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/power_save_blocker.h"
  29 #include "content/public/browser/web_contents.h"
  30 #include "content/public/common/resource_response.h"
  31 #include "net/base/io_buffer.h"
  32 #include "net/base/net_errors.h"
  33 #include "net/http/http_response_headers.h"
  34 #include "net/http/http_status_code.h"
  35 #include "net/url_request/url_request_context.h"
  36
  37 namespace content {
  38
  39 struct DownloadResourceHandler::DownloadTabInfo {
  40   GURL tab_url;
  41   GURL tab_referrer_url;
  42 };
  43
  44 namespace {
  45
  46 void CallStartedCBOnUIThread(
  47     const DownloadUrlParameters::OnStartedCallback& started_cb,
  48     DownloadItem* item,
  49     DownloadInterruptReason interrupt_reason) {
  50   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  51
  52   if (started_cb.is_null())
  53     return;
  54   started_cb.Run(item, interrupt_reason);
  55 }
  56
  57 // Static function in order to prevent any accidental accesses to
  58 // DownloadResourceHandler members from the UI thread.
  59 static void StartOnUIThread(
  60     scoped_ptr<DownloadCreateInfo> info,
  61     DownloadResourceHandler::DownloadTabInfo* tab_info,
  62     scoped_ptr<ByteStreamReader> stream,
  63     const DownloadUrlParameters::OnStartedCallback& started_cb) {
  64   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  65
  66   DownloadManager* download_manager = info->request_handle.GetDownloadManager();
  67   if (!download_manager) {
  68     // NULL in unittests or if the page closed right after starting the
  69     // download.
  70     if (!started_cb.is_null())
  71       started_cb.Run(NULL, DOWNLOAD_INTERRUPT_REASON_USER_CANCELED);
  72
  73     // |stream| gets deleted on non-FILE thread, but it's ok since
  74     // we're not using stream_writer_ yet.
  75
  76     return;
  77   }
  78
  79   info->tab_url = tab_info->tab_url;
  80   info->tab_referrer_url = tab_info->tab_referrer_url;
  81
  82   download_manager->StartDownload(info.Pass(), stream.Pass(), started_cb);
  83 }
  84
  85 void InitializeDownloadTabInfoOnUIThread(
  86     const DownloadRequestHandle& request_handle,
  87     DownloadResourceHandler::DownloadTabInfo* tab_info) {
  88   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::UI));
  89
  90   WebContents* web_contents = request_handle.GetWebContents();
  91   if (web_contents) {
  92     NavigationEntry* entry = web_contents->GetController().GetVisibleEntry();
  93     if (entry) {
  94       tab_info->tab_url = entry->GetURL();
  95       tab_info->tab_referrer_url = entry->GetReferrer().url;
  96     }
  97   }
  98 }
  99
 100 }  // namespace
 101
 102 const int DownloadResourceHandler::kDownloadByteStreamSize = 100 * 1024;
 103
 104 DownloadResourceHandler::DownloadResourceHandler(
 105     uint32 id,
 106     net::URLRequest* request,
 107     const DownloadUrlParameters::OnStartedCallback& started_cb,
 108     scoped_ptr<DownloadSaveInfo> save_info)
 109     : ResourceHandler(request),
 110       download_id_(id),
 111       started_cb_(started_cb),
 112       save_info_(save_info.Pass()),
 113       last_buffer_size_(0),
 114       bytes_read_(0),
 115       pause_count_(0),
 116       was_deferred_(false),
 117       on_response_started_called_(false) {
 118   RecordDownloadCount(UNTHROTTLED_COUNT);
 119
 120   // Do UI thread initialization asap after DownloadResourceHandler creation
 121   // since the tab could be navigated before StartOnUIThread gets called.
 122   const ResourceRequestInfoImpl* request_info = GetRequestInfo();
 123   tab_info_ = new DownloadTabInfo();
 124   BrowserThread::PostTask(
 125       BrowserThread::UI,
 126       FROM_HERE,
 127       base::Bind(&InitializeDownloadTabInfoOnUIThread,
 128                  DownloadRequestHandle(AsWeakPtr(),
 129                                        request_info->GetChildID(),
 130                                        request_info->GetRouteID(),
 131                                        request_info->GetRequestID()),
 132                  tab_info_));
 133   power_save_blocker_ = PowerSaveBlocker::Create(
 134       PowerSaveBlocker::kPowerSaveBlockPreventAppSuspension,
 135       "Download in progress");
 136 }
 137
 138 bool DownloadResourceHandler::OnUploadProgress(int request_id,
 139                                                uint64 position,
 140                                                uint64 size) {
 141   return true;
 142 }
 143
 144 bool DownloadResourceHandler::OnRequestRedirected(
 145     int request_id,
 146     const GURL& url,
 147     ResourceResponse* response,
 148     bool* defer) {
 149   // We treat a download as a main frame load, and thus update the policy URL
 150   // on redirects.
 151   request()->set_first_party_for_cookies(url);
 152   return true;
 153 }
 154
 155 // Send the download creation information to the download thread.
 156 bool DownloadResourceHandler::OnResponseStarted(
 157     int request_id,
 158     ResourceResponse* response,
 159     bool* defer) {
 160   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 161   // There can be only one (call)
 162   DCHECK(!on_response_started_called_);
 163   on_response_started_called_ = true;
 164
 165   VLOG(20) << __FUNCTION__ << "()" << DebugString()
 166            << " request_id = " << request_id;
 167   download_start_time_ = base::TimeTicks::Now();
 168
 169   // If it's a download, we don't want to poison the cache with it.
 170   request()->StopCaching();
 171
 172   // Lower priority as well, so downloads don't contend for resources
 173   // with main frames.
 174   request()->SetPriority(net::IDLE);
 175
 176   // If the content-length header is not present (or contains something other
 177   // than numbers), the incoming content_length is -1 (unknown size).
 178   // Set the content length to 0 to indicate unknown size to DownloadManager.
 179   int64 content_length =
 180       response->head.content_length > 0 ? response->head.content_length : 0;
 181
 182   const ResourceRequestInfoImpl* request_info = GetRequestInfo();
 183
 184   // Deleted in DownloadManager.
 185   scoped_ptr<DownloadCreateInfo> info(
 186       new DownloadCreateInfo(base::Time::Now(),
 187                              content_length,
 188                              request()->net_log(),
 189                              request_info->HasUserGesture(),
 190                              request_info->GetPageTransition(),
 191                              save_info_.Pass()));
 192
 193   // Create the ByteStream for sending data to the download sink.
 194   scoped_ptr<ByteStreamReader> stream_reader;
 195   CreateByteStream(
 196       base::MessageLoopProxy::current(),
 197       BrowserThread::GetMessageLoopProxyForThread(BrowserThread::FILE),
 198       kDownloadByteStreamSize, &stream_writer_, &stream_reader);
 199   stream_writer_->RegisterCallback(
 200       base::Bind(&DownloadResourceHandler::ResumeRequest, AsWeakPtr()));
 201
 202   info->download_id = download_id_;
 203   info->url_chain = request()->url_chain();
 204   info->referrer_url = GURL(request()->referrer());
 205   info->mime_type = response->head.mime_type;
 206   info->remote_address = request()->GetSocketAddress().host();
 207   request()->GetResponseHeaderByName("content-disposition",
 208                                      &info->content_disposition);
 209   RecordDownloadMimeType(info->mime_type);
 210   RecordDownloadContentDisposition(info->content_disposition);
 211
 212   info->request_handle =
 213       DownloadRequestHandle(AsWeakPtr(), request_info->GetChildID(),
 214                             request_info->GetRouteID(),
 215                             request_info->GetRequestID());
 216
 217   // Get the last modified time and etag.
 218   const net::HttpResponseHeaders* headers = request()->response_headers();
 219   if (headers) {
 220     if (headers->HasStrongValidators()) {
 221       // If we don't have strong validators as per RFC 2616 section 13.3.3, then
 222       // we neither store nor use them for range requests.
 223       if (!headers->EnumerateHeader(NULL, "Last-Modified",
 224                                     &info->last_modified))
 225         info->last_modified.clear();
 226       if (!headers->EnumerateHeader(NULL, "ETag", &info->etag))
 227         info->etag.clear();
 228     }
 229
 230     int status = headers->response_code();
 231     if (2 == status / 100  && status != net::HTTP_PARTIAL_CONTENT) {
 232       // Success & not range response; if we asked for a range, we didn't
 233       // get it--reset the file pointers to reflect that.
 234       info->save_info->offset = 0;
 235       info->save_info->hash_state = "";
 236     }
 237
 238     if (!headers->GetMimeType(&info->original_mime_type))
 239       info->original_mime_type.clear();
 240   }
 241
 242   // Blink verifies that the requester of this download is allowed to set a
 243   // suggested name for the security origin of the downlaod URL. However, this
 244   // assumption doesn't hold if there were cross origin redirects. Therefore,
 245   // clear the suggested_name for such requests.
 246   if (info->url_chain.size() > 1 &&
 247       info->url_chain.front().GetOrigin() != info->url_chain.back().GetOrigin())
 248     info->save_info->suggested_name.clear();
 249
 250   BrowserThread::PostTask(
 251       BrowserThread::UI, FROM_HERE,
 252       base::Bind(&StartOnUIThread,
 253                  base::Passed(&info),
 254                  base::Owned(tab_info_),
 255                  base::Passed(&stream_reader),
 256                  // Pass to StartOnUIThread so that variable
 257                  // access is always on IO thread but function
 258                  // is called on UI thread.
 259                  started_cb_));
 260   // Now owned by the task that was just posted.
 261   tab_info_ = NULL;
 262   // Guaranteed to be called in StartOnUIThread
 263   started_cb_.Reset();
 264
 265   return true;
 266 }
 267
 268 void DownloadResourceHandler::CallStartedCB(
 269     DownloadItem* item,
 270     DownloadInterruptReason interrupt_reason) {
 271   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 272   if (started_cb_.is_null())
 273     return;
 274   BrowserThread::PostTask(
 275       BrowserThread::UI,
 276       FROM_HERE,
 277       base::Bind(
 278           &CallStartedCBOnUIThread, started_cb_, item, interrupt_reason));
 279   started_cb_.Reset();
 280 }
 281
 282 bool DownloadResourceHandler::OnWillStart(int request_id,
 283                                           const GURL& url,
 284                                           bool* defer) {
 285   return true;
 286 }
 287
 288 bool DownloadResourceHandler::OnBeforeNetworkStart(int request_id,
 289                                                    const GURL& url,
 290                                                    bool* defer) {
 291   return true;
 292 }
 293
 294 // Create a new buffer, which will be handed to the download thread for file
 295 // writing and deletion.
 296 bool DownloadResourceHandler::OnWillRead(int request_id,
 297                                          scoped_refptr<net::IOBuffer>* buf,
 298                                          int* buf_size,
 299                                          int min_size) {
 300   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 301   DCHECK(buf && buf_size);
 302   DCHECK(!read_buffer_.get());
 303
 304   *buf_size = min_size < 0 ? kReadBufSize : min_size;
 305   last_buffer_size_ = *buf_size;
 306   read_buffer_ = new net::IOBuffer(*buf_size);
 307   *buf = read_buffer_.get();
 308   return true;
 309 }
 310
 311 // Pass the buffer to the download file writer.
 312 bool DownloadResourceHandler::OnReadCompleted(int request_id, int bytes_read,
 313                                               bool* defer) {
 314   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 315   DCHECK(read_buffer_.get());
 316
 317   base::TimeTicks now(base::TimeTicks::Now());
 318   if (!last_read_time_.is_null()) {
 319     double seconds_since_last_read = (now - last_read_time_).InSecondsF();
 320     if (now == last_read_time_)
 321       // Use 1/10 ms as a "very small number" so that we avoid
 322       // divide-by-zero error and still record a very high potential bandwidth.
 323       seconds_since_last_read = 0.00001;
 324
 325     double actual_bandwidth = (bytes_read)/seconds_since_last_read;
 326     double potential_bandwidth = last_buffer_size_/seconds_since_last_read;
 327     RecordBandwidth(actual_bandwidth, potential_bandwidth);
 328   }
 329   last_read_time_ = now;
 330
 331   if (!bytes_read)
 332     return true;
 333   bytes_read_ += bytes_read;
 334   DCHECK(read_buffer_.get());
 335
 336   // Take the data ship it down the stream.  If the stream is full, pause the
 337   // request; the stream callback will resume it.
 338   if (!stream_writer_->Write(read_buffer_, bytes_read)) {
 339     PauseRequest();
 340     *defer = was_deferred_ = true;
 341     last_stream_pause_time_ = now;
 342   }
 343
 344   read_buffer_ = NULL;  // Drop our reference.
 345
 346   if (pause_count_ > 0)
 347     *defer = was_deferred_ = true;
 348
 349   return true;
 350 }
 351
 352 void DownloadResourceHandler::OnResponseCompleted(
 353     int request_id,
 354     const net::URLRequestStatus& status,
 355     const std::string& security_info,
 356     bool* defer) {
 357   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 358   int response_code = status.is_success() ? request()->GetResponseCode() : 0;
 359   VLOG(20) << __FUNCTION__ << "()" << DebugString()
 360            << " request_id = " << request_id
 361            << " status.status() = " << status.status()
 362            << " status.error() = " << status.error()
 363            << " response_code = " << response_code;
 364
 365   net::Error error_code = net::OK;
 366   if (status.status() == net::URLRequestStatus::FAILED ||
 367       // Note cancels as failures too.
 368       status.status() == net::URLRequestStatus::CANCELED) {
 369     error_code = static_cast<net::Error>(status.error());  // Normal case.
 370     // Make sure that at least the fact of failure comes through.
 371     if (error_code == net::OK)
 372       error_code = net::ERR_FAILED;
 373   }
 374
 375   // ERR_CONTENT_LENGTH_MISMATCH and ERR_INCOMPLETE_CHUNKED_ENCODING are
 376   // allowed since a number of servers in the wild close the connection too
 377   // early by mistake. Other browsers - IE9, Firefox 11.0, and Safari 5.1.4 -
 378   // treat downloads as complete in both cases, so we follow their lead.
 379   if (error_code == net::ERR_CONTENT_LENGTH_MISMATCH ||
 380       error_code == net::ERR_INCOMPLETE_CHUNKED_ENCODING) {
 381     error_code = net::OK;
 382   }
 383   DownloadInterruptReason reason =
 384       ConvertNetErrorToInterruptReason(
 385         error_code, DOWNLOAD_INTERRUPT_FROM_NETWORK);
 386
 387   if (status.status() == net::URLRequestStatus::CANCELED &&
 388       status.error() == net::ERR_ABORTED) {
 389     // CANCELED + ERR_ABORTED == something outside of the network
 390     // stack cancelled the request.  There aren't that many things that
 391     // could do this to a download request (whose lifetime is separated from
 392     // the tab from which it came).  We map this to USER_CANCELLED as the
 393     // case we know about (system suspend because of laptop close) corresponds
 394     // to a user action.
 395     // TODO(ahendrickson) -- Find a better set of codes to use here, as
 396     // CANCELED/ERR_ABORTED can occur for reasons other than user cancel.
 397     reason = DOWNLOAD_INTERRUPT_REASON_USER_CANCELED;
 398   }
 399
 400   if (status.is_success() &&
 401       reason == DOWNLOAD_INTERRUPT_REASON_NONE &&
 402       request()->response_headers()) {
 403     // Handle server's response codes.
 404     switch(response_code) {
 405       case -1:                          // Non-HTTP request.
 406       case net::HTTP_OK:
 407       case net::HTTP_CREATED:
 408       case net::HTTP_ACCEPTED:
 409       case net::HTTP_NON_AUTHORITATIVE_INFORMATION:
 410       case net::HTTP_RESET_CONTENT:
 411       case net::HTTP_PARTIAL_CONTENT:
 412         // Expected successful codes.
 413         break;
 414       case net::HTTP_NO_CONTENT:
 415       case net::HTTP_NOT_FOUND:
 416         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_BAD_CONTENT;
 417         break;
 418       case net::HTTP_PRECONDITION_FAILED:
 419         // Failed our 'If-Unmodified-Since' or 'If-Match'; see
 420         // download_manager_impl.cc BeginDownload()
 421         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_PRECONDITION;
 422         break;
 423       case net::HTTP_REQUESTED_RANGE_NOT_SATISFIABLE:
 424         // Retry by downloading from the start automatically:
 425         // If we haven't received data when we get this error, we won't.
 426         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_NO_RANGE;
 427         break;
 428       default:    // All other errors.
 429         // Redirection and informational codes should have been handled earlier
 430         // in the stack.
 431         DCHECK_NE(3, response_code / 100);
 432         DCHECK_NE(1, response_code / 100);
 433         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_FAILED;
 434         break;
 435     }
 436   }
 437
 438   std::string accept_ranges;
 439   bool has_strong_validators = false;
 440   if (request()->response_headers()) {
 441     request()->response_headers()->EnumerateHeader(
 442         NULL, "Accept-Ranges", &accept_ranges);
 443     has_strong_validators =
 444         request()->response_headers()->HasStrongValidators();
 445   }
 446   RecordAcceptsRanges(accept_ranges, bytes_read_, has_strong_validators);
 447   RecordNetworkBlockage(base::TimeTicks::Now() - download_start_time_,
 448                         total_pause_time_);
 449
 450   CallStartedCB(NULL, reason);
 451
 452   // Send the info down the stream.  Conditional is in case we get
 453   // OnResponseCompleted without OnResponseStarted.
 454   if (stream_writer_)
 455     stream_writer_->Close(reason);
 456
 457   // If the error mapped to something unknown, record it so that
 458   // we can drill down.
 459   if (reason == DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED) {
 460     UMA_HISTOGRAM_CUSTOM_ENUMERATION("Download.MapErrorNetworkFailed",
 461                                      std::abs(status.error()),
 462                                      net::GetAllErrorCodesForUma());
 463   }
 464
 465   stream_writer_.reset();  // We no longer need the stream.
 466   read_buffer_ = NULL;
 467 }
 468
 469 void DownloadResourceHandler::OnDataDownloaded(
 470     int request_id,
 471     int bytes_downloaded) {
 472   NOTREACHED();
 473 }
 474
 475 void DownloadResourceHandler::PauseRequest() {
 476   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 477
 478   ++pause_count_;
 479 }
 480
 481 void DownloadResourceHandler::ResumeRequest() {
 482   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 483   DCHECK_LT(0, pause_count_);
 484
 485   --pause_count_;
 486
 487   if (!was_deferred_)
 488     return;
 489   if (pause_count_ > 0)
 490     return;
 491
 492   was_deferred_ = false;
 493   if (!last_stream_pause_time_.is_null()) {
 494     total_pause_time_ += (base::TimeTicks::Now() - last_stream_pause_time_);
 495     last_stream_pause_time_ = base::TimeTicks();
 496   }
 497
 498   controller()->Resume();
 499 }
 500
 501 void DownloadResourceHandler::CancelRequest() {
 502   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 503
 504   const ResourceRequestInfo* info = GetRequestInfo();
 505   ResourceDispatcherHostImpl::Get()->CancelRequest(
 506       info->GetChildID(),
 507       info->GetRequestID());
 508   // This object has been deleted.
 509 }
 510
 511 std::string DownloadResourceHandler::DebugString() const {
 512   const ResourceRequestInfo* info = GetRequestInfo();
 513   return base::StringPrintf("{"
 514                             " url_ = " "\"%s\""
 515                             " info = {"
 516                             " child_id = " "%d"
 517                             " request_id = " "%d"
 518                             " route_id = " "%d"
 519                             " }"
 520                             " }",
 521                             request() ?
 522                                 request()->url().spec().c_str() :
 523                                 "<NULL request>",
 524                             info->GetChildID(),
 525                             info->GetRequestID(),
 526                             info->GetRouteID());
 527 }
 528
 529 DownloadResourceHandler::~DownloadResourceHandler() {
 530   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
 531
 532   // This won't do anything if the callback was called before.
 533   // If it goes through, it will likely be because OnWillStart() returned
 534   // false somewhere in the chain of resource handlers.
 535   CallStartedCB(NULL, DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED);
 536
 537   // Remove output stream callback if a stream exists.
 538   if (stream_writer_)
 539     stream_writer_->RegisterCallback(base::Closure());
 540
 541   // tab_info_ must be destroyed on UI thread, since
 542   // InitializeDownloadTabInfoOnUIThread might still be using it.
 543   if (tab_info_)
 544     BrowserThread::DeleteSoon(BrowserThread::UI, FROM_HERE, tab_info_);
 545
 546   UMA_HISTOGRAM_TIMES("SB2.DownloadDuration",
 547                       base::TimeTicks::Now() - download_start_time_);
 548 }
 549
 550 }  // namespace content