Home | History | Annotate | Download | only in download 
      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(uint64 position,
    139                                                uint64 size) {
    140   return true;
    141 }
    142 
    143 bool DownloadResourceHandler::OnRequestRedirected(
    144     const GURL& url,
    145     ResourceResponse* response,
    146     bool* defer) {
    147   // We treat a download as a main frame load, and thus update the policy URL
    148   // on redirects.
    149   request()->set_first_party_for_cookies(url);
    150   return true;
    151 }
    152 
    153 // Send the download creation information to the download thread.
    154 bool DownloadResourceHandler::OnResponseStarted(
    155     ResourceResponse* response,
    156     bool* defer) {
    157   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    158   // There can be only one (call)
    159   DCHECK(!on_response_started_called_);
    160   on_response_started_called_ = true;
    161 
    162   VLOG(20) << __FUNCTION__ << "()" << DebugString();
    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   // Blink verifies that the requester of this download is allowed to set a
    239   // suggested name for the security origin of the downlaod URL. However, this
    240   // assumption doesn't hold if there were cross origin redirects. Therefore,
    241   // clear the suggested_name for such requests.
    242   if (info->url_chain.size() > 1 &&
    243       info->url_chain.front().GetOrigin() != info->url_chain.back().GetOrigin())
    244     info->save_info->suggested_name.clear();
    245 
    246   BrowserThread::PostTask(
    247       BrowserThread::UI, FROM_HERE,
    248       base::Bind(&StartOnUIThread,
    249                  base::Passed(&info),
    250                  base::Owned(tab_info_),
    251                  base::Passed(&stream_reader),
    252                  // Pass to StartOnUIThread so that variable
    253                  // access is always on IO thread but function
    254                  // is called on UI thread.
    255                  started_cb_));
    256   // Now owned by the task that was just posted.
    257   tab_info_ = NULL;
    258   // Guaranteed to be called in StartOnUIThread
    259   started_cb_.Reset();
    260 
    261   return true;
    262 }
    263 
    264 void DownloadResourceHandler::CallStartedCB(
    265     DownloadItem* item,
    266     DownloadInterruptReason interrupt_reason) {
    267   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    268   if (started_cb_.is_null())
    269     return;
    270   BrowserThread::PostTask(
    271       BrowserThread::UI,
    272       FROM_HERE,
    273       base::Bind(
    274           &CallStartedCBOnUIThread, started_cb_, item, interrupt_reason));
    275   started_cb_.Reset();
    276 }
    277 
    278 bool DownloadResourceHandler::OnWillStart(const GURL& url, bool* defer) {
    279   return true;
    280 }
    281 
    282 bool DownloadResourceHandler::OnBeforeNetworkStart(const GURL& url,
    283                                                    bool* defer) {
    284   return true;
    285 }
    286 
    287 // Create a new buffer, which will be handed to the download thread for file
    288 // writing and deletion.
    289 bool DownloadResourceHandler::OnWillRead(scoped_refptr<net::IOBuffer>* buf,
    290                                          int* buf_size,
    291                                          int min_size) {
    292   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    293   DCHECK(buf && buf_size);
    294   DCHECK(!read_buffer_.get());
    295 
    296   *buf_size = min_size < 0 ? kReadBufSize : min_size;
    297   last_buffer_size_ = *buf_size;
    298   read_buffer_ = new net::IOBuffer(*buf_size);
    299   *buf = read_buffer_.get();
    300   return true;
    301 }
    302 
    303 // Pass the buffer to the download file writer.
    304 bool DownloadResourceHandler::OnReadCompleted(int bytes_read, bool* defer) {
    305   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    306   DCHECK(read_buffer_.get());
    307 
    308   base::TimeTicks now(base::TimeTicks::Now());
    309   if (!last_read_time_.is_null()) {
    310     double seconds_since_last_read = (now - last_read_time_).InSecondsF();
    311     if (now == last_read_time_)
    312       // Use 1/10 ms as a "very small number" so that we avoid
    313       // divide-by-zero error and still record a very high potential bandwidth.
    314       seconds_since_last_read = 0.00001;
    315 
    316     double actual_bandwidth = (bytes_read)/seconds_since_last_read;
    317     double potential_bandwidth = last_buffer_size_/seconds_since_last_read;
    318     RecordBandwidth(actual_bandwidth, potential_bandwidth);
    319   }
    320   last_read_time_ = now;
    321 
    322   if (!bytes_read)
    323     return true;
    324   bytes_read_ += bytes_read;
    325   DCHECK(read_buffer_.get());
    326 
    327   // Take the data ship it down the stream.  If the stream is full, pause the
    328   // request; the stream callback will resume it.
    329   if (!stream_writer_->Write(read_buffer_, bytes_read)) {
    330     PauseRequest();
    331     *defer = was_deferred_ = true;
    332     last_stream_pause_time_ = now;
    333   }
    334 
    335   read_buffer_ = NULL;  // Drop our reference.
    336 
    337   if (pause_count_ > 0)
    338     *defer = was_deferred_ = true;
    339 
    340   return true;
    341 }
    342 
    343 void DownloadResourceHandler::OnResponseCompleted(
    344     const net::URLRequestStatus& status,
    345     const std::string& security_info,
    346     bool* defer) {
    347   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    348   int response_code = status.is_success() ? request()->GetResponseCode() : 0;
    349   VLOG(20) << __FUNCTION__ << "()" << DebugString()
    350            << " status.status() = " << status.status()
    351            << " status.error() = " << status.error()
    352            << " response_code = " << response_code;
    353 
    354   net::Error error_code = net::OK;
    355   if (status.status() == net::URLRequestStatus::FAILED ||
    356       // Note cancels as failures too.
    357       status.status() == net::URLRequestStatus::CANCELED) {
    358     error_code = static_cast<net::Error>(status.error());  // Normal case.
    359     // Make sure that at least the fact of failure comes through.
    360     if (error_code == net::OK)
    361       error_code = net::ERR_FAILED;
    362   }
    363 
    364   // ERR_CONTENT_LENGTH_MISMATCH and ERR_INCOMPLETE_CHUNKED_ENCODING are
    365   // allowed since a number of servers in the wild close the connection too
    366   // early by mistake. Other browsers - IE9, Firefox 11.0, and Safari 5.1.4 -
    367   // treat downloads as complete in both cases, so we follow their lead.
    368   if (error_code == net::ERR_CONTENT_LENGTH_MISMATCH ||
    369       error_code == net::ERR_INCOMPLETE_CHUNKED_ENCODING) {
    370     error_code = net::OK;
    371   }
    372   DownloadInterruptReason reason =
    373       ConvertNetErrorToInterruptReason(
    374         error_code, DOWNLOAD_INTERRUPT_FROM_NETWORK);
    375 
    376   if (status.status() == net::URLRequestStatus::CANCELED &&
    377       status.error() == net::ERR_ABORTED) {
    378     // CANCELED + ERR_ABORTED == something outside of the network
    379     // stack cancelled the request.  There aren't that many things that
    380     // could do this to a download request (whose lifetime is separated from
    381     // the tab from which it came).  We map this to USER_CANCELLED as the
    382     // case we know about (system suspend because of laptop close) corresponds
    383     // to a user action.
    384     // TODO(ahendrickson) -- Find a better set of codes to use here, as
    385     // CANCELED/ERR_ABORTED can occur for reasons other than user cancel.
    386     reason = DOWNLOAD_INTERRUPT_REASON_USER_CANCELED;
    387   }
    388 
    389   if (status.is_success() &&
    390       reason == DOWNLOAD_INTERRUPT_REASON_NONE &&
    391       request()->response_headers()) {
    392     // Handle server's response codes.
    393     switch(response_code) {
    394       case -1:                          // Non-HTTP request.
    395       case net::HTTP_OK:
    396       case net::HTTP_CREATED:
    397       case net::HTTP_ACCEPTED:
    398       case net::HTTP_NON_AUTHORITATIVE_INFORMATION:
    399       case net::HTTP_RESET_CONTENT:
    400       case net::HTTP_PARTIAL_CONTENT:
    401         // Expected successful codes.
    402         break;
    403       case net::HTTP_NO_CONTENT:
    404       case net::HTTP_NOT_FOUND:
    405         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_BAD_CONTENT;
    406         break;
    407       case net::HTTP_PRECONDITION_FAILED:
    408         // Failed our 'If-Unmodified-Since' or 'If-Match'; see
    409         // download_manager_impl.cc BeginDownload()
    410         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_PRECONDITION;
    411         break;
    412       case net::HTTP_REQUESTED_RANGE_NOT_SATISFIABLE:
    413         // Retry by downloading from the start automatically:
    414         // If we haven't received data when we get this error, we won't.
    415         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_NO_RANGE;
    416         break;
    417       default:    // All other errors.
    418         // Redirection and informational codes should have been handled earlier
    419         // in the stack.
    420         DCHECK_NE(3, response_code / 100);
    421         DCHECK_NE(1, response_code / 100);
    422         reason = DOWNLOAD_INTERRUPT_REASON_SERVER_FAILED;
    423         break;
    424     }
    425   }
    426 
    427   std::string accept_ranges;
    428   bool has_strong_validators = false;
    429   if (request()->response_headers()) {
    430     request()->response_headers()->EnumerateHeader(
    431         NULL, "Accept-Ranges", &accept_ranges);
    432     has_strong_validators =
    433         request()->response_headers()->HasStrongValidators();
    434   }
    435   RecordAcceptsRanges(accept_ranges, bytes_read_, has_strong_validators);
    436   RecordNetworkBlockage(base::TimeTicks::Now() - download_start_time_,
    437                         total_pause_time_);
    438 
    439   CallStartedCB(NULL, reason);
    440 
    441   // Send the info down the stream.  Conditional is in case we get
    442   // OnResponseCompleted without OnResponseStarted.
    443   if (stream_writer_)
    444     stream_writer_->Close(reason);
    445 
    446   // If the error mapped to something unknown, record it so that
    447   // we can drill down.
    448   if (reason == DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED) {
    449     UMA_HISTOGRAM_CUSTOM_ENUMERATION("Download.MapErrorNetworkFailed",
    450                                      std::abs(status.error()),
    451                                      net::GetAllErrorCodesForUma());
    452   }
    453 
    454   stream_writer_.reset();  // We no longer need the stream.
    455   read_buffer_ = NULL;
    456 }
    457 
    458 void DownloadResourceHandler::OnDataDownloaded(int bytes_downloaded) {
    459   NOTREACHED();
    460 }
    461 
    462 void DownloadResourceHandler::PauseRequest() {
    463   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    464 
    465   ++pause_count_;
    466 }
    467 
    468 void DownloadResourceHandler::ResumeRequest() {
    469   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    470   DCHECK_LT(0, pause_count_);
    471 
    472   --pause_count_;
    473 
    474   if (!was_deferred_)
    475     return;
    476   if (pause_count_ > 0)
    477     return;
    478 
    479   was_deferred_ = false;
    480   if (!last_stream_pause_time_.is_null()) {
    481     total_pause_time_ += (base::TimeTicks::Now() - last_stream_pause_time_);
    482     last_stream_pause_time_ = base::TimeTicks();
    483   }
    484 
    485   controller()->Resume();
    486 }
    487 
    488 void DownloadResourceHandler::CancelRequest() {
    489   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    490 
    491   const ResourceRequestInfo* info = GetRequestInfo();
    492   ResourceDispatcherHostImpl::Get()->CancelRequest(
    493       info->GetChildID(),
    494       info->GetRequestID());
    495   // This object has been deleted.
    496 }
    497 
    498 std::string DownloadResourceHandler::DebugString() const {
    499   const ResourceRequestInfo* info = GetRequestInfo();
    500   return base::StringPrintf("{"
    501                             " url_ = " "\"%s\""
    502                             " info = {"
    503                             " child_id = " "%d"
    504                             " request_id = " "%d"
    505                             " route_id = " "%d"
    506                             " }"
    507                             " }",
    508                             request() ?
    509                                 request()->url().spec().c_str() :
    510                                 "<NULL request>",
    511                             info->GetChildID(),
    512                             info->GetRequestID(),
    513                             info->GetRouteID());
    514 }
    515 
    516 DownloadResourceHandler::~DownloadResourceHandler() {
    517   DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));
    518 
    519   // This won't do anything if the callback was called before.
    520   // If it goes through, it will likely be because OnWillStart() returned
    521   // false somewhere in the chain of resource handlers.
    522   CallStartedCB(NULL, DOWNLOAD_INTERRUPT_REASON_NETWORK_FAILED);
    523 
    524   // Remove output stream callback if a stream exists.
    525   if (stream_writer_)
    526     stream_writer_->RegisterCallback(base::Closure());
    527 
    528   // tab_info_ must be destroyed on UI thread, since
    529   // InitializeDownloadTabInfoOnUIThread might still be using it.
    530   if (tab_info_)
    531     BrowserThread::DeleteSoon(BrowserThread::UI, FROM_HERE, tab_info_);
    532 
    533   UMA_HISTOGRAM_TIMES("SB2.DownloadDuration",
    534                       base::TimeTicks::Now() - download_start_time_);
    535 }
    536 
    537 }  // namespace content
    538