Change heuristic regex and order to match grabber-continental.

chrome/browser/autofill/credit_card_field.cc

 // Copyright (c) 2011 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 "chrome/browser/autofill/credit_card_field.h"
 
 #include <stddef.h>
 
 #include "base/logging.h"
 #include "base/memory/scoped_ptr.h"
@@ skipping 50 matching lines @@
         if (fields == 0 || credit_card_field->expiration_month_) {
           // at beginning or end
           name_pattern = l10n_util::GetStringUTF16(
               IDS_AUTOFILL_NAME_ON_CARD_RE);
         } else {
           name_pattern = l10n_util::GetStringUTF16(
               IDS_AUTOFILL_NAME_ON_CARD_CONTEXTUAL_RE);
         }
       }
 
-      if (ParseText(scanner, name_pattern, &credit_card_field->cardholder_))
+      if (ParseText(scanner, name_pattern,
+                    MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                    &credit_card_field->cardholder_))
         continue;
 
       // As a hard-coded hack for Expedia's billing pages (expedia_checkout.html
       // and ExpediaBilling.html in our test suite), recognize separate fields
       // for the cardholder's first and last name if they have the labels "cfnm"
       // and "clnm".
       scanner->SaveCursor();
       const AutofillField* first;
-      if (!is_ecml && ParseText(scanner, ASCIIToUTF16("^cfnm"), &first) &&
+      if (!is_ecml && ParseText(scanner, ASCIIToUTF16("^cfnm"),
+                                MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                                &first) &&
           ParseText(scanner, ASCIIToUTF16("^clnm"),
+                    MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
                     &credit_card_field->cardholder_last_)) {
         credit_card_field->cardholder_ = first;
         continue;
       }
       scanner->Rewind();
     }
 
     // We look for a card security code before we look for a credit
     // card number and match the general term "number".  The security code
     // has a plethora of names; we've seen "verification #",
     // "verification number", "card identification number" and others listed
     // in the |pattern| below.
     string16 pattern;
     if (is_ecml)
       pattern = GetEcmlPattern(kEcmlCardVerification);
     else
       pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_CARD_CVC_RE);
 
     if (!credit_card_field->verification_ &&
-        ParseText(scanner, pattern, &credit_card_field->verification_))
+        ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                  &credit_card_field->verification_))
       continue;
 
     // TODO(jhawkins): Parse the type select control.
 
     if (is_ecml)
       pattern = GetEcmlPattern(kEcmlCardNumber);
     else
       pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_CARD_NUMBER_RE);
 
     if (!credit_card_field->number_ &&
-        ParseText(scanner, pattern, &credit_card_field->number_))
+        ParseText(scanner, pattern, MATCH_TEXT, &credit_card_field->number_))
       continue;
 
     if (LowerCaseEqualsASCII(scanner->Cursor()->form_control_type, "month")) {
       credit_card_field->expiration_month_ = scanner->Cursor();
       scanner->Advance();
     } else {
       // "Expiration date" is the most common label here, but some pages have
       // "Expires", "exp. date" or "exp. month" and "exp. year".  We also look
       // for the field names ccmonth and ccyear, which appear on at least 4 of
       // our test pages.
       //
       // -> On at least one page (The China Shop2.html) we find only the labels
       // "month" and "year".  So for now we match these words directly; we'll
       // see if this turns out to be too general.
       //
       // Toolbar Bug 51451: indeed, simply matching "month" is too general for
       //   https://rps.fidelity.com/ftgw/rps/RtlCust/CreatePIN/Init.
       // Instead, we match only words beginning with "month".
       if (is_ecml)
         pattern = GetEcmlPattern(kEcmlCardExpireMonth);
       else
         pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_EXPIRATION_MONTH_RE);
 
       if ((!credit_card_field->expiration_month_ ||
            credit_card_field->expiration_month_->IsEmpty()) &&
-          ParseText(scanner, pattern, &credit_card_field->expiration_month_)) {
+          ParseText(scanner, pattern, MATCH_TEXT | MATCH_SELECT,
+                    &credit_card_field->expiration_month_)) {
         if (is_ecml)
           pattern = GetEcmlPattern(kEcmlCardExpireYear);
         else
           pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_EXPIRATION_DATE_RE);
 
-        if (!ParseText(scanner, pattern,
+        if (!ParseText(scanner, pattern, MATCH_TEXT | MATCH_SELECT,
                        &credit_card_field->expiration_year_)) {
           scanner->Rewind();
           return NULL;
         }
         continue;
       }
     }
 
-    if (ParseText(scanner, GetEcmlPattern(kEcmlCardExpireDay)))
+    if (ParseText(scanner,
+                  GetEcmlPattern(kEcmlCardExpireDay),
+                  MATCH_NAME | MATCH_LABEL | MATCH_TEXT | MATCH_SELECT))
       continue;
 
     // Some pages (e.g. ExpediaBilling.html) have a "card description"
     // field; we parse this field but ignore it.
     // We also ignore any other fields within a credit card block that
     // start with "card", under the assumption that they are related to
     // the credit card section being processed but are uninteresting to us.
     if (ParseText(scanner,
-                  l10n_util::GetStringUTF16(IDS_AUTOFILL_CARD_IGNORED_RE)))
+                  l10n_util::GetStringUTF16(IDS_AUTOFILL_CARD_IGNORED_RE),
+                  MATCH_NAME | MATCH_LABEL | MATCH_TEXT))
       continue;
 
     break;
   }
 
   // Some pages have a billing address field after the cardholder name field.
   // For that case, allow only just the cardholder name field.  The remaining
   // CC fields will be picked up in a following CreditCardField.
   if (credit_card_field->cardholder_)
     return credit_card_field.release();
@@ skipping 20 matching lines @@
 
 CreditCardField::CreditCardField()
     : cardholder_(NULL),
       cardholder_last_(NULL),
       type_(NULL),
       number_(NULL),
       verification_(NULL),
       expiration_month_(NULL),
       expiration_year_(NULL) {
 }