Add PDF linearization to DocumentMetadata

Meanwhile, consolidate calls to FPDFAvail_IsLinearized() to a private
helper.

Bug: 93619
Change-Id: I0b1fd52e9067e2c05e66034caf3d3ceb641411de
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2622515
Commit-Queue: Daniel Hosseinian <dhoss@chromium.org>
Reviewed-by: Hui Yingst <nigi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#842440}

pdf/document_metadata.h

@@ -30,8 +30,8 @@ enum class PdfVersion {
 // dictionary (see section 14.3.3 "Document Information Dictionary" of the ISO
 // 32000-1 standard), as well as other properties about the file.
 // TODO(crbug.com/93619): Finish adding information dictionary fields like
-// |keywords|, |creation_date|, and |mod_date|. Also add fields like
-// |size_bytes|, |is_encrypted|, and |is_linearized|.
+// `keywords`, `creation_date`, and `mod_date`. Also add fields like
+// `size_bytes` and `is_encrypted`.
 struct DocumentMetadata {
   DocumentMetadata();
   DocumentMetadata(const DocumentMetadata&) = delete;
@@ -41,6 +41,9 @@ struct DocumentMetadata {
   // Version of the document
   PdfVersion version = PdfVersion::kUnknown;
 
+  // Whether the document is optimized by linearization.
+  bool linearized = false;
+
   // The document's title.
   std::string title;
 

pdf/pdfium/pdfium_engine.cc

@@ -674,13 +674,13 @@ void PDFiumEngine::OnPendingRequestComplete() {
   if (!fpdf_availability()) {
     document_->file_access().m_FileLen = doc_loader_->GetDocumentSize();
     document_->CreateFPDFAvailability();
-    DCHECK(fpdf_availability());
+
     // Currently engine does not deal efficiently with some non-linearized
     // files.
     // See http://code.google.com/p/chromium/issues/detail?id=59400
     // To improve user experience we download entire file for non-linearized
     // PDF.
-    if (FPDFAvail_IsLinearized(fpdf_availability()) != PDF_LINEARIZED) {
+    if (!IsLinearized()) {
       // Wait complete document.
       process_when_pending_request_complete_ = false;
       document_->ResetFPDFAvailability();
@@ -2752,9 +2752,8 @@ std::vector<gfx::Size> PDFiumEngine::LoadPageSizes(
   pending_pages_.clear();
   size_t new_page_count = FPDF_GetPageCount(doc());
 
-  bool doc_complete = doc_loader_->IsDocumentComplete();
-  bool is_linear =
-      FPDFAvail_IsLinearized(fpdf_availability()) == PDF_LINEARIZED;
+  const bool doc_complete = doc_loader_->IsDocumentComplete();
+  const bool is_linear = IsLinearized();
   for (size_t i = 0; i < new_page_count; ++i) {
     // Get page availability. If |document_loaded_| == true and the page is not
     // new, then the page has been constructed already. Get page availability
@@ -2805,12 +2804,10 @@ std::vector<gfx::Size> PDFiumEngine::LoadPageSizes(
 
 void PDFiumEngine::LoadBody() {
   DCHECK(doc());
-  DCHECK(fpdf_availability());
   if (doc_loader_->IsDocumentComplete()) {
     LoadForm();
-  } else if (FPDFAvail_IsLinearized(fpdf_availability()) == PDF_LINEARIZED &&
-             FPDF_GetPageCount(doc()) == 1) {
-    // If we have only one page we should load form first, bacause it is may be
+  } else if (IsLinearized() && FPDF_GetPageCount(doc()) == 1) {
+    // If we have only one page we should load form first, because it may be an
     // XFA document. And after loading form the page count and its contents may
     // be changed.
     LoadForm();
@@ -2865,6 +2862,11 @@ void PDFiumEngine::LoadForm() {
   }
 }
 
+bool PDFiumEngine::IsLinearized() {
+  DCHECK(fpdf_availability());
+  return FPDFAvail_IsLinearized(fpdf_availability()) == PDF_LINEARIZED;
+}
+
 void PDFiumEngine::CalculateVisiblePages() {
   // Early return if the PDF isn't being loaded or if we don't have the document
   // info yet. The latter is important because otherwise as the PDF is being
@@ -3948,8 +3950,10 @@ void PDFiumEngine::LoadDocumentAttachmentInfoList() {
 void PDFiumEngine::LoadDocumentMetadata() {
   DCHECK(document_loaded_);
 
-  // Document information dictionary entries
   doc_metadata_.version = GetDocumentVersion();
+  doc_metadata_.linearized = IsLinearized();
+
+  // Document information dictionary entries
   doc_metadata_.title = GetTrimmedMetadataByField("Title");
   doc_metadata_.author = GetTrimmedMetadataByField("Author");
   doc_metadata_.subject = GetTrimmedMetadataByField("Subject");

pdf/pdfium/pdfium_engine.h

@@ -302,6 +302,9 @@ class PDFiumEngine : public PDFEngine,
 
   void LoadForm();
 
+  // Checks whether the document is optimized by linearization.
+  bool IsLinearized();
+
   // Calculates which pages should be displayed right now.
   void CalculateVisiblePages();
 

pdf/pdfium/pdfium_engine_unittest.cc

@@ -349,6 +349,7 @@ TEST_F(PDFiumEngineTest, GetDocumentMetadata) {
   const DocumentMetadata& doc_metadata = engine->GetDocumentMetadata();
 
   EXPECT_EQ(PdfVersion::k1_7, doc_metadata.version);
+  EXPECT_FALSE(doc_metadata.linearized);
   EXPECT_EQ("Sample PDF Document Info", doc_metadata.title);
   EXPECT_EQ("Chromium Authors", doc_metadata.author);
   EXPECT_EQ("Testing", doc_metadata.subject);
@@ -365,6 +366,7 @@ TEST_F(PDFiumEngineTest, GetEmptyDocumentMetadata) {
   const DocumentMetadata& doc_metadata = engine->GetDocumentMetadata();
 
   EXPECT_EQ(PdfVersion::k1_7, doc_metadata.version);
+  EXPECT_FALSE(doc_metadata.linearized);
   EXPECT_THAT(doc_metadata.title, IsEmpty());
   EXPECT_THAT(doc_metadata.author, IsEmpty());
   EXPECT_THAT(doc_metadata.subject, IsEmpty());
@@ -372,6 +374,14 @@ TEST_F(PDFiumEngineTest, GetEmptyDocumentMetadata) {
   EXPECT_THAT(doc_metadata.producer, IsEmpty());
 }
 
+TEST_F(PDFiumEngineTest, GetLinearizedDocumentMetadata) {
+  NiceMock<MockTestClient> client;
+  std::unique_ptr<PDFiumEngine> engine =
+      InitializeEngine(&client, FILE_PATH_LITERAL("linearized.pdf"));
+  ASSERT_TRUE(engine);
+  EXPECT_TRUE(engine->GetDocumentMetadata().linearized);
+}
+
 TEST_F(PDFiumEngineTest, GetBadPdfVersion) {
   NiceMock<MockTestClient> client;
   std::unique_ptr<PDFiumEngine> engine =