument, an interface for accessing the original document

when generating a document summary.

9 files changed, 100 insertions, 43 deletions

diff --git a/searchsummary/src/tests/docsummary/matched_elements_filter/matched_elements_filter_test.cpp b/searchsummary/src/tests/docsummary/matched_elements_filter/matched_elements_filter_test.cpp
index de6b704189d..675af283ee8 100644
--- a/searchsummary/src/tests/docsummary/matched_elements_filter/matched_elements_filter_test.cpp
+++ b/searchsummary/src/tests/docsummary/matched_elements_filter/matched_elements_filter_test.cpp
@@ -15,6 +15,7 @@
 #include <vespa/searchlib/common/matching_elements.h>
 #include <vespa/searchlib/common/matching_elements_fields.h>
 #include <vespa/searchlib/util/slime_output_raw_buf_adapter.h>
+#include <vespa/searchsummary/docsummary/docsum_store_document.h>
 #include <vespa/searchsummary/docsummary/docsumstate.h>
 #include <vespa/searchsummary/docsummary/idocsumenvironment.h>
 #include <vespa/searchsummary/docsummary/matched_elements_filter_dfw.h>
@@ -132,7 +133,7 @@ public:
         const char* buf;
         uint32_t buf_len;
         assert(_packer.GetDocsumBlob(&buf, &buf_len));
-        return DocsumStoreValue(buf, buf_len, std::move(doc));
+        return DocsumStoreValue(buf, buf_len, std::make_unique<DocsumStoreDocument>(std::move(doc)));
     }
 };
 
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt b/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
index 4a5648d0842..947fe9deeab 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
+++ b/searchsummary/src/vespa/searchsummary/docsummary/CMakeLists.txt
@@ -9,6 +9,7 @@ vespa_add_library(searchsummary_docsummary OBJECT
     docsumconfig.cpp
     docsumfieldwriter.cpp
     docsumstate.cpp
+    docsum_store_document.cpp
     docsumstorevalue.cpp
     docsumwriter.cpp
     dynamicteaserdfw.cpp
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp
new file mode 100644
index 00000000000..c0b894ddc0a
--- /dev/null
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.cpp
@@ -0,0 +1,31 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include "docsum_store_document.h"
+#include <vespa/document/datatype/datatype.h>
+#include <vespa/document/fieldvalue/document.h>
+
+namespace search::docsummary {
+
+DocsumStoreDocument::DocsumStoreDocument(std::unique_ptr<document::Document> document)
+    : _document(std::move(document))
+{
+}
+
+DocsumStoreDocument::~DocsumStoreDocument() = default;
+
+std::unique_ptr<document::FieldValue>
+DocsumStoreDocument::get_field_value(const vespalib::string& field_name) const
+{
+    if (_document) {
+        const document::Field& field = _document->getField(field_name);
+        auto value(field.getDataType().createFieldValue());
+        if (value) {
+            if (_document->getValue(field, *value)) {
+                return value;
+            }
+        }
+    }
+    return {};
+}
+
+}
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h
new file mode 100644
index 00000000000..4508132e7e0
--- /dev/null
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsum_store_document.h
@@ -0,0 +1,23 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include "i_docsum_store_document.h"
+
+namespace document { class Document; }
+
+namespace search::docsummary {
+
+/**
+ * Class providing access to a document retrieved from an IDocsumStore.
+ **/
+class DocsumStoreDocument : public IDocsumStoreDocument
+{
+    std::unique_ptr<document::Document> _document;
+public:
+    DocsumStoreDocument(std::unique_ptr<document::Document> document);
+    ~DocsumStoreDocument() override;
+    std::unique_ptr<document::FieldValue> get_field_value(const vespalib::string& field_name) const override;
+};
+
+}
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.cpp b/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.cpp
index 8fbbeb6db28..a60ae93b58f 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.cpp
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.cpp
@@ -1,35 +1,25 @@
 // Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 
 #include "docsumstorevalue.h"
-#include <vespa/document/fieldvalue/document.h>
+#include "i_docsum_store_document.h"
 
 namespace search::docsummary {
 
 DocsumStoreValue::DocsumStoreValue()
     : _value(static_cast<const char*>(0), 0),
-      _document(),
-      _document_ptr(nullptr)
+      _document()
 {
 }
 
 DocsumStoreValue::DocsumStoreValue(const char *pt_, uint32_t len_)
     : _value(pt_, len_),
-      _document(),
-      _document_ptr(nullptr)
+      _document()
 {
 }
 
-DocsumStoreValue::DocsumStoreValue(const char *pt_, uint32_t len_, std::unique_ptr<document::Document> document_)
+DocsumStoreValue::DocsumStoreValue(const char *pt_, uint32_t len_, std::unique_ptr<IDocsumStoreDocument> document)
     : _value(pt_, len_),
-      _document(std::move(document_)),
-      _document_ptr(_document.get())
-{
-}
-
-DocsumStoreValue::DocsumStoreValue(const char *pt_, uint32_t len_, const document::Document* document_ptr)
-    : _value(pt_, len_),
-      _document(),
-      _document_ptr(document_ptr)
+      _document(std::move(document))
 {
 }
 
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.h b/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.h
index bd6454d9c87..bdc4e7b9e19 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.h
+++ b/searchsummary/src/vespa/searchsummary/docsummary/docsumstorevalue.h
@@ -5,10 +5,10 @@
 #include <memory>
 #include <utility>
 
-namespace document { class Document; }
-
 namespace search::docsummary {
 
+class IDocsumStoreDocument;
+
 /**
  * Simple wrapper class containing the location and size of a docsum
  * blob located in memory. The memory containing the docsum blob is
@@ -19,11 +19,11 @@ class DocsumStoreValue
 {
 private:
     std::pair<const char *, uint32_t> _value;
-    // The document instance that was used to generate the docsum blob.
+    // An interface for accessing the document instance that was used
+    // to generate the docsum blob.
     // Note: This is temporary until the docsummary framework is simplified,
     //       and the docsum blob concept is removed.
-    std::unique_ptr<document::Document> _document;
-    const document::Document*           _document_ptr;
+    std::unique_ptr<IDocsumStoreDocument> _document;
 
 public:
     DocsumStoreValue(const DocsumStoreValue&) = delete;
@@ -47,18 +47,9 @@ public:
      *
      * @param pt_ docsum location
      * @param len_ docsum size
-     * @param document_ document instance used to generate the docsum blob
-     **/
-    DocsumStoreValue(const char *pt_, uint32_t len_, std::unique_ptr<document::Document> document_);
-
-    /**
-     * Construct object encapsulating the given location and size.
-     *
-     * @param pt_ docsum location
-     * @param len_ docsum size
-     * @param document_ptr document instance used to generate the docsum blob. Note: external ownership.
+     * @param document docsum store document instance
      **/
-    DocsumStoreValue(const char *pt_, uint32_t len_, const document::Document* document_ptr);
+    DocsumStoreValue(const char *pt_, uint32_t len_, std::unique_ptr<IDocsumStoreDocument> document);
 
     ~DocsumStoreValue();
 
@@ -87,7 +78,7 @@ public:
      **/
     bool valid() const { return (_value.first != 0) && (_value.second >= sizeof(uint32_t)); }
 
-    const document::Document* get_document() const { return _document_ptr; }
+    const IDocsumStoreDocument* get_document() const { return _document.get(); }
 };
 
 }
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/general_result.cpp b/searchsummary/src/vespa/searchsummary/docsummary/general_result.cpp
index 781cce8f972..12391d26ce1 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/general_result.cpp
+++ b/searchsummary/src/vespa/searchsummary/docsummary/general_result.cpp
@@ -1,6 +1,7 @@
 // Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 
 #include "general_result.h"
+#include "i_docsum_store_document.h"
 #include "resultconfig.h"
 #include <vespa/document/fieldvalue/document.h>
 #include <vespa/document/datatype/datatype.h>
@@ -73,15 +74,9 @@ std::unique_ptr<document::FieldValue>
 GeneralResult::get_field_value(const vespalib::string& field_name) const
 {
     if (_document != nullptr) {
-        const document::Field & field = _document->getField(field_name);
-        auto value(field.getDataType().createFieldValue());
-        if (value) {
-            if (_document->getValue(field, *value)) {
-                return value;
-            }
-        }
+        return _document->get_field_value(field_name);
     }
-    return std::unique_ptr<document::FieldValue>();
+    return {};
 }
 
 bool
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/general_result.h b/searchsummary/src/vespa/searchsummary/docsummary/general_result.h
index a5664a8a606..93114dbb44d 100644
--- a/searchsummary/src/vespa/searchsummary/docsummary/general_result.h
+++ b/searchsummary/src/vespa/searchsummary/docsummary/general_result.h
@@ -6,12 +6,13 @@
 #include "docsumstorevalue.h"
 
 namespace document {
-class Document;
 class FieldValue;
 }
 
 namespace search::docsummary {
 
+class DocsumStoreDocument;
+
 class GeneralResult
 {
 private:
@@ -21,7 +22,7 @@ private:
     const ResultClass      *_resClass;
     uint32_t                _entrycnt;
     ResEntry               *_entries;
-    const document::Document* _document;
+    const IDocsumStoreDocument* _document;
 
     void AllocEntries();
     void FreeEntries();
diff --git a/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h b/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h
new file mode 100644
index 00000000000..3fbf54b18a9
--- /dev/null
+++ b/searchsummary/src/vespa/searchsummary/docsummary/i_docsum_store_document.h
@@ -0,0 +1,24 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#pragma once
+
+#include <vespa/vespalib/stllike/string.h>
+#include <memory>
+
+namespace document { class FieldValue; }
+
+namespace search::docsummary {
+
+/**
+ * Interface class providing access to a document retrieved from an
+ * IDocsumStore.  Some implementations (e.g. DocsumStoreVsmDocument) might
+ * apply transforms when accessing some fields.
+ **/
+class IDocsumStoreDocument
+{
+public:
+    virtual ~IDocsumStoreDocument() = default;
+    virtual std::unique_ptr<document::FieldValue> get_field_value(const vespalib::string& field_name) const = 0;
+};
+
+}