Add word boundaries to the PDF accessibility tree.

Bug: 812304
Cq-Include-Trybots: master.tryserver.chromium.linux:linux_mojo
Change-Id: Ib729933d425af0630201e11092e406089b7084ff
Reviewed-on: https://chromium-review.googlesource.com/919543
Commit-Queue: Dominic Mazzoni <dmazzoni@chromium.org>
Reviewed-by: dsinclair <dsinclair@chromium.org>
Reviewed-by: Katie Dektar <katie@chromium.org>
Cr-Commit-Position: refs/heads/master@{#536903}

chrome/browser/pdf/pdf_extension_test.cc

@@ -766,6 +766,39 @@ IN_PROC_BROWSER_TEST_F(PDFExtensionTest, PdfAccessibilityInOOPIF) {
   ASSERT_MULTILINE_STR_MATCHES(kExpectedPDFAXTreePattern, ax_tree_dump);
 }
 
+IN_PROC_BROWSER_TEST_F(PDFExtensionTest, PdfAccessibilityWordBoundaries) {
+  content::BrowserAccessibilityState::GetInstance()->EnableAccessibility();
+
+  GURL test_pdf_url(embedded_test_server()->GetURL("/pdf/test-bookmarks.pdf"));
+  WebContents* guest_contents = LoadPdfGetGuestContents(test_pdf_url);
+  ASSERT_TRUE(guest_contents);
+
+  WaitForAccessibilityTreeToContainNodeWithName(guest_contents,
+                                                "1 First Section\r\n");
+  ui::AXTreeUpdate ax_tree = GetAccessibilityTreeSnapshot(guest_contents);
+
+  bool found = false;
+  for (auto& node : ax_tree.nodes) {
+    std::string name =
+        node.GetStringAttribute(ax::mojom::StringAttribute::kName);
+    if (node.role == ax::mojom::Role::kInlineTextBox &&
+        name == "First Section\r\n") {
+      found = true;
+      std::vector<int32_t> word_starts =
+          node.GetIntListAttribute(ax::mojom::IntListAttribute::kWordStarts);
+      std::vector<int32_t> word_ends =
+          node.GetIntListAttribute(ax::mojom::IntListAttribute::kWordEnds);
+      ASSERT_EQ(2U, word_starts.size());
+      ASSERT_EQ(2U, word_ends.size());
+      EXPECT_EQ(0, word_starts[0]);
+      EXPECT_EQ(5, word_ends[0]);
+      EXPECT_EQ(6, word_starts[1]);
+      EXPECT_EQ(13, word_ends[1]);
+    }
+  }
+  ASSERT_TRUE(found);
+}
+
 #if defined(GOOGLE_CHROME_BUILD)
 // Test a particular PDF encountered in the wild that triggered a crash
 // when accessibility is enabled.  (http://crbug.com/668724)

components/pdf/renderer/pdf_accessibility_tree.cc

@@ -5,6 +5,7 @@
 #include <algorithm>
 
 #include "base/debug/crash_logging.h"
+#include "base/i18n/break_iterator.h"
 #include "base/memory/ptr_util.h"
 #include "base/strings/stringprintf.h"
 #include "base/strings/utf_string_conversion_utils.h"
@@ -162,6 +163,7 @@ void PdfAccessibilityTree::SetAccessibilityPageInfo(
     inline_text_box_node->location = text_run_bounds;
     inline_text_box_node->AddIntListAttribute(
         ax::mojom::IntListAttribute::kCharacterOffsets, char_offsets);
+    AddWordStartsAndEnds(inline_text_box_node);
 
     para_node->location.Union(inline_text_box_node->location);
     static_text_node->location.Union(inline_text_box_node->location);
@@ -338,6 +340,28 @@ gfx::Transform* PdfAccessibilityTree::MakeTransformFromViewInfo() {
   return transform;
 }
 
+void PdfAccessibilityTree::AddWordStartsAndEnds(
+    ui::AXNodeData* inline_text_box) {
+  base::string16 text =
+      inline_text_box->GetString16Attribute(ax::mojom::StringAttribute::kName);
+  base::i18n::BreakIterator iter(text, base::i18n::BreakIterator::BREAK_WORD);
+  if (!iter.Init())
+    return;
+
+  std::vector<int32_t> word_starts;
+  std::vector<int32_t> word_ends;
+  while (iter.Advance()) {
+    if (iter.IsWord()) {
+      word_starts.push_back(iter.prev());
+      word_ends.push_back(iter.pos());
+    }
+  }
+  inline_text_box->AddIntListAttribute(ax::mojom::IntListAttribute::kWordStarts,
+                                       word_starts);
+  inline_text_box->AddIntListAttribute(ax::mojom::IntListAttribute::kWordEnds,
+                                       word_ends);
+}
+
 //
 // AXTreeSource implementation.
 //

components/pdf/renderer/pdf_accessibility_tree.h

@@ -80,6 +80,7 @@ class PdfAccessibilityTree
   float GetDeviceScaleFactor() const;
   content::RenderAccessibility* GetRenderAccessibility();
   gfx::Transform* MakeTransformFromViewInfo();
+  void AddWordStartsAndEnds(ui::AXNodeData* inline_text_box);
 
   ui::AXTree tree_;
   content::RendererPpapiHost* host_;

testing/buildbot/filters/mojo.fyi.network_browser_tests.filter

@@ -425,6 +425,7 @@
 -PDFExtensionTest.PdfAccessibilityEnableLater
 -PDFExtensionTest.PdfAccessibilityInIframe
 -PDFExtensionTest.PdfAccessibilityInOOPIF
+-PDFExtensionTest.PdfAccessibilityWordBoundaries
 -PDFExtensionTest.PdfZoomWithoutBubble
 -PDFExtensionTest.PostMessageForZeroSizedEmbed
 -PDFExtensionTest.RedirectsFailInPlugin