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/third_party/mozilla/url_parse.h"
  16
  17 using base::ASCIIToUTF16;
  18 using metrics::OmniboxEventProto;
  19
  20 TEST(AutocompleteInputTest, InputType) {
  21   struct test_data {
  22     const base::string16 input;
  23     const metrics::OmniboxInputType::Type type;
  24   } input_cases[] = {
  25     { base::string16(), metrics::OmniboxInputType::INVALID },
  26     { ASCIIToUTF16("?"), metrics::OmniboxInputType::FORCED_QUERY },
  27     { ASCIIToUTF16("?foo"), metrics::OmniboxInputType::FORCED_QUERY },
  28     { ASCIIToUTF16("?foo bar"), metrics::OmniboxInputType::FORCED_QUERY },
  29     { ASCIIToUTF16("?http://foo.com/bar"),
  30       metrics::OmniboxInputType::FORCED_QUERY },
  31     { ASCIIToUTF16("foo"), metrics::OmniboxInputType::UNKNOWN },
  32     { ASCIIToUTF16("localhost"), metrics::OmniboxInputType::URL },
  33     { ASCIIToUTF16("foo._"), metrics::OmniboxInputType::QUERY },
  34     { ASCIIToUTF16("foo.c"), metrics::OmniboxInputType::UNKNOWN },
  35     { ASCIIToUTF16("foo.com"), metrics::OmniboxInputType::URL },
  36     { ASCIIToUTF16("-foo.com"), metrics::OmniboxInputType::URL },
  37     { ASCIIToUTF16("foo-.com"), metrics::OmniboxInputType::URL },
  38     { ASCIIToUTF16("foo_.com"), metrics::OmniboxInputType::URL },
  39     { ASCIIToUTF16("foo.-com"), metrics::OmniboxInputType::QUERY },
  40     { ASCIIToUTF16("foo/"), metrics::OmniboxInputType::URL },
  41     { ASCIIToUTF16("foo/bar"), metrics::OmniboxInputType::UNKNOWN },
  42     { ASCIIToUTF16("foo/bar%00"), metrics::OmniboxInputType::QUERY },
  43     { ASCIIToUTF16("foo/bar/"), metrics::OmniboxInputType::URL },
  44     { ASCIIToUTF16("foo/bar baz\\"), metrics::OmniboxInputType::URL },
  45     { ASCIIToUTF16("foo.com/bar"), metrics::OmniboxInputType::URL },
  46     { ASCIIToUTF16("foo;bar"), metrics::OmniboxInputType::UNKNOWN },
  47     { ASCIIToUTF16("foo/bar baz"), metrics::OmniboxInputType::UNKNOWN },
  48     { ASCIIToUTF16("foo bar.com"), metrics::OmniboxInputType::QUERY },
  49     { ASCIIToUTF16("foo bar"), metrics::OmniboxInputType::QUERY },
  50     { ASCIIToUTF16("foo+bar"), metrics::OmniboxInputType::QUERY },
  51     { ASCIIToUTF16("foo+bar.com"), metrics::OmniboxInputType::UNKNOWN },
  52     { ASCIIToUTF16("\"foo:bar\""), metrics::OmniboxInputType::QUERY },
  53     { ASCIIToUTF16("link:foo.com"), metrics::OmniboxInputType::UNKNOWN },
  54     { ASCIIToUTF16("foo:81"), metrics::OmniboxInputType::URL },
  55     { ASCIIToUTF16("localhost:8080"), metrics::OmniboxInputType::URL },
  56     { ASCIIToUTF16("www.foo.com:81"), metrics::OmniboxInputType::URL },
  57     { ASCIIToUTF16("foo.com:123456"), metrics::OmniboxInputType::QUERY },
  58     { ASCIIToUTF16("foo.com:abc"), metrics::OmniboxInputType::QUERY },
  59     { ASCIIToUTF16("1.2.3.4:abc"), metrics::OmniboxInputType::QUERY },
  60     { ASCIIToUTF16("user@foo"), metrics::OmniboxInputType::UNKNOWN },
  61     { ASCIIToUTF16("user@foo.com"), metrics::OmniboxInputType::UNKNOWN },
  62     { ASCIIToUTF16("user@foo/z"), metrics::OmniboxInputType::URL },
  63     { ASCIIToUTF16("user@foo/z z"), metrics::OmniboxInputType::URL },
  64     { ASCIIToUTF16("user@foo.com/z"), metrics::OmniboxInputType::URL },
  65     { ASCIIToUTF16("user:pass@"), metrics::OmniboxInputType::UNKNOWN },
  66     { ASCIIToUTF16("user:pass@!foo.com"), metrics::OmniboxInputType::UNKNOWN },
  67     { ASCIIToUTF16("user:pass@foo"), metrics::OmniboxInputType::URL },
  68     { ASCIIToUTF16("user:pass@foo.c"), metrics::OmniboxInputType::URL },
  69     { ASCIIToUTF16("user:pass@foo.com"), metrics::OmniboxInputType::URL },
  70     { ASCIIToUTF16("user:pass@foo.com:81"), metrics::OmniboxInputType::URL },
  71     { ASCIIToUTF16("user:pass@foo:81"), metrics::OmniboxInputType::URL },
  72     { ASCIIToUTF16(".1"), metrics::OmniboxInputType::QUERY },
  73     { ASCIIToUTF16(".1/3"), metrics::OmniboxInputType::QUERY },
  74     { ASCIIToUTF16("1.2"), metrics::OmniboxInputType::QUERY },
  75     { ASCIIToUTF16(".1.2"), metrics::OmniboxInputType::UNKNOWN },
  76     { ASCIIToUTF16("1.2/"), metrics::OmniboxInputType::URL },
  77     { ASCIIToUTF16("1.2/45"), metrics::OmniboxInputType::QUERY },
  78     { ASCIIToUTF16("6008/32768"), metrics::OmniboxInputType::QUERY },
  79     { ASCIIToUTF16("12345678/"), metrics::OmniboxInputType::QUERY },
  80     { ASCIIToUTF16("123456789/"), metrics::OmniboxInputType::URL },
  81     { ASCIIToUTF16("1.2:45"), metrics::OmniboxInputType::QUERY },
  82     { ASCIIToUTF16("user@1.2:45"), metrics::OmniboxInputType::QUERY },
  83     { ASCIIToUTF16("user@foo:45"), metrics::OmniboxInputType::URL },
  84     { ASCIIToUTF16("user:pass@1.2:45"), metrics::OmniboxInputType::URL },
  85     { ASCIIToUTF16("host?query"), metrics::OmniboxInputType::UNKNOWN },
  86     { ASCIIToUTF16("host#ref"), metrics::OmniboxInputType::QUERY },
  87     { ASCIIToUTF16("host#"), metrics::OmniboxInputType::QUERY },
  88     { ASCIIToUTF16("host#ref"), metrics::OmniboxInputType::QUERY },
  89     { ASCIIToUTF16("host# ref"), metrics::OmniboxInputType::QUERY },
  90     { ASCIIToUTF16("host/#ref"), metrics::OmniboxInputType::URL },
  91     { ASCIIToUTF16("host/?#ref"), metrics::OmniboxInputType::URL },
  92     { ASCIIToUTF16("host/?#"), metrics::OmniboxInputType::URL },
  93     { ASCIIToUTF16("host.com#ref"), metrics::OmniboxInputType::URL },
  94     { ASCIIToUTF16("http://host#ref"), metrics::OmniboxInputType::URL },
  95     { ASCIIToUTF16("host/path?query"), metrics::OmniboxInputType::URL },
  96     { ASCIIToUTF16("host/path#ref"), metrics::OmniboxInputType::URL },
  97     { ASCIIToUTF16("en.wikipedia.org/wiki/Jim Beam"),
  98       metrics::OmniboxInputType::URL },
  99     // In Chrome itself, mailto: will get handled by ShellExecute, but in
 100     // unittest mode, we don't have the data loaded in the external protocol
 101     // handler to know this.
 102     // { ASCIIToUTF16("mailto:abuse@foo.com"), metrics::OmniboxInputType::URL },
 103     { ASCIIToUTF16("view-source:http://www.foo.com/"),
 104       metrics::OmniboxInputType::URL },
 105     { ASCIIToUTF16("javascript:alert(\"Hi there\");"),
 106       metrics::OmniboxInputType::URL },
 107 #if defined(OS_WIN)
 108     { ASCIIToUTF16("C:\\Program Files"), metrics::OmniboxInputType::URL },
 109     { ASCIIToUTF16("\\\\Server\\Folder\\File"),
 110       metrics::OmniboxInputType::URL },
 111 #endif  // defined(OS_WIN)
 112     { ASCIIToUTF16("http:foo"), metrics::OmniboxInputType::URL },
 113     { ASCIIToUTF16("http://foo"), metrics::OmniboxInputType::URL },
 114     { ASCIIToUTF16("http://foo._"), metrics::OmniboxInputType::UNKNOWN },
 115     { ASCIIToUTF16("http://foo.c"), metrics::OmniboxInputType::URL },
 116     { ASCIIToUTF16("http://foo.com"), metrics::OmniboxInputType::URL },
 117     { ASCIIToUTF16("http://foo_bar.com"), metrics::OmniboxInputType::URL },
 118     { ASCIIToUTF16("http://foo/bar%00"), metrics::OmniboxInputType::QUERY },
 119     { ASCIIToUTF16("http://foo/bar baz"), metrics::OmniboxInputType::URL },
 120     { ASCIIToUTF16("http://-foo.com"), metrics::OmniboxInputType::URL },
 121     { ASCIIToUTF16("http://foo-.com"), metrics::OmniboxInputType::URL },
 122     { ASCIIToUTF16("http://foo_.com"), metrics::OmniboxInputType::URL },
 123     { ASCIIToUTF16("http://foo.-com"), metrics::OmniboxInputType::UNKNOWN },
 124     { ASCIIToUTF16("http://_foo_.com"), metrics::OmniboxInputType::URL },
 125     { ASCIIToUTF16("http://foo.com:abc"), metrics::OmniboxInputType::QUERY },
 126     { ASCIIToUTF16("http://foo.com:123456"), metrics::OmniboxInputType::QUERY },
 127     { ASCIIToUTF16("http://1.2.3.4:abc"), metrics::OmniboxInputType::QUERY },
 128     { ASCIIToUTF16("http:user@foo.com"), metrics::OmniboxInputType::URL },
 129     { ASCIIToUTF16("http://user@foo.com"), metrics::OmniboxInputType::URL },
 130     { ASCIIToUTF16("http://user:pass@foo"), metrics::OmniboxInputType::URL },
 131     { ASCIIToUTF16("http:user:pass@foo.com"), metrics::OmniboxInputType::URL },
 132     { ASCIIToUTF16("http://user:pass@foo.com"),
 133       metrics::OmniboxInputType::URL },
 134     { ASCIIToUTF16("http://1.2"), metrics::OmniboxInputType::URL },
 135     { ASCIIToUTF16("http:user@1.2"), metrics::OmniboxInputType::URL },
 136     { ASCIIToUTF16("http://1.2/45"), metrics::OmniboxInputType::URL },
 137     { ASCIIToUTF16("http:ps/2 games"), metrics::OmniboxInputType::URL },
 138     { ASCIIToUTF16("https://foo.com"), metrics::OmniboxInputType::URL },
 139     { ASCIIToUTF16("127.0.0.1"), metrics::OmniboxInputType::URL },
 140     { ASCIIToUTF16("127.0.1"), metrics::OmniboxInputType::QUERY },
 141     { ASCIIToUTF16("127.0.1/"), metrics::OmniboxInputType::URL },
 142     { ASCIIToUTF16("0.0.0"), metrics::OmniboxInputType::QUERY },
 143     { ASCIIToUTF16("0.0.0.0"), metrics::OmniboxInputType::URL },
 144     { ASCIIToUTF16("0.0.0.1"), metrics::OmniboxInputType::QUERY },
 145     { ASCIIToUTF16("http://0.0.0.1/"), metrics::OmniboxInputType::QUERY },
 146     { ASCIIToUTF16("browser.tabs.closeButtons"),
 147       metrics::OmniboxInputType::UNKNOWN },
 148     { base::WideToUTF16(L"\u6d4b\u8bd5"), metrics::OmniboxInputType::UNKNOWN },
 149     { ASCIIToUTF16("[2001:]"), metrics::OmniboxInputType::QUERY },
 150     { ASCIIToUTF16("[2001:dB8::1]"), metrics::OmniboxInputType::URL },
 151     { ASCIIToUTF16("192.168.0.256"), metrics::OmniboxInputType::QUERY },
 152     { ASCIIToUTF16("[foo.com]"), metrics::OmniboxInputType::QUERY },
 153     { ASCIIToUTF16("filesystem:http://a.com/t/bar"),
 154       metrics::OmniboxInputType::URL },
 155     { ASCIIToUTF16("filesystem:http://a.com/"),
 156       metrics::OmniboxInputType::QUERY },
 157     { ASCIIToUTF16("filesystem:file://"), metrics::OmniboxInputType::QUERY },
 158     { ASCIIToUTF16("filesystem:http"), metrics::OmniboxInputType::QUERY },
 159     { ASCIIToUTF16("filesystem:"), metrics::OmniboxInputType::QUERY },
 160     { ASCIIToUTF16("chrome-search://"), metrics::OmniboxInputType::QUERY },
 161     { ASCIIToUTF16("chrome-devtools:"), metrics::OmniboxInputType::QUERY },
 162     { ASCIIToUTF16("about://f;"), metrics::OmniboxInputType::QUERY },
 163     { ASCIIToUTF16("://w"), metrics::OmniboxInputType::QUERY },
 164     { ASCIIToUTF16(":w"), metrics::OmniboxInputType::QUERY },
 165     { base::WideToUTF16(L".\u062A"), metrics::OmniboxInputType::UNKNOWN },
 166   };
 167
 168   for (size_t i = 0; i < arraysize(input_cases); ++i) {
 169     SCOPED_TRACE(input_cases[i].input);
 170     AutocompleteInput input(input_cases[i].input, base::string16::npos,
 171                             std::string(), GURL(),
 172                             OmniboxEventProto::INVALID_SPEC, true, false, true,
 173                             true, false, TestSchemeClassifier());
 174     EXPECT_EQ(input_cases[i].type, input.type());
 175   }
 176 }
 177
 178 TEST(AutocompleteInputTest, InputTypeWithDesiredTLD) {
 179   struct test_data {
 180     const base::string16 input;
 181     const metrics::OmniboxInputType::Type type;
 182     const std::string spec;  // Unused if not a URL.
 183   } input_cases[] = {
 184     { ASCIIToUTF16("401k"), metrics::OmniboxInputType::URL,
 185         std::string("http://www.401k.com/") },
 186     { ASCIIToUTF16("56"), metrics::OmniboxInputType::URL,
 187         std::string("http://www.56.com/") },
 188     { ASCIIToUTF16("1.2"), metrics::OmniboxInputType::URL,
 189         std::string("http://www.1.2.com/") },
 190     { ASCIIToUTF16("1.2/3.4"), metrics::OmniboxInputType::URL,
 191         std::string("http://www.1.2.com/3.4") },
 192     { ASCIIToUTF16("192.168.0.1"), metrics::OmniboxInputType::URL,
 193         std::string("http://www.192.168.0.1.com/") },
 194     { ASCIIToUTF16("999999999999999"), metrics::OmniboxInputType::URL,
 195         std::string("http://www.999999999999999.com/") },
 196     { ASCIIToUTF16("x@y"), metrics::OmniboxInputType::URL,
 197         std::string("http://x@www.y.com/") },
 198     { ASCIIToUTF16("x@y.com"), metrics::OmniboxInputType::URL,
 199         std::string("http://x@y.com/") },
 200     { ASCIIToUTF16("y/z z"), metrics::OmniboxInputType::URL,
 201         std::string("http://www.y.com/z%20z") },
 202     { ASCIIToUTF16("abc.com"), metrics::OmniboxInputType::URL,
 203         std::string("http://abc.com/") },
 204     { ASCIIToUTF16("foo bar"), metrics::OmniboxInputType::QUERY,
 205         std::string() },
 206   };
 207
 208   for (size_t i = 0; i < arraysize(input_cases); ++i) {
 209     SCOPED_TRACE(input_cases[i].input);
 210     AutocompleteInput input(input_cases[i].input, base::string16::npos, "com",
 211                             GURL(), OmniboxEventProto::INVALID_SPEC, true,
 212                             false, true, true, false, TestSchemeClassifier());
 213     EXPECT_EQ(input_cases[i].type, input.type());
 214     if (input_cases[i].type == metrics::OmniboxInputType::URL)
 215       EXPECT_EQ(input_cases[i].spec, input.canonicalized_url().spec());
 216   }
 217 }
 218
 219 // This tests for a regression where certain input in the omnibox caused us to
 220 // crash. As long as the test completes without crashing, we're fine.
 221 TEST(AutocompleteInputTest, InputCrash) {
 222   AutocompleteInput input(base::WideToUTF16(L"\uff65@s"), base::string16::npos,
 223                           std::string(), GURL(),
 224                           OmniboxEventProto::INVALID_SPEC, true, false, true,
 225                           true, false, TestSchemeClassifier());
 226 }
 227
 228 TEST(AutocompleteInputTest, ParseForEmphasizeComponent) {
 229   using url::Component;
 230   Component kInvalidComponent(0, -1);
 231   struct test_data {
 232     const base::string16 input;
 233     const Component scheme;
 234     const Component host;
 235   } input_cases[] = {
 236     { base::string16(), kInvalidComponent, kInvalidComponent },
 237     { ASCIIToUTF16("?"), kInvalidComponent, kInvalidComponent },
 238     { ASCIIToUTF16("?http://foo.com/bar"), kInvalidComponent,
 239         kInvalidComponent },
 240     { ASCIIToUTF16("foo/bar baz"), kInvalidComponent, Component(0, 3) },
 241     { ASCIIToUTF16("http://foo/bar baz"), Component(0, 4), Component(7, 3) },
 242     { ASCIIToUTF16("link:foo.com"), Component(0, 4), kInvalidComponent },
 243     { ASCIIToUTF16("www.foo.com:81"), kInvalidComponent, Component(0, 11) },
 244     { base::WideToUTF16(L"\u6d4b\u8bd5"), kInvalidComponent, Component(0, 2) },
 245     { ASCIIToUTF16("view-source:http://www.foo.com/"), Component(12, 4),
 246         Component(19, 11) },
 247     { ASCIIToUTF16("view-source:https://example.com/"),
 248       Component(12, 5), Component(20, 11) },
 249     { ASCIIToUTF16("view-source:www.foo.com"), kInvalidComponent,
 250         Component(12, 11) },
 251     { ASCIIToUTF16("view-source:"), Component(0, 11), kInvalidComponent },
 252     { ASCIIToUTF16("view-source:garbage"), kInvalidComponent,
 253         Component(12, 7) },
 254     { ASCIIToUTF16("view-source:http://http://foo"), Component(12, 4),
 255         Component(19, 4) },
 256     { ASCIIToUTF16("view-source:view-source:http://example.com/"),
 257         Component(12, 11), kInvalidComponent }
 258   };
 259
 260   for (size_t i = 0; i < arraysize(input_cases); ++i) {
 261     SCOPED_TRACE(input_cases[i].input);
 262     Component scheme, host;
 263     AutocompleteInput::ParseForEmphasizeComponents(input_cases[i].input,
 264                                                    TestSchemeClassifier(),
 265                                                    &scheme,
 266                                                    &host);
 267     AutocompleteInput input(input_cases[i].input, base::string16::npos,
 268                             std::string(), GURL(),
 269                             OmniboxEventProto::INVALID_SPEC, true, false, true,
 270                             true, false, TestSchemeClassifier());
 271     EXPECT_EQ(input_cases[i].scheme.begin, scheme.begin);
 272     EXPECT_EQ(input_cases[i].scheme.len, scheme.len);
 273     EXPECT_EQ(input_cases[i].host.begin, host.begin);
 274     EXPECT_EQ(input_cases[i].host.len, host.len);
 275   }
 276 }
 277
 278 TEST(AutocompleteInputTest, InputTypeWithCursorPosition) {
 279   struct test_data {
 280     const base::string16 input;
 281     size_t cursor_position;
 282     const base::string16 normalized_input;
 283     size_t normalized_cursor_position;
 284   } input_cases[] = {
 285     { ASCIIToUTF16("foo bar"), base::string16::npos,
 286       ASCIIToUTF16("foo bar"), base::string16::npos },
 287
 288     // regular case, no changes.
 289     { ASCIIToUTF16("foo bar"), 3, ASCIIToUTF16("foo bar"), 3 },
 290
 291     // extra leading space.
 292     { ASCIIToUTF16("  foo bar"), 3, ASCIIToUTF16("foo bar"), 1 },
 293     { ASCIIToUTF16("      foo bar"), 3, ASCIIToUTF16("foo bar"), 0 },
 294     { ASCIIToUTF16("      foo bar   "), 2, ASCIIToUTF16("foo bar   "), 0 },
 295
 296     // forced query.
 297     { ASCIIToUTF16("?foo bar"), 2, ASCIIToUTF16("foo bar"), 1 },
 298     { ASCIIToUTF16("  ?foo bar"), 4, ASCIIToUTF16("foo bar"), 1 },
 299     { ASCIIToUTF16("?  foo bar"), 4, ASCIIToUTF16("foo bar"), 1 },
 300     { ASCIIToUTF16("  ?  foo bar"), 6, ASCIIToUTF16("foo bar"), 1 },
 301   };
 302
 303   for (size_t i = 0; i < arraysize(input_cases); ++i) {
 304     SCOPED_TRACE(input_cases[i].input);
 305     AutocompleteInput input(input_cases[i].input,
 306                             input_cases[i].cursor_position, std::string(),
 307                             GURL(), OmniboxEventProto::INVALID_SPEC, true,
 308                             false, true, true, false, TestSchemeClassifier());
 309     EXPECT_EQ(input_cases[i].normalized_input, input.text());
 310     EXPECT_EQ(input_cases[i].normalized_cursor_position,
 311               input.cursor_position());
 312   }
 313 }