net/base/filename_util_internal.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 "net/base/filename_util.h"
   6
   7 #include "base/files/file_path.h"
   8 #include "base/files/file_util.h"
   9 #include "base/path_service.h"
  10 #include "base/strings/string_util.h"
  11 #include "base/strings/sys_string_conversions.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "base/threading/thread_restrictions.h"
  14 #include "net/base/escape.h"
  15 #include "net/base/filename_util_internal.h"
  16 #include "net/base/mime_util.h"
  17 #include "net/base/net_string_util.h"
  18 #include "net/http/http_content_disposition.h"
  19 #include "url/gurl.h"
  20
  21 namespace net {
  22
  23 void SanitizeGeneratedFileName(base::FilePath::StringType* filename,
  24                                bool replace_trailing) {
  25   const base::FilePath::CharType kReplace[] = FILE_PATH_LITERAL("-");
  26   if (filename->empty())
  27     return;
  28   if (replace_trailing) {
  29     // Handle CreateFile() stripping trailing dots and spaces on filenames
  30     // http://support.microsoft.com/kb/115827
  31     size_t length = filename->size();
  32     size_t pos = filename->find_last_not_of(FILE_PATH_LITERAL(" ."));
  33     filename->resize((pos == std::string::npos) ? 0 : (pos + 1));
  34     base::TrimWhitespace(*filename, base::TRIM_TRAILING, filename);
  35     if (filename->empty())
  36       return;
  37     size_t trimmed = length - filename->size();
  38     if (trimmed)
  39       filename->insert(filename->end(), trimmed, kReplace[0]);
  40   }
  41   base::TrimString(*filename, FILE_PATH_LITERAL("."), filename);
  42   if (filename->empty())
  43     return;
  44   // Replace any path information by changing path separators.
  45   ReplaceSubstringsAfterOffset(filename, 0, FILE_PATH_LITERAL("/"), kReplace);
  46   ReplaceSubstringsAfterOffset(filename, 0, FILE_PATH_LITERAL("\\"), kReplace);
  47 }
  48
  49 // Returns the filename determined from the last component of the path portion
  50 // of the URL.  Returns an empty string if the URL doesn't have a path or is
  51 // invalid. If the generated filename is not reliable,
  52 // |should_overwrite_extension| will be set to true, in which case a better
  53 // extension should be determined based on the content type.
  54 std::string GetFileNameFromURL(const GURL& url,
  55                                const std::string& referrer_charset,
  56                                bool* should_overwrite_extension) {
  57   // about: and data: URLs don't have file names, but esp. data: URLs may
  58   // contain parts that look like ones (i.e., contain a slash).  Therefore we
  59   // don't attempt to divine a file name out of them.
  60   if (!url.is_valid() || url.SchemeIs("about") || url.SchemeIs("data"))
  61     return std::string();
  62
  63   const std::string unescaped_url_filename = UnescapeURLComponent(
  64       url.ExtractFileName(),
  65       UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
  66
  67   // The URL's path should be escaped UTF-8, but may not be.
  68   std::string decoded_filename = unescaped_url_filename;
  69   if (!base::IsStringUTF8(decoded_filename)) {
  70     // TODO(jshin): this is probably not robust enough. To be sure, we need
  71     // encoding detection.
  72     base::string16 utf16_output;
  73     if (!referrer_charset.empty() &&
  74         net::ConvertToUTF16(
  75             unescaped_url_filename, referrer_charset.c_str(), &utf16_output)) {
  76       decoded_filename = base::UTF16ToUTF8(utf16_output);
  77     } else {
  78       decoded_filename =
  79           base::WideToUTF8(base::SysNativeMBToWide(unescaped_url_filename));
  80     }
  81   }
  82   // If the URL contains a (possibly empty) query, assume it is a generator, and
  83   // allow the determined extension to be overwritten.
  84   *should_overwrite_extension = !decoded_filename.empty() && url.has_query();
  85
  86   return decoded_filename;
  87 }
  88
  89 // Returns whether the specified extension is automatically integrated into the
  90 // windows shell.
  91 bool IsShellIntegratedExtension(const base::FilePath::StringType& extension) {
  92   base::FilePath::StringType extension_lower =
  93       base::StringToLowerASCII(extension);
  94
  95   // http://msdn.microsoft.com/en-us/library/ms811694.aspx
  96   // Right-clicking on shortcuts can be magical.
  97   if ((extension_lower == FILE_PATH_LITERAL("local")) ||
  98       (extension_lower == FILE_PATH_LITERAL("lnk")))
  99     return true;
 100
 101   // http://www.juniper.net/security/auto/vulnerabilities/vuln2612.html
 102   // Files become magical if they end in a CLSID, so block such extensions.
 103   if (!extension_lower.empty() &&
 104       (extension_lower[0] == FILE_PATH_LITERAL('{')) &&
 105       (extension_lower[extension_lower.length() - 1] == FILE_PATH_LITERAL('}')))
 106     return true;
 107   return false;
 108 }
 109
 110 // Returns whether the specified file name is a reserved name on windows.
 111 // This includes names like "com2.zip" (which correspond to devices) and
 112 // desktop.ini and thumbs.db which have special meaning to the windows shell.
 113 bool IsReservedName(const base::FilePath::StringType& filename) {
 114   // This list is taken from the MSDN article "Naming a file"
 115   // http://msdn2.microsoft.com/en-us/library/aa365247(VS.85).aspx
 116   // I also added clock$ because GetSaveFileName seems to consider it as a
 117   // reserved name too.
 118   static const char* const known_devices[] = {
 119       "con",  "prn",  "aux",  "nul",  "com1", "com2", "com3",  "com4",
 120       "com5", "com6", "com7", "com8", "com9", "lpt1", "lpt2",  "lpt3",
 121       "lpt4", "lpt5", "lpt6", "lpt7", "lpt8", "lpt9", "clock$"};
 122 #if defined(OS_WIN)
 123   std::string filename_lower =
 124       base::StringToLowerASCII(base::WideToUTF8(filename));
 125 #elif defined(OS_POSIX)
 126   std::string filename_lower = base::StringToLowerASCII(filename);
 127 #endif
 128
 129   for (size_t i = 0; i < arraysize(known_devices); ++i) {
 130     // Exact match.
 131     if (filename_lower == known_devices[i])
 132       return true;
 133     // Starts with "DEVICE.".
 134     if (filename_lower.find(std::string(known_devices[i]) + ".") == 0)
 135       return true;
 136   }
 137
 138   static const char* const magic_names[] = {
 139     // These file names are used by the "Customize folder" feature of the shell.
 140     "desktop.ini",
 141     "thumbs.db",
 142   };
 143
 144   for (size_t i = 0; i < arraysize(magic_names); ++i) {
 145     if (filename_lower == magic_names[i])
 146       return true;
 147   }
 148
 149   return false;
 150 }
 151
 152 // Examines the current extension in |file_name| and modifies it if necessary in
 153 // order to ensure the filename is safe.  If |file_name| doesn't contain an
 154 // extension or if |ignore_extension| is true, then a new extension will be
 155 // constructed based on the |mime_type|.
 156 //
 157 // We're addressing two things here:
 158 //
 159 // 1) Usability.  If there is no reliable file extension, we want to guess a
 160 //    reasonable file extension based on the content type.
 161 //
 162 // 2) Shell integration.  Some file extensions automatically integrate with the
 163 //    shell.  We block these extensions to prevent a malicious web site from
 164 //    integrating with the user's shell.
 165 void EnsureSafeExtension(const std::string& mime_type,
 166                          bool ignore_extension,
 167                          base::FilePath* file_name) {
 168   // See if our file name already contains an extension.
 169   base::FilePath::StringType extension = file_name->Extension();
 170   if (!extension.empty())
 171     extension.erase(extension.begin());  // Erase preceding '.'.
 172
 173   if ((ignore_extension || extension.empty()) && !mime_type.empty()) {
 174     base::FilePath::StringType preferred_mime_extension;
 175     std::vector<base::FilePath::StringType> all_mime_extensions;
 176     net::GetPreferredExtensionForMimeType(mime_type, &preferred_mime_extension);
 177     net::GetExtensionsForMimeType(mime_type, &all_mime_extensions);
 178     // If the existing extension is in the list of valid extensions for the
 179     // given type, use it. This avoids doing things like pointlessly renaming
 180     // "foo.jpg" to "foo.jpeg".
 181     if (std::find(all_mime_extensions.begin(),
 182                   all_mime_extensions.end(),
 183                   extension) != all_mime_extensions.end()) {
 184       // leave |extension| alone
 185     } else if (!preferred_mime_extension.empty()) {
 186       extension = preferred_mime_extension;
 187     }
 188   }
 189
 190 #if defined(OS_WIN)
 191   static const base::FilePath::CharType default_extension[] =
 192       FILE_PATH_LITERAL("download");
 193
 194   // Rename shell-integrated extensions.
 195   // TODO(asanka): Consider stripping out the bad extension and replacing it
 196   // with the preferred extension for the MIME type if one is available.
 197   if (IsShellIntegratedExtension(extension))
 198     extension.assign(default_extension);
 199 #endif
 200
 201   *file_name = file_name->ReplaceExtension(extension);
 202 }
 203
 204 bool FilePathToString16(const base::FilePath& path, base::string16* converted) {
 205 #if defined(OS_WIN)
 206   return base::WideToUTF16(
 207       path.value().c_str(), path.value().size(), converted);
 208 #elif defined(OS_POSIX)
 209   std::string component8 = path.AsUTF8Unsafe();
 210   return !component8.empty() &&
 211          base::UTF8ToUTF16(component8.c_str(), component8.size(), converted);
 212 #endif
 213 }
 214
 215 base::string16 GetSuggestedFilenameImpl(
 216     const GURL& url,
 217     const std::string& content_disposition,
 218     const std::string& referrer_charset,
 219     const std::string& suggested_name,
 220     const std::string& mime_type,
 221     const std::string& default_name,
 222     ReplaceIllegalCharactersCallback replace_illegal_characters_callback) {
 223   // TODO: this function to be updated to match the httpbis recommendations.
 224   // Talk to abarth for the latest news.
 225
 226   // We don't translate this fallback string, "download". If localization is
 227   // needed, the caller should provide localized fallback in |default_name|.
 228   static const base::FilePath::CharType kFinalFallbackName[] =
 229       FILE_PATH_LITERAL("download");
 230   std::string filename;  // In UTF-8
 231   bool overwrite_extension = false;
 232
 233   // Try to extract a filename from content-disposition first.
 234   if (!content_disposition.empty()) {
 235     HttpContentDisposition header(content_disposition, referrer_charset);
 236     filename = header.filename();
 237   }
 238
 239   // Then try to use the suggested name.
 240   if (filename.empty() && !suggested_name.empty())
 241     filename = suggested_name;
 242
 243   // Now try extracting the filename from the URL.  GetFileNameFromURL() only
 244   // looks at the last component of the URL and doesn't return the hostname as a
 245   // failover.
 246   if (filename.empty())
 247     filename = GetFileNameFromURL(url, referrer_charset, &overwrite_extension);
 248
 249   // Finally try the URL hostname, but only if there's no default specified in
 250   // |default_name|.  Some schemes (e.g.: file:, about:, data:) do not have a
 251   // host name.
 252   if (filename.empty() && default_name.empty() && url.is_valid() &&
 253       !url.host().empty()) {
 254     // TODO(jungshik) : Decode a 'punycoded' IDN hostname. (bug 1264451)
 255     filename = url.host();
 256   }
 257
 258   bool replace_trailing = false;
 259   base::FilePath::StringType result_str, default_name_str;
 260 #if defined(OS_WIN)
 261   replace_trailing = true;
 262   result_str = base::UTF8ToUTF16(filename);
 263   default_name_str = base::UTF8ToUTF16(default_name);
 264 #else
 265   result_str = filename;
 266   default_name_str = default_name;
 267 #endif
 268   SanitizeGeneratedFileName(&result_str, replace_trailing);
 269   if (result_str.find_last_not_of(FILE_PATH_LITERAL("-_")) ==
 270       base::FilePath::StringType::npos) {
 271     result_str = !default_name_str.empty()
 272                      ? default_name_str
 273                      : base::FilePath::StringType(kFinalFallbackName);
 274     overwrite_extension = false;
 275   }
 276   replace_illegal_characters_callback.Run(&result_str, '-');
 277   base::FilePath result(result_str);
 278   GenerateSafeFileName(mime_type, overwrite_extension, &result);
 279
 280   base::string16 result16;
 281   if (!FilePathToString16(result, &result16)) {
 282     result = base::FilePath(default_name_str);
 283     if (!FilePathToString16(result, &result16)) {
 284       result = base::FilePath(kFinalFallbackName);
 285       FilePathToString16(result, &result16);
 286     }
 287   }
 288   return result16;
 289 }
 290
 291 base::FilePath GenerateFileNameImpl(
 292     const GURL& url,
 293     const std::string& content_disposition,
 294     const std::string& referrer_charset,
 295     const std::string& suggested_name,
 296     const std::string& mime_type,
 297     const std::string& default_file_name,
 298     ReplaceIllegalCharactersCallback replace_illegal_characters_callback) {
 299   base::string16 file_name =
 300       GetSuggestedFilenameImpl(url,
 301                                content_disposition,
 302                                referrer_charset,
 303                                suggested_name,
 304                                mime_type,
 305                                default_file_name,
 306                                replace_illegal_characters_callback);
 307
 308 #if defined(OS_WIN)
 309   base::FilePath generated_name(file_name);
 310 #else
 311   base::FilePath generated_name(
 312       base::SysWideToNativeMB(base::UTF16ToWide(file_name)));
 313 #endif
 314
 315   DCHECK(!generated_name.empty());
 316
 317   return generated_name;
 318 }
 319
 320 }  // namespace net