Add text after overlapping annotations in accessibility tree

This CL fixes the issue where text runs after overlapping annotations
are not added to the accessibility tree. In the loop where text runs are
added to the accessibility tree, some text runs are skipped due to
incorrect normalization after overlapping annotations are added to the
tree.

Tests have been added to validate the scenario.

Fixed: 1066802
Change-Id: I6e7b3c72d9c25c45033a057d44261e888e9e7796
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2131808Reviewed-by: Kevin Babbitt <kbabbitt@microsoft.com>
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Ankit Kumar 🌪️ <ankk@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#755928}

chrome/browser/pdf/pdf_extension_test.cc

@@ -2564,6 +2564,12 @@ IN_PROC_BROWSER_TEST_F(PDFExtensionAccessibilityTextExtractionTest,
   RunTextExtractionTest(FILE_PATH_LITERAL("text-image-link.pdf"));
 }
 
+// Test data of inline text boxes for PDF with overlapping annotations.
+IN_PROC_BROWSER_TEST_F(PDFExtensionAccessibilityTextExtractionTest,
+                       OverlappingAnnots) {
+  RunTextExtractionTest(FILE_PATH_LITERAL("overlapping-annots.pdf"));
+}
+
 class PDFExtensionAccessibilityTreeDumpTest
     : public PDFExtensionTest,
       public ::testing::WithParamInterface<size_t> {

chrome/test/data/pdf/accessibility/overlapping-annots-expected.txt

+Overlapping link text,
+¶
+link text,
+¶
+ some text after link texts.
\ No newline at end of file

chrome/test/data/pdf/accessibility/overlapping-annots.in

+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+>>
+endobj
+{{object 2 0}} <<
+  /Type /Pages
+  /MediaBox [0 0 612 792]
+  /CropBox [0 0 612 792]
+  /Count 1
+  /Kids [3 0 R]
+  /Resources <<
+    /Font <<
+      /F1 4 0 R
+    >>
+  >>
+>>
+endobj
+{{object 3 0}} <<
+  /Type /Page
+  /Parent 2 0 R
+  /Contents 5 0 R
+  /Annots [6 0 R 7 0 R]
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /BaseFont /Times-Roman
+>>
+endobj
+{{object 5 0}} <<
+  {{streamlen}}
+>>
+stream
+BT
+70 700 Td
+/F1 12 Tf
+(Overlapping link text, some text after link texts.) Tj
+ET
+endstream
+endobj
+{{object 6 0}} <<
+  /Type /Annot
+  /Subtype /Link
+  /Rect [70 698 175 710]
+  /A <<
+    /Type /Action
+    /S /URI
+    /URI (https://www.bing.com)
+  >>
+  /F 4
+>>
+endobj
+{{object 7 0}} <<
+  /Type /Annot
+  /Subtype /Link
+  /Rect [133 698 175 710]
+  /A <<
+    /Type /Action
+    /S /URI
+    /URI (https://www.adobe.com)
+  >>
+  /F 4
+>>
+endobj
+{{xref}}
+{{trailer}}
+{{startxref}}
+%%EOF
\ No newline at end of file

components/pdf/renderer/pdf_accessibility_tree.cc

@@ -282,6 +282,13 @@ bool IsObjectInTextRun(const std::vector<T>& objects,
           objects[object_index].text_run_index <= text_run_index);
 }
 
+size_t NormalizeTextRunIndex(uint32_t object_end_text_run_index,
+                             size_t current_text_run_index) {
+  return std::max<size_t>(
+      object_end_text_run_index,
+      current_text_run_index ? current_text_run_index - 1 : 0);
+}
+
 bool IsTextRenderModeFill(const PP_TextRenderingMode& mode) {
   switch (mode) {
     case PP_TEXTRENDERINGMODE_FILL:
@@ -567,7 +574,7 @@ void PdfAccessibilityTree::AddPageContent(
           link_node->relative_bounds.bounds);
 
       text_run_index =
-          std::max<size_t>(link_end_text_run_index, text_run_index);
+          NormalizeTextRunIndex(link_end_text_run_index, text_run_index);
     } else if (IsObjectInTextRun(images, current_image_index, text_run_index)) {
       FinishStaticNode(&static_text_node, &static_text);
       // If the |text_run_index| is less than or equal to the image's text run
@@ -604,7 +611,7 @@ void PdfAccessibilityTree::AddPageContent(
           highlight_node->relative_bounds.bounds);
 
       text_run_index =
-          std::max<size_t>(highlight_end_text_run_index, text_run_index);
+          NormalizeTextRunIndex(highlight_end_text_run_index, text_run_index);
     } else if (IsObjectInTextRun(text_fields, current_text_field_index,
                                  text_run_index) &&
                base::FeatureList::IsEnabled(

components/pdf/renderer/pdf_accessibility_tree_browsertest.cc

@@ -348,6 +348,99 @@ TEST_F(PdfAccessibilityTreeTest, TestPdfAccessibilityTreeCreation) {
             image_node->GetStringAttribute(ax::mojom::StringAttribute::kName));
 }
 
+TEST_F(PdfAccessibilityTreeTest, TestOverlappingAnnots) {
+  text_runs_.emplace_back(kFirstRunMultiLine);
+  text_runs_.emplace_back(kSecondRunMultiLine);
+  text_runs_.emplace_back(kThirdRunMultiLine);
+  text_runs_.emplace_back(kFourthRunMultiLine);
+  chars_.insert(chars_.end(), std::begin(kDummyCharsData),
+                std::end(kDummyCharsData));
+
+  {
+    ppapi::PdfAccessibilityLinkInfo link;
+    link.bounds = PP_MakeFloatRectFromXYWH(1.0f, 1.0f, 5.0f, 6.0f);
+    link.url = kChromiumTestUrl;
+    link.text_run_index = 0;
+    link.text_run_count = 3;
+    page_objects_.links.push_back(std::move(link));
+  }
+
+  {
+    ppapi::PdfAccessibilityLinkInfo link;
+    link.bounds = PP_MakeFloatRectFromXYWH(1.0f, 2.0f, 5.0f, 6.0f);
+    link.url = kChromiumTestUrl;
+    link.text_run_index = 1;
+    link.text_run_count = 2;
+    page_objects_.links.push_back(std::move(link));
+  }
+
+  page_info_.text_run_count = text_runs_.size();
+  page_info_.char_count = chars_.size();
+
+  content::RenderFrame* render_frame = view_->GetMainRenderFrame();
+  ASSERT_TRUE(render_frame);
+  render_frame->SetAccessibilityModeForTest(ui::AXMode::kWebContents);
+  ASSERT_TRUE(render_frame->GetRenderAccessibility());
+
+  FakeRendererPpapiHost host(view_->GetMainRenderFrame());
+  PP_Instance instance = 0;
+  pdf::PdfAccessibilityTree pdf_accessibility_tree(&host, instance);
+
+  pdf_accessibility_tree.SetAccessibilityViewportInfo(viewport_info_);
+  pdf_accessibility_tree.SetAccessibilityDocInfo(doc_info_);
+  pdf_accessibility_tree.SetAccessibilityPageInfo(page_info_, text_runs_,
+                                                  chars_, page_objects_);
+
+  /*
+   * Expected tree structure
+   * Document
+   * ++ Region
+   * ++++ Paragraph
+   * ++++++ Link
+   * ++++++ Link
+   * ++++++ Static Text
+   */
+
+  ui::AXNode* root_node = pdf_accessibility_tree.GetRoot();
+  ASSERT_TRUE(root_node);
+  EXPECT_EQ(ax::mojom::Role::kDocument, root_node->data().role);
+  ASSERT_EQ(1u, root_node->children().size());
+
+  ui::AXNode* page_node = root_node->children()[0];
+  ASSERT_TRUE(page_node);
+  EXPECT_EQ(ax::mojom::Role::kRegion, page_node->data().role);
+  ASSERT_EQ(1u, page_node->children().size());
+
+  ui::AXNode* paragraph_node = page_node->children()[0];
+  ASSERT_TRUE(paragraph_node);
+  EXPECT_EQ(ax::mojom::Role::kParagraph, paragraph_node->data().role);
+  const std::vector<ui::AXNode*>& child_nodes = paragraph_node->children();
+  ASSERT_EQ(3u, child_nodes.size());
+
+  ui::AXNode* link_node = child_nodes[0];
+  ASSERT_TRUE(link_node);
+  EXPECT_EQ(kChromiumTestUrl,
+            link_node->GetStringAttribute(ax::mojom::StringAttribute::kUrl));
+  EXPECT_EQ(ax::mojom::Role::kLink, link_node->data().role);
+  EXPECT_EQ(gfx::RectF(1.0f, 1.0f, 5.0f, 6.0f),
+            link_node->data().relative_bounds.bounds);
+  ASSERT_EQ(1u, link_node->children().size());
+
+  link_node = child_nodes[1];
+  ASSERT_TRUE(link_node);
+  EXPECT_EQ(kChromiumTestUrl,
+            link_node->GetStringAttribute(ax::mojom::StringAttribute::kUrl));
+  EXPECT_EQ(ax::mojom::Role::kLink, link_node->data().role);
+  EXPECT_EQ(gfx::RectF(1.0f, 2.0f, 5.0f, 6.0f),
+            link_node->data().relative_bounds.bounds);
+  ASSERT_EQ(1u, link_node->children().size());
+
+  ui::AXNode* static_text_node = child_nodes[2];
+  ASSERT_TRUE(static_text_node);
+  EXPECT_EQ(ax::mojom::Role::kStaticText, static_text_node->data().role);
+  ASSERT_EQ(1u, static_text_node->children().size());
+}
+
 TEST_F(PdfAccessibilityTreeTest, TestHighlightCreation) {
   // Enable feature flag
   base::test::ScopedFeatureList scoped_feature_list;