Introduce a wrapper class to handle text node in TextIterator

third_party/WebKit/Source/core/editing/iterators/TextIterator.h

Index: third_party/WebKit/Source/core/editing/iterators/TextIterator.h
diff --git a/third_party/WebKit/Source/core/editing/iterators/TextIterator.h b/third_party/WebKit/Source/core/editing/iterators/TextIterator.h
index a95717d39c9491697f1566f9e924614a32408f9f..2e04c44b3616722074b9974323812fd9c9a7617f 100644
--- a/third_party/WebKit/Source/core/editing/iterators/TextIterator.h
+++ b/third_party/WebKit/Source/core/editing/iterators/TextIterator.h
@@ -49,6 +49,77 @@ PlainText(const EphemeralRange&,
 String PlainText(const EphemeralRangeInFlatTree&,
                  const TextIteratorBehavior& = TextIteratorBehavior());
 
+// TODO(xiaochengh): Move the class to dedicated files.
+class CORE_EXPORT TextNodeContentExtractor {

[yosin_UTC9, 2017/05/25 06:44:09]

- Please add class comment.
- s/CORE_EXPORT// since we use TextNodeContentExtract only in TI, at this time.

[Xiaocheng, 2017/05/25 23:47:58]

Done.

+  STACK_ALLOCATED();
+
+ public:
+  TextNodeContentExtractor(const TextIteratorBehavior&, TextIteratorTextState&);

[yosin_UTC9, 2017/05/25 06:44:09]

nit: s/TextIteratorState&/TextIterator*/

[Xiaocheng, 2017/05/25 23:47:58]

Is there a style guide? I've seen a lot of classes passing non-const references
to their constructors:

https://cs.chromium.org/search/?q=LocalFrame%26&type=cs

+
+  void SetBoundaries(Node* start_container,
+                     int start_offset,
+                     Node* end_container,
+                     int end_offset);
+
+  // Returns true if more text is emitted without traversing to the next node.
+  bool HandleRemainingTextRuns();
+
+  // Returns true if a leading white space is emitted before a replaced element.
+  bool FixLeadingWhiteSpaceForReplacedElement(Node*);
+
+  bool HandleTextNode(Text*);

[yosin_UTC9, 2017/05/25 06:44:09]

Please add a comment for |bool| return value.

[Xiaocheng, 2017/05/25 23:47:58]

Done.

+
+ private:
+  void HandlePreFormattedTextNode();
+  void HandleTextBox();
+  void HandleTextNodeFirstLetter(LayoutTextFragment*);
+  bool ShouldHandleFirstLetter(const LayoutText&) const;
+  bool ShouldProceedToRemainingText() const;
+  void ProceedToRemainingText();
+  size_t RestoreCollapsedTrailingSpace(InlineTextBox* next_text_box,
+                                       size_t subrun_end);
+
+  // Used when the visibility of the style should not affect text gathering.
+  bool IgnoresStyleVisibility() const {
+    return behavior_.IgnoresStyleVisibility();
+  }
+
+  // The range.
+  Member<Node> start_container_;
+  int start_offset_;
+  Member<Node> end_container_;
+  int end_offset_;
+
+  // The current text node and offset, from which text is being emitted.
+  Member<Text> text_node_;
+  int offset_;
+
+  InlineTextBox* text_box_;
+
+  // Remember if we are in the middle of handling a pre-formatted text node.
+  bool needs_handle_pre_formatted_text_node_;
+  // Used when deciding text fragment created by :first-letter should be looked
+  // into.
+  bool handled_first_letter_;
+  // Used when iteration over :first-letter text to save pointer to
+  // remaining text box.
+  InlineTextBox* remaining_text_box_;
+  // Used to point to LayoutText object for :first-letter.
+  LayoutText* first_letter_text_;
+
+  // Used to do the whitespace collapsing logic.
+  bool last_text_node_ended_with_collapsed_space_;
+
+  // Used when text boxes are out of order (Hebrew/Arabic w/ embeded LTR text)
+  Vector<InlineTextBox*> sorted_text_boxes_;
+  size_t sorted_text_boxes_position_;
+
+  const TextIteratorBehavior behavior_;
+
+  // Contains state of emitted text.
+  TextIteratorTextState& text_state_;
+};
+
 // Iterates through the DOM range, returning all the text, and 0-length
 // boundaries at points where replaced elements break up the text flow.  The
 // text comes back in chunks so as to optimize for performance of the iteration.
@@ -141,22 +212,8 @@ class CORE_TEMPLATE_CLASS_EXPORT TextIteratorAlgorithm {
   // not always clearly control the iteration progress. Should consider removing
   // the return values and control the iteration in a cleaner way.
   bool HandleTextNode();
-  void HandlePreFormattedTextNode();
   bool HandleReplacedElement();
   bool HandleNonTextNode();
-
-  void HandleTextBox();
-  void HandleTextNodeFirstLetter(LayoutTextFragment*);
-  bool ShouldHandleFirstLetter(const LayoutText&) const;
-  bool ShouldProceedToRemainingText() const;
-  void ProceedToRemainingText();
-
-  // Returns true if more text is emitted without traversing to the next node.
-  bool HandleRemainingTextRuns();
-
-  // Returns true if a leading white space is emitted before a replaced element.
-  bool FixLeadingWhiteSpaceForReplacedElement(Node* parent);
-
   void SpliceBuffer(UChar,
                     Node* text_node,
                     Node* offset_base_node,
@@ -166,8 +223,6 @@ class CORE_TEMPLATE_CLASS_EXPORT TextIteratorAlgorithm {
                 LayoutText* layout_object,
                 int text_start_offset,
                 int text_end_offset);
-  size_t RestoreCollapsedTrailingSpace(InlineTextBox* next_text_box,
-                                       size_t subrun_end);
 
   // Used by selection preservation code. There should be one character emitted
   // between every VisiblePosition in the Range used to create the TextIterator.
@@ -222,7 +277,6 @@ class CORE_TEMPLATE_CLASS_EXPORT TextIteratorAlgorithm {
   // Current position, not necessarily of the text being returned, but position
   // as we walk through the DOM tree.
   Member<Node> node_;
-  int offset_;
   IterationProgress iteration_progress_;
   FullyClippedStateStackAlgorithm<Strategy> fully_clipped_stack_;
   int shadow_depth_;
@@ -237,34 +291,14 @@ class CORE_TEMPLATE_CLASS_EXPORT TextIteratorAlgorithm {
   Member<Node> end_node_;
   Member<Node> past_end_node_;
 
-  // The current text node, from which text is being emitted.
-  Member<Text> text_node_;
-
   // Used when there is still some pending text from the current node; when
   // these are false and 0, we go back to normal iterating.
   bool needs_another_newline_;
-  InlineTextBox* text_box_;
-  // Used when iteration over :first-letter text to save pointer to
-  // remaining text box.
-  InlineTextBox* remaining_text_box_;
-  // Used to point to LayoutText object for :first-letter.
-  LayoutText* first_letter_text_;
 
-  // Used to do the whitespace collapsing logic.
   Member<Text> last_text_node_;
-  bool last_text_node_ended_with_collapsed_space_;
-
-  // Used when text boxes are out of order (Hebrew/Arabic w/ embeded LTR text)
-  Vector<InlineTextBox*> sorted_text_boxes_;
-  size_t sorted_text_boxes_position_;
 
   const TextIteratorBehavior behavior_;
 
-  // Remember if we are in the middle of handling a pre-formatted text node.
-  bool needs_handle_pre_formatted_text_node_;
-  // Used when deciding text fragment created by :first-letter should be looked
-  // into.
-  bool handled_first_letter_;
   // Used when stopsOnFormControls() is true to determine if the iterator should
   // keep advancing.
   bool should_stop_;
@@ -274,6 +308,9 @@ class CORE_TEMPLATE_CLASS_EXPORT TextIteratorAlgorithm {
 
   // Contains state of emitted text.
   TextIteratorTextState text_state_;
+
+  // Helper for extracting text content from text nodes.
+  TextNodeContentExtractor text_extractor_;
 };
 
 extern template class CORE_EXTERN_TEMPLATE_EXPORT