[Password Manager] HTML parsing based client-side form type classifier

chrome/renderer/autofill/form_classifier_browsertest.cc

Index: chrome/renderer/autofill/form_classifier_browsertest.cc
diff --git a/chrome/renderer/autofill/form_classifier_browsertest.cc b/chrome/renderer/autofill/form_classifier_browsertest.cc
new file mode 100644
index 0000000000000000000000000000000000000000..8634221d7541a1627be5d07e93fbcf8335560696
--- /dev/null
+++ b/chrome/renderer/autofill/form_classifier_browsertest.cc
@@ -0,0 +1,270 @@
+// Copyright 2016 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 "base/macros.h"
+#include "base/strings/utf_string_conversions.h"
+#include "chrome/test/base/chrome_render_view_test.h"
+#include "components/autofill/content/renderer/form_classifier.h"
+#include "testing/gtest/include/gtest/gtest.h"
+#include "third_party/WebKit/public/web/WebDocument.h"
+#include "third_party/WebKit/public/web/WebFormElement.h"
+#include "third_party/WebKit/public/web/WebLocalFrame.h"
+
+namespace autofill {
+
+class FormClassifierTest : public ChromeRenderViewTest {
+ public:
+  FormClassifierTest() {}
+
+  void TearDown() override {
+    LoadHTML("");
+    ChromeRenderViewTest::TearDown();
+  }
+
+  bool GetGenerationField(std::string* generation_field) {
+    blink::WebDocument document = GetMainFrame()->document();
+    blink::WebFormElement form =
+        document.getElementById("test_form").to<blink::WebFormElement>();
+    base::string16 generation_field16;
+    bool generation_availalbe =
+        ClassifyFormAndFindGenerationField(form, &generation_field16);
+    *generation_field = base::UTF16ToUTF8(generation_field16);
+    return generation_availalbe;
+  }
+
+ private:
+  DISALLOW_COPY_AND_ASSIGN(FormClassifierTest);
+};
+
+const char kSigninFormHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <SELECT id='account_type'>"
+    "    <OPTION value = 'personal'>"
+    "    <OPTION value = 'corporate'>"
+    "  </SELECT>"
+    "  <INPUT type = 'text' id = 'username'/>"
+    "  <INPUT type = 'password' id = 'password'/>"
+    "  <INPUT type = 'checkbox' id = 'remember_me'/>"
+    "  <INPUT type = 'checkbox' id = 'secure_login'/>"
+    "  <INPUT type = 'submit' id = 'signin' />"
+    "  <INPUT type = 'hidden' id = 'ignore_this' />"
+    "  <INPUT type = 'hidden' id = 'ignore_this_too' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithSeveralTextFieldsFormHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <INPUT type = 'text' id = 'full_name'/>"
+    "  <INPUT type = 'text' id = 'username'/>"
+    "  <INPUT type = 'password' id = 'password'/>"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithSeveralPasswordFieldsHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <INPUT type = 'password' id = 'password'/>"
+    "  <INPUT type = 'password' id = 'confirm_password'/>"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithManyCheckboxesHTML[] =
+    "<FORM id = 'test_form'> "
+    "  </SELECT>"
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'checkbox' id = 'subscribe_science' />"
+    "  <INPUT type = 'checkbox' id = 'subscribe_music' />"
+    "  <INPUT type = 'checkbox' id = 'subscribe_sport' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithOtherFieldsHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'color' id = 'account_color' />"
+    "  <INPUT type = 'date' id = 'date_of_birth' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSigninFormWithTextFeatureInInputElementHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <INPUT type = 'text' id = 'username' class = 'sign-in_field' />"
+    "  <INPUT type = 'password' id = 'password' class = 'sign-in_field' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSigninFormWithTextFeatureInFormTagHTML[] =
+    "<FORM id = 'test_form' some_attribute='auth_form' > "
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSigninAndSignupFormsInOneDivHTML[] =
+    "<DIV class = 'signup_signin_container'>"
+    "  <FORM id = 'test_form' class = 'unknown_form_type' > "
+    "    <INPUT type = 'text' id = 'username' />"
+    "    <INPUT type = 'password' id = 'password' />"
+    "    <INPUT type = 'submit' id = 'submit' />"
+    "  </FORM>"
+    "  <FORM id = 'another_form' class = 'unknown_form_type' > "
+    "    <INPUT type = 'text' id = 'username' />"
+    "    <INPUT type = 'password' id = 'password' />"
+    "    <INPUT type = 'submit' id = 'submit' />"
+    "  </FORM>"
+    "</DIV>";
+
+const char kSigninFormWithTextFeatureInEnclosingParentHTML[] =
+    "<DIV id = 'signup_signin_container'>"
+    "  <DIV id = 'signin_wraper'>"
+    "    <FORM id = 'test_form' > "
+    "      <INPUT type = 'text' id = 'username' />"
+    "      <INPUT type = 'password' id = 'password' />"
+    "      <INPUT type = 'submit' id = 'submit' />"
+    "    </FORM>"
+    "  </DIV>"
+    "</DIV>";
+
+const char kSigninFormWithInvisibleFieldsHTML[] =
+    "<FORM id = 'test_form'> "
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'input' hidden id = 'hidden_field1' />"
+    "  <INPUT type = 'password' hidden id = 'hidden_field2'/>"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithSigninButtonHTML[] =
+    "<FORM id = 'test_form' >"
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'password' id = 'confirm_password' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "  <INPUT type = 'button' id = 'goto_signin_form' />"
+    "  <INPUT type = 'image' id = 'goto_auth_form' />"
+    "</FORM>";
+
+const char kSomeFormWithoutPasswordFields[] =
+    "<FORM id = 'test_form' >"
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'text' id = 'fullname' />"
+    "  <INPUT type = 'text' id = 'address' />"
+    "  <INPUT type = 'text' id = 'phone' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kSignupFormWithSigninTextFeatureAndManyFieldsHTML[] =
+    "<FORM id = 'test_form' class = 'log-on_container'> "
+    "  <INPUT type = 'text' id = 'fullname' />"
+    "  <INPUT type = 'text' id = 'username' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+const char kChangeFormWithTreePasswordFieldsHTML[] =
+    "<FORM id = 'test_form' >"
+    "  <INPUT type = 'password' id = 'old_password' />"
+    "  <INPUT type = 'password' id = 'password' />"
+    "  <INPUT type = 'password' id = 'confirm_password' />"
+    "  <INPUT type = 'submit' id = 'submit' />"
+    "</FORM>";
+
+TEST_F(FormClassifierTest, SigninForm) {
+  // Signin form with as many as possible visible elements,
+  // i.e. if one more text/password/checkbox/other field is added,  the form

[vabr (Chromium), 2016/06/10 13:22:11]

nit: Please keep horizontal spaces to one character and also fix the line breaks
to make most of the 80 character per line limit.

[kolos1, 2016/06/13 14:27:34]

Removed double space. 

Did I exceed 80 character limit somewhere?

[vabr (Chromium), 2016/06/13 16:01:57]

No, you did not, quite the opposite :), which was what I was commenting about.
In other words, a good part of line 176 could have been fitted at the end of
line 175.

+  // will be recognized as a signup form.
+  LoadHTML(kSigninFormHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SignupFormWithSeveralTextFields) {
+  LoadHTML(kSignupFormWithSeveralTextFieldsFormHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, SignupFormWithSeveralPasswordFieldsHTML) {
+  LoadHTML(kSignupFormWithSeveralPasswordFieldsHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, SignupFormWithManyCheckboxesHTML) {
+  LoadHTML(kSignupFormWithManyCheckboxesHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, SignupFormWithOtherFieldsHTML) {
+  LoadHTML(kSignupFormWithOtherFieldsHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, SigninFormWithTextFeatureInInputElementHTML) {
+  LoadHTML(kSigninFormWithTextFeatureInInputElementHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SigninFormWithTextFeatureInFormTagHTML) {
+  LoadHTML(kSigninFormWithTextFeatureInFormTagHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SigninAndSignupFormsInOneDivHTML) {
+  LoadHTML(kSigninAndSignupFormsInOneDivHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SigninFormWithTextFeatureInEnclosingParentHTML) {
+  LoadHTML(kSigninFormWithTextFeatureInEnclosingParentHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SigninFormWithInvisibleFieldsHTML) {
+  LoadHTML(kSigninFormWithInvisibleFieldsHTML);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SignupFormWithSigninButtonHTML) {
+  LoadHTML(kSignupFormWithSigninButtonHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, SomeFormWithoutPasswordFields) {
+  LoadHTML(kSomeFormWithoutPasswordFields);
+  std::string generation_field;
+  EXPECT_FALSE(GetGenerationField(&generation_field));
+}
+
+TEST_F(FormClassifierTest, SignupFormWithSigninTextFeatureAndManyFieldsHTML) {
+  // Even if there is signin text feature, the number of fields is more reliable
+  // signal of signup form. So, this form should classified as signup.

[vabr (Chromium), 2016/06/10 13:22:11]

typo: should classified -> should be classified

[kolos1, 2016/06/13 14:27:34]

Done.

+  LoadHTML(kSignupFormWithSigninTextFeatureAndManyFieldsHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+TEST_F(FormClassifierTest, kChangeFormWithTreePasswordFieldsHTML) {
+  LoadHTML(kChangeFormWithTreePasswordFieldsHTML);
+  std::string generation_field;
+  EXPECT_TRUE(GetGenerationField(&generation_field));
+  EXPECT_EQ("password", generation_field);
+}
+
+}  // namespace autofill