components/omnibox/autocomplete_input_unittest.cc

   1 // Copyright 2014 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "components/omnibox/autocomplete_input.h"
   6
   7 #include "base/basictypes.h"
   8 #include "base/strings/string16.h"
   9 #include "base/strings/utf_string_conversions.h"
  10 #include "build/build_config.h"
  11 #include "components/metrics/proto/omnibox_event.pb.h"
  12 #include "components/metrics/proto/omnibox_input_type.pb.h"
  13 #include "components/omnibox/test_scheme_classifier.h"
  14 #include "testing/gtest/include/gtest/gtest.h"
  15 #include "url/url_parse.h"
  16 #include "url/url_util.h"
  17
  18 using base::ASCIIToUTF16;
  19 using metrics::OmniboxEventProto;
  20
  21 TEST(AutocompleteInputTest, InputType) {
  22   // TODO(mukai): Fix this scheme setup.
  23   url::AddStandardScheme("chrome");
  24   url::AddStandardScheme("chrome-devtools");
  25   url::AddStandardScheme("chrome-search");
  26
  27   struct test_data {
  28     const base::string16 input;
  29     const metrics::OmniboxInputType::Type type;
  30   } input_cases[] = {
  31     { base::string16(), metrics::OmniboxInputType::INVALID },
  32     { ASCIIToUTF16("?"), metrics::OmniboxInputType::FORCED_QUERY },
  33     { ASCIIToUTF16("?foo"), metrics::OmniboxInputType::FORCED_QUERY },
  34     { ASCIIToUTF16("?foo bar"), metrics::OmniboxInputType::FORCED_QUERY },
  35     { ASCIIToUTF16("?http://foo.com/bar"),
  36       metrics::OmniboxInputType::FORCED_QUERY },
  37     { ASCIIToUTF16("foo"), metrics::OmniboxInputType::UNKNOWN },
  38     { ASCIIToUTF16("localhost"), metrics::OmniboxInputType::URL },
  39     { ASCIIToUTF16("foo.c"), metrics::OmniboxInputType::UNKNOWN },
  40     { ASCIIToUTF16("foo.com"), metrics::OmniboxInputType::URL },
  41     { ASCIIToUTF16("-foo.com"), metrics::OmniboxInputType::URL },
  42     { ASCIIToUTF16("foo-.com"), metrics::OmniboxInputType::URL },
  43     { ASCIIToUTF16("foo_.com"), metrics::OmniboxInputType::UNKNOWN },
  44     { ASCIIToUTF16("foo.-com"), metrics::OmniboxInputType::QUERY },
  45     { ASCIIToUTF16("foo/"), metrics::OmniboxInputType::URL },
  46     { ASCIIToUTF16("foo/bar"), metrics::OmniboxInputType::UNKNOWN },
  47     { ASCIIToUTF16("foo/bar%00"), metrics::OmniboxInputType::QUERY },
  48     { ASCIIToUTF16("foo/bar/"), metrics::OmniboxInputType::URL },
  49     { ASCIIToUTF16("foo/bar baz\\"), metrics::OmniboxInputType::URL },
  50     { ASCIIToUTF16("foo.com/bar"), metrics::OmniboxInputType::URL },
  51     { ASCIIToUTF16("foo;bar"), metrics::OmniboxInputType::UNKNOWN },
  52     { ASCIIToUTF16("foo/bar baz"), metrics::OmniboxInputType::UNKNOWN },
  53     { ASCIIToUTF16("foo bar.com"), metrics::OmniboxInputType::QUERY },
  54     { ASCIIToUTF16("foo bar"), metrics::OmniboxInputType::QUERY },
  55     { ASCIIToUTF16("foo+bar"), metrics::OmniboxInputType::QUERY },
  56     { ASCIIToUTF16("foo+bar.com"), metrics::OmniboxInputType::UNKNOWN },
  57     { ASCIIToUTF16("\"foo:bar\""), metrics::OmniboxInputType::QUERY },
  58     { ASCIIToUTF16("link:foo.com"), metrics::OmniboxInputType::UNKNOWN },
  59     { ASCIIToUTF16("foo:81"), metrics::OmniboxInputType::URL },
  60     { ASCIIToUTF16("localhost:8080"), metrics::OmniboxInputType::URL },
  61     { ASCIIToUTF16("www.foo.com:81"), metrics::OmniboxInputType::URL },
  62     { ASCIIToUTF16("foo.com:123456"), metrics::OmniboxInputType::QUERY },
  63     { ASCIIToUTF16("foo.com:abc"), metrics::OmniboxInputType::QUERY },
  64     { ASCIIToUTF16("1.2.3.4:abc"), metrics::OmniboxInputType::QUERY },
  65     { ASCIIToUTF16("user@foo.com"), metrics::OmniboxInputType::UNKNOWN },
  66     { ASCIIToUTF16("user@foo/z"), metrics::OmniboxInputType::URL },
  67     { ASCIIToUTF16("user@foo/z z"), metrics::OmniboxInputType::URL },
  68     { ASCIIToUTF16("user@foo.com/z"), metrics::OmniboxInputType::URL },
  69     { ASCIIToUTF16("user:pass@"), metrics::OmniboxInputType::UNKNOWN },
  70     { ASCIIToUTF16("user:pass@!foo.com"), metrics::OmniboxInputType::UNKNOWN },
  71     { ASCIIToUTF16("user:pass@foo"), metrics::OmniboxInputType::URL },
  72     { ASCIIToUTF16("user:pass@foo.c"), metrics::OmniboxInputType::URL },
  73     { ASCIIToUTF16("user:pass@foo.com"), metrics::OmniboxInputType::URL },
  74     { ASCIIToUTF16("user:pass@foo.com:81"), metrics::OmniboxInputType::URL },
  75     { ASCIIToUTF16("user:pass@foo:81"), metrics::OmniboxInputType::URL },
  76     { ASCIIToUTF16("1.2"), metrics::OmniboxInputType::UNKNOWN },
  77     { ASCIIToUTF16("1.2/45"), metrics::OmniboxInputType::UNKNOWN },
  78     { ASCIIToUTF16("1.2:45"), metrics::OmniboxInputType::UNKNOWN },
  79     { ASCIIToUTF16("user@1.2:45"), metrics::OmniboxInputType::URL },
  80     { ASCIIToUTF16("user@foo:45"), metrics::OmniboxInputType::URL },
  81     { ASCIIToUTF16("user:pass@1.2:45"), metrics::OmniboxInputType::URL },
  82     { ASCIIToUTF16("host?query"), metrics::OmniboxInputType::UNKNOWN },
  83     { ASCIIToUTF16("host#ref"), metrics::OmniboxInputType::UNKNOWN },
  84     { ASCIIToUTF16("host/path?query"), metrics::OmniboxInputType::URL },
  85     { ASCIIToUTF16("host/path#ref"), metrics::OmniboxInputType::URL },
  86     { ASCIIToUTF16("en.wikipedia.org/wiki/Jim Beam"),
  87       metrics::OmniboxInputType::URL },
  88     // In Chrome itself, mailto: will get handled by ShellExecute, but in
  89     // unittest mode, we don't have the data loaded in the external protocol
  90     // handler to know this.
  91     // { ASCIIToUTF16("mailto:abuse@foo.com"), metrics::OmniboxInputType::URL },
  92     { ASCIIToUTF16("view-source:http://www.foo.com/"),
  93       metrics::OmniboxInputType::URL },
  94     { ASCIIToUTF16("javascript:alert(\"Hi there\");"),
  95       metrics::OmniboxInputType::URL },
  96 #if defined(OS_WIN)
  97     { ASCIIToUTF16("C:\\Program Files"), metrics::OmniboxInputType::URL },
  98     { ASCIIToUTF16("\\\\Server\\Folder\\File"),
  99       metrics::OmniboxInputType::URL },
 100 #endif  // defined(OS_WIN)
 101     { ASCIIToUTF16("http:foo"), metrics::OmniboxInputType::URL },
 102     { ASCIIToUTF16("http://foo"), metrics::OmniboxInputType::URL },
 103     { ASCIIToUTF16("http://foo.c"), metrics::OmniboxInputType::URL },
 104     { ASCIIToUTF16("http://foo.com"), metrics::OmniboxInputType::URL },
 105     { ASCIIToUTF16("http://foo_bar.com"), metrics::OmniboxInputType::URL },
 106     { ASCIIToUTF16("http://foo/bar%00"), metrics::OmniboxInputType::QUERY },
 107     { ASCIIToUTF16("http://foo/bar baz"), metrics::OmniboxInputType::URL },
 108     { ASCIIToUTF16("http://-foo.com"), metrics::OmniboxInputType::URL },
 109     { ASCIIToUTF16("http://foo-.com"), metrics::OmniboxInputType::URL },
 110     { ASCIIToUTF16("http://foo_.com"), metrics::OmniboxInputType::UNKNOWN },
 111     { ASCIIToUTF16("http://foo.-com"), metrics::OmniboxInputType::UNKNOWN },
 112     { ASCIIToUTF16("http://_foo_.com"), metrics::OmniboxInputType::UNKNOWN },
 113     { ASCIIToUTF16("http://foo.com:abc"), metrics::OmniboxInputType::QUERY },
 114     { ASCIIToUTF16("http://foo.com:123456"), metrics::OmniboxInputType::QUERY },
 115     { ASCIIToUTF16("http://1.2.3.4:abc"), metrics::OmniboxInputType::QUERY },
 116     { ASCIIToUTF16("http:user@foo.com"), metrics::OmniboxInputType::URL },
 117     { ASCIIToUTF16("http://user@foo.com"), metrics::OmniboxInputType::URL },
 118     { ASCIIToUTF16("http:user:pass@foo.com"), metrics::OmniboxInputType::URL },
 119     { ASCIIToUTF16("http://user:pass@foo.com"),
 120       metrics::OmniboxInputType::URL },
 121     { ASCIIToUTF16("http://1.2"), metrics::OmniboxInputType::URL },
 122     { ASCIIToUTF16("http://1.2/45"), metrics::OmniboxInputType::URL },
 123     { ASCIIToUTF16("http:ps/2 games"), metrics::OmniboxInputType::URL },
 124     { ASCIIToUTF16("https://foo.com"), metrics::OmniboxInputType::URL },
 125     { ASCIIToUTF16("127.0.0.1"), metrics::OmniboxInputType::URL },
 126     { ASCIIToUTF16("127.0.1"), metrics::OmniboxInputType::UNKNOWN },
 127     { ASCIIToUTF16("127.0.1/"), metrics::OmniboxInputType::URL },
 128     { ASCIIToUTF16("browser.tabs.closeButtons"),
 129       metrics::OmniboxInputType::UNKNOWN },
 130     { base::WideToUTF16(L"\u6d4b\u8bd5"), metrics::OmniboxInputType::UNKNOWN },
 131     { ASCIIToUTF16("[2001:]"), metrics::OmniboxInputType::QUERY },
 132     { ASCIIToUTF16("[2001:dB8::1]"), metrics::OmniboxInputType::URL },
 133     { ASCIIToUTF16("192.168.0.256"), metrics::OmniboxInputType::QUERY },
 134     { ASCIIToUTF16("[foo.com]"), metrics::OmniboxInputType::QUERY },
 135     { ASCIIToUTF16("filesystem:http://a.com/t/bar"),
 136       metrics::OmniboxInputType::URL },
 137     { ASCIIToUTF16("filesystem:http://a.com/"),
 138       metrics::OmniboxInputType::QUERY },
 139     { ASCIIToUTF16("filesystem:file://"), metrics::OmniboxInputType::QUERY },
 140     { ASCIIToUTF16("filesystem:http"), metrics::OmniboxInputType::QUERY },
 141     { ASCIIToUTF16("filesystem:"), metrics::OmniboxInputType::QUERY },
 142     { ASCIIToUTF16("chrome-search://"), metrics::OmniboxInputType::QUERY },
 143     { ASCIIToUTF16("chrome-devtools:"), metrics::OmniboxInputType::QUERY },
 144     { ASCIIToUTF16("about://f;"), metrics::OmniboxInputType::QUERY },
 145     { ASCIIToUTF16("://w"), metrics::OmniboxInputType::QUERY },
 146     { ASCIIToUTF16(":w"), metrics::OmniboxInputType::QUERY },
 147   };
 148
 149   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(input_cases); ++i) {
 150     SCOPED_TRACE(input_cases[i].input);
 151     AutocompleteInput input(input_cases[i].input, base::string16::npos,
 152                             base::string16(), GURL(),
 153                             OmniboxEventProto::INVALID_SPEC, true, false, true,
 154                             true, TestSchemeClassifier());
 155     EXPECT_EQ(input_cases[i].type, input.type());
 156   }
 157 }
 158
 159 TEST(AutocompleteInputTest, InputTypeWithDesiredTLD) {
 160   struct test_data {
 161     const base::string16 input;
 162     const metrics::OmniboxInputType::Type type;
 163     const std::string spec;  // Unused if not a URL.
 164   } input_cases[] = {
 165     { ASCIIToUTF16("401k"), metrics::OmniboxInputType::URL,
 166         std::string("http://www.401k.com/") },
 167     { ASCIIToUTF16("999999999999999"), metrics::OmniboxInputType::URL,
 168         std::string("http://www.999999999999999.com/") },
 169     { ASCIIToUTF16("x@y"), metrics::OmniboxInputType::URL,
 170         std::string("http://x@www.y.com/") },
 171     { ASCIIToUTF16("y/z z"), metrics::OmniboxInputType::URL,
 172         std::string("http://www.y.com/z%20z") },
 173     { ASCIIToUTF16("abc.com"), metrics::OmniboxInputType::URL,
 174         std::string("http://abc.com/") },
 175     { ASCIIToUTF16("foo bar"), metrics::OmniboxInputType::QUERY,
 176         std::string() },
 177   };
 178
 179   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(input_cases); ++i) {
 180     SCOPED_TRACE(input_cases[i].input);
 181     AutocompleteInput input(input_cases[i].input, base::string16::npos,
 182                             ASCIIToUTF16("com"), GURL(),
 183                             OmniboxEventProto::INVALID_SPEC, true, false, true,
 184                             true, TestSchemeClassifier());
 185     EXPECT_EQ(input_cases[i].type, input.type());
 186     if (input_cases[i].type == metrics::OmniboxInputType::URL)
 187       EXPECT_EQ(input_cases[i].spec, input.canonicalized_url().spec());
 188   }
 189 }
 190
 191 // This tests for a regression where certain input in the omnibox caused us to
 192 // crash. As long as the test completes without crashing, we're fine.
 193 TEST(AutocompleteInputTest, InputCrash) {
 194   AutocompleteInput input(base::WideToUTF16(L"\uff65@s"), base::string16::npos,
 195                           base::string16(), GURL(),
 196                           OmniboxEventProto::INVALID_SPEC, true, false,
 197                           true, true, TestSchemeClassifier());
 198 }
 199
 200 TEST(AutocompleteInputTest, ParseForEmphasizeComponent) {
 201   using url::Component;
 202   Component kInvalidComponent(0, -1);
 203   struct test_data {
 204     const base::string16 input;
 205     const Component scheme;
 206     const Component host;
 207   } input_cases[] = {
 208     { base::string16(), kInvalidComponent, kInvalidComponent },
 209     { ASCIIToUTF16("?"), kInvalidComponent, kInvalidComponent },
 210     { ASCIIToUTF16("?http://foo.com/bar"), kInvalidComponent,
 211         kInvalidComponent },
 212     { ASCIIToUTF16("foo/bar baz"), kInvalidComponent, Component(0, 3) },
 213     { ASCIIToUTF16("http://foo/bar baz"), Component(0, 4), Component(7, 3) },
 214     { ASCIIToUTF16("link:foo.com"), Component(0, 4), kInvalidComponent },
 215     { ASCIIToUTF16("www.foo.com:81"), kInvalidComponent, Component(0, 11) },
 216     { base::WideToUTF16(L"\u6d4b\u8bd5"), kInvalidComponent, Component(0, 2) },
 217     { ASCIIToUTF16("view-source:http://www.foo.com/"), Component(12, 4),
 218         Component(19, 11) },
 219     { ASCIIToUTF16("view-source:https://example.com/"),
 220       Component(12, 5), Component(20, 11) },
 221     { ASCIIToUTF16("view-source:www.foo.com"), kInvalidComponent,
 222         Component(12, 11) },
 223     { ASCIIToUTF16("view-source:"), Component(0, 11), kInvalidComponent },
 224     { ASCIIToUTF16("view-source:garbage"), kInvalidComponent,
 225         Component(12, 7) },
 226     { ASCIIToUTF16("view-source:http://http://foo"), Component(12, 4),
 227         Component(19, 4) },
 228     { ASCIIToUTF16("view-source:view-source:http://example.com/"),
 229         Component(12, 11), kInvalidComponent }
 230   };
 231
 232   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(input_cases); ++i) {
 233     SCOPED_TRACE(input_cases[i].input);
 234     Component scheme, host;
 235     AutocompleteInput::ParseForEmphasizeComponents(input_cases[i].input,
 236                                                    TestSchemeClassifier(),
 237                                                    &scheme,
 238                                                    &host);
 239     AutocompleteInput input(input_cases[i].input, base::string16::npos,
 240                             base::string16(), GURL(),
 241                             OmniboxEventProto::INVALID_SPEC, true,
 242                             false, true, true, TestSchemeClassifier());
 243     EXPECT_EQ(input_cases[i].scheme.begin, scheme.begin);
 244     EXPECT_EQ(input_cases[i].scheme.len, scheme.len);
 245     EXPECT_EQ(input_cases[i].host.begin, host.begin);
 246     EXPECT_EQ(input_cases[i].host.len, host.len);
 247   }
 248 }
 249
 250 TEST(AutocompleteInputTest, InputTypeWithCursorPosition) {
 251   struct test_data {
 252     const base::string16 input;
 253     size_t cursor_position;
 254     const base::string16 normalized_input;
 255     size_t normalized_cursor_position;
 256   } input_cases[] = {
 257     { ASCIIToUTF16("foo bar"), base::string16::npos,
 258       ASCIIToUTF16("foo bar"), base::string16::npos },
 259
 260     // regular case, no changes.
 261     { ASCIIToUTF16("foo bar"), 3, ASCIIToUTF16("foo bar"), 3 },
 262
 263     // extra leading space.
 264     { ASCIIToUTF16("  foo bar"), 3, ASCIIToUTF16("foo bar"), 1 },
 265     { ASCIIToUTF16("      foo bar"), 3, ASCIIToUTF16("foo bar"), 0 },
 266     { ASCIIToUTF16("      foo bar   "), 2, ASCIIToUTF16("foo bar   "), 0 },
 267
 268     // forced query.
 269     { ASCIIToUTF16("?foo bar"), 2, ASCIIToUTF16("foo bar"), 1 },
 270     { ASCIIToUTF16("  ?foo bar"), 4, ASCIIToUTF16("foo bar"), 1 },
 271     { ASCIIToUTF16("?  foo bar"), 4, ASCIIToUTF16("foo bar"), 1 },
 272     { ASCIIToUTF16("  ?  foo bar"), 6, ASCIIToUTF16("foo bar"), 1 },
 273   };
 274
 275   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(input_cases); ++i) {
 276     SCOPED_TRACE(input_cases[i].input);
 277     AutocompleteInput input(input_cases[i].input,
 278                             input_cases[i].cursor_position,
 279                             base::string16(), GURL(),
 280                             OmniboxEventProto::INVALID_SPEC,
 281                             true, false, true, true, TestSchemeClassifier());
 282     EXPECT_EQ(input_cases[i].normalized_input, input.text());
 283     EXPECT_EQ(input_cases[i].normalized_cursor_position,
 284               input.cursor_position());
 285   }
 286 }