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/strings/stringprintf.h"
  14 #include "content/browser/byte_stream.h"
  15 #include "content/browser/download/download_create_info.h"
  16 #include "content/browser/download/download_interrupt_reasons_impl.h"
  17 #include "content/browser/download/download_manager_impl.h"
  18 #include "content/browser/download/download_request_handle.h"
  19 #include "content/browser/download/download_stats.h"
  20 #include "content/browser/loader/resource_dispatcher_host_impl.h"
  21 #include "content/browser/loader/resource_request_info_impl.h"
  22 #include "content/public/browser/browser_thread.h"
  23 #include "content/public/browser/download_interrupt_reasons.h"
  24 #include "content/public/browser/download_item.h"
  25 #include "content/public/browser/download_manager_delegate.h"
  26 #include "content/public/browser/navigation_entry.h"
  27 #include "content/public/browser/power_save_blocker.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_CURRENTLY_ON(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_CURRENTLY_ON(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_CURRENTLY_ON(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   power_save_blocker_ = PowerSaveBlocker::Create(
 133       PowerSaveBlocker::kPowerSaveBlockPreventAppSuspension,
 134       PowerSaveBlocker::kReasonOther, "Download in progress");
 135 }
 136
 137 bool DownloadResourceHandler::OnUploadProgress(uint64 position,
 138                                                uint64 size) {
 139   return true;
 140 }
 141
 142 bool DownloadResourceHandler::OnRequestRedirected(
 143     const net::RedirectInfo& redirect_info,
 144     ResourceResponse* response,
 145     bool* defer) {
 146   return true;
 147 }
 148
 149 // Send the download creation information to the download thread.
 150 bool DownloadResourceHandler::OnResponseStarted(
 151     ResourceResponse* response,
 152     bool* defer) {
 153   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 154   // There can be only one (call)
 155   DCHECK(!on_response_started_called_);
 156   on_response_started_called_ = true;
 157
 158   DVLOG(20) << __FUNCTION__ << "()" << DebugString();
 159   download_start_time_ = base::TimeTicks::Now();
 160
 161   // If it's a download, we don't want to poison the cache with it.
 162   request()->StopCaching();
 163
 164   // Lower priority as well, so downloads don't contend for resources
 165   // with main frames.
 166   request()->SetPriority(net::IDLE);
 167
 168   // If the content-length header is not present (or contains something other
 169   // than numbers), the incoming content_length is -1 (unknown size).
 170   // Set the content length to 0 to indicate unknown size to DownloadManager.
 171   int64 content_length =
 172       response->head.content_length > 0 ? response->head.content_length : 0;
 173
 174   const ResourceRequestInfoImpl* request_info = GetRequestInfo();
 175
 176   // Deleted in DownloadManager.
 177   scoped_ptr<DownloadCreateInfo> info(
 178       new DownloadCreateInfo(base::Time::Now(),
 179                              content_length,
 180                              request()->net_log(),
 181                              request_info->HasUserGesture(),
 182                              request_info->GetPageTransition(),
 183                              save_info_.Pass()));
 184
 185   // Create the ByteStream for sending data to the download sink.
 186   scoped_ptr<ByteStreamReader> stream_reader;
 187   CreateByteStream(
 188       base::MessageLoopProxy::current(),
 189       BrowserThread::GetMessageLoopProxyForThread(BrowserThread::FILE),
 190       kDownloadByteStreamSize, &stream_writer_, &stream_reader);
 191   stream_writer_->RegisterCallback(
 192       base::Bind(&DownloadResourceHandler::ResumeRequest, AsWeakPtr()));
 193
 194   info->download_id = download_id_;
 195   info->url_chain = request()->url_chain();
 196   info->referrer_url = GURL(request()->referrer());
 197   info->mime_type = response->head.mime_type;
 198   info->remote_address = request()->GetSocketAddress().host();
 199   request()->GetResponseHeaderByName("content-disposition",
 200                                      &info->content_disposition);
 201   RecordDownloadMimeType(info->mime_type);
 202   RecordDownloadContentDisposition(info->content_disposition);
 203
 204   info->request_handle =
 205       DownloadRequestHandle(AsWeakPtr(), request_info->GetChildID(),
 206                             request_info->GetRouteID(),
 207                             request_info->GetRequestID());
 208
 209   // Get the last modified time and etag.
 210   const net::HttpResponseHeaders* headers = request()->response_headers();
 211   if (headers) {
 212     if (headers->HasStrongValidators()) {
 213       // If we don't have strong validators as per RFC 2616 section 13.3.3, then
 214       // we neither store nor use them for range requests.
 215       if (!headers->EnumerateHeader(NULL, "Last-Modified",
 216                                     &info->last_modified))
 217         info->last_modified.clear();
 218       if (!headers->EnumerateHeader(NULL, "ETag", &info->etag))
 219         info->etag.clear();
 220     }
 221
 222     int status = headers->response_code();
 223     if (2 == status / 100  && status != net::HTTP_PARTIAL_CONTENT) {
 224       // Success & not range response; if we asked for a range, we didn't
 225       // get it--reset the file pointers to reflect that.
 226       info->save_info->offset = 0;
 227       info->save_info->hash_state = "";
 228     }
 229
 230     if (!headers->GetMimeType(&info->original_mime_type))
 231       info->original_mime_type.clear();
 232   }
 233
 234   // Blink verifies that the requester of this download is allowed to set a
 235   // suggested name for the security origin of the downlaod URL. However, this
 236   // assumption doesn't hold if there were cross origin redirects. Therefore,
 237   // clear the suggested_name for such requests.
 238   if (info->url_chain.size() > 1 &&
 239       info->url_chain.front().GetOrigin() != info->url_chain.back().GetOrigin())
 240     info->save_info->suggested_name.clear();
 241
 242   BrowserThread::PostTask(
 243       BrowserThread::UI, FROM_HERE,
 244       base::Bind(&StartOnUIThread,
 245                  base::Passed(&info),
 246                  base::Owned(tab_info_),
 247                  base::Passed(&stream_reader),
 248                  // Pass to StartOnUIThread so that variable
 249                  // access is always on IO thread but function
 250                  // is called on UI thread.
 251                  started_cb_));
 252   // Now owned by the task that was just posted.
 253   tab_info_ = NULL;
 254   // Guaranteed to be called in StartOnUIThread
 255   started_cb_.Reset();
 256
 257   return true;
 258 }
 259
 260 void DownloadResourceHandler::CallStartedCB(
 261     DownloadItem* item,
 262     DownloadInterruptReason interrupt_reason) {
 263   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 264   if (started_cb_.is_null())
 265     return;
 266   BrowserThread::PostTask(
 267       BrowserThread::UI,
 268       FROM_HERE,
 269       base::Bind(
 270           &CallStartedCBOnUIThread, started_cb_, item, interrupt_reason));
 271   started_cb_.Reset();
 272 }
 273
 274 bool DownloadResourceHandler::OnWillStart(const GURL& url, bool* defer) {
 275   return true;
 276 }
 277
 278 bool DownloadResourceHandler::OnBeforeNetworkStart(const GURL& url,
 279                                                    bool* defer) {
 280   return true;
 281 }
 282
 283 // Create a new buffer, which will be handed to the download thread for file
 284 // writing and deletion.
 285 bool DownloadResourceHandler::OnWillRead(scoped_refptr<net::IOBuffer>* buf,
 286                                          int* buf_size,
 287                                          int min_size) {
 288   DCHECK_CURRENTLY_ON(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 bytes_read, bool* defer) {
 301   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 302   DCHECK(read_buffer_.get());
 303
 304   base::TimeTicks now(base::TimeTicks::Now());
 305   if (!last_read_time_.is_null()) {
 306     double seconds_since_last_read = (now - last_read_time_).InSecondsF();
 307     if (now == last_read_time_)
 308       // Use 1/10 ms as a "very small number" so that we avoid
 309       // divide-by-zero error and still record a very high potential bandwidth.
 310       seconds_since_last_read = 0.00001;
 311
 312     double actual_bandwidth = (bytes_read)/seconds_since_last_read;
 313     double potential_bandwidth = last_buffer_size_/seconds_since_last_read;
 314     RecordBandwidth(actual_bandwidth, potential_bandwidth);
 315   }
 316   last_read_time_ = now;
 317
 318   if (!bytes_read)
 319     return true;
 320   bytes_read_ += bytes_read;
 321   DCHECK(read_buffer_.get());
 322
 323   // Take the data ship it down the stream.  If the stream is full, pause the
 324   // request; the stream callback will resume it.
 325   if (!stream_writer_->Write(read_buffer_, bytes_read)) {
 326     PauseRequest();
 327     *defer = was_deferred_ = true;
 328     last_stream_pause_time_ = now;
 329   }
 330
 331   read_buffer_ = NULL;  // Drop our reference.
 332
 333   if (pause_count_ > 0)
 334     *defer = was_deferred_ = true;
 335
 336   return true;
 337 }
 338
 339 void DownloadResourceHandler::OnResponseCompleted(
 340     const net::URLRequestStatus& status,
 341     const std::string& security_info,
 342     bool* defer) {
 343   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 344   int response_code = status.is_success() ? request()->GetResponseCode() : 0;
 345   DVLOG(20) << __FUNCTION__ << "()" << DebugString()
 346             << " status.status() = " << status.status()
 347             << " status.error() = " << status.error()
 348             << " response_code = " << response_code;
 349
 350   net::Error error_code = net::OK;
 351   if (status.status() == net::URLRequestStatus::FAILED ||
 352       // Note cancels as failures too.
 353       status.status() == net::URLRequestStatus::CANCELED) {
 354     error_code = static_cast<net::Error>(status.error());  // Normal case.
 355     // Make sure that at least the fact of failure comes through.
 356     if (error_code == net::OK)
 357       error_code = net::ERR_FAILED;
 358   }
 359
 360   // ERR_CONTENT_LENGTH_MISMATCH and ERR_INCOMPLETE_CHUNKED_ENCODING are
 361   // allowed since a number of servers in the wild close the connection too
 362   // early by mistake. Other browsers - IE9, Firefox 11.0, and Safari 5.1.4 -
 363   // treat downloads as complete in both cases, so we follow their lead.
 364   if (error_code == net::ERR_CONTENT_LENGTH_MISMATCH ||
 365       error_code == net::ERR_INCOMPLETE_CHUNKED_ENCODING) {
 366     error_code = net::OK;
 367   }
 368   DownloadInterruptReason reason =
 369       ConvertNetErrorToInterruptReason(
 370         error_code, DOWNLOAD_INTERRUPT_FROM_NETWORK);
 371
 372   if (status.status() == net::URLRequestStatus::CANCELED &&
 373       status.error() == net::ERR_ABORTED) {
 374     // CANCELED + ERR_ABORTED == something outside of the network
 375     // stack cancelled the request.  There aren't that many things that
 376     // could do this to a download request (whose lifetime is separated from
 377     // the tab from which it came).  We map this to USER_CANCELLED as the
 378     // case we know about (system suspend because of laptop close) corresponds
 379     // to a user action.
 380     // TODO(ahendrickson) -- Find a better set of codes to use here, as
 381     // CANCELED/ERR_ABORTED can occur for reasons other than user cancel.
 382     if (net::IsCertStatusError(request()->ssl_info().cert_status))
 383       reason = DOWNLOAD_INTERRUPT_REASON_SERVER_CERT_PROBLEM;
 384     else
 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       case net::HTTP_UNAUTHORIZED:
 417         // Server didn't authorize this request.
 418         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_UNAUTHORIZED;
 419         break;
 420       default:    // All other errors.
 421         // Redirection and informational codes should have been handled earlier
 422         // in the stack.
 423         DCHECK_NE(3, response_code / 100);
 424         DCHECK_NE(1, response_code / 100);
 425         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_FAILED;
 426         break;
 427     }
 428   }
 429
 430   std::string accept_ranges;
 431   bool has_strong_validators = false;
 432   if (request()->response_headers()) {
 433     request()->response_headers()->EnumerateHeader(
 434         NULL, "Accept-Ranges", &accept_ranges);
 435     has_strong_validators =
 436         request()->response_headers()->HasStrongValidators();
 437   }
 438   RecordAcceptsRanges(accept_ranges, bytes_read_, has_strong_validators);
 439   RecordNetworkBlockage(base::TimeTicks::Now() - download_start_time_,
 440                         total_pause_time_);
 441
 442   CallStartedCB(NULL, reason);
 443
 444   // Send the info down the stream.  Conditional is in case we get
 445   // OnResponseCompleted without OnResponseStarted.
 446   if (stream_writer_)
 447     stream_writer_->Close(reason);
 448
 449   // If the error mapped to something unknown, record it so that
 450   // we can drill down.
 451   if (reason == DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED) {
 452     UMA_HISTOGRAM_CUSTOM_ENUMERATION("Download.MapErrorNetworkFailed",
 453                                      std::abs(status.error()),
 454                                      net::GetAllErrorCodesForUma());
 455   }
 456
 457   stream_writer_.reset();  // We no longer need the stream.
 458   read_buffer_ = NULL;
 459 }
 460
 461 void DownloadResourceHandler::OnDataDownloaded(int bytes_downloaded) {
 462   NOTREACHED();
 463 }
 464
 465 void DownloadResourceHandler::PauseRequest() {
 466   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 467
 468   ++pause_count_;
 469 }
 470
 471 void DownloadResourceHandler::ResumeRequest() {
 472   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 473   DCHECK_LT(0, pause_count_);
 474
 475   --pause_count_;
 476
 477   if (!was_deferred_)
 478     return;
 479   if (pause_count_ > 0)
 480     return;
 481
 482   was_deferred_ = false;
 483   if (!last_stream_pause_time_.is_null()) {
 484     total_pause_time_ += (base::TimeTicks::Now() - last_stream_pause_time_);
 485     last_stream_pause_time_ = base::TimeTicks();
 486   }
 487
 488   controller()->Resume();
 489 }
 490
 491 void DownloadResourceHandler::CancelRequest() {
 492   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 493
 494   const ResourceRequestInfo* info = GetRequestInfo();
 495   ResourceDispatcherHostImpl::Get()->CancelRequest(
 496       info->GetChildID(),
 497       info->GetRequestID());
 498   // This object has been deleted.
 499 }
 500
 501 std::string DownloadResourceHandler::DebugString() const {
 502   const ResourceRequestInfo* info = GetRequestInfo();
 503   return base::StringPrintf("{"
 504                             " url_ = " "\"%s\""
 505                             " info = {"
 506                             " child_id = " "%d"
 507                             " request_id = " "%d"
 508                             " route_id = " "%d"
 509                             " }"
 510                             " }",
 511                             request() ?
 512                                 request()->url().spec().c_str() :
 513                                 "<NULL request>",
 514                             info->GetChildID(),
 515                             info->GetRequestID(),
 516                             info->GetRouteID());
 517 }
 518
 519 DownloadResourceHandler::~DownloadResourceHandler() {
 520   DCHECK_CURRENTLY_ON(BrowserThread::IO);
 521
 522   // This won't do anything if the callback was called before.
 523   // If it goes through, it will likely be because OnWillStart() returned
 524   // false somewhere in the chain of resource handlers.
 525   CallStartedCB(NULL, DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED);
 526
 527   // Remove output stream callback if a stream exists.
 528   if (stream_writer_)
 529     stream_writer_->RegisterCallback(base::Closure());
 530
 531   // tab_info_ must be destroyed on UI thread, since
 532   // InitializeDownloadTabInfoOnUIThread might still be using it.
 533   if (tab_info_)
 534     BrowserThread::DeleteSoon(BrowserThread::UI, FROM_HERE, tab_info_);
 535
 536   UMA_HISTOGRAM_TIMES("SB2.DownloadDuration",
 537                       base::TimeTicks::Now() - download_start_time_);
 538 }
 539
 540 }  // namespace content