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