net/url_request/url_request_file_job.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 // For loading files, we make use of overlapped i/o to ensure that reading from
   6 // the filesystem (e.g., a network filesystem) does not block the calling
   7 // thread.  An alternative approach would be to use a background thread or pool
   8 // of threads, but it seems better to leverage the operating system's ability
   9 // to do background file reads for us.
  10 //
  11 // Since overlapped reads require a 'static' buffer for the duration of the
  12 // asynchronous read, the URLRequestFileJob keeps a buffer as a member var.  In
  13 // URLRequestFileJob::Read, data is simply copied from the object's buffer into
  14 // the given buffer.  If there is no data to copy, the URLRequestFileJob
  15 // attempts to read more from the file to fill its buffer.  If reading from the
  16 // file does not complete synchronously, then the URLRequestFileJob waits for a
  17 // signal from the OS that the overlapped read has completed.  It does so by
  18 // leveraging the MessageLoop::WatchObject API.
  19
  20 #include "net/url_request/url_request_file_job.h"
  21
  22 #include "base/bind.h"
  23 #include "base/compiler_specific.h"
  24 #include "base/file_util.h"
  25 #include "base/message_loop/message_loop.h"
  26 #include "base/strings/string_util.h"
  27 #include "base/synchronization/lock.h"
  28 #include "base/task_runner.h"
  29 #include "base/threading/thread_restrictions.h"
  30 #include "build/build_config.h"
  31 #include "net/base/file_stream.h"
  32 #include "net/base/io_buffer.h"
  33 #include "net/base/load_flags.h"
  34 #include "net/base/mime_util.h"
  35 #include "net/base/net_errors.h"
  36 #include "net/base/net_util.h"
  37 #include "net/filter/filter.h"
  38 #include "net/http/http_util.h"
  39 #include "net/url_request/url_request_error_job.h"
  40 #include "net/url_request/url_request_file_dir_job.h"
  41 #include "url/gurl.h"
  42
  43 #if defined(OS_WIN)
  44 #include "base/win/shortcut.h"
  45 #endif
  46
  47 namespace net {
  48
  49 URLRequestFileJob::FileMetaInfo::FileMetaInfo()
  50     : file_size(0),
  51       mime_type_result(false),
  52       file_exists(false),
  53       is_directory(false) {
  54 }
  55
  56 URLRequestFileJob::URLRequestFileJob(
  57     URLRequest* request,
  58     NetworkDelegate* network_delegate,
  59     const base::FilePath& file_path,
  60     const scoped_refptr<base::TaskRunner>& file_task_runner)
  61     : URLRequestJob(request, network_delegate),
  62       file_path_(file_path),
  63       stream_(new FileStream(NULL, file_task_runner)),
  64       file_task_runner_(file_task_runner),
  65       remaining_bytes_(0),
  66       weak_ptr_factory_(this) {}
  67
  68 void URLRequestFileJob::Start() {
  69   FileMetaInfo* meta_info = new FileMetaInfo();
  70   file_task_runner_->PostTaskAndReply(
  71       FROM_HERE,
  72       base::Bind(&URLRequestFileJob::FetchMetaInfo, file_path_,
  73                  base::Unretained(meta_info)),
  74       base::Bind(&URLRequestFileJob::DidFetchMetaInfo,
  75                  weak_ptr_factory_.GetWeakPtr(),
  76                  base::Owned(meta_info)));
  77 }
  78
  79 void URLRequestFileJob::Kill() {
  80   stream_.reset();
  81   weak_ptr_factory_.InvalidateWeakPtrs();
  82
  83   URLRequestJob::Kill();
  84 }
  85
  86 bool URLRequestFileJob::ReadRawData(IOBuffer* dest, int dest_size,
  87                                     int *bytes_read) {
  88   DCHECK_NE(dest_size, 0);
  89   DCHECK(bytes_read);
  90   DCHECK_GE(remaining_bytes_, 0);
  91
  92   if (remaining_bytes_ < dest_size)
  93     dest_size = static_cast<int>(remaining_bytes_);
  94
  95   // If we should copy zero bytes because |remaining_bytes_| is zero, short
  96   // circuit here.
  97   if (!dest_size) {
  98     *bytes_read = 0;
  99     return true;
 100   }
 101
 102   int rv = stream_->Read(dest, dest_size,
 103                          base::Bind(&URLRequestFileJob::DidRead,
 104                                     weak_ptr_factory_.GetWeakPtr()));
 105   if (rv >= 0) {
 106     // Data is immediately available.
 107     *bytes_read = rv;
 108     remaining_bytes_ -= rv;
 109     DCHECK_GE(remaining_bytes_, 0);
 110     return true;
 111   }
 112
 113   // Otherwise, a read error occured.  We may just need to wait...
 114   if (rv == ERR_IO_PENDING) {
 115     SetStatus(URLRequestStatus(URLRequestStatus::IO_PENDING, 0));
 116   } else {
 117     NotifyDone(URLRequestStatus(URLRequestStatus::FAILED, rv));
 118   }
 119   return false;
 120 }
 121
 122 bool URLRequestFileJob::IsRedirectResponse(GURL* location,
 123                                            int* http_status_code) {
 124   if (meta_info_.is_directory) {
 125     // This happens when we discovered the file is a directory, so needs a
 126     // slash at the end of the path.
 127     std::string new_path = request_->url().path();
 128     new_path.push_back('/');
 129     GURL::Replacements replacements;
 130     replacements.SetPathStr(new_path);
 131
 132     *location = request_->url().ReplaceComponents(replacements);
 133     *http_status_code = 301;  // simulate a permanent redirect
 134     return true;
 135   }
 136
 137 #if defined(OS_WIN)
 138   // Follow a Windows shortcut.
 139   // We just resolve .lnk file, ignore others.
 140   if (!LowerCaseEqualsASCII(file_path_.Extension(), ".lnk"))
 141     return false;
 142
 143   base::FilePath new_path = file_path_;
 144   bool resolved;
 145   resolved = base::win::ResolveShortcut(new_path, &new_path, NULL);
 146
 147   // If shortcut is not resolved succesfully, do not redirect.
 148   if (!resolved)
 149     return false;
 150
 151   *location = FilePathToFileURL(new_path);
 152   *http_status_code = 301;
 153   return true;
 154 #else
 155   return false;
 156 #endif
 157 }
 158
 159 Filter* URLRequestFileJob::SetupFilter() const {
 160   // Bug 9936 - .svgz files needs to be decompressed.
 161   return LowerCaseEqualsASCII(file_path_.Extension(), ".svgz")
 162       ? Filter::GZipFactory() : NULL;
 163 }
 164
 165 bool URLRequestFileJob::GetMimeType(std::string* mime_type) const {
 166   DCHECK(request_);
 167   if (meta_info_.mime_type_result) {
 168     *mime_type = meta_info_.mime_type;
 169     return true;
 170   }
 171   return false;
 172 }
 173
 174 void URLRequestFileJob::SetExtraRequestHeaders(
 175     const HttpRequestHeaders& headers) {
 176   std::string range_header;
 177   if (headers.GetHeader(HttpRequestHeaders::kRange, &range_header)) {
 178     // We only care about "Range" header here.
 179     std::vector<HttpByteRange> ranges;
 180     if (HttpUtil::ParseRangeHeader(range_header, &ranges)) {
 181       if (ranges.size() == 1) {
 182         byte_range_ = ranges[0];
 183       } else {
 184         // We don't support multiple range requests in one single URL request,
 185         // because we need to do multipart encoding here.
 186         // TODO(hclam): decide whether we want to support multiple range
 187         // requests.
 188         NotifyDone(URLRequestStatus(URLRequestStatus::FAILED,
 189                                     ERR_REQUEST_RANGE_NOT_SATISFIABLE));
 190       }
 191     }
 192   }
 193 }
 194
 195 URLRequestFileJob::~URLRequestFileJob() {
 196 }
 197
 198 void URLRequestFileJob::FetchMetaInfo(const base::FilePath& file_path,
 199                                       FileMetaInfo* meta_info) {
 200   base::File::Info file_info;
 201   meta_info->file_exists = base::GetFileInfo(file_path, &file_info);
 202   if (meta_info->file_exists) {
 203     meta_info->file_size = file_info.size;
 204     meta_info->is_directory = file_info.is_directory;
 205   }
 206   // On Windows GetMimeTypeFromFile() goes to the registry. Thus it should be
 207   // done in WorkerPool.
 208   meta_info->mime_type_result = GetMimeTypeFromFile(file_path,
 209                                                     &meta_info->mime_type);
 210 }
 211
 212 void URLRequestFileJob::DidFetchMetaInfo(const FileMetaInfo* meta_info) {
 213   meta_info_ = *meta_info;
 214
 215   // We use URLRequestFileJob to handle files as well as directories without
 216   // trailing slash.
 217   // If a directory does not exist, we return ERR_FILE_NOT_FOUND. Otherwise,
 218   // we will append trailing slash and redirect to FileDirJob.
 219   // A special case is "\" on Windows. We should resolve as invalid.
 220   // However, Windows resolves "\" to "C:\", thus reports it as existent.
 221   // So what happens is we append it with trailing slash and redirect it to
 222   // FileDirJob where it is resolved as invalid.
 223   if (!meta_info_.file_exists) {
 224     DidOpen(ERR_FILE_NOT_FOUND);
 225     return;
 226   }
 227   if (meta_info_.is_directory) {
 228     DidOpen(OK);
 229     return;
 230   }
 231
 232   int flags = base::PLATFORM_FILE_OPEN |
 233               base::PLATFORM_FILE_READ |
 234               base::PLATFORM_FILE_ASYNC;
 235   int rv = stream_->Open(file_path_, flags,
 236                          base::Bind(&URLRequestFileJob::DidOpen,
 237                                     weak_ptr_factory_.GetWeakPtr()));
 238   if (rv != ERR_IO_PENDING)
 239     DidOpen(rv);
 240 }
 241
 242 void URLRequestFileJob::DidOpen(int result) {
 243   if (result != OK) {
 244     NotifyDone(URLRequestStatus(URLRequestStatus::FAILED, result));
 245     return;
 246   }
 247
 248   if (!byte_range_.ComputeBounds(meta_info_.file_size)) {
 249     NotifyDone(URLRequestStatus(URLRequestStatus::FAILED,
 250                ERR_REQUEST_RANGE_NOT_SATISFIABLE));
 251     return;
 252   }
 253
 254   remaining_bytes_ = byte_range_.last_byte_position() -
 255                      byte_range_.first_byte_position() + 1;
 256   DCHECK_GE(remaining_bytes_, 0);
 257
 258   if (remaining_bytes_ > 0 && byte_range_.first_byte_position() != 0) {
 259     int rv = stream_->Seek(FROM_BEGIN, byte_range_.first_byte_position(),
 260                            base::Bind(&URLRequestFileJob::DidSeek,
 261                                       weak_ptr_factory_.GetWeakPtr()));
 262     if (rv != ERR_IO_PENDING) {
 263       // stream_->Seek() failed, so pass an intentionally erroneous value
 264       // into DidSeek().
 265       DidSeek(-1);
 266     }
 267   } else {
 268     // We didn't need to call stream_->Seek() at all, so we pass to DidSeek()
 269     // the value that would mean seek success. This way we skip the code
 270     // handling seek failure.
 271     DidSeek(byte_range_.first_byte_position());
 272   }
 273 }
 274
 275 void URLRequestFileJob::DidSeek(int64 result) {
 276   if (result != byte_range_.first_byte_position()) {
 277     NotifyDone(URLRequestStatus(URLRequestStatus::FAILED,
 278                                 ERR_REQUEST_RANGE_NOT_SATISFIABLE));
 279     return;
 280   }
 281
 282   set_expected_content_size(remaining_bytes_);
 283   NotifyHeadersComplete();
 284 }
 285
 286 void URLRequestFileJob::DidRead(int result) {
 287   if (result > 0) {
 288     SetStatus(URLRequestStatus());  // Clear the IO_PENDING status
 289   } else if (result == 0) {
 290     NotifyDone(URLRequestStatus());
 291   } else {
 292     NotifyDone(URLRequestStatus(URLRequestStatus::FAILED, result));
 293   }
 294
 295   remaining_bytes_ -= result;
 296   DCHECK_GE(remaining_bytes_, 0);
 297
 298   NotifyReadComplete(result);
 299 }
 300
 301 }  // namespace net