AutoFill: Fix scraping a label from a div table.

BUG=61439
TEST=FormManagerTest.LabelsInferredFromDivTable

Review URL: http://codereview.chromium.org/4248002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@64691 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 79 insertions, 5 deletions

diff --git a/chrome/browser/autofill/form_field.h b/chrome/browser/autofill/form_field.h
index cf57350..38b2ec2 100644
--- a/chrome/browser/autofill/form_field.h
+++ b/chrome/browser/autofill/form_field.h
@@ -126,7 +126,7 @@ class FormField {
   // Checkout field name limitation. All ECML compliant web forms will be
   // recognized correctly as such however the restrictions on having exactly
   // ECML compliant names have been loosened to only require that field names
-  // be prefixed with an ECML compiant name in order to accommodate checkout.
+  // be prefixed with an ECML compliant name in order to accommodate checkout.
   // Additionally we allow the use of '.' as a word delimiter in addition to the
   // ECML standard '_' (see FormField::FormField for details).
   static string16 GetEcmlPattern(const char* ecml_name);
diff --git a/chrome/renderer/form_manager.cc b/chrome/renderer/form_manager.cc
index 52bc6cf..265d0c2 100644
--- a/chrome/renderer/form_manager.cc
+++ b/chrome/renderer/form_manager.cc
@@ -190,6 +190,26 @@ string16 InferLabelFromTable(
 }
 
 // Helper for |InferLabelForElement()| that infers a label, if possible, from
+// a surrounding div table.
+// Eg. <div>Some Text<span><input ...></span></div>
+string16 InferLabelFromDivTable(
+    const WebFormControlElement& element) {
+  WebNode parent = element.parentNode();
+  while (!parent.isNull() && parent.isElementNode() &&
+         !parent.to<WebElement>().hasTagName("div"))
+    parent = parent.parentNode();
+
+  if (parent.isNull() || !parent.isElementNode())
+    return string16();
+
+  WebElement e = parent.to<WebElement>();
+  if (e.isNull() || !e.hasTagName("div"))
+    return string16();
+
+  return FindChildText(e);
+}
+
+// Helper for |InferLabelForElement()| that infers a label, if possible, from
 // a surrounding definition list.
 // Eg. <dl><dt>Some Text</dt><dd><input ...></dd></dl>
 // Eg. <dl><dt><b>Some Text</b></dt><dd><b><input ...></b></dd></dl>
@@ -702,14 +722,16 @@ string16 FormManager::InferLabelForElement(
   string16 inferred_label = InferLabelFromPrevious(element);
 
   // If we didn't find a label, check for table cell case.
-  if (inferred_label.empty()) {
+  if (inferred_label.empty())
     inferred_label = InferLabelFromTable(element);
-  }
+
+  // If we didn't find a label, check for div table case.
+  if (inferred_label.empty())
+    inferred_label = InferLabelFromDivTable(element);
 
   // If we didn't find a label, check for definition list case.
-  if (inferred_label.empty()) {
+  if (inferred_label.empty())
     inferred_label = InferLabelFromDefinitionList(element);
-  }
 
   return inferred_label;
 }
diff --git a/chrome/renderer/form_manager_browsertest.cc b/chrome/renderer/form_manager_browsertest.cc
index 7e6bc6a..76ff574 100644
--- a/chrome/renderer/form_manager_browsertest.cc
+++ b/chrome/renderer/form_manager_browsertest.cc
@@ -1381,6 +1381,58 @@ TEST_F(FormManagerTest, LabelsInferredWithImageTags) {
             fields[5]);
 }
 
+TEST_F(FormManagerTest, LabelsInferredFromDivTable) {
+  LoadHTML("<FORM name=\"TestForm\" action=\"http://cnn.com\" method=\"post\">"
+           "<DIV>First Name:<BR>"
+           "  <SPAN>"
+           "    <INPUT type=\"text\" name=\"firstname\" value=\"John\">"
+           "  </SPAN>"
+           "</DIV>"
+           "<DIV>Last Name:<BR>"
+           "  <SPAN>"
+           "    <INPUT type=\"text\" name=\"lastname\" value=\"Doe\">"
+           "  </SPAN>"
+           "</DIV>"
+           "<input type=\"submit\" name=\"reply-send\" value=\"Send\">"
+           "</FORM>");
+
+  WebFrame* web_frame = GetMainFrame();
+  ASSERT_NE(static_cast<WebFrame*>(NULL), web_frame);
+
+  FormManager form_manager;
+  form_manager.ExtractForms(web_frame);
+
+  std::vector<FormData> forms;
+  form_manager.GetFormsInFrame(web_frame, FormManager::REQUIRE_NONE, &forms);
+  ASSERT_EQ(1U, forms.size());
+
+  const FormData& form = forms[0];
+  EXPECT_EQ(ASCIIToUTF16("TestForm"), form.name);
+  EXPECT_EQ(GURL(web_frame->url()), form.origin);
+  EXPECT_EQ(GURL("http://cnn.com"), form.action);
+
+  const std::vector<FormField>& fields = form.fields;
+  ASSERT_EQ(3U, fields.size());
+  EXPECT_EQ(FormField(ASCIIToUTF16("First Name:"),
+                      ASCIIToUTF16("firstname"),
+                      ASCIIToUTF16("John"),
+                      ASCIIToUTF16("text"),
+                      20),
+            fields[0]);
+  EXPECT_EQ(FormField(ASCIIToUTF16("Last Name:"),
+                      ASCIIToUTF16("lastname"),
+                      string16(),
+                      ASCIIToUTF16("text"),
+                      20),
+            fields[1]);
+  EXPECT_EQ(FormField(string16(),
+                      ASCIIToUTF16("reply-send"),
+                      ASCIIToUTF16("Send"),
+                      ASCIIToUTF16("submit"),
+                      0),
+            fields[2]);
+}
+
 TEST_F(FormManagerTest, FillFormMaxLength) {
   LoadHTML("<FORM name=\"TestForm\" action=\"http://buh.com\" method=\"post\">"
            "  <INPUT type=\"text\" id=\"firstname\" maxlength=\"5\"/>"