net/base/filename_util.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/net_string_util.h"
  17 #include "net/http/http_content_disposition.h"
  18 #include "url/gurl.h"
  19
  20 namespace net {
  21
  22 // Prefix to prepend to get a file URL.
  23 static const base::FilePath::CharType kFileURLPrefix[] =
  24     FILE_PATH_LITERAL("file:///");
  25
  26 GURL FilePathToFileURL(const base::FilePath& path) {
  27   // Produce a URL like "file:///C:/foo" for a regular file, or
  28   // "file://///server/path" for UNC. The URL canonicalizer will fix up the
  29   // latter case to be the canonical UNC form: "file://server/path"
  30   base::FilePath::StringType url_string(kFileURLPrefix);
  31   if (!path.IsAbsolute()) {
  32     base::FilePath current_dir;
  33     PathService::Get(base::DIR_CURRENT, &current_dir);
  34     url_string.append(current_dir.value());
  35     url_string.push_back(base::FilePath::kSeparators[0]);
  36   }
  37   url_string.append(path.value());
  38
  39   // Now do replacement of some characters. Since we assume the input is a
  40   // literal filename, anything the URL parser might consider special should
  41   // be escaped here.
  42
  43   // must be the first substitution since others will introduce percents as the
  44   // escape character
  45   ReplaceSubstringsAfterOffset(
  46       &url_string, 0, FILE_PATH_LITERAL("%"), FILE_PATH_LITERAL("%25"));
  47
  48   // semicolon is supposed to be some kind of separator according to RFC 2396
  49   ReplaceSubstringsAfterOffset(
  50       &url_string, 0, FILE_PATH_LITERAL(";"), FILE_PATH_LITERAL("%3B"));
  51
  52   ReplaceSubstringsAfterOffset(
  53       &url_string, 0, FILE_PATH_LITERAL("#"), FILE_PATH_LITERAL("%23"));
  54
  55   ReplaceSubstringsAfterOffset(
  56       &url_string, 0, FILE_PATH_LITERAL("?"), FILE_PATH_LITERAL("%3F"));
  57
  58 #if defined(OS_POSIX)
  59   ReplaceSubstringsAfterOffset(
  60       &url_string, 0, FILE_PATH_LITERAL("\\"), FILE_PATH_LITERAL("%5C"));
  61 #endif
  62
  63   return GURL(url_string);
  64 }
  65
  66 bool FileURLToFilePath(const GURL& url, base::FilePath* file_path) {
  67   *file_path = base::FilePath();
  68   base::FilePath::StringType& file_path_str =
  69       const_cast<base::FilePath::StringType&>(file_path->value());
  70   file_path_str.clear();
  71
  72   if (!url.is_valid())
  73     return false;
  74
  75 #if defined(OS_WIN)
  76   std::string path;
  77   std::string host = url.host();
  78   if (host.empty()) {
  79     // URL contains no host, the path is the filename. In this case, the path
  80     // will probably be preceeded with a slash, as in "/C:/foo.txt", so we
  81     // trim out that here.
  82     path = url.path();
  83     size_t first_non_slash = path.find_first_not_of("/\\");
  84     if (first_non_slash != std::string::npos && first_non_slash > 0)
  85       path.erase(0, first_non_slash);
  86   } else {
  87     // URL contains a host: this means it's UNC. We keep the preceeding slash
  88     // on the path.
  89     path = "\\\\";
  90     path.append(host);
  91     path.append(url.path());
  92   }
  93   std::replace(path.begin(), path.end(), '/', '\\');
  94 #else  // defined(OS_WIN)
  95   // Firefox seems to ignore the "host" of a file url if there is one. That is,
  96   // file://foo/bar.txt maps to /bar.txt.
  97   // TODO(dhg): This should probably take into account UNCs which could
  98   // include a hostname other than localhost or blank
  99   std::string path = url.path();
 100 #endif  // !defined(OS_WIN)
 101
 102   if (path.empty())
 103     return false;
 104
 105   // GURL stores strings as percent-encoded 8-bit, this will undo if possible.
 106   path = UnescapeURLComponent(
 107       path, UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS);
 108
 109 #if defined(OS_WIN)
 110   if (base::IsStringUTF8(path)) {
 111     file_path_str.assign(base::UTF8ToWide(path));
 112     // We used to try too hard and see if |path| made up entirely of
 113     // the 1st 256 characters in the Unicode was a zero-extended UTF-16.
 114     // If so, we converted it to 'Latin-1' and checked if the result was UTF-8.
 115     // If the check passed, we converted the result to UTF-8.
 116     // Otherwise, we treated the result as the native OS encoding.
 117     // However, that led to http://crbug.com/4619 and http://crbug.com/14153
 118   } else {
 119     // Not UTF-8, assume encoding is native codepage and we're done. We know we
 120     // are giving the conversion function a nonempty string, and it may fail if
 121     // the given string is not in the current encoding and give us an empty
 122     // string back. We detect this and report failure.
 123     file_path_str = base::SysNativeMBToWide(path);
 124   }
 125 #else  // defined(OS_WIN)
 126   // Collapse multiple path slashes into a single path slash.
 127   std::string new_path;
 128   do {
 129     new_path = path;
 130     ReplaceSubstringsAfterOffset(&new_path, 0, "//", "/");
 131     path.swap(new_path);
 132   } while (new_path != path);
 133
 134   file_path_str.assign(path);
 135 #endif  // !defined(OS_WIN)
 136
 137   return !file_path_str.empty();
 138 }
 139
 140 void GenerateSafeFileName(const std::string& mime_type,
 141                           bool ignore_extension,
 142                           base::FilePath* file_path) {
 143   // Make sure we get the right file extension
 144   EnsureSafeExtension(mime_type, ignore_extension, file_path);
 145
 146 #if defined(OS_WIN)
 147   // Prepend "_" to the file name if it's a reserved name
 148   base::FilePath::StringType leaf_name = file_path->BaseName().value();
 149   DCHECK(!leaf_name.empty());
 150   if (IsReservedName(leaf_name)) {
 151     leaf_name = base::FilePath::StringType(FILE_PATH_LITERAL("_")) + leaf_name;
 152     *file_path = file_path->DirName();
 153     if (file_path->value() == base::FilePath::kCurrentDirectory) {
 154       *file_path = base::FilePath(leaf_name);
 155     } else {
 156       *file_path = file_path->Append(leaf_name);
 157     }
 158   }
 159 #endif
 160 }
 161
 162 }  // namespace net