Change heuristic regex and order to match grabber-continental.

chrome/browser/autofill/address_field.cc

Index: chrome/browser/autofill/address_field.cc
diff --git a/chrome/browser/autofill/address_field.cc b/chrome/browser/autofill/address_field.cc
index 4624fa515ce2e8279e9cb88e09c9506e71942716..2b64ddcdffa5578e0defe20b95ca40f2463c697b 100644
--- a/chrome/browser/autofill/address_field.cc
+++ b/chrome/browser/autofill/address_field.cc
@@ -13,6 +13,8 @@
 #include "base/utf_string_conversions.h"
 #include "chrome/browser/autofill/autofill_field.h"
 #include "chrome/browser/autofill/autofill_scanner.h"
+#include "chrome/browser/autofill/email_field.h"
+#include "chrome/browser/autofill/phone_field.h"
 #include "grit/autofill_resources.h"
 #include "ui/base/l10n/l10n_util.h"
 
@@ -78,6 +80,22 @@ AddressField* AddressField::Parse(AutofillScanner* scanner, bool is_ecml) {
 
   // Allow address fields to appear in any order.
   while (!scanner->IsEnd()) {
+    // Every loop, we have to parse email and phone, and if we find it, we
+    // break the loop even while continuing address. Because we want to give
+    // email and phone more priority than address. We have to rewind |scanner|
+    // position after the email and phone parsing.
+    size_t cursor_position = scanner->SaveCursor();
+    bool is_email = EmailField::Parse(scanner, is_ecml);
+    scanner->RewindTo(cursor_position);
+    if (is_email)
+      break;
+
+    cursor_position = scanner->SaveCursor();
+    bool is_phone = PhoneField::Parse(scanner, is_ecml);
+    scanner->RewindTo(cursor_position);
+    if (is_phone)
+      break;
+
     if (ParseCompany(scanner, is_ecml, address_field.get()) ||
         ParseAddressLines(scanner, is_ecml, address_field.get()) ||
         ParseCity(scanner, is_ecml, address_field.get()) ||
@@ -85,8 +103,10 @@ AddressField* AddressField::Parse(AutofillScanner* scanner, bool is_ecml) {
         ParseZipCode(scanner, is_ecml, address_field.get()) ||
         ParseCountry(scanner, is_ecml, address_field.get())) {
       continue;
-    } else if (ParseText(scanner, attention_ignored) ||
-               ParseText(scanner, region_ignored)) {
+    } else if (ParseText(scanner, attention_ignored,
+               MATCH_NAME | MATCH_LABEL | MATCH_TEXT) ||
+               ParseText(scanner, region_ignored,
+               MATCH_NAME | MATCH_LABEL | MATCH_TEXT)) {
       // We ignore the following:
       // * Attention.
       // * Province/Region/Other.
@@ -166,7 +186,8 @@ bool AddressField::ParseCompany(AutofillScanner* scanner,
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_COMPANY_RE);
   }
 
-  return ParseText(scanner, pattern, &address_field->company_);
+  return ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                   &address_field->company_);
 }
 
 // static
@@ -187,15 +208,18 @@ bool AddressField::ParseAddressLines(AutofillScanner* scanner,
   string16 pattern;
   if (is_ecml) {
     pattern = GetEcmlPattern(kEcmlShipToAddress1, kEcmlBillToAddress1, '|');
-    if (!ParseText(scanner, pattern, &address_field->address1_))
+    if (!ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                   &address_field->address1_))
       return false;
   } else {
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_ADDRESS_LINE_1_RE);
     string16 label_pattern =
         l10n_util::GetStringUTF16(IDS_AUTOFILL_ADDRESS_LINE_1_LABEL_RE);
 
-    if (!ParseText(scanner, pattern, &address_field->address1_) &&
-        !ParseLabelText(scanner, label_pattern, &address_field->address1_))
+    if (!ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                   &address_field->address1_) &&
+        !ParseText(scanner, label_pattern, MATCH_LABEL | MATCH_TEXT,
+                   &address_field->address1_))
       return false;
   }
 
@@ -205,25 +229,29 @@ bool AddressField::ParseAddressLines(AutofillScanner* scanner,
   if (is_ecml) {
     pattern = GetEcmlPattern(kEcmlShipToAddress2, kEcmlBillToAddress2, '|');
     if (!ParseEmptyText(scanner, &address_field->address2_))
-      ParseText(scanner, pattern, &address_field->address2_);
+      ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                &address_field->address2_);
   } else {
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_ADDRESS_LINE_2_RE);
     string16 label_pattern =
         l10n_util::GetStringUTF16(IDS_AUTOFILL_ADDRESS_LINE_1_LABEL_RE);
     if (!ParseEmptyText(scanner, &address_field->address2_) &&
-        !ParseText(scanner, pattern, &address_field->address2_))
-      ParseLabelText(scanner, label_pattern, &address_field->address2_);
+        !ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                   &address_field->address2_))
+      ParseText(scanner, label_pattern, MATCH_LABEL | MATCH_TEXT,
+                &address_field->address2_);
   }
 
   // Try for a third line, which we will promptly discard.
   if (address_field->address2_ != NULL) {
     if (is_ecml) {
       pattern = GetEcmlPattern(kEcmlShipToAddress3, kEcmlBillToAddress3, '|');
-      ParseText(scanner, pattern);
+      ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT);
     } else {
       pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_ADDRESS_LINE_3_RE);
       if (!ParseEmptyText(scanner, NULL))
-        ParseText(scanner, pattern, NULL);
+        ParseText(scanner, pattern,
+                  MATCH_NAME | MATCH_LABEL | MATCH_TEXT, NULL);
     }
   }
 
@@ -246,7 +274,9 @@ bool AddressField::ParseCountry(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_COUNTRY_RE);
 
-  return ParseText(scanner, pattern, &address_field->country_);
+  return ParseText(scanner, pattern,
+                   MATCH_NAME | MATCH_LABEL | MATCH_TEXT | MATCH_SELECT,
+                   &address_field->country_);
 }
 
 // static
@@ -285,7 +315,8 @@ bool AddressField::ParseZipCode(AutofillScanner* scanner,
     tempType = kGenericAddress;
   }
 
-  if (!ParseText(scanner, pattern, &address_field->zip_))
+  if (!ParseText(scanner, pattern, MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
+                 &address_field->zip_))
     return false;
 
   address_field->type_ = tempType;
@@ -294,6 +325,7 @@ bool AddressField::ParseZipCode(AutofillScanner* scanner,
     // the substring "zip".
     ParseText(scanner,
               l10n_util::GetStringUTF16(IDS_AUTOFILL_ZIP_4_RE),
+              MATCH_NAME | MATCH_LABEL | MATCH_TEXT,
               &address_field->zip4_);
   }
 
@@ -315,7 +347,9 @@ bool AddressField::ParseCity(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_CITY_RE);
 
-  return ParseText(scanner, pattern, &address_field->city_);
+  return ParseText(scanner, pattern,
+                   MATCH_NAME | MATCH_LABEL | MATCH_TEXT | MATCH_SELECT,
+                   &address_field->city_);
 }
 
 // static
@@ -331,7 +365,9 @@ bool AddressField::ParseState(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_STATE_RE);
 
-  return ParseText(scanner, pattern, &address_field->state_);
+  return ParseText(scanner, pattern,
+                   MATCH_NAME | MATCH_LABEL | MATCH_TEXT | MATCH_SELECT,
+                   &address_field->state_);
 }
 
 AddressType AddressField::AddressTypeFromText(const string16 &text) {