Autofill address line 2 is not applied to ebay.com

Fixes heuristics to include "suite" and "unit" when matching address line 2.

BUG=48197
TEST=FormStructureTest.TwoAddressLinesSkipSuite, FormStructureTest.TwoAddressLinesDontSkipSuite

Review URL: http://codereview.chromium.org/3134042

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@57596 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 97 insertions, 14 deletions

diff --git a/chrome/browser/autofill/address_field.cc b/chrome/browser/autofill/address_field.cc
index 4d52a4c..f21f532 100644
--- a/chrome/browser/autofill/address_field.cc
+++ b/chrome/browser/autofill/address_field.cc
@@ -210,12 +210,6 @@ bool AddressField::ParseAddressLines(
         return false;
   }
 
-  // Some pages (e.g. expedia_checkout.html) have an apartment or
-  // suite number at this point.  The occasional page (e.g.
-  // Ticketmaster3.html) calls this a unit number.  We ignore this
-  // field since we can't fill it yet.
-  ParseText(iter, ASCIIToUTF16("suite|unit"));
-
   // Optionally parse more address lines, which may have empty labels.
   // Some pages have 3 address lines (eg SharperImageModifyAccount.html)
   // Some pages even have 4 address lines (e.g. uk/ShoesDirect2.html)!
@@ -225,7 +219,7 @@ bool AddressField::ParseAddressLines(
     if (!ParseEmptyText(iter, &address_field->address2_))
       ParseText(iter, pattern, &address_field->address2_);
   } else {
-    pattern = ASCIIToUTF16("address2|street|street_line2|addr2");
+    pattern = ASCIIToUTF16("address2|street|street_line2|addr2|suite|unit");
     string16 label_pattern = ASCIIToUTF16("address");
     if (!ParseEmptyText(iter, &address_field->address2_))
       if (!ParseText(iter, pattern, &address_field->address2_))
@@ -239,8 +233,11 @@ bool AddressField::ParseAddressLines(
                                kEcmlBillToAddress3, '|');
       ParseText(iter, pattern);
     } else {
-      pattern = ASCIIToUTF16("line3");
-      ParseLabelText(iter, pattern, NULL);
+      pattern = ASCIIToUTF16("address3|street|street_line3|addr3|line3");
+      string16 label_pattern = ASCIIToUTF16("address");
+      if (!ParseEmptyText(iter, NULL))
+        if (!ParseText(iter, pattern, NULL))
+          ParseLabelText(iter, label_pattern, NULL);
     }
   }
 
diff --git a/chrome/browser/autofill/autofill_download_unittest.cc b/chrome/browser/autofill/autofill_download_unittest.cc
index 4e81323..baecd97 100644
--- a/chrome/browser/autofill/autofill_download_unittest.cc
+++ b/chrome/browser/autofill/autofill_download_unittest.cc
@@ -148,7 +148,7 @@ TEST(AutoFillDownloadTest, QueryAndUploadTest) {
                                                ASCIIToUTF16("text"),
                                                0));
   form.fields.push_back(webkit_glue::FormField(ASCIIToUTF16("city"),
-                                               ASCIIToUTF16("address2"),
+                                               ASCIIToUTF16("city"),
                                                string16(),
                                                ASCIIToUTF16("text"),
                                                0));
diff --git a/chrome/browser/autofill/form_structure_unittest.cc b/chrome/browser/autofill/form_structure_unittest.cc
index 9055578..b2d0435 100644
--- a/chrome/browser/autofill/form_structure_unittest.cc
+++ b/chrome/browser/autofill/form_structure_unittest.cc
@@ -741,6 +741,94 @@ TEST(FormStructureTest, ThreeAddressLines) {
   EXPECT_EQ(ADDRESS_HOME_CITY, form_structure->field(3)->heuristic_type());
 }
 
+// This example comes from expedia.com where they use a "Suite" label to
+// indicate a suite or apartment number.  We interpret this as address line 2.
+// And the following "Street address second line" we interpret as address line
+// 3 and discard.
+// See http://crbug.com/48197 for details.
+TEST(FormStructureTest, ThreeAddressLinesExpedia) {
+  scoped_ptr<FormStructure> form_structure;
+  FormData form;
+
+  form.method = ASCIIToUTF16("post");
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("Street:"),
+                             ASCIIToUTF16("FOPIH_RgWebCC_0_IHAddress_ads1"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("Suite or Apt:"),
+                             ASCIIToUTF16("FOPIH_RgWebCC_0_IHAddress_adap"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("Street address second line"),
+                             ASCIIToUTF16("FOPIH_RgWebCC_0_IHAddress_ads2"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("City:"),
+                             ASCIIToUTF16("FOPIH_RgWebCC_0_IHAddress_adct"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form_structure.reset(new FormStructure(form));
+  EXPECT_TRUE(form_structure->IsAutoFillable());
+  ASSERT_EQ(4U, form_structure->field_count());
+  ASSERT_EQ(3U, form_structure->autofill_count());
+
+  // Address Line 1.
+  EXPECT_EQ(ADDRESS_HOME_LINE1, form_structure->field(0)->heuristic_type());
+  // Suite / Apt.
+  EXPECT_EQ(ADDRESS_HOME_LINE2, form_structure->field(1)->heuristic_type());
+  // Address Line 3.
+  EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(2)->heuristic_type());
+  // City.
+  EXPECT_EQ(ADDRESS_HOME_CITY, form_structure->field(3)->heuristic_type());
+}
+
+// This example comes from ebay.com where the word "suite" appears in the label
+// and the name "address2" clearly indicates that this is the address line 2.
+// See http://crbug.com/48197 for details.
+TEST(FormStructureTest, TwoAddressLinesEbay) {
+  scoped_ptr<FormStructure> form_structure;
+  FormData form;
+
+  form.method = ASCIIToUTF16("post");
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("Address Line1"),
+                             ASCIIToUTF16("address1"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("Floor number, suite number, etc"),
+                             ASCIIToUTF16("address2"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form.fields.push_back(
+      webkit_glue::FormField(ASCIIToUTF16("City"),
+                             ASCIIToUTF16("city"),
+                             string16(),
+                             ASCIIToUTF16("text"),
+                             0));
+  form_structure.reset(new FormStructure(form));
+  EXPECT_TRUE(form_structure->IsAutoFillable());
+  ASSERT_EQ(3U, form_structure->field_count());
+  ASSERT_EQ(3U, form_structure->autofill_count());
+
+  // Address Line 1.
+  EXPECT_EQ(ADDRESS_HOME_LINE1, form_structure->field(0)->heuristic_type());
+  // Address Line 2.
+  EXPECT_EQ(ADDRESS_HOME_LINE2, form_structure->field(1)->heuristic_type());
+  // City.
+  EXPECT_EQ(ADDRESS_HOME_CITY, form_structure->field(2)->heuristic_type());
+}
+
 TEST(FormStructureTest, HeuristicsStateWithProvince) {
   scoped_ptr<FormStructure> form_structure;
   FormData form;
@@ -852,15 +940,13 @@ TEST(FormStructureTest, HeuristicsWithBilling) {
   form_structure.reset(new FormStructure(form));
   EXPECT_TRUE(form_structure->IsAutoFillable());
   ASSERT_EQ(11U, form_structure->field_count());
-  ASSERT_EQ(10U, form_structure->autofill_count());
+  ASSERT_EQ(11U, 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(COMPANY_NAME, form_structure->field(2)->heuristic_type());
   EXPECT_EQ(ADDRESS_BILLING_LINE1, form_structure->field(3)->heuristic_type());
-  // Note: We'd expect this to match ADDRESS_BILLING_LINE2, but due to toolbar
-  // heuristics for other pages we skip fields with label including "suite".
-  EXPECT_EQ(UNKNOWN_TYPE, form_structure->field(4)->heuristic_type());
+  EXPECT_EQ(ADDRESS_BILLING_LINE2, form_structure->field(4)->heuristic_type());
   EXPECT_EQ(ADDRESS_BILLING_CITY, form_structure->field(5)->heuristic_type());
   EXPECT_EQ(ADDRESS_BILLING_STATE, form_structure->field(6)->heuristic_type());
   EXPECT_EQ(ADDRESS_BILLING_COUNTRY,