Change heuristic regex and order to match grabber-continental.

chrome/browser/autofill/form_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/form_field.h"
 
 #include <stddef.h>
 #include <string>
 #include <utility>
 
@@ skipping 217 matching lines @@
   field = CreditCardField::Parse(scanner, is_ecml);
   if (field)
     return field;
 
   // We search for a |NameField| last since it matches the word "name", which is
   // relatively general.
   return NameField::Parse(scanner, is_ecml);
 }
 
 // static
-bool FormField::ParseText(AutofillScanner* scanner, const string16& pattern) {
-  const AutofillField* field;
-  return ParseText(scanner, pattern, &field);
-}
-
-// static
 bool FormField::ParseText(AutofillScanner* scanner,
                           const string16& pattern,
+                          int match_type) {
+  const AutofillField* field;
+  return ParseText(scanner, pattern, match_type, &field);
+}
+
+// static
+bool FormField::ParseText(AutofillScanner* scanner,
+                          const string16& pattern,
+                          int match_type,
                           const AutofillField** dest) {
-  return ParseText(scanner, pattern, dest, false);
+  return ParseText(scanner, pattern, dest, match_type);
 }
 
 // static
 bool FormField::ParseEmptyText(AutofillScanner* scanner,
                                const AutofillField** dest) {
-  return ParseLabelText(scanner, ASCIIToUTF16("^$"), dest);
-}
-
-// static
-bool FormField::ParseLabelText(AutofillScanner* scanner,
-                               const string16& pattern,
-                               const AutofillField** dest) {
-  return ParseText(scanner, pattern, dest, true);
+  return ParseText(scanner, ASCIIToUTF16("^$"),
+                   MATCH_LABEL | MATCH_TEXT | MATCH_SELECT, dest);
 }
 
 // static
 bool FormField::ParseText(AutofillScanner* scanner,
                           const string16& pattern,
                           const AutofillField** dest,
-                          bool match_label_only) {
+                          int match_type) {
   if (scanner->IsEnd())
     return false;
 
   const AutofillField* field = scanner->Cursor();
+
+  if (!(match_type & MATCH_SELECT) &&
+      field->form_control_type == ASCIIToUTF16("select-one"))
+    return false;
+
+  if (!(match_type & MATCH_TEXT) &&
+      field->form_control_type == ASCIIToUTF16("text"))

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

What about 'email', 'tel', and so on?  Are there any clients for which
MATCH_TEXT is not set?

[honten.org, 2011/05/14 07:30:05]

Yeah, you are right.

I completely forgot HTML5 tags;-)

Let me think about it.

On 2011/05/14 06:59:12, Ilya Sherman wrote:

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

As I used to did in form_manager, maybe we should also add isTextField() in
FormField, down't we?

And we define
MATCH_ANY_TEXT,
MATCH_TEXT,
MATCH_EMAIL,
MATCH_DATE,
etc...

MATCH_ANY_TEXT checks isTextField flag and others check form_controy_type.

What do you think?

On 2011/05/14 07:30:05, honten wrote:

+    return false;
+
+  bool match_label_only = (match_type & MATCH_LABEL) &&
+                          !(match_type & MATCH_NAME);

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

This computation doesn't seem right to me -- if MATCH_LABEL isn't set, then we
end up matching against both the name and the label?  We should refactor Match()
to also take a match type -- or perhaps refactor the code that calls into the
Match() method to call the submethods directly.

[honten.org, 2011/05/14 07:30:05]

Actually, I wanted to refactor Match(), but it is used heavily in tests. And it
take a while to rewrite everything...

So can I prepare new function like MatchWithBitmap() to refactor?

On 2011/05/14 06:59:12, Ilya Sherman wrote:

   if (Match(field, pattern, match_label_only)) {
     if (dest)
       *dest = field;
     scanner->Advance();
     return true;
   }
 
   return false;
 }
 
 // static
-bool FormField::ParseLabelAndName(AutofillScanner* scanner,
-                                  const string16& pattern,
-                                  const AutofillField** dest) {
+bool FormField::ParseEmpty(AutofillScanner* scanner) {
+  // TODO(jhawkins): Handle select fields.
+  const string16 pattern(ASCIIToUTF16("^$"));
   const AutofillField* field = scanner->Cursor();
   if (MatchLabel(field, pattern) && MatchName(field, pattern)) {
-    if (dest)
-      *dest = field;
     scanner->Advance();
     return true;
   }
 
   return false;
 }
 
 // static
-bool FormField::ParseEmpty(AutofillScanner* scanner) {
-  // TODO(jhawkins): Handle select fields.
-  return ParseLabelAndName(scanner, ASCIIToUTF16("^$"), NULL);
-}
-
-// static
 bool FormField::Add(FieldTypeMap* field_type_map,
                     const AutofillField* field,
                     AutofillFieldType type) {
   // Several fields are optional.
   if (!field)
     return true;
 
   // TODO(isherman): Is this the intent?
   return field_type_map->insert(make_pair(field->unique_name(), type)).second;
 }
 
 string16 FormField::GetEcmlPattern(const char* ecml_name) {
   return ASCIIToUTF16(std::string("^") + ecml_name);
 }
 
 string16 FormField::GetEcmlPattern(const char* ecml_name1,
                                    const char* ecml_name2,
                                    char pattern_operator) {
   return ASCIIToUTF16(StringPrintf("^%s%c^%s",
                                    ecml_name1, pattern_operator, ecml_name2));
 }