uri-test.cpp

   1 #include "uri.hpp"
   2
   3 #include <glog/logging.h>
   4
   5 int test_good()
   6 {
   7   struct test_case {
   8     char const* uri;
   9     uri::components parts;
  10   };
  11
  12   // clang-format off
  13   test_case tests[] = {
  14   {"foo://dude@example.com:8042/over/there?name=ferret#nose",
  15   {"foo", "dude@example.com:8042", "dude", "example.com", "8042", "/over/there", "name=ferret", "nose", }, },
  16
  17   {"foo://example.com:8042/over/there?name=ferret#nose",
  18   {"foo", "example.com:8042", {}, "example.com", "8042", "/over/there", "name=ferret", "nose", }, },
  19
  20   {"ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm",
  21   {"ftp", "cnn.example.com&story=breaking_news@10.0.0.1", "cnn.example.com&story=breaking_news", "10.0.0.1", {}, "/top_story.htm", {}, {}, }, },
  22
  23   {"ftp://foo.bar/baz",
  24   {"ftp", "foo.bar", {}, "foo.bar", {}, "/baz", {}, {}, }, },
  25
  26   {"ftp://ftp.is.co.za/rfc/rfc1808.txt",
  27   {"ftp", "ftp.is.co.za", {}, "ftp.is.co.za", {}, "/rfc/rfc1808.txt", {}, {}, }, },
  28
  29   {"http://-.~_!$&'()*+,;=:%40:80%2f::::::@example.com",
  30   {"http", "-.~_!$&'()*+,;=:%40:80%2f::::::@example.com", "-.~_!$&'()*+,;=:%40:80%2f::::::", "example.com", {}, "", {}, {}, }, },
  31
  32   {"http://1337.net",
  33   {"http", "1337.net", {}, "1337.net", {}, "", {}, {}, }, },
  34
  35   {"http://142.42.1.1/",
  36   {"http", "142.42.1.1", {}, "142.42.1.1", {}, "/", {}, {}, }, },
  37
  38   {"http://142.42.1.1:8080/",
  39   {"http", "142.42.1.1:8080", {}, "142.42.1.1", "8080", "/", {}, {}, }, },
  40
  41   {"http://223.255.255.254",
  42   {"http", "223.255.255.254", {}, "223.255.255.254", {}, "", {}, {}, }, },
  43
  44   {"http://a.b-c.de",
  45   {"http", "a.b-c.de", {}, "a.b-c.de", {}, "", {}, {}, }, },
  46
  47   {"http://code.google.com/events/#&product=browser",
  48   {"http", "code.google.com", {}, "code.google.com", {}, "/events/", {}, "&product=browser", }, },
  49
  50   {"http://example.com",
  51   {"http", "example.com", {}, "example.com", {}, "", {}, {}, }, },
  52
  53   {"http://example.com/",
  54   {"http", "example.com", {}, "example.com", {}, "/", {}, {}, }, },
  55
  56   {"http://example.com:",
  57   {"http", "example.com:", {}, "example.com", "", "", {}, {}, }, },
  58
  59   {"http://example.com:/",
  60   {"http", "example.com:", {}, "example.com", "", "/", {}, {}, }, },
  61
  62   {"http://example.com:80",
  63   {"http", "example.com:80", {}, "example.com", "80", "", {}, {}, }, },
  64
  65   {"http://example.com:80/",
  66   {"http", "example.com:80", {}, "example.com", "80", "/", {}, {}, }, },
  67
  68   {"http://foo.bar/?q=Test%20URL-encoded%20stuff",
  69   {"http", "foo.bar", {}, "foo.bar", {}, "/", "q=Test%20URL-encoded%20stuff", {}, }, },
  70
  71   {"http://foo.com/(something)?after=parens",
  72   {"http", "foo.com", {}, "foo.com", {}, "/(something)", "after=parens", {}, }, },
  73
  74   {"http://foo.com/blah_(wikipedia)#cite-1",
  75   {"http", "foo.com", {}, "foo.com", {}, "/blah_(wikipedia)", {}, "cite-1", }, },
  76
  77   {"http://foo.com/blah_(wikipedia)_blah#cite-1",
  78   {"http", "foo.com", {}, "foo.com", {}, "/blah_(wikipedia)_blah", {}, "cite-1", }, },
  79
  80   {"http://foo.com/blah_blah",
  81   {"http", "foo.com", {}, "foo.com", {}, "/blah_blah", {}, {}, }, },
  82
  83   {"http://foo.com/blah_blah/",
  84   {"http", "foo.com", {}, "foo.com", {}, "/blah_blah/", {}, {}, }, },
  85
  86   {"http://foo.com/blah_blah_(wikipedia)",
  87   {"http", "foo.com", {}, "foo.com", {}, "/blah_blah_(wikipedia)", {}, {}, }, },
  88
  89   {"http://foo.com/blah_blah_(wikipedia)_(again)",
  90   {"http", "foo.com", {}, "foo.com", {}, "/blah_blah_(wikipedia)_(again)", {}, {}, }, },
  91
  92   {"http://foo.com/unicode_(✪)_in_parens",
  93   {"http", "foo.com", {}, "foo.com", {}, "/unicode_(✪)_in_parens", {}, {}, }, },
  94
  95   {"http://j.mp",
  96   {"http", "j.mp", {}, "j.mp", {}, "", {}, {}, }, },
  97
  98   {"http://userid:password@example.com",
  99   {"http", "userid:password@example.com", "userid:password", "example.com", {}, "", {}, {}, }, },
 100
 101   {"http://userid:password@example.com/",
 102   {"http", "userid:password@example.com", "userid:password", "example.com", {}, "/", {}, {}, }, },
 103
 104   {"http://userid:password@example.com:8080",
 105   {"http", "userid:password@example.com:8080", "userid:password", "example.com", "8080", "", {}, {}, }, },
 106
 107   {"http://userid:password@example.com:8080/",
 108   {"http", "userid:password@example.com:8080", "userid:password", "example.com", "8080", "/", {}, {}, }, },
 109
 110   {"http://userid@example.com",
 111   {"http", "userid@example.com", "userid", "example.com", {}, "", {}, {}, }, },
 112
 113   {"http://userid@example.com/",
 114   {"http", "userid@example.com", "userid", "example.com", {}, "/", {}, {}, }, },
 115
 116   {"http://userid@example.com:8080",
 117   {"http", "userid@example.com:8080", "userid", "example.com", "8080", "", {}, {}, }, },
 118
 119   {"http://userid@example.com:8080/",
 120   {"http", "userid@example.com:8080", "userid", "example.com", "8080", "/", {}, {}, }, },
 121
 122   {"http://www.example.com/wpstyle/?p=364",
 123   {"http", "www.example.com", {}, "www.example.com", {}, "/wpstyle/", "p=364", {}, }, },
 124
 125   {"http://www.ics.uci.edu/pub/ietf/uri/#Related",
 126   {"http", "www.ics.uci.edu", {}, "www.ics.uci.edu", {}, "/pub/ietf/uri/", {}, "Related", }, },
 127
 128   {"http://www.ietf.org/rfc/rfc2396.txt",
 129   {"http", "www.ietf.org", {}, "www.ietf.org", {}, "/rfc/rfc2396.txt", {}, {}, }, },
 130
 131   {"http://مثال.إختبار",
 132   {"http", "مثال.إختبار", {}, "مثال.إختبار", {}, "", {}, {}, }, },
 133
 134   {"http://उदाहरण.परीक्षा",
 135   {"http", "उदाहरण.परीक्षा", {}, "उदाहरण.परीक्षा", {}, "", {}, {}, }, },
 136
 137   {"http://⌘.ws",
 138   {"http", "⌘.ws", {}, "⌘.ws", {}, "", {}, {}, }, },
 139
 140   {"http://⌘.ws/",
 141   {"http", "⌘.ws", {}, "⌘.ws", {}, "/", {}, {}, }, },
 142
 143   {"http://☺.damowmow.com/",
 144   {"http", "☺.damowmow.com", {}, "☺.damowmow.com", {}, "/", {}, {}, }, },
 145
 146   {"http://✪df.ws/123",
 147   {"http", "✪df.ws", {}, "✪df.ws", {}, "/123", {}, {}, }, },
 148
 149   {"http://➡.ws/䨹",
 150   {"http", "➡.ws", {}, "➡.ws", {}, "/䨹", {}, {}, }, },
 151
 152   {"http://例子.测试",
 153   {"http", "例子.测试", {}, "例子.测试", {}, "", {}, {}, }, },
 154
 155   {"https://tools.ietf.org/html/rfc3986#appendix-B",
 156   {"https", "tools.ietf.org", {}, "tools.ietf.org", {}, "/html/rfc3986", {}, "appendix-B", }, },
 157
 158   {"https://www.example.com/foo/?bar=baz&inga=42&quux",
 159   {"https", "www.example.com", {}, "www.example.com", {}, "/foo/", "bar=baz&inga=42&quux", {}, }, },
 160
 161   {"https://xn%2D%2Dui8h%2Edigilicious%2Ecom/",
 162   {"https", "xn%2D%2Dui8h%2Edigilicious%2Ecom", {}, "xn%2D%2Dui8h%2Edigilicious%2Ecom", {}, "/", {}, {}, }, },
 163
 164   {"https://xn--ui8h%2Edigilicious%2Ecom/",
 165   {"https", "xn--ui8h%2Edigilicious%2Ecom", {}, "xn--ui8h%2Edigilicious%2Ecom", {}, "/", {}, {}, }, },
 166
 167   {"https://xn--ui8h.digilicious.com/",
 168   {"https", "xn--ui8h.digilicious.com", {}, "xn--ui8h.digilicious.com", {}, "/", {}, {}, }, },
 169
 170   {"https://🍔.digilicious.com/",
 171   {"https", "🍔.digilicious.com", {}, "🍔.digilicious.com", {}, "/", {}, {}, }, },
 172
 173   {"ldap://[2001:db8::7]/c=GB?objectClass?one",
 174   {"ldap", "[2001:db8::7]", {}, "[2001:db8::7]", {}, "/c=GB", "objectClass?one", {}, }, },
 175
 176   {"mailto:John.Doe@example.com",
 177   {"mailto", {}, {}, {}, {}, "John.Doe@example.com", {}, {}, }, },
 178
 179   {"news:comp.infosystems.www.servers.unix",
 180   {"news", {}, {}, {}, {}, "comp.infosystems.www.servers.unix", {}, {}, }, },
 181
 182   {"tel:+1-816-555-1212",
 183   {"tel", {}, {}, {}, {}, "+1-816-555-1212", {}, {}, }, },
 184
 185   {"telnet://192.0.2.16:80/",
 186   {"telnet", "192.0.2.16:80", {}, "192.0.2.16", "80", "/", {}, {}, }, },
 187
 188   {"urn:example:animal:ferret:nose",
 189   {"urn", {}, {}, {}, {}, "example:animal:ferret:nose", {}, {}, }, },
 190
 191   {"urn:oasis:names:specification:docbook:dtd:xml:4.1.2",
 192   {"urn", {}, {}, {}, {}, "oasis:names:specification:docbook:dtd:xml:4.1.2", {}, {}, }, },
 193   };
 194   // clang-format on
 195
 196   auto failures = 0;
 197
 198   for (auto&& test : tests) {
 199     uri::generic u{test.uri};
 200     if (test.parts != u.parts()) {
 201       std::cerr << test.uri << " failed to check\n";
 202
 203       std::cout << "URL:\n";
 204       if (u.scheme())
 205         std::cout << "scheme()     == " << *u.scheme() << '\n';
 206       if (u.authority())
 207         std::cout << "authority()  == " << *u.authority() << '\n';
 208       if (u.userinfo())
 209         std::cout << "userinfo()   == " << *u.userinfo() << '\n';
 210       if (u.host())
 211         std::cout << "host()       == " << *u.host() << '\n';
 212       if (u.port())
 213         std::cout << "port()       == " << *u.port() << '\n';
 214       if (u.path())
 215         std::cout << "path()       == " << *u.path() << '\n';
 216       if (u.query())
 217         std::cout << "query()      == " << *u.query() << '\n';
 218       if (u.fragment())
 219         std::cout << "fragment()   == " << *u.fragment() << '\n';
 220
 221       std::cout << "\ntest:\n";
 222
 223       if (test.parts.scheme)
 224         std::cout << "scheme     == " << *test.parts.scheme << '\n';
 225       if (test.parts.authority)
 226         std::cout << "authority  == " << *test.parts.authority << '\n';
 227       if (test.parts.userinfo)
 228         std::cout << "userinfo   == " << *test.parts.userinfo << '\n';
 229       if (test.parts.host)
 230         std::cout << "host       == " << *test.parts.host << '\n';
 231       if (test.parts.port)
 232         std::cout << "port       == " << *test.parts.port << '\n';
 233       if (test.parts.path)
 234         std::cout << "path       == " << *test.parts.path << '\n';
 235       if (test.parts.query)
 236         std::cout << "query      == " << *test.parts.query << '\n';
 237       if (test.parts.fragment)
 238         std::cout << "fragment   == " << *test.parts.fragment << '\n';
 239
 240       std::cout << "\n";
 241
 242       ++failures;
 243     }
 244   }
 245
 246   return failures;
 247 }
 248
 249 int test_bad()
 250 {
 251   // Verify a bunch of bad URIs all throw exceptions.
 252   auto failures = 0;
 253
 254   char const* bad_uris[]{
 255       "http://",
 256       "http://.",
 257       "http://..",
 258       "http://../",
 259       "http://?",
 260       "http://?\?",
 261       "http://?\?/",
 262       "http://#",
 263       "http://##",
 264       "http://##/",
 265       "http://foo.bar?q=Spaces should be encoded",
 266       "//",
 267       "//a",
 268       "///a",
 269       "///",
 270       "http:///a",
 271       "foo.com",
 272       "http:// shouldfail.com",
 273       ":// should fail",
 274       "http://foo.bar/foo(bar)baz quux",
 275       "http://-error-.invalid/",
 276       "http://-a.b.co",
 277       "http://a.b-.co",
 278       "http://1.1.1.1.1",
 279       "http://.www.foo.bar/",
 280       "http://.www.foo.bar./",
 281   };
 282
 283   // I have to confess, I don't know what's wrong with these:
 284
 285   //  "ftps://foo.bar/",
 286   //  "http://a.b--c.de/",
 287   //  "rdar://1234",
 288   //  "h://test",
 289   //  "http://0.0.0.0",
 290   //  "http://10.1.1.0",
 291   //  "http://10.1.1.255",
 292   //  "http://224.1.1.1",
 293   //  "http://123.123.123",
 294   //  "http://3628126748",
 295   //  "http://10.1.1.1",
 296   //  "http://10.1.1.254",
 297   //  "http://www.foo.bar./",
 298
 299   for (auto uri : bad_uris) {
 300     try {
 301       uri::generic u{uri};
 302       LOG(ERROR) << "should not parse \"" << uri << "\" as \"" << u << "\"\n";
 303       ++failures;
 304     }
 305     catch (uri::syntax_error const& e) {
 306       // all good
 307     }
 308     catch (std::exception const& e) {
 309       LOG(FATAL) << "unexpected exception " << e.what() << '\n';
 310     }
 311   }
 312
 313   return failures;
 314 }
 315
 316 int test_resolution()
 317 {
 318   struct test_case {
 319     char const* ref;
 320     char const* resolved;
 321   };
 322
 323   // clang-format off
 324   test_case tests[] = {
 325
 326       // 5.4.1.  Normal Examples
 327
 328       {"g:h",     "g:h"},
 329       {"g",       "http://a/b/c/g"},
 330       {"./g",     "http://a/b/c/g"},
 331       {"g/",      "http://a/b/c/g/"},
 332       {"/g",      "http://a/g"},
 333       {"//g",     "http://g"},
 334       {"?y",      "http://a/b/c/d;p?y"},
 335       {"g?y",     "http://a/b/c/g?y"},
 336       {"#s",      "http://a/b/c/d;p?q#s"},
 337       {"g#s",     "http://a/b/c/g#s"},
 338       {"g?y#s",   "http://a/b/c/g?y#s"},
 339       {";x",      "http://a/b/c/;x"},
 340       {"g;x",     "http://a/b/c/g;x"},
 341       {"g;x?y#s", "http://a/b/c/g;x?y#s"},
 342       {"",        "http://a/b/c/d;p?q"},
 343       {".",       "http://a/b/c/"},
 344       {"./",      "http://a/b/c/"},
 345       {"..",      "http://a/b/"},
 346       {"../",     "http://a/b/"},
 347       {"../g",    "http://a/b/g"},
 348       {"../..",   "http://a/"},
 349       {"../../",  "http://a/"},
 350       {"../../g", "http://a/g"},
 351
 352       // 5.4.2.  Abnormal Examples
 353
 354       {"../../../g",    "http://a/g"},
 355       {"../../../../g", "http://a/g"},
 356
 357       {"/./g",          "http://a/g"},
 358       {"/../g",         "http://a/g"},
 359       {"g.",            "http://a/b/c/g."},
 360       {".g",            "http://a/b/c/.g"},
 361       {"g..",           "http://a/b/c/g.."},
 362       {"..g",           "http://a/b/c/..g"},
 363
 364       {"g?y/./x",       "http://a/b/c/g?y/./x"},
 365       {"g?y/../x",      "http://a/b/c/g?y/../x"},
 366       {"g#s/./x",       "http://a/b/c/g#s/./x"},
 367       {"g#s/../x",      "http://a/b/c/g#s/../x"},
 368
 369       {"http:g",        "http:g"}, // for strict parsers
 370   };
 371   // clang-format on
 372
 373   // 5.4.  Reference Resolution Examples
 374
 375   uri::absolute base("http://a/b/c/d;p?q");
 376
 377   auto failures = 0;
 378
 379   for (auto&& test : tests) {
 380     uri::reference ref(test.ref);
 381     auto resolved = uri::resolve_ref(base, ref);
 382     auto resolved_str = uri::to_string(resolved);
 383     if (resolved_str != test.resolved) {
 384       LOG(ERROR) << "##### Failure #####";
 385       LOG(ERROR) << "for input == " << test.ref << '\n';
 386       LOG(ERROR) << "ref == " << ref;
 387       LOG(ERROR) << "resolved == " << resolved;
 388       LOG(ERROR) << "should match == " << test.resolved;
 389       ++failures;
 390     }
 391   }
 392
 393   return failures;
 394 }
 395
 396 int main(int argc, char* argv[])
 397 {
 398   auto failures = 0;
 399
 400   failures += test_good();
 401   failures += test_bad();
 402   failures += test_resolution();
 403
 404   {
 405     // the two explicit tests from RFC 3986:
 406     uri::components parts;
 407     parts.path = "/a/b/c/./../../g";
 408     CHECK_EQ(uri::normalize(parts), "/a/g");
 409     parts.path = "mid/content=5/../6";
 410     CHECK_EQ(uri::normalize(parts), "mid/6");
 411   }
 412
 413   // Parse command line args as URIs.
 414
 415   for (auto i = 1; i < argc; ++i) {
 416     uri::reference u{argv[i]};
 417
 418     std::cout << "scheme == " << (u.scheme() ? *u.scheme() : "{}") << '\n'
 419               << "auth   == " << (u.authority() ? *u.authority() : "{}") << '\n'
 420               << "user   == " << (u.userinfo() ? *u.userinfo() : "{}") << '\n'
 421               << "host   == " << (u.host() ? *u.host() : "{}") << '\n'
 422               << "port   == " << (u.port() ? *u.port() : "{}") << '\n'
 423               << "path   == " << (u.path() ? *u.path() : "{}") << '\n'
 424               << "query  == " << (u.query() ? *u.query() : "{}") << '\n'
 425               << "frag   == " << (u.fragment() ? *u.fragment() : "{}") << '\n';
 426
 427     std::cout << "string == " << uri::to_string(u) << '\n';
 428
 429     std::cout << "norm   == " << uri::normalize(u.parts()) << '\n';
 430   }
 431
 432   if (failures)
 433     LOG(FATAL) << "one or more tests failed";
 434 }