Implement NameField, a FormField that matches a name field in a form.

BUG=none
TEST=none
Review URL: http://codereview.chromium.org/504030

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@34784 0039d316-1c4b-4281-b951-d872f2087c98

4 files changed, 236 insertions, 4 deletions

diff --git a/chrome/browser/autofill/form_field.cc b/chrome/browser/autofill/form_field.cc
index ec15120..e6e375b 100644
--- a/chrome/browser/autofill/form_field.cc
+++ b/chrome/browser/autofill/form_field.cc
@@ -7,6 +7,7 @@
 #include "chrome/browser/autofill/address_field.h"
 #include "chrome/browser/autofill/autofill_field.h"
 #include "chrome/browser/autofill/credit_card_field.h"
+#include "chrome/browser/autofill/name_field.h"
 #include "chrome/browser/autofill/phone_field.h"
 #include "third_party/WebKit/WebKit/chromium/public/WebRegularExpression.h"
 #include "third_party/WebKit/WebKit/chromium/public/WebString.h"
@@ -77,10 +78,9 @@ FormField* FormField::ParseFormField(
   if (field != NULL)
     return field;
 
-  // TODO(jhawkins):
-  //  - NameField
-
-  return NULL;
+  // We search for a NameField last since it matches the word "name", which is
+  // relatively general.
+  return NameField::Parse(iter, is_ecml);
 }
 
 // static
diff --git a/chrome/browser/autofill/name_field.cc b/chrome/browser/autofill/name_field.cc
new file mode 100644
index 0000000..8a32dd6
--- /dev/null
+++ b/chrome/browser/autofill/name_field.cc
@@ -0,0 +1,157 @@
+// Copyright (c) 2009 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/name_field.h"
+
+#include "base/string_util.h"
+#include "chrome/browser/autofill/autofill_type.h"
+
+NameField* NameField::Parse(std::vector<AutoFillField*>::const_iterator* iter,
+                            bool is_ecml) {
+  // Try FirstLastNameField first since it's more specific.
+  NameField* field = FirstLastNameField::Parse(iter, is_ecml);
+  if (field == NULL && !is_ecml)
+    field = FullNameField::Parse(iter);
+  return field;
+}
+
+FullNameField* FullNameField::Parse(
+    std::vector<AutoFillField*>::const_iterator* iter) {
+  // Exclude labels containing the string "username", which typically
+  // denotes a login ID rather than the user's actual name.
+  AutoFillField* field = **iter;
+  if (Match(field, ASCIIToUTF16("username")))
+    return NULL;
+
+  // Searching for any label containing the word "name" is too general;
+  // for example, Travelocity_Edit travel profile.html contains a field
+  // "Travel Profile Name".
+  if (ParseText(iter, ASCIIToUTF16("^name|full name|your name|customer name"),
+                &field))
+    return new FullNameField(field);
+
+  return NULL;
+}
+
+FirstLastNameField* FirstLastNameField::Parse1(
+    std::vector<AutoFillField*>::const_iterator* iter) {
+  // Some pages (e.g. Overstock_comBilling.html, SmithsonianCheckout.html)
+  // have the label "Name" followed by two or three text fields.
+  FirstLastNameField v;
+  std::vector<AutoFillField*>::const_iterator q = *iter;
+
+  AutoFillField* next;
+  if (ParseText(&q, ASCIIToUTF16("^name"), &v.first_name_) &&
+      ParseText(&q, ASCIIToUTF16(""), &next)) {
+    if (ParseText(&q, ASCIIToUTF16(""), &v.last_name_)) {
+      // There are three name fields; assume that the middle one is a
+      // middle initial (it is, at least, on SmithsonianCheckout.html).
+      v.middle_name_ = next;
+      v.middle_initial_ = true;
+    } else {  // only two name fields
+      v.last_name_ = next;
+    }
+
+    *iter = q;
+    return new FirstLastNameField(v);
+  }
+
+  return NULL;
+}
+
+FirstLastNameField* FirstLastNameField::Parse2(
+    std::vector<AutoFillField*>::const_iterator* iter) {
+  FirstLastNameField v;
+  std::vector<AutoFillField*>::const_iterator q = *iter;
+
+  // A fair number of pages use the names "fname" and "lname" for naming
+  // first and last name fields (examples from the test suite:
+  // BESTBUY_COM - Sign In2.html; Crate and Barrel Check Out.html;
+  // dell_checkout1.html).  At least one UK page (The China Shop2.html)
+  // asks, in stuffy English style, for just initials and a surname,
+  // so we match "initials" here (and just fill in a first name there,
+  // American-style).
+  if (!ParseText(&q, ASCIIToUTF16("first name|initials|fname"), &v.first_name_))
+    return NULL;
+
+  // We check for a middle initial before checking for a middle name
+  // because at least one page (PC Connection.html) has a field marked
+  // as both (the label text is "MI" and the element name is
+  // "txtmiddlename"); such a field probably actually represents a
+  // middle initial.
+  if (ParseText(&q, ASCIIToUTF16("^mi|middle initial|m.i."), &v.middle_name_)) {
+    v.middle_initial_ = true;
+  } else {
+    ParseText(&q, ASCIIToUTF16("middle name|mname"), &v.middle_name_);
+  }
+
+  if (!ParseText(&q, ASCIIToUTF16("last name|lname|surname"), &v.last_name_))
+    return NULL;
+
+  *iter = q;
+  return new FirstLastNameField(v);
+}
+
+FirstLastNameField* FirstLastNameField::ParseEcmlName(
+    std::vector<AutoFillField*>::const_iterator* iter) {
+  FirstLastNameField field;
+  std::vector<AutoFillField*>::const_iterator q = *iter;
+
+  string16 pattern = GetEcmlPattern(kEcmlShipToFirstName,
+                                    kEcmlBillToFirstName, '|');
+  if (!ParseText(&q, pattern, &field.first_name_))
+    return NULL;
+
+  pattern = GetEcmlPattern(kEcmlShipToMiddleName, kEcmlBillToMiddleName, '|');
+  ParseText(&q, pattern, &field.middle_name_);
+
+  pattern = GetEcmlPattern(kEcmlShipToLastName, kEcmlBillToLastName, '|');
+  if (ParseText(&q, pattern, &field.last_name_)) {
+    *iter = q;
+    return new FirstLastNameField(field);
+  }
+
+  return NULL;
+}
+
+FirstLastNameField* FirstLastNameField::Parse(
+    std::vector<AutoFillField*>::const_iterator* iter,
+    bool is_ecml) {
+  if (is_ecml) {
+    return ParseEcmlName(iter);
+  } else {
+    FirstLastNameField* v = Parse1(iter);
+    if (v != NULL)
+      return v;
+
+    return Parse2(iter);
+  }
+}
+
+bool FirstLastNameField::GetFieldInfo(FieldTypeMap* field_type_map) const {
+  bool ok = Add(field_type_map, first_name_, AutoFillType(NAME_FIRST));
+  DCHECK(ok);
+  ok = ok && Add(field_type_map, last_name_, AutoFillType(NAME_LAST));
+  DCHECK(ok);
+  AutoFillType type = middle_initial_ ?
+      AutoFillType(NAME_MIDDLE_INITIAL) : AutoFillType(NAME_MIDDLE);
+  ok = ok && Add(field_type_map, middle_name_, type);
+  DCHECK(ok);
+
+  return ok;
+}
+
+FirstLastNameField::FirstLastNameField()
+    : first_name_(NULL),
+      middle_name_(NULL),
+      last_name_(NULL),
+      middle_initial_(false) {
+}
+
+FirstLastNameField::FirstLastNameField(const FirstLastNameField& field)
+    : first_name_(field.first_name_),
+      middle_name_(field.middle_name_),
+      last_name_(field.last_name_),
+      middle_initial_(field.middle_initial_) {
+}
diff --git a/chrome/browser/autofill/name_field.h b/chrome/browser/autofill/name_field.h
new file mode 100644
index 0000000..c827d6c
--- /dev/null
+++ b/chrome/browser/autofill/name_field.h
@@ -0,0 +1,73 @@
+// Copyright (c) 2009 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.
+
+#ifndef CHROME_BROWSER_AUTOFILL_NAME_FIELD_H_
+#define CHROME_BROWSER_AUTOFILL_NAME_FIELD_H_
+
+#include <vector>
+
+#include "base/logging.h"
+#include "chrome/browser/autofill/autofill_field.h"
+#include "chrome/browser/autofill/form_field.h"
+
+// A form field that can parse either a FullNameField or a FirstLastNameField.
+class NameField : public FormField {
+ public:
+  static NameField* Parse(std::vector<AutoFillField*>::const_iterator* iter,
+                          bool is_ecml);
+
+  virtual int priority() const { return 0; }
+
+ protected:
+  NameField() {}
+
+ private:
+  DISALLOW_COPY_AND_ASSIGN(NameField);
+};
+
+// A form field that can parse a full name field.
+class FullNameField : public NameField {
+ public:
+  virtual bool GetFieldInfo(FieldTypeMap* field_type_map) const {
+    bool ok = Add(field_type_map, field_, AutoFillType(NAME_FULL));
+    DCHECK(ok);
+    return true;
+  }
+
+  static FullNameField* Parse(
+      std::vector<AutoFillField*>::const_iterator* iter);
+
+ private:
+  explicit FullNameField(AutoFillField* field) : field_(field) {}
+
+  AutoFillField* field_;
+  DISALLOW_COPY_AND_ASSIGN(FullNameField);
+};
+
+// A form field that can parse a first and last name field.
+class FirstLastNameField : public NameField {
+ public:
+  static FirstLastNameField* Parse1(
+      std::vector<AutoFillField*>::const_iterator* iter);
+  static FirstLastNameField* Parse2(
+      std::vector<AutoFillField*>::const_iterator* iter);
+  static FirstLastNameField* ParseEcmlName(
+      std::vector<AutoFillField*>::const_iterator* iter);
+  static FirstLastNameField* Parse(
+      std::vector<AutoFillField*>::const_iterator* iter, bool is_ecml);
+
+  virtual bool GetFieldInfo(FieldTypeMap* field_type_map) const;
+
+ private:
+  FirstLastNameField();
+  explicit FirstLastNameField(const FirstLastNameField& field);
+  void operator=(const FirstLastNameField& field);
+
+  AutoFillField* first_name_;
+  AutoFillField* middle_name_;  // Optional.
+  AutoFillField* last_name_;
+  bool middle_initial_;  // True if middle_name_ is a middle initial.
+};
+
+#endif  // CHROME_BROWSER_AUTOFILL_NAME_FIELD_H_
diff --git a/chrome/chrome_browser.gypi b/chrome/chrome_browser.gypi
index 5004d95..71224bd 100755
--- a/chrome/chrome_browser.gypi
+++ b/chrome/chrome_browser.gypi
@@ -97,6 +97,8 @@
         'browser/autofill/form_field.h',
         'browser/autofill/form_structure.cc',
         'browser/autofill/form_structure.h',
+        'browser/autofill/name_field.cc',
+        'browser/autofill/name_field.h',
         'browser/autofill/personal_data_manager.cc',
         'browser/autofill/personal_data_manager.h',
         'browser/autofill/phone_field.cc',