[Autofill] Update the autocomplete types implementation to match the current HTML spec.

chrome/browser/autofill/form_structure_unittest.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 "base/memory/scoped_ptr.h"
 #include "base/string_util.h"
 #include "base/utf_string_conversions.h"
 #include "chrome/browser/autofill/form_structure.h"
 #include "chrome/common/form_data.h"
 #include "chrome/common/form_field_data.h"
@@ skipping 311 matching lines @@
   // Address.
   EXPECT_EQ(ADDRESS_HOME_LINE1, form_structure->field(4)->heuristic_type());
   // City.
   EXPECT_EQ(ADDRESS_HOME_CITY, form_structure->field(5)->heuristic_type());
   // Zip.
   EXPECT_EQ(ADDRESS_HOME_ZIP, form_structure->field(6)->heuristic_type());
   // Submit.
   EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(7)->heuristic_type());
 }
 
-// Verify that we can correctly process the |autocompletetype| attribute.
-TEST(FormStructureTest, HeuristicsAutocompletetype) {
+// Verify that we can correctly process the |autocomplete| attribute.
+TEST(FormStructureTest, HeuristicsAutocompleteAttribute) {
   scoped_ptr<FormStructure> form_structure;
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
   field.label = string16();
   field.name = ASCIIToUTF16("field1");
-  field.autocomplete_type = ASCIIToUTF16("given-name");
+  field.autocomplete_attribute = ASCIIToUTF16("given-name");
   form.fields.push_back(field);
 
   field.label = string16();
   field.name = ASCIIToUTF16("field2");
-  field.autocomplete_type = ASCIIToUTF16("surname");
+  field.autocomplete_attribute = ASCIIToUTF16("family-name");
   form.fields.push_back(field);
 
   field.label = string16();
   field.name = ASCIIToUTF16("field3");
-  field.autocomplete_type = ASCIIToUTF16("email");
+  field.autocomplete_attribute = ASCIIToUTF16("email");
   form.fields.push_back(field);
 
   form_structure.reset(new FormStructure(form));
   form_structure->DetermineHeuristicTypes();
   EXPECT_TRUE(form_structure->IsAutofillable(true));
 
   // Expect the correct number of fields.
   ASSERT_EQ(3U, form_structure->field_count());
   ASSERT_EQ(3U, form_structure->autofill_count());
 
   EXPECT_EQ(NAME_FIRST, form_structure->field(0)->heuristic_type());
   EXPECT_EQ(NAME_LAST, form_structure->field(1)->heuristic_type());
   EXPECT_EQ(EMAIL_ADDRESS, form_structure->field(2)->heuristic_type());
 }
 
-// Verify that we can correctly process the |autocompletetype| attribute for
-// phone number types (especially phone prefixes and suffixes).
-TEST(FormStructureTest, HeuristicsAutocompletetypePhones) {
+// Verify that we can correctly process the 'autocomplete' attribute for phone
+// number types (especially phone prefixes and suffixes).
+TEST(FormStructureTest, HeuristicsAutocompleteAttributePhoneTypes) {
   scoped_ptr<FormStructure> form_structure;
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
   field.label = string16();
   field.name = ASCIIToUTF16("field1");
-  field.autocomplete_type = ASCIIToUTF16("phone-local");
+  field.autocomplete_attribute = ASCIIToUTF16("tel-local");
   form.fields.push_back(field);
 
   field.label = string16();
   field.name = ASCIIToUTF16("field2");
-  field.autocomplete_type = ASCIIToUTF16("phone-local-prefix");
+  field.autocomplete_attribute = ASCIIToUTF16("tel-local-prefix");
   form.fields.push_back(field);
 
   field.label = string16();
   field.name = ASCIIToUTF16("field3");
-  field.autocomplete_type = ASCIIToUTF16("phone-local-suffix");
+  field.autocomplete_attribute = ASCIIToUTF16("tel-local-suffix");
   form.fields.push_back(field);
 
   form_structure.reset(new FormStructure(form));
   form_structure->DetermineHeuristicTypes();
   EXPECT_TRUE(form_structure->IsAutofillable(true));
 
   // Expect the correct number of fields.
   ASSERT_EQ(3U, form_structure->field_count());
   EXPECT_EQ(3U, form_structure->autofill_count());
 
   EXPECT_EQ(PHONE_HOME_NUMBER, form_structure->field(0)->heuristic_type());
   EXPECT_EQ(AutofillField::IGNORED, form_structure->field(0)->phone_part());
   EXPECT_EQ(PHONE_HOME_NUMBER, form_structure->field(1)->heuristic_type());
   EXPECT_EQ(AutofillField::PHONE_PREFIX,
             form_structure->field(1)->phone_part());
   EXPECT_EQ(PHONE_HOME_NUMBER, form_structure->field(2)->heuristic_type());
   EXPECT_EQ(AutofillField::PHONE_SUFFIX,
             form_structure->field(2)->phone_part());
 }
 
-// If at least one field includes the |autocompletetype| attribute, we should
-// not try to apply any other heuristics.
-TEST(FormStructureTest, AutocompletetypeOverridesOtherHeuristics) {
+// If at least one field includes type hints in the 'autocomplete' attribute, we
+// should not try to apply any other heuristics.
+TEST(FormStructureTest, AutocompleteAttributeOverridesOtherHeuristics) {
   scoped_ptr<FormStructure> form_structure;
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   // Start with a regular contact form.
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
   field.label = ASCIIToUTF16("First Name");
   field.name = ASCIIToUTF16("firstname");
@@ skipping 12 matching lines @@
   EXPECT_TRUE(form_structure->IsAutofillable(true));
   EXPECT_TRUE(form_structure->ShouldBeCrowdsourced());
 
   ASSERT_EQ(3U, form_structure->field_count());
   ASSERT_EQ(3U, form_structure->autofill_count());
 
   EXPECT_EQ(NAME_FIRST, form_structure->field(0)->heuristic_type());
   EXPECT_EQ(NAME_LAST, form_structure->field(1)->heuristic_type());
   EXPECT_EQ(EMAIL_ADDRESS, form_structure->field(2)->heuristic_type());
 
-  // Now update the first form field to include an 'autocompletetype' attribute.
-  form.fields.front().autocomplete_type = ASCIIToUTF16("x-other");
+  // Now update the first form field to include an 'autocomplete' attribute.
+  form.fields.front().autocomplete_attribute = ASCIIToUTF16("x-other");
   form_structure.reset(new FormStructure(form));
   form_structure->DetermineHeuristicTypes();
   EXPECT_FALSE(form_structure->IsAutofillable(true));
   EXPECT_FALSE(form_structure->ShouldBeCrowdsourced());
 
   ASSERT_EQ(3U, form_structure->field_count());
   ASSERT_EQ(0U, form_structure->autofill_count());
 
   EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(0)->heuristic_type());
   EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(1)->heuristic_type());
   EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(2)->heuristic_type());
 }
 
 // Verify that we can correctly process sections listed in the |autocomplete|
 // attribute.
-TEST(FormStructureTest, HeuristicsAutocompletetypeWithSections) {
-  scoped_ptr<FormStructure> form_structure;
+TEST(FormStructureTest, HeuristicsAutocompleteAttributeWithSections) {
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
-  // We expect "shipping" and "billing" to be the most common sections.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field1");
-  field.autocomplete_type = ASCIIToUTF16("section-shipping given-name");
-  form.fields.push_back(field);
-
-  // Some field will have no section specified.  These fall into the default
-  // section, with an empty name.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field2");
-  field.autocomplete_type = ASCIIToUTF16("surname");
+  // Some fields will have no section specified.  These fall into the default
+  // section.
+  field.autocomplete_attribute = ASCIIToUTF16("email");
   form.fields.push_back(field);
 
   // We allow arbitrary section names.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field3");
-  field.autocomplete_type = ASCIIToUTF16("section-foo address-line1");
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo email");
   form.fields.push_back(field);
 
-  // Specifying "section-" is equivalent to not specifying a section.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field4");
-  field.autocomplete_type = ASCIIToUTF16("section- address-line2");
+  // "shipping" and "billing" are special section tokens that don't require the
+  // "section-" prefix.
+  field.autocomplete_attribute = ASCIIToUTF16("shipping email");
+  form.fields.push_back(field);
+  field.autocomplete_attribute = ASCIIToUTF16("billing email");
+  form.fields.push_back(field);
+
+  // "shipping" and "billing" can be combined with other section names.
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo shipping email");
+  form.fields.push_back(field);
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo billing email");
   form.fields.push_back(field);
 
   // We don't do anything clever to try to coalesce sections; it's up to site
   // authors to avoid typos.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field5");
-  field.autocomplete_type = ASCIIToUTF16("section--shipping locality");
+  field.autocomplete_attribute = ASCIIToUTF16("section--foo email");
+  form.fields.push_back(field);
+
+  // "shipping email" and "section--shipping" email should be parsed as
+  // different sections.  This is only an interesting test due to how we
+  // implement implicit section names from attributes like "shipping email"; see
+  // the implementation for more details.
+  field.autocomplete_attribute = ASCIIToUTF16("section--shipping email");
   form.fields.push_back(field);
 
   // Credit card fields are implicitly in a separate section from other fields.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field6");
-  field.autocomplete_type = ASCIIToUTF16("section-shipping cc-number");
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo cc-number");
   form.fields.push_back(field);
 
-  form_structure.reset(new FormStructure(form));
-  form_structure->DetermineHeuristicTypes();
-  EXPECT_TRUE(form_structure->IsAutofillable(true));
+  FormStructure form_structure(form);
+  form_structure.DetermineHeuristicTypes();
+  EXPECT_TRUE(form_structure.IsAutofillable(true));
 
   // Expect the correct number of fields.
-  ASSERT_EQ(6U, form_structure->field_count());
-  ASSERT_EQ(6U, form_structure->autofill_count());
+  ASSERT_EQ(9U, form_structure.field_count());
+  EXPECT_EQ(9U, form_structure.autofill_count());
 
-  EXPECT_EQ(NAME_FIRST, form_structure->field(0)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("shipping-default"),
-            form_structure->field(0)->section());
-  EXPECT_EQ(NAME_LAST, form_structure->field(1)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("-default"), form_structure->field(1)->section());
-  EXPECT_EQ(ADDRESS_HOME_LINE1, form_structure->field(2)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("foo-default"), form_structure->field(2)->section());
-  EXPECT_EQ(ADDRESS_HOME_LINE2, form_structure->field(3)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("-default"), form_structure->field(3)->section());
-  EXPECT_EQ(ADDRESS_HOME_CITY, form_structure->field(4)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("-shipping-default"),
-            form_structure->field(4)->section());
-  EXPECT_EQ(CREDIT_CARD_NUMBER, form_structure->field(5)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("shipping-cc"), form_structure->field(5)->section());
+  // All of the fields in this form should be parsed as belonging to different
+  // sections.
+  std::set<string16> section_names;
+  for (size_t i = 0; i < 9; ++i) {
+    section_names.insert(form_structure.field(i)->section());
+  }
+  EXPECT_EQ(9U, section_names.size());
 }
 
-// Verify that we can correctly process fallback types listed in the
-// |autocompletetype| attribute.
-TEST(FormStructureTest, HeuristicsAutocompletetypeWithFallbacks) {
-  scoped_ptr<FormStructure> form_structure;
+// Verify that we can correctly process a degenerate section listed in the
+// |autocomplete| attribute.
+TEST(FormStructureTest, HeuristicsAutocompleteAttributeWithSectionsDegenerate) {
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
-  // Skip over any sections and "x"-prefixed types.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field1");
-  field.autocomplete_type =
-      ASCIIToUTF16("section-full-name x-given-name-initial given-name");
+  // Some fields will have no section specified.  These fall into the default
+  // section.
+  field.autocomplete_attribute = ASCIIToUTF16("email");
   form.fields.push_back(field);
 
-  // Stop processing once we see a known type.
-  field.label = string16();
-  field.name = ASCIIToUTF16("field2");
-  field.autocomplete_type = ASCIIToUTF16("section-full-name surname full-name");
+  // Specifying "section-" is equivalent to not specifying a section.
+  field.autocomplete_attribute = ASCIIToUTF16("section- email");
   form.fields.push_back(field);
 
-  // Skip over unknown types even if they are not prefixed with "x-".
-  field.label = string16();
-  field.name = ASCIIToUTF16("field3");
-  field.autocomplete_type =
-      ASCIIToUTF16("section-shipping mobile-phone-full phone-full");
+  FormStructure form_structure(form);
+  form_structure.DetermineHeuristicTypes();
+
+  // Expect the correct number of fields.
+  ASSERT_EQ(2U, form_structure.field_count());
+  EXPECT_EQ(2U, form_structure.autofill_count());
+
+  // All of the fields in this form should be parsed as belonging to the same
+  // sections.
+  std::set<string16> section_names;
+  for (size_t i = 0; i < 2; ++i) {
+    section_names.insert(form_structure.field(i)->section());
+  }
+  EXPECT_EQ(1U, section_names.size());
+}
+
+// Verify that we can correctly process repeated sections listed in the
+// |autocomplete| attribute.
+TEST(FormStructureTest, HeuristicsAutocompleteAttributeWithSectionsRepeated) {
+  FormData form;
+  form.method = ASCIIToUTF16("post");
+
+  FormFieldData field;
+  field.form_control_type = ASCIIToUTF16("text");
+
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo email");
+  form.fields.push_back(field);
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo street-address");
   form.fields.push_back(field);
 
-  form_structure.reset(new FormStructure(form));
-  form_structure->DetermineHeuristicTypes();
-  EXPECT_TRUE(form_structure->IsAutofillable(true));
+  FormStructure form_structure(form);
+  form_structure.DetermineHeuristicTypes();
 
   // Expect the correct number of fields.
-  ASSERT_EQ(3U, form_structure->field_count());
-  ASSERT_EQ(3U, form_structure->autofill_count());
+  ASSERT_EQ(2U, form_structure.field_count());
+  EXPECT_EQ(2U, form_structure.autofill_count());
 
-  EXPECT_EQ(NAME_FIRST, form_structure->field(0)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("full-name-default"),
-            form_structure->field(0)->section());
-  EXPECT_EQ(NAME_LAST, form_structure->field(1)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("full-name-default"),
-            form_structure->field(1)->section());
-  EXPECT_EQ(PHONE_HOME_WHOLE_NUMBER,
-            form_structure->field(2)->heuristic_type());
-  EXPECT_EQ(ASCIIToUTF16("shipping-default"),
-            form_structure->field(2)->section());
+  // All of the fields in this form should be parsed as belonging to the same
+  // sections.
+  std::set<string16> section_names;
+  for (size_t i = 0; i < 2; ++i) {
+    section_names.insert(form_structure.field(i)->section());
+  }
+  EXPECT_EQ(1U, section_names.size());
+}
+
+// Verify that we do not override the author-specified sections from a form with
+// local heuristics.
+TEST(FormStructureTest, HeuristicsDontOverrideAutocompleteAttributeSections) {
+  FormData form;
+  form.method = ASCIIToUTF16("post");
+
+  FormFieldData field;
+  field.form_control_type = ASCIIToUTF16("text");
+
+  field.name = ASCIIToUTF16("one");
+  field.autocomplete_attribute = ASCIIToUTF16("street-address");
+  form.fields.push_back(field);
+  field.name = string16();
+  field.autocomplete_attribute = ASCIIToUTF16("section-foo email");
+  form.fields.push_back(field);
+  field.name = string16();
+  field.autocomplete_attribute = ASCIIToUTF16("name");
+  form.fields.push_back(field);
+  field.name = ASCIIToUTF16("two");
+  field.autocomplete_attribute = ASCIIToUTF16("street-address");
+  form.fields.push_back(field);
+
+  FormStructure form_structure(form);
+  form_structure.DetermineHeuristicTypes();
+
+  // Expect the correct number of fields.
+  ASSERT_EQ(4U, form_structure.field_count());
+  EXPECT_EQ(4U, form_structure.autofill_count());
+
+  // Normally, the two separate address fields would cause us to detect two
+  // separate sections; but because there is an author-specified section in this
+  // form, we do not apply these usual heuristics.
+  EXPECT_EQ(ASCIIToUTF16("one"), form_structure.field(0)->name);
+  EXPECT_EQ(ASCIIToUTF16("two"), form_structure.field(3)->name);
+  EXPECT_EQ(form_structure.field(0)->section(),
+            form_structure.field(3)->section());
 }
 
 TEST(FormStructureTest, HeuristicsSample8) {
   scoped_ptr<FormStructure> form_structure;
   FormData form;
   form.method = ASCIIToUTF16("post");
 
   FormFieldData field;
   field.form_control_type = ASCIIToUTF16("text");
 
@@ skipping 1391 matching lines @@
   EXPECT_EQ(FormStructureTest::Hash64Bit(
       std::string("https://login.facebook.com&&email&first")),
       form_structure->FormSignature());
 
   form.name = ASCIIToUTF16("login_form");
   form_structure.reset(new FormStructure(form));
   EXPECT_EQ(FormStructureTest::Hash64Bit(
       std::string("https://login.facebook.com&login_form&email&first")),
       form_structure->FormSignature());
 }