net/http/http_content_disposition_unittest.cc

   1 // Copyright (c) 2012 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 "net/http/http_content_disposition.h"
   6
   7 #include "base/strings/utf_string_conversions.h"
   8 #include "testing/gtest/include/gtest/gtest.h"
   9
  10 namespace net {
  11
  12 namespace {
  13
  14 struct FileNameCDCase {
  15   const char* header;
  16   const char* referrer_charset;
  17   const wchar_t* expected;
  18 };
  19
  20 }  // anonymous namespace
  21
  22 TEST(HttpContentDispositionTest, Filename) {
  23   const FileNameCDCase tests[] = {
  24     // Test various forms of C-D header fields emitted by web servers.
  25     {"inline; filename=\"abcde.pdf\"", "", L"abcde.pdf"},
  26     {"inline; name=\"abcde.pdf\"", "", L"abcde.pdf"},
  27     {"attachment; filename=abcde.pdf", "", L"abcde.pdf"},
  28     {"attachment; name=abcde.pdf", "", L"abcde.pdf"},
  29     {"attachment; filename=abc,de.pdf", "", L"abc,de.pdf"},
  30     {"filename=abcde.pdf", "", L"abcde.pdf"},
  31     {"filename= abcde.pdf", "", L"abcde.pdf"},
  32     {"filename =abcde.pdf", "", L"abcde.pdf"},
  33     {"filename = abcde.pdf", "", L"abcde.pdf"},
  34     {"filename\t=abcde.pdf", "", L"abcde.pdf"},
  35     {"filename \t\t  =abcde.pdf", "", L"abcde.pdf"},
  36     {"name=abcde.pdf", "", L"abcde.pdf"},
  37     {"inline; filename=\"abc%20de.pdf\"", "",
  38      L"abc de.pdf"},
  39     // Unbalanced quotation mark
  40     {"filename=\"abcdef.pdf", "", L"abcdef.pdf"},
  41     // Whitespaces are converted to a space.
  42     {"inline; filename=\"abc  \t\nde.pdf\"", "",
  43      L"abc    de.pdf"},
  44     // %-escaped UTF-8
  45     {"attachment; filename=\"%EC%98%88%EC%88%A0%20"
  46      "%EC%98%88%EC%88%A0.jpg\"", "", L"\xc608\xc220 \xc608\xc220.jpg"},
  47     {"attachment; filename=\"%F0%90%8C%B0%F0%90%8C%B1"
  48      "abc.jpg\"", "", L"\U00010330\U00010331abc.jpg"},
  49     {"attachment; filename=\"%EC%98%88%EC%88%A0 \n"
  50      "%EC%98%88%EC%88%A0.jpg\"", "", L"\xc608\xc220  \xc608\xc220.jpg"},
  51     // RFC 2047 with various charsets and Q/B encodings
  52     {"attachment; filename=\"=?EUC-JP?Q?=B7=DD=BD="
  53      "D13=2Epng?=\"", "", L"\x82b8\x8853" L"3.png"},
  54     {"attachment; filename==?eUc-Kr?b?v7m8+iAzLnBuZw==?=",
  55      "", L"\xc608\xc220 3.png"},
  56     {"attachment; filename==?utf-8?Q?=E8=8A=B8=E8"
  57      "=A1=93_3=2Epng?=", "", L"\x82b8\x8853 3.png"},
  58     {"attachment; filename==?utf-8?Q?=F0=90=8C=B0"
  59      "_3=2Epng?=", "", L"\U00010330 3.png"},
  60     {"inline; filename=\"=?iso88591?Q?caf=e9_=2epng?=\"",
  61      "", L"caf\x00e9 .png"},
  62     // Space after an encoded word should be removed.
  63     {"inline; filename=\"=?iso88591?Q?caf=E9_?= .png\"",
  64      "", L"caf\x00e9 .png"},
  65     // Two encoded words with different charsets (not very likely to be emitted
  66     // by web servers in the wild). Spaces between them are removed.
  67     {"inline; filename=\"=?euc-kr?b?v7m8+iAz?="
  68      " =?ksc5601?q?=BF=B9=BC=FA=2Epng?=\"", "",
  69      L"\xc608\xc220 3\xc608\xc220.png"},
  70     {"attachment; filename=\"=?windows-1252?Q?caf=E9?="
  71      "  =?iso-8859-7?b?4eI=?= .png\"", "", L"caf\x00e9\x03b1\x03b2.png"},
  72     // Non-ASCII string is passed through and treated as UTF-8 as long as
  73     // it's valid as UTF-8 and regardless of |referrer_charset|.
  74     {"attachment; filename=caf\xc3\xa9.png",
  75      "iso-8859-1", L"caf\x00e9.png"},
  76     {"attachment; filename=caf\xc3\xa9.png",
  77      "", L"caf\x00e9.png"},
  78     // Non-ASCII/Non-UTF-8 string. Fall back to the referrer charset.
  79     {"attachment; filename=caf\xe5.png",
  80      "windows-1253", L"caf\x03b5.png"},
  81 #if 0
  82     // Non-ASCII/Non-UTF-8 string. Fall back to the native codepage.
  83     // TODO(jungshik): We need to set the OS default codepage
  84     // to a specific value before testing. On Windows, we can use
  85     // SetThreadLocale().
  86     {"attachment; filename=\xb0\xa1\xb0\xa2.png",
  87      "", L"\xac00\xac01.png"},
  88 #endif
  89     // Failure cases
  90     // Invalid hex-digit "G"
  91     {"attachment; filename==?iiso88591?Q?caf=EG?=", "",
  92      L""},
  93     // Incomplete RFC 2047 encoded-word (missing '='' at the end)
  94     {"attachment; filename==?iso88591?Q?caf=E3?", "", L""},
  95     // Extra character at the end of an encoded word
  96     {"attachment; filename==?iso88591?Q?caf=E3?==",
  97      "", L""},
  98     // Extra token at the end of an encoded word
  99     {"attachment; filename==?iso88591?Q?caf=E3?=?",
 100      "", L""},
 101     {"attachment; filename==?iso88591?Q?caf=E3?=?=",
 102      "",  L""},
 103     // Incomplete hex-escaped chars
 104     {"attachment; filename==?windows-1252?Q?=63=61=E?=",
 105      "", L""},
 106     {"attachment; filename=%EC%98%88%EC%88%A", "", L""},
 107     // %-escaped non-UTF-8 encoding is an "error"
 108     {"attachment; filename=%B7%DD%BD%D1.png", "", L""},
 109     // Two RFC 2047 encoded words in a row without a space is an error.
 110     {"attachment; filename==?windows-1252?Q?caf=E3?="
 111      "=?iso-8859-7?b?4eIucG5nCg==?=", "", L""},
 112
 113     // RFC 5987 tests with Filename*  : see http://tools.ietf.org/html/rfc5987
 114     {"attachment; filename*=foo.html", "", L""},
 115     {"attachment; filename*=foo'.html", "", L""},
 116     {"attachment; filename*=''foo'.html", "", L""},
 117     {"attachment; filename*=''foo.html'", "", L""},
 118     {"attachment; filename*=''f\"oo\".html'", "", L""},
 119     {"attachment; filename*=bogus_charset''foo.html'",
 120      "", L""},
 121     {"attachment; filename*='en'foo.html'", "", L""},
 122     {"attachment; filename*=iso-8859-1'en'foo.html", "",
 123       L"foo.html"},
 124     {"attachment; filename*=utf-8'en'foo.html", "",
 125       L"foo.html"},
 126     // charset cannot be omitted.
 127     {"attachment; filename*='es'f\xfa.html'", "", L""},
 128     // Non-ASCII bytes are not allowed.
 129     {"attachment; filename*=iso-8859-1'es'f\xfa.html", "",
 130       L""},
 131     {"attachment; filename*=utf-8'es'f\xce\xba.html", "",
 132       L""},
 133     // TODO(jshin): Space should be %-encoded, but currently, we allow
 134     // spaces.
 135     {"inline; filename*=iso88591''cafe foo.png", "",
 136       L"cafe foo.png"},
 137
 138     // Filename* tests converted from Q-encoded tests above.
 139     {"attachment; filename*=EUC-JP''%B7%DD%BD%D13%2Epng",
 140      "", L"\x82b8\x8853" L"3.png"},
 141     {"attachment; filename*=utf-8''"
 142       "%E8%8A%B8%E8%A1%93%203%2Epng", "", L"\x82b8\x8853 3.png"},
 143     {"attachment; filename*=utf-8''%F0%90%8C%B0 3.png", "",
 144       L"\U00010330 3.png"},
 145     {"inline; filename*=Euc-Kr'ko'%BF%B9%BC%FA%2Epng", "",
 146      L"\xc608\xc220.png"},
 147     {"attachment; filename*=windows-1252''caf%E9.png", "",
 148       L"caf\x00e9.png"},
 149
 150     // Multiple filename, filename*, name parameters specified.
 151     {"attachment; name=\"foo\"; filename=\"bar\"", "", L"bar"},
 152     {"attachment; filename=\"bar\"; name=\"foo\"", "", L"bar"},
 153     {"attachment; filename=\"bar\"; filename*=utf-8''baz", "", L"baz"},
 154
 155     // http://greenbytes.de/tech/tc2231/ filename* test cases.
 156     // attwithisofn2231iso
 157     {"attachment; filename*=iso-8859-1''foo-%E4.html", "",
 158       L"foo-\xe4.html"},
 159     // attwithfn2231utf8
 160     {"attachment; filename*="
 161       "UTF-8''foo-%c3%a4-%e2%82%ac.html", "", L"foo-\xe4-\x20ac.html"},
 162     // attwithfn2231noc : no encoding specified but UTF-8 is used.
 163     {"attachment; filename*=''foo-%c3%a4-%e2%82%ac.html",
 164       "", L""},
 165     // attwithfn2231utf8comp
 166     {"attachment; filename*=UTF-8''foo-a%cc%88.html", "",
 167       L"foo-\xe4.html"},
 168 #ifdef ICU_SHOULD_FAIL_CONVERSION_ON_INVALID_CHARACTER
 169     // This does not work because we treat ISO-8859-1 synonymous with
 170     // Windows-1252 per HTML5. For HTTP, in theory, we're not
 171     // supposed to.
 172     // attwithfn2231utf8-bad
 173     {"attachment; filename*="
 174       "iso-8859-1''foo-%c3%a4-%e2%82%ac.html", "", L""},
 175 #endif
 176     // attwithfn2231ws1
 177     {"attachment; filename *=UTF-8''foo-%c3%a4.html", "",
 178       L""},
 179     // attwithfn2231ws2
 180     {"attachment; filename*= UTF-8''foo-%c3%a4.html", "",
 181       L"foo-\xe4.html"},
 182     // attwithfn2231ws3
 183     {"attachment; filename* =UTF-8''foo-%c3%a4.html", "",
 184       L"foo-\xe4.html"},
 185     // attwithfn2231quot
 186     {"attachment; filename*=\"UTF-8''foo-%c3%a4.html\"",
 187       "", L""},
 188     // attfnboth
 189     {"attachment; filename=\"foo-ae.html\"; "
 190       "filename*=UTF-8''foo-%c3%a4.html", "", L"foo-\xe4.html"},
 191     // attfnboth2
 192     {"attachment; filename*=UTF-8''foo-%c3%a4.html; "
 193       "filename=\"foo-ae.html\"", "", L"foo-\xe4.html"},
 194     // attnewandfn
 195     {"attachment; foobar=x; filename=\"foo.html\"", "",
 196       L"foo.html"},
 197   };
 198   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(tests); ++i) {
 199     HttpContentDisposition header(tests[i].header, tests[i].referrer_charset);
 200     EXPECT_EQ(tests[i].expected,
 201         base::UTF8ToWide(header.filename()))
 202         << "Failed on input: " << tests[i].header;
 203   }
 204 }
 205
 206 // Test cases from http://greenbytes.de/tech/tc2231/
 207 TEST(HttpContentDispositionTest, tc2231) {
 208   const struct FileNameCDCase {
 209     const char* header;
 210     net::HttpContentDisposition::Type expected_type;
 211     const wchar_t* expected_filename;
 212   } tests[] = {
 213     // http://greenbytes.de/tech/tc2231/#inlonly
 214     { "inline",
 215       net::HttpContentDisposition::INLINE,
 216       L""
 217     },
 218     // http://greenbytes.de/tech/tc2231/#inlonlyquoted
 219     { "\"inline\"",
 220       net::HttpContentDisposition::INLINE,
 221       L""
 222     },
 223     // http://greenbytes.de/tech/tc2231/#inlwithasciifilename
 224     { "inline; filename=\"foo.html\"",
 225       net::HttpContentDisposition::INLINE,
 226       L"foo.html"
 227     },
 228     // http://greenbytes.de/tech/tc2231/#inlwithfnattach
 229     { "inline; filename=\"Not an attachment!\"",
 230       net::HttpContentDisposition::INLINE,
 231       L"Not an attachment!"
 232     },
 233     // http://greenbytes.de/tech/tc2231/#inlwithasciifilenamepdf
 234     { "inline; filename=\"foo.pdf\"",
 235       net::HttpContentDisposition::INLINE,
 236       L"foo.pdf"
 237     },
 238     // http://greenbytes.de/tech/tc2231/#attonly
 239     { "attachment",
 240       net::HttpContentDisposition::ATTACHMENT,
 241       L""
 242     },
 243     // http://greenbytes.de/tech/tc2231/#attonlyquoted
 244     { "\"attachment\"",
 245       net::HttpContentDisposition::INLINE,
 246       L""
 247     },
 248     // http://greenbytes.de/tech/tc2231/#attonly403
 249     // TODO(abarth): This isn't testable in this unit test.
 250     // http://greenbytes.de/tech/tc2231/#attonlyucase
 251     { "ATTACHMENT",
 252       net::HttpContentDisposition::ATTACHMENT,
 253       L""
 254     },
 255     // http://greenbytes.de/tech/tc2231/#attwithasciifilename
 256     { "attachment; filename=\"foo.html\"",
 257       net::HttpContentDisposition::ATTACHMENT,
 258       L"foo.html"
 259     },
 260     // http://greenbytes.de/tech/tc2231/#attwithasciifnescapedchar
 261     { "attachment; filename=\"f\\oo.html\"",
 262       net::HttpContentDisposition::ATTACHMENT,
 263       L"foo.html"
 264     },
 265     // http://greenbytes.de/tech/tc2231/#attwithasciifnescapedquote
 266     { "attachment; filename=\"\\\"quoting\\\" tested.html\"",
 267       net::HttpContentDisposition::ATTACHMENT,
 268       L"\"quoting\" tested.html"
 269     },
 270     // http://greenbytes.de/tech/tc2231/#attwithquotedsemicolon
 271     { "attachment; filename=\"Here's a semicolon;.html\"",
 272       net::HttpContentDisposition::ATTACHMENT,
 273       L"Here's a semicolon;.html"
 274     },
 275     // http://greenbytes.de/tech/tc2231/#attwithfilenameandextparam
 276     { "attachment; foo=\"bar\"; filename=\"foo.html\"",
 277       net::HttpContentDisposition::ATTACHMENT,
 278       L"foo.html"
 279     },
 280     // http://greenbytes.de/tech/tc2231/#attwithfilenameandextparamescaped
 281     { "attachment; foo=\"\\\"\\\\\";filename=\"foo.html\"",
 282       net::HttpContentDisposition::ATTACHMENT,
 283       L"foo.html"
 284     },
 285     // http://greenbytes.de/tech/tc2231/#attwithasciifilenameucase
 286     { "attachment; FILENAME=\"foo.html\"",
 287       net::HttpContentDisposition::ATTACHMENT,
 288       L"foo.html"
 289     },
 290     // http://greenbytes.de/tech/tc2231/#attwithasciifilenamenq
 291     { "attachment; filename=foo.html",
 292       net::HttpContentDisposition::ATTACHMENT,
 293       L"foo.html"
 294     },
 295     // http://greenbytes.de/tech/tc2231/#attwithasciifilenamenqs
 296     // Note: tc2231 says we should fail to parse this header.
 297     { "attachment; filename=foo.html ;",
 298       net::HttpContentDisposition::ATTACHMENT,
 299       L"foo.html"
 300     },
 301     // http://greenbytes.de/tech/tc2231/#attemptyparam
 302     // Note: tc2231 says we should fail to parse this header.
 303     { "attachment; ;filename=foo",
 304       net::HttpContentDisposition::ATTACHMENT,
 305       L"foo"
 306     },
 307     // http://greenbytes.de/tech/tc2231/#attwithasciifilenamenqws
 308     // Note: tc2231 says we should fail to parse this header.
 309     { "attachment; filename=foo bar.html",
 310       net::HttpContentDisposition::ATTACHMENT,
 311       L"foo bar.html"
 312     },
 313     // http://greenbytes.de/tech/tc2231/#attwithfntokensq
 314     { "attachment; filename='foo.bar'",
 315       net::HttpContentDisposition::ATTACHMENT,
 316       L"foo.bar"  // Should be L"'foo.bar'"
 317     },
 318 #ifdef ICU_SHOULD_FAIL_CONVERSION_ON_INVALID_CHARACTER
 319     // http://greenbytes.de/tech/tc2231/#attwithisofnplain
 320     { "attachment; filename=\"foo-\xE4html\"",
 321       net::HttpContentDisposition::ATTACHMENT,
 322       L""  // Should be L"foo-\xE4.html"
 323     },
 324 #endif
 325     // http://greenbytes.de/tech/tc2231/#attwithutf8fnplain
 326     // Note: We'll UTF-8 decode the file name, even though tc2231 says not to.
 327     { "attachment; filename=\"foo-\xC3\xA4.html\"",
 328       net::HttpContentDisposition::ATTACHMENT,
 329       L"foo-\xE4.html"
 330     },
 331     // http://greenbytes.de/tech/tc2231/#attwithfnrawpctenca
 332     { "attachment; filename=\"foo-%41.html\"",
 333       net::HttpContentDisposition::ATTACHMENT,
 334       L"foo-A.html"  // Should be L"foo-%41.html"
 335     },
 336     // http://greenbytes.de/tech/tc2231/#attwithfnusingpct
 337     { "attachment; filename=\"50%.html\"",
 338       net::HttpContentDisposition::ATTACHMENT,
 339       L"50%.html"
 340     },
 341     // http://greenbytes.de/tech/tc2231/#attwithfnrawpctencaq
 342     { "attachment; filename=\"foo-%\\41.html\"",
 343       net::HttpContentDisposition::ATTACHMENT,
 344       L"foo-A.html"  // Should be L"foo-%41.html"
 345     },
 346     // http://greenbytes.de/tech/tc2231/#attwithnamepct
 347     { "attachment; name=\"foo-%41.html\"",
 348       net::HttpContentDisposition::ATTACHMENT,
 349       L"foo-A.html"  // Should be L"foo-%41.html"
 350     },
 351 #ifdef ICU_SHOULD_FAIL_CONVERSION_ON_INVALID_CHARACTER
 352     // http://greenbytes.de/tech/tc2231/#attwithfilenamepctandiso
 353     { "attachment; filename=\"\xE4-%41.html\"",
 354       net::HttpContentDisposition::ATTACHMENT,
 355       L""  // Should be L"\xE4-%41.htm"
 356     },
 357 #endif
 358     // http://greenbytes.de/tech/tc2231/#attwithfnrawpctenclong
 359     { "attachment; filename=\"foo-%c3%a4-%e2%82%ac.html\"",
 360       net::HttpContentDisposition::ATTACHMENT,
 361       L"foo-\xE4-\u20AC.html"  // Should be L"foo-%c3%a4-%e2%82%ac.html"
 362     },
 363     // http://greenbytes.de/tech/tc2231/#attwithasciifilenamews1
 364     { "attachment; filename =\"foo.html\"",
 365       net::HttpContentDisposition::ATTACHMENT,
 366       L"foo.html"
 367     },
 368     // http://greenbytes.de/tech/tc2231/#attwith2filenames
 369     // Note: tc2231 says we should fail to parse this header.
 370     { "attachment; filename=\"foo.html\"; filename=\"bar.html\"",
 371       net::HttpContentDisposition::ATTACHMENT,
 372       L"foo.html"
 373     },
 374     // http://greenbytes.de/tech/tc2231/#attfnbrokentoken
 375     // Note: tc2231 says we should fail to parse this header.
 376     { "attachment; filename=foo[1](2).html",
 377       net::HttpContentDisposition::ATTACHMENT,
 378       L"foo[1](2).html"
 379     },
 380 #ifdef ICU_SHOULD_FAIL_CONVERSION_ON_INVALID_CHARACTER
 381     // http://greenbytes.de/tech/tc2231/#attfnbrokentokeniso
 382     // Note: tc2231 says we should fail to parse this header.
 383     { "attachment; filename=foo-\xE4.html",
 384       net::HttpContentDisposition::ATTACHMENT,
 385       L""
 386     },
 387 #endif
 388     // http://greenbytes.de/tech/tc2231/#attfnbrokentokenutf
 389     // Note: tc2231 says we should fail to parse this header.
 390     { "attachment; filename=foo-\xC3\xA4.html",
 391       net::HttpContentDisposition::ATTACHMENT,
 392       L"foo-\xE4.html"
 393     },
 394     // http://greenbytes.de/tech/tc2231/#attmissingdisposition
 395     // Note: tc2231 says we should fail to parse this header.
 396     { "filename=foo.html",
 397       net::HttpContentDisposition::INLINE,
 398       L"foo.html"
 399     },
 400     // http://greenbytes.de/tech/tc2231/#attmissingdisposition2
 401     // Note: tc2231 says we should fail to parse this header.
 402     { "x=y; filename=foo.html",
 403       net::HttpContentDisposition::INLINE,
 404       L"foo.html"
 405     },
 406     // http://greenbytes.de/tech/tc2231/#attmissingdisposition3
 407     // Note: tc2231 says we should fail to parse this header.
 408     { "\"foo; filename=bar;baz\"; filename=qux",
 409       net::HttpContentDisposition::INLINE,
 410       L""  // Firefox gets qux
 411     },
 412     // http://greenbytes.de/tech/tc2231/#attmissingdisposition4
 413     // Note: tc2231 says we should fail to parse this header.
 414     { "filename=foo.html, filename=bar.html",
 415       net::HttpContentDisposition::INLINE,
 416       L"foo.html, filename=bar.html"
 417     },
 418     // http://greenbytes.de/tech/tc2231/#emptydisposition
 419     // Note: tc2231 says we should fail to parse this header.
 420     { "; filename=foo.html",
 421       net::HttpContentDisposition::INLINE,
 422       L"foo.html"
 423     },
 424     // http://greenbytes.de/tech/tc2231/#attandinline
 425     // Note: tc2231 says we should fail to parse this header.
 426     { "inline; attachment; filename=foo.html",
 427       net::HttpContentDisposition::INLINE,
 428       L""
 429     },
 430     // http://greenbytes.de/tech/tc2231/#attandinline2
 431     // Note: tc2231 says we should fail to parse this header.
 432     { "attachment; inline; filename=foo.html",
 433       net::HttpContentDisposition::ATTACHMENT,
 434       L""
 435     },
 436     // http://greenbytes.de/tech/tc2231/#attbrokenquotedfn
 437     // Note: tc2231 says we should fail to parse this header.
 438     { "attachment; filename=\"foo.html\".txt",
 439       net::HttpContentDisposition::ATTACHMENT,
 440       L"foo.html\".txt"
 441     },
 442     // http://greenbytes.de/tech/tc2231/#attbrokenquotedfn2
 443     // Note: tc2231 says we should fail to parse this header.
 444     { "attachment; filename=\"bar",
 445       net::HttpContentDisposition::ATTACHMENT,
 446       L"bar"
 447     },
 448     // http://greenbytes.de/tech/tc2231/#attbrokenquotedfn3
 449     // Note: tc2231 says we should fail to parse this header.
 450     { "attachment; filename=foo\"bar;baz\"qux",
 451       net::HttpContentDisposition::ATTACHMENT,
 452       L"foo\"bar;baz\"qux"
 453     },
 454     // http://greenbytes.de/tech/tc2231/#attmultinstances
 455     // Note: tc2231 says we should fail to parse this header.
 456     { "attachment; filename=foo.html, attachment; filename=bar.html",
 457       net::HttpContentDisposition::ATTACHMENT,
 458       L"foo.html, attachment"
 459     },
 460     // http://greenbytes.de/tech/tc2231/#attmissingdelim
 461     { "attachment; foo=foo filename=bar",
 462       net::HttpContentDisposition::ATTACHMENT,
 463       L""
 464     },
 465     // http://greenbytes.de/tech/tc2231/#attreversed
 466     // Note: tc2231 says we should fail to parse this header.
 467     { "filename=foo.html; attachment",
 468       net::HttpContentDisposition::INLINE,
 469       L"foo.html"
 470     },
 471     // http://greenbytes.de/tech/tc2231/#attconfusedparam
 472     { "attachment; xfilename=foo.html",
 473       net::HttpContentDisposition::ATTACHMENT,
 474       L""
 475     },
 476     // http://greenbytes.de/tech/tc2231/#attabspath
 477     { "attachment; filename=\"/foo.html\"",
 478       net::HttpContentDisposition::ATTACHMENT,
 479       L"/foo.html"
 480     },
 481     // http://greenbytes.de/tech/tc2231/#attabspathwin
 482     { "attachment; filename=\"\\\\foo.html\"",
 483       net::HttpContentDisposition::ATTACHMENT,
 484       L"\\foo.html"
 485     },
 486     // http://greenbytes.de/tech/tc2231/#dispext
 487     { "foobar",
 488       net::HttpContentDisposition::ATTACHMENT,
 489       L""
 490     },
 491     // http://greenbytes.de/tech/tc2231/#dispextbadfn
 492     { "attachment; example=\"filename=example.txt\"",
 493       net::HttpContentDisposition::ATTACHMENT,
 494       L""
 495     },
 496     // http://greenbytes.de/tech/tc2231/#attnewandfn
 497     { "attachment; foobar=x; filename=\"foo.html\"",
 498       net::HttpContentDisposition::ATTACHMENT,
 499       L"foo.html"
 500     },
 501     // TODO(abarth): Add the filename* tests, but check
 502     //              HttpContentDispositionTest.Filename for overlap.
 503     // TODO(abarth): http://greenbytes.de/tech/tc2231/#attrfc2047token
 504     // TODO(abarth): http://greenbytes.de/tech/tc2231/#attrfc2047quoted
 505   };
 506   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(tests); ++i) {
 507     HttpContentDisposition header(tests[i].header, std::string());
 508     EXPECT_EQ(tests[i].expected_type, header.type())
 509         << "Failed on input: " << tests[i].header;
 510     EXPECT_EQ(tests[i].expected_filename, base::UTF8ToWide(header.filename()))
 511         << "Failed on input: " << tests[i].header;
 512   }
 513 }
 514
 515 TEST(HttpContentDispositionTest, ParseResult) {
 516   const struct ParseResultTestCase {
 517     const char* header;
 518     int expected_flags;
 519   } kTestCases[] = {
 520     // Basic feature tests
 521     { "", HttpContentDisposition::INVALID },
 522     { "example=x", HttpContentDisposition::INVALID },
 523     { "attachment; filename=", HttpContentDisposition::HAS_DISPOSITION_TYPE },
 524     { "attachment; name=", HttpContentDisposition::HAS_DISPOSITION_TYPE },
 525     { "attachment; filename*=", HttpContentDisposition::HAS_DISPOSITION_TYPE },
 526     { "attachment; filename==?utf-8?Q?\?=",
 527       HttpContentDisposition::HAS_DISPOSITION_TYPE },
 528     { "filename=x", HttpContentDisposition::HAS_FILENAME },
 529     { "example; filename=x",
 530       HttpContentDisposition::HAS_DISPOSITION_TYPE |
 531       HttpContentDisposition::HAS_UNKNOWN_DISPOSITION_TYPE |
 532       HttpContentDisposition::HAS_FILENAME},
 533     { "attachment; filename=x",
 534       HttpContentDisposition::HAS_DISPOSITION_TYPE |
 535       HttpContentDisposition::HAS_FILENAME },
 536     { "attachment; filename=x; name=y",
 537       HttpContentDisposition::HAS_DISPOSITION_TYPE |
 538       HttpContentDisposition::HAS_FILENAME |
 539       HttpContentDisposition::HAS_NAME },
 540     { "attachment; name=y; filename*=utf-8''foo; name=x",
 541       HttpContentDisposition::HAS_DISPOSITION_TYPE |
 542       HttpContentDisposition::HAS_EXT_FILENAME |
 543       HttpContentDisposition::HAS_NAME },
 544
 545     // Feature tests for 'filename' attribute.
 546     { "filename=foo\xcc\x88",
 547       HttpContentDisposition::HAS_FILENAME |
 548       HttpContentDisposition::HAS_NON_ASCII_STRINGS },
 549     { "filename=foo%cc%88",
 550       HttpContentDisposition::HAS_FILENAME |
 551       HttpContentDisposition::HAS_PERCENT_ENCODED_STRINGS },
 552     { "filename==?utf-8?Q?foo?=",
 553       HttpContentDisposition::HAS_FILENAME |
 554       HttpContentDisposition::HAS_RFC2047_ENCODED_STRINGS },
 555     { "filename=\"=?utf-8?Q?foo?=\"",
 556       HttpContentDisposition::HAS_FILENAME |
 557       HttpContentDisposition::HAS_RFC2047_ENCODED_STRINGS },
 558     { "filename==?utf-8?Q?foo?", HttpContentDisposition::INVALID },
 559     { "name=foo\xcc\x88",
 560       HttpContentDisposition::HAS_NAME },
 561
 562     // Shouldn't set |has_non_ascii_strings| based on 'name' attribute.
 563     { "filename=x; name=foo\xcc\x88",
 564       HttpContentDisposition::HAS_FILENAME |
 565       HttpContentDisposition::HAS_NAME },
 566     { "filename=foo\xcc\x88 foo%cc%88 =?utf-8?Q?foo?=",
 567       HttpContentDisposition::HAS_FILENAME |
 568       HttpContentDisposition::HAS_NON_ASCII_STRINGS |
 569       HttpContentDisposition::HAS_PERCENT_ENCODED_STRINGS |
 570       HttpContentDisposition::HAS_RFC2047_ENCODED_STRINGS },
 571
 572     // If 'filename' attribute is invalid, should set any flags based on it.
 573     { "filename=foo\xcc\x88 foo%cc%88 =?utf-8?Q?foo?",
 574       HttpContentDisposition::INVALID },
 575     { "filename=foo\xcc\x88 foo%cc%88 =?utf-8?Q?foo?; name=x",
 576       HttpContentDisposition::HAS_NAME },
 577   };
 578
 579   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(kTestCases); ++i) {
 580     const ParseResultTestCase& test_case = kTestCases[i];
 581     HttpContentDisposition content_disposition(test_case.header, "utf-8");
 582     int result = content_disposition.parse_result_flags();
 583
 584     SCOPED_TRACE(testing::Message() << "Test case " << i
 585                                     << " with header " << test_case.header);
 586     EXPECT_EQ(test_case.expected_flags, result);
 587   }
 588 }
 589
 590 }  // namespace net