Allow word separators in a few more Autofill regexes.

This also fixes an issue where inferred sections were not being updated after server types were returned.

BUG=93595
TEST=browser_tests --gtest_filter=*DataDrivenHeuristics*16*


Review URL: http://codereview.chromium.org/7981047

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@102251 0039d316-1c4b-4281-b951-d872f2087c98

chrome/browser/autofill/autofill_regex_constants.cc.utf8

@@ -122,7 +122,7 @@ const char kShippingDesignatorRe[] = "ship";
 // credit_card_field.cc
 /////////////////////////////////////////////////////////////////////////////
 const char kNameOnCardRe[] =
-    "card.?holder|name.?on.?card|ccname|ccfullname|owner"
+    "card.?holder|name.?on.?card|cc.?name|cc.?full.?name|owner"
     "|karteninhaber"  // de-DE
     "|nombre.*tarjeta"  // es
     "|nom.*carte"  // fr-FR
@@ -134,7 +134,7 @@ const char kNameOnCardRe[] =
 const char kNameOnCardContextualRe[] =
     "name";
 const char kCardNumberRe[] =
-    "card.?number|card.?#|card.?no|ccnum|acctnum"
+    "card.?number|card.?#|card.?no|cc.?num|acct.?num"
     "|nummer"  // de-DE
     "|credito|numero|número"  // es
     "|numéro"  // fr-FR

chrome/browser/autofill/form_structure.cc

@@ -279,9 +279,9 @@ void FormStructure::DetermineHeuristicTypes() {
   // First, try to detect field types based on the fields' |autocompletetype|
   // attributes.  If there is at least one form field with this attribute, don't
   // try to apply other heuristics to match fields in this form.
-  bool found_sections;
+  bool has_author_specified_sections;
   ParseAutocompletetypeAttributes(&has_author_specified_types_,
-                                  &found_sections);
+                                  &has_author_specified_sections);
 
   if (!has_author_specified_types_) {
     FieldTypeMap field_type_map;
@@ -295,21 +295,7 @@ void FormStructure::DetermineHeuristicTypes() {
   }
 
   UpdateAutofillCount();
-
-  if (!found_sections)
-    IdentifySections();
-
-  // Ensure that credit card and address fields are in separate sections.
-  // This simplifies the section-aware logic in autofill_manager.cc.
-  for (std::vector<AutofillField*>::iterator field = fields_->begin();
-       field != fields_->end(); ++field) {
-    AutofillType::FieldTypeGroup field_type_group =
-        AutofillType((*field)->type()).group();
-    if (field_type_group == AutofillType::CREDIT_CARD)
-      (*field)->set_section((*field)->section() + ASCIIToUTF16("-cc"));
-    else
-      (*field)->set_section((*field)->section() + ASCIIToUTF16("-default"));
-  }
+  IdentifySections(has_author_specified_sections);
 }
 
 bool FormStructure::EncodeUploadRequest(
@@ -464,6 +450,7 @@ void FormStructure::ParseQueryResponse(const std::string& response_xml,
     }
 
     form->UpdateAutofillCount();
+    form->IdentifySections(false);
   }
 
   AutofillMetrics::ServerQueryMetric metric;
@@ -923,50 +910,65 @@ void FormStructure::ParseAutocompletetypeAttributes(bool* found_attribute,
   }
 }
 
-void FormStructure::IdentifySections() {
+void FormStructure::IdentifySections(bool has_author_specified_sections) {
   if (fields_.empty())
     return;
 
-  // Name sections after the first field in the section.
-  string16 current_section = fields_->front()->unique_name();
+  if (!has_author_specified_sections) {
+    // Name sections after the first field in the section.
+    string16 current_section = fields_->front()->unique_name();
+
+    // Keep track of the types we've seen in this section.
+    std::set<AutofillFieldType> seen_types;
+    AutofillFieldType previous_type = UNKNOWN_TYPE;
+
+    for (std::vector<AutofillField*>::iterator field = fields_->begin();
+         field != fields_->end(); ++field) {
+      const AutofillFieldType current_type =
+          AutofillType::GetEquivalentFieldType((*field)->type());
+
+      bool already_saw_current_type = seen_types.count(current_type) > 0;
+
+      // Forms often ask for multiple phone numbers -- e.g. both a daytime and
+      // evening phone number.  Our phone number detection is also generally a
+      // little off.  Hence, ignore this field type as a signal here.
+      if (AutofillType(current_type).group() == AutofillType::PHONE_HOME)
+        already_saw_current_type = false;
+
+      // Some forms have adjacent fields of the same type.  Two common examples:
+      //  * Forms with two email fields, where the second is meant to "confirm"
+      //    the first.
+      //  * Forms with a <select> menu for states in some countries, and a
+      //    freeform <input> field for states in other countries.  (Usually,
+      //    only one of these two will be visible for any given choice of
+      //    country.)
+      // Generally, adjacent fields of the same type belong in the same logical
+      // section.
+      if (current_type == previous_type)
+        already_saw_current_type = false;
+
+      previous_type = current_type;
+
+      if (current_type != UNKNOWN_TYPE && already_saw_current_type) {
+        // We reached the end of a section, so start a new section.
+        seen_types.clear();
+        current_section = (*field)->unique_name();
+      }
 
-  // Keep track of the types we've seen in this section.
-  std::set<AutofillFieldType> seen_types;
-  AutofillFieldType previous_type = UNKNOWN_TYPE;
+      seen_types.insert(current_type);
+      (*field)->set_section(current_section);
+    }
+  }
 
+  // Ensure that credit card and address fields are in separate sections.
+  // This simplifies the section-aware logic in autofill_manager.cc.
   for (std::vector<AutofillField*>::iterator field = fields_->begin();
        field != fields_->end(); ++field) {
-    const AutofillFieldType current_type =
-        AutofillType::GetEquivalentFieldType((*field)->type());
-
-    bool already_saw_current_type = seen_types.count(current_type) > 0;
-
-    // Forms often ask for multiple phone numbers -- e.g. both a daytime and
-    // evening phone number.  Our phone number detection is also generally a
-    // little off.  Hence, ignore this field type as a signal here.
-    if (AutofillType(current_type).group() == AutofillType::PHONE_HOME)
-      already_saw_current_type = false;
-
-    // Some forms have adjacent fields of the same type.  Two common examples:
-    //  * Forms with two email fields, where the second is meant to "confirm"
-    //    the first.
-    //  * Forms with a <select> menu for states in some countries, and a
-    //    freeform <input> field for states in other countries.  (Usually, only
-    //    one of these two will be visible for any given choice of country.)
-    // Generally, adjacent fields of the same type belong in the same logical
-    // section.
-    if (current_type == previous_type)
-      already_saw_current_type = false;
-
-    previous_type = current_type;
-
-    if (current_type != UNKNOWN_TYPE && already_saw_current_type) {
-      // We reached the end of a section, so start a new section.
-      seen_types.clear();
-      current_section = (*field)->unique_name();
-    }
-
-    seen_types.insert(current_type);
-    (*field)->set_section(current_section);
+    AutofillType::FieldTypeGroup field_type_group =
+        AutofillType((*field)->type()).group();
+    if (field_type_group == AutofillType::CREDIT_CARD)
+      (*field)->set_section((*field)->section() + ASCIIToUTF16("-cc"));
+    else
+      (*field)->set_section((*field)->section() + ASCIIToUTF16("-default"));
   }
 }

chrome/browser/autofill/form_structure.h

@@ -173,8 +173,11 @@ class FormStructure {
   // Classifies each field in |fields_| into a logical section.
   // Sections are identified by the heuristic that a logical section should not
   // include multiple fields of the same autofill type (with some exceptions, as
-  // described in the implementation).
-  void IdentifySections();
+  // described in the implementation).  Sections are furthermore distinguished
+  // as either credit card or non-credit card sections.
+  // If |has_author_specified_sections| is true, only the second pass --
+  // distinguishing credit card sections from non-credit card ones -- is made.
+  void IdentifySections(bool has_author_specified_sections);
 
   // The name of the form.
   string16 form_name_;

chrome/test/data/autofill/heuristics/input/16_crbug_93595.html

+<!-- https://www.jbox.com/checkout/payment# -->
+<form name="paymentForm" method="post" action="#" id="paymentForm">
+  <div id="errors"></div>
+  <fieldset id="paymentchoose">
+    <legend>Choose a payment method</legend>
+    <div>
+      <select id="paymentoption" accesskeyclass="element select medium" name="payment_method" onchange="changePayment()">
+        <option selected="selected">
+          Visa
+        </option>
+        <option>
+          Mastercard
+        </option>
+        <option>
+          American Express
+        </option>
+        <option>
+          Discover
+        </option>
+        <option>
+          Card on File (provide CVV2)
+        </option>
+        <option>
+          Check / Money Order
+        </option>
+        <option>
+          Gift Certificate / Credit
+        </option>
+        <option>
+          Western Union (we'll send instructions)
+        </option>
+        <option value="PayPal">
+          PayPal
+        </option>
+        <option value="Google Checkout">
+          Google Checkout
+        </option>
+      </select>
+    </div>
+  </fieldset>
+  <fieldset id="ccinput">
+    <legend>Credit Card Details</legend>
+    <div>
+      <input name="cc_number" class="element text medium" type="text" maxlength="255" value=""> <label for="address1"><b>Credit Card Number</b></label>
+    </div>
+    <div>
+      <input name="cc_name" class="element text medium" type="text" maxlength="60" value=""> <label for="address1">Name On Card | Ex: <i>Peter Payne</i></label>
+    </div>
+    <div class="left">
+      <span><input name="cc_exp" class="element text small" type="text" maxlength="255" value=""> <label for="cc_exp">Expiration date<br>
+      (MM/YY)</label></span>
+    </div>
+    <div>
+      <span><input name="cc_cvv2" class="element text small" type="text" maxlength="255" value=""> <label for="cc_cvv2">CVV2 / Security Code:</label></span>
+    </div>
+  </fieldset>
+  <fieldset>
+    <legend>If an item is not available:</legend> <select name="field_17">
+      <option value="Select an option...">
+        Select an option...
+      </option>
+      <option value="Cancel the item">
+        Cancel the item
+      </option>
+      <option value="Wait for all items to be available">
+        Wait for all items to be available
+      </option>
+      <option value="Split and send separately (additional fees apply)">
+        Split and send separately (additional fees apply)
+      </option>
+    </select>
+  </fieldset>
+  <fieldset>
+    <legend>Please confirm your age</legend> <input name="age" class="element text small" type="text" maxlength="255" value="">
+  </fieldset>
+  <fieldset>
+    <legend>Comment or Special Instructions (if any)</legend>
+    <textarea id="element_6" name="comments" class="element textarea small"></textarea>
+  </fieldset><input type="hidden" name="paymentForm" value="259215"> <input id="placeorder" class="button_text" type="submit" name="submit" value="Place Order">
+</form>

chrome/test/data/autofill/heuristics/output/16_crbug_93595.out

+UNKNOWN_TYPE | payment_method |  | Visa
+CREDIT_CARD_NUMBER | cc_number |  | 
+CREDIT_CARD_NAME | cc_name |  | 
+CREDIT_CARD_EXP_DATE_2_DIGIT_YEAR | cc_exp | Expiration date (MM/YY) | 
+UNKNOWN_TYPE | cc_cvv2 | CVV2 / Security Code: | 
+UNKNOWN_TYPE | field_17 |  | Select an option...
+UNKNOWN_TYPE | age |  |