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