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..437d413ebf3c1b6caf7e8c7d68acdc6a5a23751f 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;

[Ilya Sherman, 2011/05/14 06:59:12]

David and I discussed this a bit offline today -- I'll try to rehash/summarize
that here:  Rather than changing the parsing for address fields, let's try
making multiple passes over the form in FormField::ParseFormFields().  In the
first pass, we can check for any email fields; in the second pass, we can check
for any phone fields; in the third pass, for any address fields; and so on. 
This should both add a bit more clarity to the code; and also give us a bit more
flexibility for parsing.

[Ilya Sherman, 2011/05/14 07:01:33]

Whoops, didn't mean to send this comment out -- wanted to let David think about
this some more and then add his thoughts.  So, I guess for now keep this in mind
as a possibility, but not necessarily the best one.

[honten.org, 2011/05/14 07:41:20]

I see, it might be better idea.

But, at first, I want to refactor ParseText() with bitmap. I want to commit
gradually;-)

On 2011/05/14 07:01:33, Ilya Sherman wrote:

+
     if (ParseCompany(scanner, is_ecml, address_field.get()) ||
         ParseAddressLines(scanner, is_ecml, address_field.get()) ||
         ParseCity(scanner, is_ecml, address_field.get()) ||
@@ -246,7 +264,8 @@ bool AddressField::ParseCountry(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_COUNTRY_RE);
 
-  return ParseText(scanner, pattern, &address_field->country_);
+  return ParseText(scanner, FormField::Pattern(pattern, true),
+                   &address_field->country_);
 }
 
 // static
@@ -315,7 +334,8 @@ bool AddressField::ParseCity(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_CITY_RE);
 
-  return ParseText(scanner, pattern, &address_field->city_);
+  return ParseText(scanner, FormField::Pattern(pattern, true),
+                   &address_field->city_);
 }
 
 // static
@@ -331,7 +351,8 @@ bool AddressField::ParseState(AutofillScanner* scanner,
   else
     pattern = l10n_util::GetStringUTF16(IDS_AUTOFILL_STATE_RE);
 
-  return ParseText(scanner, pattern, &address_field->state_);
+  return ParseText(scanner, FormField::Pattern(pattern, true),
+                   &address_field->state_);
 }
 
 AddressType AddressField::AddressTypeFromText(const string16 &text) {