Autofill: Improve the order of heuristics to apply when inferring labels

components/autofill/content/renderer/form_autofill_util.cc

 // Copyright 2013 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "components/autofill/content/renderer/form_autofill_util.h"
 
 #include <map>
 #include <set>
 
 #include "base/command_line.h"
@@ skipping 111 matching lines @@
 
     WebElement cur_element = parent_node.to<WebElement>();
     if (cur_element.hasHTMLTagName("form") ||
         cur_element.hasHTMLTagName("fieldset")) {
       return true;
     }
   }
   return false;
 }
 
+// Returns true if |node| is an element and it is a container type that
+// InferLabelForElement() can traverse.
+bool IsTraversableContainerElement(const WebNode& node) {
+  if (!node.isElementNode())
+    return false;
+
+  std::string tag_name = node.toConst<WebElement>().tagName().utf8();
+  return (tag_name == "DD" ||

[Evan Stade, 2015/03/18 01:51:29]

nit: parens are unnecessary

[Lei Zhang, 2015/03/18 02:02:18]

but the parens tells my editor to line up the entries when auto-indenting

+          tag_name == "DIV" ||
+          tag_name == "FIELDSET" ||
+          tag_name == "LI" ||
+          tag_name == "TD" ||
+          tag_name == "TABLE");
+}
+
 // Check whether the given field satisfies the REQUIRE_AUTOCOMPLETE requirement.
 bool SatisfiesRequireAutocomplete(const WebInputElement& input_element) {
   return input_element.autoComplete();
 }
 
 // Returns the colspan for a <td> / <th>. Defaults to 1.
 size_t CalculateTableCellColumnSpan(const WebElement& element) {
   DCHECK(element.hasHTMLTagName("td") || element.hasHTMLTagName("th"));
 
   size_t span = 1;
@@ skipping 397 matching lines @@
 // Because this is already traversing the <div> structure, if it finds a <label>
 // sibling along the way, infer from that <label>.
 base::string16 InferLabelFromDivTable(const WebFormControlElement& element) {
   WebNode node = element.parentNode();
   bool looking_for_parent = true;
   std::set<WebNode> divs_to_skip;
 
   // Search the sibling and parent <div>s until we find a candidate label.
   base::string16 inferred_label;
   CR_DEFINE_STATIC_LOCAL(WebString, kDiv, ("div"));
-  CR_DEFINE_STATIC_LOCAL(WebString, kTable, ("table"));
-  CR_DEFINE_STATIC_LOCAL(WebString, kFieldSet, ("fieldset"));
   CR_DEFINE_STATIC_LOCAL(WebString, kLabel, ("label"));
   while (inferred_label.empty() && !node.isNull()) {
     if (HasTagName(node, kDiv)) {
       if (looking_for_parent)
         inferred_label = FindChildTextWithIgnoreList(node, divs_to_skip);
       else
         inferred_label = FindChildText(node);
 
       // Avoid sibling DIVs that contain autofillable fields.
       if (!looking_for_parent && !inferred_label.empty()) {
         CR_DEFINE_STATIC_LOCAL(WebString, kSelector,
                                ("input, select, textarea"));
         blink::WebExceptionCode ec = 0;
         WebElement result_element = node.querySelector(kSelector, ec);
         if (!result_element.isNull()) {
           inferred_label.clear();
           divs_to_skip.insert(node);
         }
       }
 
       looking_for_parent = false;
     } else if (!looking_for_parent && HasTagName(node, kLabel)) {
       WebLabelElement label_element = node.to<WebLabelElement>();
       if (label_element.correspondingControl().isNull())
         inferred_label = FindChildText(node);
-    } else if (looking_for_parent &&
-               (HasTagName(node, kTable) || HasTagName(node, kFieldSet))) {
-      // If the element is in a table or fieldset, its label most likely is too.
+    } else if (looking_for_parent && IsTraversableContainerElement(node)) {
+      // If the element is in a non-div container, its label most likely is too.
       break;
     }
 
     if (node.previousSibling().isNull()) {
       // If there are no more siblings, continue walking up the tree.
       looking_for_parent = true;
     }
 
     node = looking_for_parent ? node.parentNode() : node.previousSibling();
   }
@@ skipping 21 matching lines @@
   while (!previous.isNull() && previous.isTextNode())
     previous = previous.previousSibling();
 
   CR_DEFINE_STATIC_LOCAL(WebString, kDefinitionTag, ("dt"));
   if (previous.isNull() || !HasTagName(previous, kDefinitionTag))
     return base::string16();
 
   return FindChildText(previous);
 }
 
-// Returns true if the closest ancestor is a <div> and not a <td>.
-// Returns false if the closest ancestor is a <td> tag,
-// or if there is no <div> or <td> ancestor.
-bool ClosestAncestorIsDivAndNotTD(const WebFormControlElement& element) {
+// Returns the element type for all ancestor nodes in CAPS, starting with the
+// parent node.
+std::vector<std::string> AncestorTagNames(
+    const WebFormControlElement& element) {
+  std::vector<std::string> tag_names;
   for (WebNode parent_node = element.parentNode();
        !parent_node.isNull();
        parent_node = parent_node.parentNode()) {
     if (!parent_node.isElementNode())
       continue;
 
-    WebElement cur_element = parent_node.to<WebElement>();
-    if (cur_element.hasHTMLTagName("div"))
-      return true;
-    if (cur_element.hasHTMLTagName("td"))
-      return false;
+    tag_names.push_back(parent_node.to<WebElement>().tagName().utf8());
   }
-  return false;
+  return tag_names;
 }
 
 // Infers corresponding label for |element| from surrounding context in the DOM,
 // e.g. the contents of the preceding <p> tag or text element.
 base::string16 InferLabelForElement(const WebFormControlElement& element) {
   base::string16 inferred_label;
   if (IsCheckableElement(toWebInputElement(&element))) {
     inferred_label = InferLabelFromNext(element);
     if (!inferred_label.empty())
       return inferred_label;
   }
 
   inferred_label = InferLabelFromPrevious(element);
   if (!inferred_label.empty())
     return inferred_label;
 
   // If we didn't find a label, check for placeholder text.
   inferred_label = InferLabelFromPlaceholder(element);
   if (!inferred_label.empty())
     return inferred_label;
 
-  // If we didn't find a label, check for list item case.
-  inferred_label = InferLabelFromListItem(element);
-  if (!inferred_label.empty())
-    return inferred_label;
+  // For all other searches that involve traversing up the tree, the search
+  // order is based on which tag is the closest ancestor to |element|.
+  std::vector<std::string> tag_names = AncestorTagNames(element);
+  std::set<std::string> seen_tag_names;
+  for (const std::string& tag_name : tag_names) {
+    if (ContainsKey(seen_tag_names, tag_name))
+      continue;
 
-  // If we didn't find a label, check for definition list case.
-  inferred_label = InferLabelFromDefinitionList(element);
-  if (!inferred_label.empty())
-    return inferred_label;
+    seen_tag_names.insert(tag_name);
+    if (tag_name == "DIV") {
+      inferred_label = InferLabelFromDivTable(element);
+    } else if (tag_name == "TD") {
+      inferred_label = InferLabelFromTableColumn(element);
+      if (inferred_label.empty())
+        inferred_label = InferLabelFromTableRow(element);
+    } else if (tag_name == "DD") {
+      inferred_label = InferLabelFromDefinitionList(element);
+    } else if (tag_name == "LI") {
+      inferred_label = InferLabelFromListItem(element);
+    } else if (tag_name == "FIELDSET") {
+      break;
+    }
 
-  bool check_div_first = ClosestAncestorIsDivAndNotTD(element);
-  if (check_div_first) {
-    // If we didn't find a label, check for div table case first since it's the
-    // closest ancestor.
-    inferred_label = InferLabelFromDivTable(element);
     if (!inferred_label.empty())
-      return inferred_label;
+      break;
   }
 
-  // If we didn't find a label, check for table cell case.
-  inferred_label = InferLabelFromTableColumn(element);
-  if (!inferred_label.empty())
-    return inferred_label;
-
-  // If we didn't find a label, check for table row case.
-  inferred_label = InferLabelFromTableRow(element);
-  if (!inferred_label.empty())
-    return inferred_label;
-
-  if (!check_div_first) {
-    // If we didn't find a label from the table, check for div table case if we
-    // haven't already.
-    inferred_label = InferLabelFromDivTable(element);
-  }
   return inferred_label;
 }
 
 // Fills |option_strings| with the values of the <option> elements present in
 // |select_element|.
 void GetOptionStringsFromElement(const WebSelectElement& select_element,
                                  std::vector<base::string16>* option_values,
                                  std::vector<base::string16>* option_contents) {
   DCHECK(!select_element.isNull());
 
@@ skipping 836 matching lines @@
 
 gfx::RectF GetScaledBoundingBox(float scale, WebElement* element) {
   gfx::Rect bounding_box(element->boundsInViewportSpace());
   return gfx::RectF(bounding_box.x() * scale,
                     bounding_box.y() * scale,
                     bounding_box.width() * scale,
                     bounding_box.height() * scale);
 }
 
 }  // namespace autofill