src/content/browser/download/download_resource_handler.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 "content/browser/download/download_resource_handler.h"
   6
   7 #include <string>
   8
   9 #include "base/bind.h"
  10 #include "base/logging.h"
  11 #include "base/message_loop/message_loop_proxy.h"
  12 #include "base/metrics/histogram.h"
  13 #include "base/metrics/stats_counters.h"
  14 #include "base/strings/stringprintf.h"
  15 #include "content/browser/byte_stream.h"
  16 #include "content/browser/download/download_create_info.h"
  17 #include "content/browser/download/download_interrupt_reasons_impl.h"
  18 #include "content/browser/download/download_manager_impl.h"
  19 #include "content/browser/download/download_request_handle.h"
  20 #include "content/browser/download/download_stats.h"
  21 #include "content/browser/loader/resource_dispatcher_host_impl.h"
  22 #include "content/browser/loader/resource_request_info_impl.h"
  23 #include "content/public/browser/browser_thread.h"
  24 #include "content/public/browser/download_interrupt_reasons.h"
  25 #include "content/public/browser/download_item.h"
  26 #include "content/public/browser/download_manager_delegate.h"
  27 #include "content/public/browser/navigation_entry.h"
  28 #include "content/public/browser/web_contents.h"
  29 #include "content/public/common/resource_response.h"
  30 #include "net/base/io_buffer.h"
  31 #include "net/base/net_errors.h"
  32 #include "net/http/http_response_headers.h"
  33 #include "net/http/http_status_code.h"
  34 #include "net/url_request/url_request_context.h"
  35
  36 namespace content {
  37
  38 struct DownloadResourceHandler::DownloadTabInfo {
  39   GURL tab_url;
  40   GURL tab_referrer_url;
  41 };
  42
  43 namespace {
  44
  45 void CallStartedCBOnUIThread(
  46     const DownloadUrlParameters::OnStartedCallback& started_cb,
  47     DownloadItem* item,
  48     DownloadInterruptReason interrupt_reason) {
  49   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  50
  51   if (started_cb.is_null())
  52     return;
  53   started_cb.Run(item, interrupt_reason);
  54 }
  55
  56 // Static function in order to prevent any accidental accesses to
  57 // DownloadResourceHandler members from the UI thread.
  58 static void StartOnUIThread(
  59     scoped_ptr<DownloadCreateInfo> info,
  60     DownloadResourceHandler::DownloadTabInfo* tab_info,
  61     scoped_ptr<ByteStreamReader> stream,
  62     const DownloadUrlParameters::OnStartedCallback& started_cb) {
  63   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  64
  65   DownloadManager* download_manager = info->request_handle.GetDownloadManager();
  66   if (!download_manager) {
  67     // NULL in unittests or if the page closed right after starting the
  68     // download.
  69     if (!started_cb.is_null())
  70       started_cb.Run(NULL, DOWNLOAD_INTERRUPT_REASON_USER_CANCELED);
  71
  72     // |stream| gets deleted on non-FILE thread, but it's ok since
  73     // we're not using stream_writer_ yet.
  74
  75     return;
  76   }
  77
  78   info->tab_url = tab_info->tab_url;
  79   info->tab_referrer_url = tab_info->tab_referrer_url;
  80
  81   download_manager->StartDownload(info.Pass(), stream.Pass(), started_cb);
  82 }
  83
  84 void InitializeDownloadTabInfoOnUIThread(
  85     const DownloadRequestHandle& request_handle,
  86     DownloadResourceHandler::DownloadTabInfo* tab_info) {
  87   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  88
  89   WebContents* web_contents = request_handle.GetWebContents();
  90   if (web_contents) {
  91     NavigationEntry* entry = web_contents->GetController().GetVisibleEntry();
  92     if (entry) {
  93       tab_info->tab_url = entry->GetURL();
  94       tab_info->tab_referrer_url = entry->GetReferrer().url;
  95     }
  96   }
  97 }
  98
  99 }  // namespace
 100
 101 const int DownloadResourceHandler::kDownloadByteStreamSize = 100 * 1024;
 102
 103 DownloadResourceHandler::DownloadResourceHandler(
 104     uint32 id,
 105     net::URLRequest* request,
 106     const DownloadUrlParameters::OnStartedCallback& started_cb,
 107     scoped_ptr<DownloadSaveInfo> save_info)
 108     : ResourceHandler(request),
 109       download_id_(id),
 110       started_cb_(started_cb),
 111       save_info_(save_info.Pass()),
 112       last_buffer_size_(0),
 113       bytes_read_(0),
 114       pause_count_(0),
 115       was_deferred_(false),
 116       on_response_started_called_(false) {
 117   RecordDownloadCount(UNTHROTTLED_COUNT);
 118
 119   // Do UI thread initialization asap after DownloadResourceHandler creation
 120   // since the tab could be navigated before StartOnUIThread gets called.
 121   const ResourceRequestInfoImpl* request_info = GetRequestInfo();
 122   tab_info_ = new DownloadTabInfo();
 123   BrowserThread::PostTask(
 124       BrowserThread::UI,
 125       FROM_HERE,
 126       base::Bind(&InitializeDownloadTabInfoOnUIThread,
 127                  DownloadRequestHandle(AsWeakPtr(),
 128                                        request_info->GetChildID(),
 129                                        request_info->GetRouteID(),
 130                                        request_info->GetRequestID()),
 131                  tab_info_));
 132 }
 133
 134 bool DownloadResourceHandler::OnUploadProgress(int request_id,
 135                                                uint64 position,
 136                                                uint64 size) {
 137   return true;
 138 }
 139
 140 bool DownloadResourceHandler::OnRequestRedirected(
 141     int request_id,
 142     const GURL& url,
 143     ResourceResponse* response,
 144     bool* defer) {
 145   // We treat a download as a main frame load, and thus update the policy URL
 146   // on redirects.
 147   request()->set_first_party_for_cookies(url);
 148   return true;
 149 }
 150
 151 // Send the download creation information to the download thread.
 152 bool DownloadResourceHandler::OnResponseStarted(
 153     int request_id,
 154     ResourceResponse* response,
 155     bool* defer) {
 156   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 157   // There can be only one (call)
 158   DCHECK(!on_response_started_called_);
 159   on_response_started_called_ = true;
 160
 161   VLOG(20) << __FUNCTION__ << "()" << DebugString()
 162            << " request_id = " << request_id;
 163   download_start_time_ = base::TimeTicks::Now();
 164
 165   // If it's a download, we don't want to poison the cache with it.
 166   request()->StopCaching();
 167
 168   // Lower priority as well, so downloads don't contend for resources
 169   // with main frames.
 170   request()->SetPriority(net::IDLE);
 171
 172   // If the content-length header is not present (or contains something other
 173   // than numbers), the incoming content_length is -1 (unknown size).
 174   // Set the content length to 0 to indicate unknown size to DownloadManager.
 175   int64 content_length =
 176       response->head.content_length > 0 ? response->head.content_length : 0;
 177
 178   const ResourceRequestInfoImpl* request_info = GetRequestInfo();
 179
 180   // Deleted in DownloadManager.
 181   scoped_ptr<DownloadCreateInfo> info(
 182       new DownloadCreateInfo(base::Time::Now(),
 183                              content_length,
 184                              request()->net_log(),
 185                              request_info->HasUserGesture(),
 186                              request_info->GetPageTransition(),
 187                              save_info_.Pass()));
 188
 189   // Create the ByteStream for sending data to the download sink.
 190   scoped_ptr<ByteStreamReader> stream_reader;
 191   CreateByteStream(
 192       base::MessageLoopProxy::current(),
 193       BrowserThread::GetMessageLoopProxyForThread(BrowserThread::FILE),
 194       kDownloadByteStreamSize, &stream_writer_, &stream_reader);
 195   stream_writer_->RegisterCallback(
 196       base::Bind(&DownloadResourceHandler::ResumeRequest, AsWeakPtr()));
 197
 198   info->download_id = download_id_;
 199   info->url_chain = request()->url_chain();
 200   info->referrer_url = GURL(request()->referrer());
 201   info->mime_type = response->head.mime_type;
 202   info->remote_address = request()->GetSocketAddress().host();
 203   request()->GetResponseHeaderByName("content-disposition",
 204                                      &info->content_disposition);
 205   RecordDownloadMimeType(info->mime_type);
 206   RecordDownloadContentDisposition(info->content_disposition);
 207
 208   info->request_handle =
 209       DownloadRequestHandle(AsWeakPtr(), request_info->GetChildID(),
 210                             request_info->GetRouteID(),
 211                             request_info->GetRequestID());
 212
 213   // Get the last modified time and etag.
 214   const net::HttpResponseHeaders* headers = request()->response_headers();
 215   if (headers) {
 216     if (headers->HasStrongValidators()) {
 217       // If we don't have strong validators as per RFC 2616 section 13.3.3, then
 218       // we neither store nor use them for range requests.
 219       if (!headers->EnumerateHeader(NULL, "Last-Modified",
 220                                     &info->last_modified))
 221         info->last_modified.clear();
 222       if (!headers->EnumerateHeader(NULL, "ETag", &info->etag))
 223         info->etag.clear();
 224     }
 225
 226     int status = headers->response_code();
 227     if (2 == status / 100  && status != net::HTTP_PARTIAL_CONTENT) {
 228       // Success & not range response; if we asked for a range, we didn't
 229       // get it--reset the file pointers to reflect that.
 230       info->save_info->offset = 0;
 231       info->save_info->hash_state = "";
 232     }
 233
 234     if (!headers->GetMimeType(&info->original_mime_type))
 235       info->original_mime_type.clear();
 236   }
 237
 238   BrowserThread::PostTask(
 239       BrowserThread::UI, FROM_HERE,
 240       base::Bind(&StartOnUIThread,
 241                  base::Passed(&info),
 242                  base::Owned(tab_info_),
 243                  base::Passed(&stream_reader),
 244                  // Pass to StartOnUIThread so that variable
 245                  // access is always on IO thread but function
 246                  // is called on UI thread.
 247                  started_cb_));
 248   // Now owned by the task that was just posted.
 249   tab_info_ = NULL;
 250   // Guaranteed to be called in StartOnUIThread
 251   started_cb_.Reset();
 252
 253   return true;
 254 }
 255
 256 void DownloadResourceHandler::CallStartedCB(
 257     DownloadItem* item,
 258     DownloadInterruptReason interrupt_reason) {
 259   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 260   if (started_cb_.is_null())
 261     return;
 262   BrowserThread::PostTask(
 263       BrowserThread::UI,
 264       FROM_HERE,
 265       base::Bind(
 266           &CallStartedCBOnUIThread, started_cb_, item, interrupt_reason));
 267   started_cb_.Reset();
 268 }
 269
 270 bool DownloadResourceHandler::OnWillStart(int request_id,
 271                                           const GURL& url,
 272                                           bool* defer) {
 273   return true;
 274 }
 275
 276 bool DownloadResourceHandler::OnBeforeNetworkStart(int request_id,
 277                                                    const GURL& url,
 278                                                    bool* defer) {
 279   return true;
 280 }
 281
 282 // Create a new buffer, which will be handed to the download thread for file
 283 // writing and deletion.
 284 bool DownloadResourceHandler::OnWillRead(int request_id,
 285                                          scoped_refptr<net::IOBuffer>* buf,
 286                                          int* buf_size,
 287                                          int min_size) {
 288   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 289   DCHECK(buf && buf_size);
 290   DCHECK(!read_buffer_.get());
 291
 292   *buf_size = min_size < 0 ? kReadBufSize : min_size;
 293   last_buffer_size_ = *buf_size;
 294   read_buffer_ = new net::IOBuffer(*buf_size);
 295   *buf = read_buffer_.get();
 296   return true;
 297 }
 298
 299 // Pass the buffer to the download file writer.
 300 bool DownloadResourceHandler::OnReadCompleted(int request_id, int bytes_read,
 301                                               bool* defer) {
 302   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 303   DCHECK(read_buffer_.get());
 304
 305   base::TimeTicks now(base::TimeTicks::Now());
 306   if (!last_read_time_.is_null()) {
 307     double seconds_since_last_read = (now - last_read_time_).InSecondsF();
 308     if (now == last_read_time_)
 309       // Use 1/10 ms as a "very small number" so that we avoid
 310       // divide-by-zero error and still record a very high potential bandwidth.
 311       seconds_since_last_read = 0.00001;
 312
 313     double actual_bandwidth = (bytes_read)/seconds_since_last_read;
 314     double potential_bandwidth = last_buffer_size_/seconds_since_last_read;
 315     RecordBandwidth(actual_bandwidth, potential_bandwidth);
 316   }
 317   last_read_time_ = now;
 318
 319   if (!bytes_read)
 320     return true;
 321   bytes_read_ += bytes_read;
 322   DCHECK(read_buffer_.get());
 323
 324   // Take the data ship it down the stream.  If the stream is full, pause the
 325   // request; the stream callback will resume it.
 326   if (!stream_writer_->Write(read_buffer_, bytes_read)) {
 327     PauseRequest();
 328     *defer = was_deferred_ = true;
 329     last_stream_pause_time_ = now;
 330   }
 331
 332   read_buffer_ = NULL;  // Drop our reference.
 333
 334   if (pause_count_ > 0)
 335     *defer = was_deferred_ = true;
 336
 337   return true;
 338 }
 339
 340 void DownloadResourceHandler::OnResponseCompleted(
 341     int request_id,
 342     const net::URLRequestStatus& status,
 343     const std::string& security_info,
 344     bool* defer) {
 345   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 346   int response_code = status.is_success() ? request()->GetResponseCode() : 0;
 347   VLOG(20) << __FUNCTION__ << "()" << DebugString()
 348            << " request_id = " << request_id
 349            << " status.status() = " << status.status()
 350            << " status.error() = " << status.error()
 351            << " response_code = " << response_code;
 352
 353   net::Error error_code = net::OK;
 354   if (status.status() == net::URLRequestStatus::FAILED ||
 355       // Note cancels as failures too.
 356       status.status() == net::URLRequestStatus::CANCELED) {
 357     error_code = static_cast<net::Error>(status.error());  // Normal case.
 358     // Make sure that at least the fact of failure comes through.
 359     if (error_code == net::OK)
 360       error_code = net::ERR_FAILED;
 361   }
 362
 363   // ERR_CONTENT_LENGTH_MISMATCH and ERR_INCOMPLETE_CHUNKED_ENCODING are
 364   // allowed since a number of servers in the wild close the connection too
 365   // early by mistake. Other browsers - IE9, Firefox 11.0, and Safari 5.1.4 -
 366   // treat downloads as complete in both cases, so we follow their lead.
 367   if (error_code == net::ERR_CONTENT_LENGTH_MISMATCH ||
 368       error_code == net::ERR_INCOMPLETE_CHUNKED_ENCODING) {
 369     error_code = net::OK;
 370   }
 371   DownloadInterruptReason reason =
 372       ConvertNetErrorToInterruptReason(
 373         error_code, DOWNLOAD_INTERRUPT_FROM_NETWORK);
 374
 375   if (status.status() == net::URLRequestStatus::CANCELED &&
 376       status.error() == net::ERR_ABORTED) {
 377     // CANCELED + ERR_ABORTED == something outside of the network
 378     // stack cancelled the request.  There aren't that many things that
 379     // could do this to a download request (whose lifetime is separated from
 380     // the tab from which it came).  We map this to USER_CANCELLED as the
 381     // case we know about (system suspend because of laptop close) corresponds
 382     // to a user action.
 383     // TODO(ahendrickson) -- Find a better set of codes to use here, as
 384     // CANCELED/ERR_ABORTED can occur for reasons other than user cancel.
 385     reason = DOWNLOAD_INTERRUPT_REASON_USER_CANCELED;
 386   }
 387
 388   if (status.is_success() &&
 389       reason == DOWNLOAD_INTERRUPT_REASON_NONE &&
 390       request()->response_headers()) {
 391     // Handle server's response codes.
 392     switch(response_code) {
 393       case -1:                          // Non-HTTP request.
 394       case net::HTTP_OK:
 395       case net::HTTP_CREATED:
 396       case net::HTTP_ACCEPTED:
 397       case net::HTTP_NON_AUTHORITATIVE_INFORMATION:
 398       case net::HTTP_RESET_CONTENT:
 399       case net::HTTP_PARTIAL_CONTENT:
 400         // Expected successful codes.
 401         break;
 402       case net::HTTP_NO_CONTENT:
 403       case net::HTTP_NOT_FOUND:
 404         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_BAD_CONTENT;
 405         break;
 406       case net::HTTP_PRECONDITION_FAILED:
 407         // Failed our 'If-Unmodified-Since' or 'If-Match'; see
 408         // download_manager_impl.cc BeginDownload()
 409         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_PRECONDITION;
 410         break;
 411       case net::HTTP_REQUESTED_RANGE_NOT_SATISFIABLE:
 412         // Retry by downloading from the start automatically:
 413         // If we haven't received data when we get this error, we won't.
 414         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_NO_RANGE;
 415         break;
 416       default:    // All other errors.
 417         // Redirection and informational codes should have been handled earlier
 418         // in the stack.
 419         DCHECK_NE(3, response_code / 100);
 420         DCHECK_NE(1, response_code / 100);
 421         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_FAILED;
 422         break;
 423     }
 424   }
 425
 426   std::string accept_ranges;
 427   bool has_strong_validators = false;
 428   if (request()->response_headers()) {
 429     request()->response_headers()->EnumerateHeader(
 430         NULL, "Accept-Ranges", &accept_ranges);
 431     has_strong_validators =
 432         request()->response_headers()->HasStrongValidators();
 433   }
 434   RecordAcceptsRanges(accept_ranges, bytes_read_, has_strong_validators);
 435   RecordNetworkBlockage(base::TimeTicks::Now() - download_start_time_,
 436                         total_pause_time_);
 437
 438   CallStartedCB(NULL, reason);
 439
 440   // Send the info down the stream.  Conditional is in case we get
 441   // OnResponseCompleted without OnResponseStarted.
 442   if (stream_writer_)
 443     stream_writer_->Close(reason);
 444
 445   // If the error mapped to something unknown, record it so that
 446   // we can drill down.
 447   if (reason == DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED) {
 448     UMA_HISTOGRAM_CUSTOM_ENUMERATION("Download.MapErrorNetworkFailed",
 449                                      std::abs(status.error()),
 450                                      net::GetAllErrorCodesForUma());
 451   }
 452
 453   stream_writer_.reset();  // We no longer need the stream.
 454   read_buffer_ = NULL;
 455 }
 456
 457 void DownloadResourceHandler::OnDataDownloaded(
 458     int request_id,
 459     int bytes_downloaded) {
 460   NOTREACHED();
 461 }
 462
 463 void DownloadResourceHandler::PauseRequest() {
 464   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 465
 466   ++pause_count_;
 467 }
 468
 469 void DownloadResourceHandler::ResumeRequest() {
 470   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 471   DCHECK_LT(0, pause_count_);
 472
 473   --pause_count_;
 474
 475   if (!was_deferred_)
 476     return;
 477   if (pause_count_ > 0)
 478     return;
 479
 480   was_deferred_ = false;
 481   if (!last_stream_pause_time_.is_null()) {
 482     total_pause_time_ += (base::TimeTicks::Now() - last_stream_pause_time_);
 483     last_stream_pause_time_ = base::TimeTicks();
 484   }
 485
 486   controller()->Resume();
 487 }
 488
 489 void DownloadResourceHandler::CancelRequest() {
 490   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 491
 492   const ResourceRequestInfo* info = GetRequestInfo();
 493   ResourceDispatcherHostImpl::Get()->CancelRequest(
 494       info->GetChildID(),
 495       info->GetRequestID());
 496   // This object has been deleted.
 497 }
 498
 499 std::string DownloadResourceHandler::DebugString() const {
 500   const ResourceRequestInfo* info = GetRequestInfo();
 501   return base::StringPrintf("{"
 502                             " url_ = " "\"%s\""
 503                             " info = {"
 504                             " child_id = " "%d"
 505                             " request_id = " "%d"
 506                             " route_id = " "%d"
 507                             " }"
 508                             " }",
 509                             request() ?
 510                                 request()->url().spec().c_str() :
 511                                 "<NULL request>",
 512                             info->GetChildID(),
 513                             info->GetRequestID(),
 514                             info->GetRouteID());
 515 }
 516
 517 DownloadResourceHandler::~DownloadResourceHandler() {
 518   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 519
 520   // This won't do anything if the callback was called before.
 521   // If it goes through, it will likely be because OnWillStart() returned
 522   // false somewhere in the chain of resource handlers.
 523   CallStartedCB(NULL, DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED);
 524
 525   // Remove output stream callback if a stream exists.
 526   if (stream_writer_)
 527     stream_writer_->RegisterCallback(base::Closure());
 528
 529   // tab_info_ must be destroyed on UI thread, since
 530   // InitializeDownloadTabInfoOnUIThread might still be using it.
 531   if (tab_info_)
 532     BrowserThread::DeleteSoon(BrowserThread::UI, FROM_HERE, tab_info_);
 533
 534   UMA_HISTOGRAM_TIMES("SB2.DownloadDuration",
 535                       base::TimeTicks::Now() - download_start_time_);
 536 }
 537
 538 }  // namespace content