1// Copyright (c) 2010 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 <stdlib.h>
6
7#include "base/basictypes.h"
8#include "base/file_util.h"
9#include "base/path_service.h"
10#include "base/string_util.h"
11#include "base/utf_string_conversions.h"
12#include "chrome/browser/net/url_fixer_upper.h"
13#include "chrome/common/chrome_paths.h"
14#include "googleurl/src/url_parse.h"
15#include "googleurl/src/gurl.h"
16#include "net/base/net_util.h"
17#include "testing/gtest/include/gtest/gtest.h"
18
19namespace {
20  class URLFixerUpperTest : public testing::Test {
21  };
22};
23
24namespace url_parse {
25
26std::ostream& operator<<(std::ostream& os, const Component& part) {
27  return os << "(begin=" << part.begin << ", len=" << part.len << ")";
28}
29
30}  // namespace url_parse
31
32struct segment_case {
33  const std::string input;
34  const std::string result;
35  const url_parse::Component scheme;
36  const url_parse::Component username;
37  const url_parse::Component password;
38  const url_parse::Component host;
39  const url_parse::Component port;
40  const url_parse::Component path;
41  const url_parse::Component query;
42  const url_parse::Component ref;
43};
44
45static const segment_case segment_cases[] = {
46  { "http://www.google.com/", "http",
47    url_parse::Component(0, 4), // scheme
48    url_parse::Component(), // username
49    url_parse::Component(), // password
50    url_parse::Component(7, 14), // host
51    url_parse::Component(), // port
52    url_parse::Component(21, 1), // path
53    url_parse::Component(), // query
54    url_parse::Component(), // ref
55  },
56  { "aBoUt:vErSiOn", "about",
57    url_parse::Component(0, 5), // scheme
58    url_parse::Component(), // username
59    url_parse::Component(), // password
60    url_parse::Component(), // host
61    url_parse::Component(), // port
62    url_parse::Component(), // path
63    url_parse::Component(), // query
64    url_parse::Component(), // ref
65  },
66  { "    www.google.com:124?foo#", "http",
67    url_parse::Component(), // scheme
68    url_parse::Component(), // username
69    url_parse::Component(), // password
70    url_parse::Component(4, 14), // host
71    url_parse::Component(19, 3), // port
72    url_parse::Component(), // path
73    url_parse::Component(23, 3), // query
74    url_parse::Component(27, 0), // ref
75  },
76  { "user@www.google.com", "http",
77    url_parse::Component(), // scheme
78    url_parse::Component(0, 4), // username
79    url_parse::Component(), // password
80    url_parse::Component(5, 14), // host
81    url_parse::Component(), // port
82    url_parse::Component(), // path
83    url_parse::Component(), // query
84    url_parse::Component(), // ref
85  },
86  { "ftp:/user:P:a$$Wd@..ftp.google.com...::23///pub?foo#bar", "ftp",
87    url_parse::Component(0, 3), // scheme
88    url_parse::Component(5, 4), // username
89    url_parse::Component(10, 7), // password
90    url_parse::Component(18, 20), // host
91    url_parse::Component(39, 2), // port
92    url_parse::Component(41, 6), // path
93    url_parse::Component(48, 3), // query
94    url_parse::Component(52, 3), // ref
95  },
96  { "[2001:db8::1]/path", "http",
97    url_parse::Component(), // scheme
98    url_parse::Component(), // username
99    url_parse::Component(), // password
100    url_parse::Component(0, 13), // host
101    url_parse::Component(), // port
102    url_parse::Component(13, 5), // path
103    url_parse::Component(), // query
104    url_parse::Component(), // ref
105  },
106  { "[::1]", "http",
107    url_parse::Component(), // scheme
108    url_parse::Component(), // username
109    url_parse::Component(), // password
110    url_parse::Component(0, 5), // host
111    url_parse::Component(), // port
112    url_parse::Component(), // path
113    url_parse::Component(), // query
114    url_parse::Component(), // ref
115  },
116  // Incomplete IPv6 addresses (will not canonicalize).
117  { "[2001:4860:", "http",
118    url_parse::Component(), // scheme
119    url_parse::Component(), // username
120    url_parse::Component(), // password
121    url_parse::Component(0, 11), // host
122    url_parse::Component(), // port
123    url_parse::Component(), // path
124    url_parse::Component(), // query
125    url_parse::Component(), // ref
126  },
127  { "[2001:4860:/foo", "http",
128    url_parse::Component(), // scheme
129    url_parse::Component(), // username
130    url_parse::Component(), // password
131    url_parse::Component(0, 11), // host
132    url_parse::Component(), // port
133    url_parse::Component(11, 4), // path
134    url_parse::Component(), // query
135    url_parse::Component(), // ref
136  },
137  { "http://:b005::68]", "http",
138    url_parse::Component(0, 4), // scheme
139    url_parse::Component(), // username
140    url_parse::Component(), // password
141    url_parse::Component(7, 10), // host
142    url_parse::Component(), // port
143    url_parse::Component(), // path
144    url_parse::Component(), // query
145    url_parse::Component(), // ref
146  },
147  // Can't do anything useful with this.
148  { ":b005::68]", "",
149    url_parse::Component(0, 0), // scheme
150    url_parse::Component(), // username
151    url_parse::Component(), // password
152    url_parse::Component(), // host
153    url_parse::Component(), // port
154    url_parse::Component(), // path
155    url_parse::Component(), // query
156    url_parse::Component(), // ref
157  },
158};
159
160TEST(URLFixerUpperTest, SegmentURL) {
161  std::string result;
162  url_parse::Parsed parts;
163
164  for (size_t i = 0; i < arraysize(segment_cases); ++i) {
165    segment_case value = segment_cases[i];
166    result = URLFixerUpper::SegmentURL(value.input, &parts);
167    EXPECT_EQ(value.result, result);
168    EXPECT_EQ(value.scheme, parts.scheme);
169    EXPECT_EQ(value.username, parts.username);
170    EXPECT_EQ(value.password, parts.password);
171    EXPECT_EQ(value.host, parts.host);
172    EXPECT_EQ(value.port, parts.port);
173    EXPECT_EQ(value.path, parts.path);
174    EXPECT_EQ(value.query, parts.query);
175    EXPECT_EQ(value.ref, parts.ref);
176  }
177}
178
179// Creates a file and returns its full name as well as the decomposed
180// version. Example:
181//    full_path = "c:\foo\bar.txt"
182//    dir = "c:\foo"
183//    file_name = "bar.txt"
184static bool MakeTempFile(const FilePath& dir,
185                         const FilePath& file_name,
186                         FilePath* full_path) {
187  *full_path = dir.Append(file_name);
188  return file_util::WriteFile(*full_path, "", 0) == 0;
189}
190
191// Returns true if the given URL is a file: URL that matches the given file
192static bool IsMatchingFileURL(const std::string& url,
193                              const FilePath& full_file_path) {
194  if (url.length() <= 8)
195    return false;
196  if (std::string("file:///") != url.substr(0, 8))
197    return false; // no file:/// prefix
198  if (url.find('\\') != std::string::npos)
199    return false; // contains backslashes
200
201  FilePath derived_path;
202  net::FileURLToFilePath(GURL(url), &derived_path);
203
204  return FilePath::CompareEqualIgnoreCase(derived_path.value(),
205                                          full_file_path.value());
206}
207
208struct fixup_case {
209  const std::string input;
210  const std::string desired_tld;
211  const std::string output;
212} fixup_cases[] = {
213  {"www.google.com", "", "http://www.google.com/"},
214  {" www.google.com     ", "", "http://www.google.com/"},
215  {" foo.com/asdf  bar", "", "http://foo.com/asdf%20%20bar"},
216  {"..www.google.com..", "", "http://www.google.com./"},
217  {"http://......", "", "http://....../"},
218  {"http://host.com:ninety-two/", "", "http://host.com:ninety-two/"},
219  {"http://host.com:ninety-two?foo", "", "http://host.com:ninety-two/?foo"},
220  {"google.com:123", "", "http://google.com:123/"},
221  {"about:", "", "about:"},
222  {"about:version", "", "about:version"},
223  {"www:123", "", "http://www:123/"},
224  {"   www:123", "", "http://www:123/"},
225  {"www.google.com?foo", "", "http://www.google.com/?foo"},
226  {"www.google.com#foo", "", "http://www.google.com/#foo"},
227  {"www.google.com?", "", "http://www.google.com/?"},
228  {"www.google.com#", "", "http://www.google.com/#"},
229  {"www.google.com:123?foo#bar", "", "http://www.google.com:123/?foo#bar"},
230  {"user@www.google.com", "", "http://user@www.google.com/"},
231  {"\xE6\xB0\xB4.com" , "", "http://xn--1rw.com/"},
232  // It would be better if this next case got treated as http, but I don't see
233  // a clean way to guess this isn't the new-and-exciting "user" scheme.
234  {"user:passwd@www.google.com:8080/", "", "user:passwd@www.google.com:8080/"},
235  // {"file:///c:/foo/bar%20baz.txt", "", "file:///C:/foo/bar%20baz.txt"},
236  {"ftp.google.com", "", "ftp://ftp.google.com/"},
237  {"    ftp.google.com", "", "ftp://ftp.google.com/"},
238  {"FTP.GooGle.com", "", "ftp://ftp.google.com/"},
239  {"ftpblah.google.com", "", "http://ftpblah.google.com/"},
240  {"ftp", "", "http://ftp/"},
241  {"google.ftp.com", "", "http://google.ftp.com/"},
242  // URLs which end with 0x85 (NEL in ISO-8859).
243  { "http://google.com/search?q=\xd0\x85", "",
244    "http://google.com/search?q=%D0%85"
245  },
246  { "http://google.com/search?q=\xec\x97\x85", "",
247    "http://google.com/search?q=%EC%97%85"
248  },
249  { "http://google.com/search?q=\xf0\x90\x80\x85", "",
250    "http://google.com/search?q=%F0%90%80%85"
251  },
252  // URLs which end with 0xA0 (non-break space in ISO-8859).
253  { "http://google.com/search?q=\xd0\xa0", "",
254    "http://google.com/search?q=%D0%A0"
255  },
256  { "http://google.com/search?q=\xec\x97\xa0", "",
257    "http://google.com/search?q=%EC%97%A0"
258  },
259  { "http://google.com/search?q=\xf0\x90\x80\xa0", "",
260    "http://google.com/search?q=%F0%90%80%A0"
261  },
262  // URLs containing IPv6 literals.
263  {"[2001:db8::2]", "", "http://[2001:db8::2]/"},
264  {"[::]:80", "", "http://[::]/"},
265  {"[::]:80/path", "", "http://[::]/path"},
266  {"[::]:180/path", "", "http://[::]:180/path"},
267  // TODO(pmarks): Maybe we should parse bare IPv6 literals someday.
268  {"::1", "", "::1"},
269};
270
271TEST(URLFixerUpperTest, FixupURL) {
272  for (size_t i = 0; i < arraysize(fixup_cases); ++i) {
273    fixup_case value = fixup_cases[i];
274    EXPECT_EQ(value.output, URLFixerUpper::FixupURL(value.input,
275        value.desired_tld).possibly_invalid_spec());
276  }
277
278  // Check the TLD-appending functionality
279  fixup_case tld_cases[] = {
280    {"google", "com", "http://www.google.com/"},
281    {"google.", "com", "http://www.google.com/"},
282    {"google..", "com", "http://www.google.com/"},
283    {".google", "com", "http://www.google.com/"},
284    {"www.google", "com", "http://www.google.com/"},
285    {"google.com", "com", "http://google.com/"},
286    {"http://google", "com", "http://www.google.com/"},
287    {"..google..", "com", "http://www.google.com/"},
288    {"http://www.google", "com", "http://www.google.com/"},
289    {"9999999999999999", "com", "http://www.9999999999999999.com/"},
290    {"google/foo", "com", "http://www.google.com/foo"},
291    {"google.com/foo", "com", "http://google.com/foo"},
292    {"google/?foo=.com", "com", "http://www.google.com/?foo=.com"},
293    {"www.google/?foo=www.", "com", "http://www.google.com/?foo=www."},
294    {"google.com/?foo=.com", "com", "http://google.com/?foo=.com"},
295    {"http://www.google.com", "com", "http://www.google.com/"},
296    {"google:123", "com", "http://www.google.com:123/"},
297    {"http://google:123", "com", "http://www.google.com:123/"},
298  };
299  for (size_t i = 0; i < arraysize(tld_cases); ++i) {
300    fixup_case value = tld_cases[i];
301    EXPECT_EQ(value.output, URLFixerUpper::FixupURL(value.input,
302        value.desired_tld).possibly_invalid_spec());
303  }
304}
305
306// Test different types of file inputs to URIFixerUpper::FixupURL. This
307// doesn't go into the nice array of fixups above since the file input
308// has to exist.
309TEST(URLFixerUpperTest, FixupFile) {
310  // this "original" filename is the one we tweak to get all the variations
311  FilePath dir;
312  FilePath original;
313  ASSERT_TRUE(PathService::Get(chrome::DIR_APP, &dir));
314  ASSERT_TRUE(MakeTempFile(
315      dir,
316      FilePath(FILE_PATH_LITERAL("url fixer upper existing file.txt")),
317      &original));
318
319  // reference path
320  GURL golden(net::FilePathToFileURL(original));
321
322  // c:\foo\bar.txt -> file:///c:/foo/bar.txt (basic)
323#if defined(OS_WIN)
324  GURL fixedup(URLFixerUpper::FixupURL(WideToUTF8(original.value()),
325                                       std::string()));
326#elif defined(OS_POSIX)
327  GURL fixedup(URLFixerUpper::FixupURL(original.value(), std::string()));
328#endif
329  EXPECT_EQ(golden, fixedup);
330
331  // TODO(port): Make some equivalent tests for posix.
332#if defined(OS_WIN)
333  // c|/foo\bar.txt -> file:///c:/foo/bar.txt (pipe allowed instead of colon)
334  std::string cur(WideToUTF8(original.value()));
335  EXPECT_EQ(':', cur[1]);
336  cur[1] = '|';
337  EXPECT_EQ(golden, URLFixerUpper::FixupURL(cur, std::string()));
338
339  fixup_case file_cases[] = {
340    {"c:\\This%20is a non-existent file.txt", "",
341     "file:///C:/This%2520is%20a%20non-existent%20file.txt"},
342
343    // \\foo\bar.txt -> file://foo/bar.txt
344    // UNC paths, this file won't exist, but since there are no escapes, it
345    // should be returned just converted to a file: URL.
346    {"\\\\SomeNonexistentHost\\foo\\bar.txt", "",
347     "file://somenonexistenthost/foo/bar.txt"},
348    // We do this strictly, like IE8, which only accepts this form using
349    // backslashes and not forward ones.  Turning "//foo" into "http" matches
350    // Firefox and IE, silly though it may seem (it falls out of adding "http"
351    // as the default protocol if you haven't entered one).
352    {"//SomeNonexistentHost\\foo/bar.txt", "",
353     "http://somenonexistenthost/foo/bar.txt"},
354    {"file:///C:/foo/bar", "", "file:///C:/foo/bar"},
355
356    // Much of the work here comes from GURL's canonicalization stage.
357    {"file://C:/foo/bar", "", "file:///C:/foo/bar"},
358    {"file:c:", "", "file:///C:/"},
359    {"file:c:WINDOWS", "", "file:///C:/WINDOWS"},
360    {"file:c|Program Files", "", "file:///C:/Program%20Files"},
361    {"file:/file", "", "file://file/"},
362    {"file:////////c:\\foo", "", "file:///C:/foo"},
363    {"file://server/folder/file", "", "file://server/folder/file"},
364
365    // These are fixups we don't do, but could consider:
366    //
367    //   {"file:///foo:/bar", "", "file://foo/bar"},
368    //   {"file:/\\/server\\folder/file", "", "file://server/folder/file"},
369  };
370#elif defined(OS_POSIX)
371
372#if defined(OS_MACOSX)
373#define HOME "/Users/"
374#else
375#define HOME "/home/"
376#endif
377  URLFixerUpper::home_directory_override = "/foo";
378  fixup_case file_cases[] = {
379    // File URLs go through GURL, which tries to escape intelligently.
380    {"/This%20is a non-existent file.txt", "",
381     "file:///This%2520is%20a%20non-existent%20file.txt"},
382    // A plain "/" refers to the root.
383    {"/", "",
384     "file:///"},
385
386    // These rely on the above home_directory_override.
387    {"~", "",
388     "file:///foo"},
389    {"~/bar", "",
390     "file:///foo/bar"},
391
392    // References to other users' homedirs.
393    {"~foo", "",
394     "file://" HOME "foo"},
395    {"~x/blah", "",
396     "file://" HOME "x/blah"},
397  };
398#endif
399  for (size_t i = 0; i < arraysize(file_cases); i++) {
400    EXPECT_EQ(file_cases[i].output, URLFixerUpper::FixupURL(file_cases[i].input,
401        file_cases[i].desired_tld).possibly_invalid_spec());
402  }
403
404  EXPECT_TRUE(file_util::Delete(original, false));
405}
406
407TEST(URLFixerUpperTest, FixupRelativeFile) {
408  FilePath full_path, dir;
409  FilePath file_part(FILE_PATH_LITERAL("url_fixer_upper_existing_file.txt"));
410  ASSERT_TRUE(PathService::Get(chrome::DIR_APP, &dir));
411  ASSERT_TRUE(MakeTempFile(dir, file_part, &full_path));
412  ASSERT_TRUE(file_util::AbsolutePath(&full_path));
413
414  // make sure we pass through good URLs
415  for (size_t i = 0; i < arraysize(fixup_cases); ++i) {
416    fixup_case value = fixup_cases[i];
417#if defined(OS_WIN)
418    FilePath input(UTF8ToWide(value.input));
419#elif defined(OS_POSIX)
420    FilePath input(value.input);
421#endif
422    EXPECT_EQ(value.output,
423        URLFixerUpper::FixupRelativeFile(dir, input).possibly_invalid_spec());
424  }
425
426  // make sure the existing file got fixed-up to a file URL, and that there
427  // are no backslashes
428  EXPECT_TRUE(IsMatchingFileURL(URLFixerUpper::FixupRelativeFile(dir,
429      file_part).possibly_invalid_spec(), full_path));
430  EXPECT_TRUE(file_util::Delete(full_path, false));
431
432  // create a filename we know doesn't exist and make sure it doesn't get
433  // fixed up to a file URL
434  FilePath nonexistent_file(
435      FILE_PATH_LITERAL("url_fixer_upper_nonexistent_file.txt"));
436  std::string fixedup(URLFixerUpper::FixupRelativeFile(dir,
437      nonexistent_file).possibly_invalid_spec());
438  EXPECT_NE(std::string("file:///"), fixedup.substr(0, 8));
439  EXPECT_FALSE(IsMatchingFileURL(fixedup, nonexistent_file));
440
441  // make a subdir to make sure relative paths with directories work, also
442  // test spaces:
443  // "app_dir\url fixer-upper dir\url fixer-upper existing file.txt"
444  FilePath sub_dir(FILE_PATH_LITERAL("url fixer-upper dir"));
445  FilePath sub_file(FILE_PATH_LITERAL("url fixer-upper existing file.txt"));
446  FilePath new_dir = dir.Append(sub_dir);
447  file_util::CreateDirectory(new_dir);
448  ASSERT_TRUE(MakeTempFile(new_dir, sub_file, &full_path));
449  ASSERT_TRUE(file_util::AbsolutePath(&full_path));
450
451  // test file in the subdir
452  FilePath relative_file = sub_dir.Append(sub_file);
453  EXPECT_TRUE(IsMatchingFileURL(URLFixerUpper::FixupRelativeFile(dir,
454      relative_file).possibly_invalid_spec(), full_path));
455
456  // test file in the subdir with different slashes and escaping.
457  FilePath::StringType relative_file_str = sub_dir.value() +
458      FILE_PATH_LITERAL("/") + sub_file.value();
459  ReplaceSubstringsAfterOffset(&relative_file_str, 0,
460      FILE_PATH_LITERAL(" "), FILE_PATH_LITERAL("%20"));
461  EXPECT_TRUE(IsMatchingFileURL(URLFixerUpper::FixupRelativeFile(dir,
462      FilePath(relative_file_str)).possibly_invalid_spec(), full_path));
463
464  // test relative directories and duplicate slashes
465  // (should resolve to the same file as above)
466  relative_file_str = sub_dir.value() + FILE_PATH_LITERAL("/../") +
467      sub_dir.value() + FILE_PATH_LITERAL("///./") + sub_file.value();
468  EXPECT_TRUE(IsMatchingFileURL(URLFixerUpper::FixupRelativeFile(dir,
469      FilePath(relative_file_str)).possibly_invalid_spec(), full_path));
470
471  // done with the subdir
472  EXPECT_TRUE(file_util::Delete(full_path, false));
473  EXPECT_TRUE(file_util::Delete(new_dir, true));
474}
475