Make WebFrameContentDumper use Element.innerText

This patch changes WebFrameContentDumper to use Element.innerText
instead of its ad-hoc TextDumper, so that its output is more
predictable.

This patch also allows all WebFrameSwapTest to pass in LayoutNG.

Bug: 781434
Change-Id: I961fa43c1821e2a0e7eae6b01a6affe884eacc72
Reviewed-on: https://chromium-review.googlesource.com/c/1371145Reviewed-by: Kent Tamura <tkent@chromium.org>
Commit-Queue: Xiaocheng Hu <xiaochengh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#615560}

third_party/blink/renderer/core/exported/web_frame_content_dumper.cc

@@ -26,138 +26,6 @@ namespace blink {
 
 namespace {
 
-const int text_dumper_max_depth = 512;
-
-bool IsRenderedAndVisible(const Node& node) {
-  if (node.GetLayoutObject() &&
-      node.GetLayoutObject()->Style()->Visibility() == EVisibility::kVisible)
-    return true;
-  if (node.IsElementNode() && ToElement(node).HasDisplayContentsStyle())
-    return true;
-  return false;
-}
-
-size_t RequiredLineBreaksAround(const Node& node) {
-  if (!IsRenderedAndVisible(node))
-    return 0;
-  if (node.IsTextNode())
-    return 0;
-  if (IsHTMLParagraphElement(node))
-    return 2;
-  if (LayoutObject* layout_object = node.GetLayoutObject()) {
-    if (!layout_object->Style()->IsDisplayInlineType())
-      return 1;
-    if (layout_object->Style()->Display() == EDisplay::kTableCaption)
-      return 1;
-  }
-  return 0;
-}
-
-// This class dumps innerText of a node into a StringBuilder, following the spec
-// [*] but with a simplified whitespace handling algorithm when processing text
-// nodes: only leading and trailing collapsed whitespaces are removed; all other
-// whitespace characters are left as-is, without any collapsing or conversion.
-// For example, from HTML <p>\na\n\nb\n</p>, we get text dump "a\n\nb".
-// [*] https://developer.mozilla.org/en-US/docs/Web/API/Node/innerText
-class TextDumper final {
-  STACK_ALLOCATED();
-
- public:
-  TextDumper(StringBuilder& builder, size_t max_length)
-      : builder_(builder), max_length_(max_length) {}
-
-  void DumpTextFrom(const Node& node) {
-    DCHECK(!has_emitted_);
-    DCHECK(!required_line_breaks_);
-    HandleNode(node, 0);
-  }
-
- private:
-  void HandleNode(const Node& node, int depth) {
-    const size_t required_line_breaks_around = RequiredLineBreaksAround(node);
-    AddRequiredLineBreaks(required_line_breaks_around);
-
-    if (depth < text_dumper_max_depth) {
-      for (const Node& child : NodeTraversal::ChildrenOf(node)) {
-        HandleNode(child, depth + 1);
-        if (builder_.length() >= max_length_)
-          return;
-      }
-    }
-
-    if (!IsRenderedAndVisible(node))
-      return;
-
-    if (node.IsTextNode())
-      return HandleTextNode(ToText(node));
-
-    if (IsHTMLBRElement(node))
-      return DumpText("\n");
-
-    if (LayoutObject* layout_object = node.GetLayoutObject()) {
-      if (layout_object->IsTableCell() &&
-          ToLayoutTableCell(layout_object)->NextCell())
-        return DumpText("\t");
-      if (layout_object->IsTableRow() &&
-          ToLayoutTableRow(layout_object)->NextRow())
-        return DumpText("\n");
-    }
-
-    AddRequiredLineBreaks(required_line_breaks_around);
-  }
-
-  void HandleTextNode(const Text& node) {
-    const LayoutText* layout_text = node.GetLayoutObject();
-    if (!layout_text)
-      return;
-    if (layout_text->IsTextFragment() &&
-        ToLayoutTextFragment(layout_text)->IsRemainingTextLayoutObject()) {
-      const LayoutText* first_letter =
-          ToLayoutText(AssociatedLayoutObjectOf(node, 0));
-      if (first_letter && first_letter != layout_text)
-        HandleLayoutText(*first_letter);
-    }
-    HandleLayoutText(*layout_text);
-  }
-
-  void HandleLayoutText(const LayoutText& text) {
-    if (!text.HasNonCollapsedText())
-      return;
-    size_t text_start = text.CaretMinOffset();
-    size_t text_end = text.CaretMaxOffset();
-    String dump = text.GetText().Substring(text_start, text_end - text_start);
-    DumpText(dump);
-  }
-
-  void AddRequiredLineBreaks(size_t required) {
-    required_line_breaks_ = std::max(required, required_line_breaks_);
-  }
-
-  void DumpText(String text) {
-    if (!text.length())
-      return;
-
-    if (has_emitted_ && required_line_breaks_) {
-      for (size_t i = 0; i < required_line_breaks_; ++i)
-        builder_.Append('\n');
-    }
-    required_line_breaks_ = 0;
-    builder_.Append(text);
-    has_emitted_ = true;
-
-    if (builder_.length() > max_length_)
-      builder_.Resize(max_length_);
-  }
-
-  bool has_emitted_ = false;
-  size_t required_line_breaks_ = 0;
-
-  StringBuilder& builder_;
-  const size_t max_length_;
-
-  DISALLOW_COPY_AND_ASSIGN(TextDumper);
-};
-
 void FrameContentAsPlainText(size_t max_chars,
                              LocalFrame* frame,
                              StringBuilder& output) {
@@ -171,8 +39,11 @@ void FrameContentAsPlainText(size_t max_chars,
   DCHECK(!frame->View()->NeedsLayout());
   DCHECK(!document->NeedsLayoutTreeUpdate());
 
-  if (document->documentElement())
-    TextDumper(output, max_chars).DumpTextFrom(*document->documentElement());
+  if (document->documentElement()) {
+    output.Append(document->documentElement()->innerText());
+    if (output.length() >= max_chars)
+      output.Resize(max_chars);
+  }
 
   // The separator between frames when the frames are converted to plain text.
   const LChar kFrameSeparator[] = {'\n', '\n'};

third_party/blink/renderer/core/exported/web_frame_test.cc

@@ -9219,7 +9219,7 @@ TEST_F(WebFrameSwapTest, SwapFirstChild) {
   frame_test_helpers::LoadFrame(local_frame, base_url_ + "subframe-hello.html");
   std::string content =
       WebFrameContentDumper::DumpWebViewAsText(WebView(), 1024).Utf8();
-  EXPECT_EQ("\n\nhello\n\nb\n\n\na\n\nc", content);
+  EXPECT_EQ("  \n\nhello\n\nb \n\na\n\nc", content);
 }
 
 void WebFrameTest::SwapAndVerifyMiddleChildConsistency(
@@ -9255,7 +9255,7 @@ TEST_F(WebFrameSwapTest, SwapMiddleChild) {
   frame_test_helpers::LoadFrame(local_frame, base_url_ + "subframe-hello.html");
   std::string content =
       WebFrameContentDumper::DumpWebViewAsText(WebView(), 1024).Utf8();
-  EXPECT_EQ("\n\na\n\nhello\n\nc", content);
+  EXPECT_EQ("  \n\na\n\nhello\n\nc", content);
 }
 
 void WebFrameTest::SwapAndVerifyLastChildConsistency(const char* const message,
@@ -9285,7 +9285,7 @@ TEST_F(WebFrameSwapTest, SwapLastChild) {
   frame_test_helpers::LoadFrame(local_frame, base_url_ + "subframe-hello.html");
   std::string content =
       WebFrameContentDumper::DumpWebViewAsText(WebView(), 1024).Utf8();
-  EXPECT_EQ("\n\na\n\nb\n\n\na\n\nhello", content);
+  EXPECT_EQ("  \n\na\n\nb \n\na\n\nhello", content);
 }
 
 TEST_F(WebFrameSwapTest, DetachProvisionalFrame) {
@@ -9398,7 +9398,7 @@ TEST_F(WebFrameSwapTest, SwapParentShouldDetachChildren) {
   frame_test_helpers::LoadFrame(local_frame, base_url_ + "subframe-hello.html");
   std::string content =
       WebFrameContentDumper::DumpWebViewAsText(WebView(), 1024).Utf8();
-  EXPECT_EQ("\n\na\n\nhello\n\nc", content);
+  EXPECT_EQ("  \n\na\n\nhello\n\nc", content);
 }
 
 TEST_F(WebFrameSwapTest, SwapPreservesGlobalContext) {