Home | History | Annotate | Download | only in url_request
      1 // Copyright (c) 2006-2009 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 "net/url_request/url_request.h"
      6 
      7 #include "base/compiler_specific.h"
      8 #include "base/message_loop.h"
      9 #include "base/singleton.h"
     10 #include "base/stats_counters.h"
     11 #include "net/base/load_flags.h"
     12 #include "net/base/load_log.h"
     13 #include "net/base/net_errors.h"
     14 #include "net/base/ssl_cert_request_info.h"
     15 #include "net/base/upload_data.h"
     16 #include "net/http/http_response_headers.h"
     17 #include "net/http/http_util.h"
     18 #include "net/url_request/url_request_context.h"
     19 #include "net/url_request/url_request_job.h"
     20 #include "net/url_request/url_request_job_manager.h"
     21 
     22 using base::Time;
     23 using net::UploadData;
     24 using std::string;
     25 using std::wstring;
     26 
     27 // Max number of http redirects to follow.  Same number as gecko.
     28 static const int kMaxRedirects = 20;
     29 
     30 static URLRequestJobManager* GetJobManager() {
     31   return Singleton<URLRequestJobManager>::get();
     32 }
     33 
     34 ///////////////////////////////////////////////////////////////////////////////
     35 // URLRequest
     36 
     37 URLRequest::URLRequest(const GURL& url, Delegate* delegate)
     38     : url_(url),
     39       original_url_(url),
     40       method_("GET"),
     41       load_flags_(net::LOAD_NORMAL),
     42       delegate_(delegate),
     43       is_pending_(false),
     44       enable_profiling_(false),
     45       redirect_limit_(kMaxRedirects),
     46       final_upload_progress_(0),
     47       priority_(net::LOWEST),
     48       ALLOW_THIS_IN_INITIALIZER_LIST(request_tracker_node_(this)) {
     49   SIMPLE_STATS_COUNTER("URLRequestCount");
     50 
     51   // Sanity check out environment.
     52   DCHECK(MessageLoop::current()) <<
     53       "The current MessageLoop must exist";
     54   DCHECK_EQ(MessageLoop::TYPE_IO, MessageLoop::current()->type()) <<
     55       "The current MessageLoop must be TYPE_IO";
     56 }
     57 
     58 URLRequest::~URLRequest() {
     59   Cancel();
     60 
     61   if (job_)
     62     OrphanJob();
     63 
     64   set_context(NULL);
     65 }
     66 
     67 // static
     68 URLRequest::ProtocolFactory* URLRequest::RegisterProtocolFactory(
     69     const string& scheme, ProtocolFactory* factory) {
     70   return GetJobManager()->RegisterProtocolFactory(scheme, factory);
     71 }
     72 
     73 // static
     74 void URLRequest::RegisterRequestInterceptor(Interceptor* interceptor) {
     75   GetJobManager()->RegisterRequestInterceptor(interceptor);
     76 }
     77 
     78 // static
     79 void URLRequest::UnregisterRequestInterceptor(Interceptor* interceptor) {
     80   GetJobManager()->UnregisterRequestInterceptor(interceptor);
     81 }
     82 
     83 void URLRequest::AppendBytesToUpload(const char* bytes, int bytes_len) {
     84   DCHECK(bytes_len > 0 && bytes);
     85   if (!upload_)
     86     upload_ = new UploadData();
     87   upload_->AppendBytes(bytes, bytes_len);
     88 }
     89 
     90 void URLRequest::AppendFileRangeToUpload(const FilePath& file_path,
     91                                          uint64 offset, uint64 length) {
     92   DCHECK(file_path.value().length() > 0 && length > 0);
     93   if (!upload_)
     94     upload_ = new UploadData();
     95   upload_->AppendFileRange(file_path, offset, length);
     96 }
     97 
     98 void URLRequest::set_upload(net::UploadData* upload) {
     99   upload_ = upload;
    100 }
    101 
    102 // Get the upload data directly.
    103 net::UploadData* URLRequest::get_upload() {
    104   return upload_.get();
    105 }
    106 
    107 bool URLRequest::has_upload() const {
    108   return upload_ != NULL;
    109 }
    110 
    111 void URLRequest::SetExtraRequestHeaderById(int id, const string& value,
    112                                            bool overwrite) {
    113   DCHECK(!is_pending_);
    114   NOTREACHED() << "implement me!";
    115 }
    116 
    117 void URLRequest::SetExtraRequestHeaderByName(const string& name,
    118                                              const string& value,
    119                                              bool overwrite) {
    120   DCHECK(!is_pending_);
    121   NOTREACHED() << "implement me!";
    122 }
    123 
    124 void URLRequest::SetExtraRequestHeaders(const string& headers) {
    125   DCHECK(!is_pending_);
    126   if (headers.empty()) {
    127     extra_request_headers_.clear();
    128   } else {
    129 #ifndef NDEBUG
    130     size_t crlf = headers.rfind("\r\n", headers.size() - 1);
    131     DCHECK(crlf != headers.size() - 2) << "headers must not end with CRLF";
    132 #endif
    133     extra_request_headers_ = headers + "\r\n";
    134   }
    135 
    136   // NOTE: This method will likely become non-trivial once the other setters
    137   // for request headers are implemented.
    138 }
    139 
    140 net::LoadState URLRequest::GetLoadState() const {
    141   return job_ ? job_->GetLoadState() : net::LOAD_STATE_IDLE;
    142 }
    143 
    144 uint64 URLRequest::GetUploadProgress() const {
    145   if (!job_) {
    146     // We haven't started or the request was cancelled
    147     return 0;
    148   }
    149   if (final_upload_progress_) {
    150     // The first job completed and none of the subsequent series of
    151     // GETs when following redirects will upload anything, so we return the
    152     // cached results from the initial job, the POST.
    153     return final_upload_progress_;
    154   }
    155   return job_->GetUploadProgress();
    156 }
    157 
    158 void URLRequest::GetResponseHeaderById(int id, string* value) {
    159   DCHECK(job_);
    160   NOTREACHED() << "implement me!";
    161 }
    162 
    163 void URLRequest::GetResponseHeaderByName(const string& name, string* value) {
    164   DCHECK(value);
    165   if (response_info_.headers) {
    166     response_info_.headers->GetNormalizedHeader(name, value);
    167   } else {
    168     value->clear();
    169   }
    170 }
    171 
    172 void URLRequest::GetAllResponseHeaders(string* headers) {
    173   DCHECK(headers);
    174   if (response_info_.headers) {
    175     response_info_.headers->GetNormalizedHeaders(headers);
    176   } else {
    177     headers->clear();
    178   }
    179 }
    180 
    181 net::HttpResponseHeaders* URLRequest::response_headers() const {
    182   return response_info_.headers.get();
    183 }
    184 
    185 bool URLRequest::GetResponseCookies(ResponseCookies* cookies) {
    186   DCHECK(job_);
    187   return job_->GetResponseCookies(cookies);
    188 }
    189 
    190 void URLRequest::GetMimeType(string* mime_type) {
    191   DCHECK(job_);
    192   job_->GetMimeType(mime_type);
    193 }
    194 
    195 void URLRequest::GetCharset(string* charset) {
    196   DCHECK(job_);
    197   job_->GetCharset(charset);
    198 }
    199 
    200 int URLRequest::GetResponseCode() {
    201   DCHECK(job_);
    202   return job_->GetResponseCode();
    203 }
    204 
    205 // static
    206 bool URLRequest::IsHandledProtocol(const std::string& scheme) {
    207   return GetJobManager()->SupportsScheme(scheme);
    208 }
    209 
    210 // static
    211 bool URLRequest::IsHandledURL(const GURL& url) {
    212   if (!url.is_valid()) {
    213     // We handle error cases.
    214     return true;
    215   }
    216 
    217   return IsHandledProtocol(url.scheme());
    218 }
    219 
    220 void URLRequest::set_first_party_for_cookies(
    221     const GURL& first_party_for_cookies) {
    222   first_party_for_cookies_ = first_party_for_cookies;
    223 }
    224 
    225 void URLRequest::set_method(const std::string& method) {
    226   DCHECK(!is_pending_);
    227   method_ = method;
    228 }
    229 
    230 void URLRequest::set_referrer(const std::string& referrer) {
    231   DCHECK(!is_pending_);
    232   referrer_ = referrer;
    233 }
    234 
    235 GURL URLRequest::GetSanitizedReferrer() const {
    236   GURL ret(referrer());
    237 
    238   // Ensure that we do not send username and password fields in the referrer.
    239   if (ret.has_username() || ret.has_password()) {
    240     GURL::Replacements referrer_mods;
    241     referrer_mods.ClearUsername();
    242     referrer_mods.ClearPassword();
    243     ret = ret.ReplaceComponents(referrer_mods);
    244   }
    245 
    246   return ret;
    247 }
    248 
    249 void URLRequest::Start() {
    250   StartJob(GetJobManager()->CreateJob(this));
    251 }
    252 
    253 ///////////////////////////////////////////////////////////////////////////////
    254 
    255 void URLRequest::StartJob(URLRequestJob* job) {
    256   DCHECK(!is_pending_);
    257   DCHECK(!job_);
    258 
    259   net::LoadLog::BeginEvent(load_log_, net::LoadLog::TYPE_URL_REQUEST_START);
    260 
    261   job_ = job;
    262   job_->SetExtraRequestHeaders(extra_request_headers_);
    263 
    264   if (upload_.get())
    265     job_->SetUpload(upload_.get());
    266 
    267   is_pending_ = true;
    268 
    269   response_info_.request_time = Time::Now();
    270   response_info_.was_cached = false;
    271 
    272   // Don't allow errors to be sent from within Start().
    273   // TODO(brettw) this may cause NotifyDone to be sent synchronously,
    274   // we probably don't want this: they should be sent asynchronously so
    275   // the caller does not get reentered.
    276   job_->Start();
    277 }
    278 
    279 void URLRequest::Restart() {
    280   // Should only be called if the original job didn't make any progress.
    281   DCHECK(job_ && !job_->has_response_started());
    282   RestartWithJob(GetJobManager()->CreateJob(this));
    283 }
    284 
    285 void URLRequest::RestartWithJob(URLRequestJob *job) {
    286   DCHECK(job->request() == this);
    287   PrepareToRestart();
    288   StartJob(job);
    289 }
    290 
    291 void URLRequest::Cancel() {
    292   DoCancel(net::ERR_ABORTED, net::SSLInfo());
    293 }
    294 
    295 void URLRequest::SimulateError(int os_error) {
    296   DoCancel(os_error, net::SSLInfo());
    297 }
    298 
    299 void URLRequest::SimulateSSLError(int os_error, const net::SSLInfo& ssl_info) {
    300   // This should only be called on a started request.
    301   if (!is_pending_ || !job_ || job_->has_response_started()) {
    302     NOTREACHED();
    303     return;
    304   }
    305   DoCancel(os_error, ssl_info);
    306 }
    307 
    308 void URLRequest::DoCancel(int os_error, const net::SSLInfo& ssl_info) {
    309   DCHECK(os_error < 0);
    310 
    311   // If the URL request already has an error status, then canceling is a no-op.
    312   // Plus, we don't want to change the error status once it has been set.
    313   if (status_.is_success()) {
    314     status_.set_status(URLRequestStatus::CANCELED);
    315     status_.set_os_error(os_error);
    316     response_info_.ssl_info = ssl_info;
    317   }
    318 
    319   // There's nothing to do if we are not waiting on a Job.
    320   if (!is_pending_ || !job_)
    321     return;
    322 
    323   job_->Kill();
    324 
    325   // The Job will call our NotifyDone method asynchronously.  This is done so
    326   // that the Delegate implementation can call Cancel without having to worry
    327   // about being called recursively.
    328 }
    329 
    330 bool URLRequest::Read(net::IOBuffer* dest, int dest_size, int *bytes_read) {
    331   DCHECK(job_);
    332   DCHECK(bytes_read);
    333   DCHECK(!job_->is_done());
    334   *bytes_read = 0;
    335 
    336   if (dest_size == 0) {
    337     // Caller is not too bright.  I guess we've done what they asked.
    338     return true;
    339   }
    340 
    341   // Once the request fails or is cancelled, read will just return 0 bytes
    342   // to indicate end of stream.
    343   if (!status_.is_success()) {
    344     return true;
    345   }
    346 
    347   return job_->Read(dest, dest_size, bytes_read);
    348 }
    349 
    350 void URLRequest::ReceivedRedirect(const GURL& location, bool* defer_redirect) {
    351   URLRequestJob* job = GetJobManager()->MaybeInterceptRedirect(this, location);
    352   if (job) {
    353     RestartWithJob(job);
    354   } else if (delegate_) {
    355     delegate_->OnReceivedRedirect(this, location, defer_redirect);
    356   }
    357 }
    358 
    359 void URLRequest::ResponseStarted() {
    360   if (!status_.is_success())
    361     net::LoadLog::AddErrorCode(load_log_, status_.os_error());
    362 
    363   net::LoadLog::EndEvent(load_log_, net::LoadLog::TYPE_URL_REQUEST_START);
    364 
    365   URLRequestJob* job = GetJobManager()->MaybeInterceptResponse(this);
    366   if (job) {
    367     RestartWithJob(job);
    368   } else if (delegate_) {
    369     delegate_->OnResponseStarted(this);
    370   }
    371 }
    372 
    373 void URLRequest::FollowDeferredRedirect() {
    374   DCHECK(job_);
    375   DCHECK(status_.is_success());
    376 
    377   job_->FollowDeferredRedirect();
    378 }
    379 
    380 void URLRequest::SetAuth(const wstring& username, const wstring& password) {
    381   DCHECK(job_);
    382   DCHECK(job_->NeedsAuth());
    383 
    384   job_->SetAuth(username, password);
    385 }
    386 
    387 void URLRequest::CancelAuth() {
    388   DCHECK(job_);
    389   DCHECK(job_->NeedsAuth());
    390 
    391   job_->CancelAuth();
    392 }
    393 
    394 void URLRequest::ContinueWithCertificate(net::X509Certificate* client_cert) {
    395   DCHECK(job_);
    396 
    397   job_->ContinueWithCertificate(client_cert);
    398 }
    399 
    400 void URLRequest::ContinueDespiteLastError() {
    401   DCHECK(job_);
    402 
    403   job_->ContinueDespiteLastError();
    404 }
    405 
    406 void URLRequest::PrepareToRestart() {
    407   DCHECK(job_);
    408 
    409   job_->Kill();
    410   OrphanJob();
    411 
    412   response_info_ = net::HttpResponseInfo();
    413   status_ = URLRequestStatus();
    414   is_pending_ = false;
    415 }
    416 
    417 void URLRequest::OrphanJob() {
    418   job_->Kill();
    419   job_->DetachRequest();  // ensures that the job will not call us again
    420   job_ = NULL;
    421 }
    422 
    423 // static
    424 std::string URLRequest::StripPostSpecificHeaders(const std::string& headers) {
    425   // These are headers that may be attached to a POST.
    426   static const char* const kPostHeaders[] = {
    427       "content-type",
    428       "content-length",
    429       "origin"
    430   };
    431   return net::HttpUtil::StripHeaders(
    432       headers, kPostHeaders, arraysize(kPostHeaders));
    433 }
    434 
    435 int URLRequest::Redirect(const GURL& location, int http_status_code) {
    436   if (net::LoadLog::IsUnbounded(load_log_)) {
    437     net::LoadLog::AddString(load_log_, StringPrintf("Redirected (%d) to %s",
    438         http_status_code, location.spec().c_str()));
    439   }
    440   if (redirect_limit_ <= 0) {
    441     DLOG(INFO) << "disallowing redirect: exceeds limit";
    442     return net::ERR_TOO_MANY_REDIRECTS;
    443   }
    444 
    445   if (!location.is_valid())
    446     return net::ERR_INVALID_URL;
    447 
    448   if (!job_->IsSafeRedirect(location)) {
    449     DLOG(INFO) << "disallowing redirect: unsafe protocol";
    450     return net::ERR_UNSAFE_REDIRECT;
    451   }
    452 
    453   bool strip_post_specific_headers = false;
    454   if (http_status_code != 307) {
    455     // NOTE: Even though RFC 2616 says to preserve the request method when
    456     // following a 302 redirect, normal browsers don't do that.  Instead, they
    457     // all convert a POST into a GET in response to a 302 and so shall we.  For
    458     // 307 redirects, browsers preserve the method.  The RFC says to prompt the
    459     // user to confirm the generation of a new POST request, but IE omits this
    460     // prompt and so shall we.
    461     strip_post_specific_headers = method_ == "POST";
    462     method_ = "GET";
    463     upload_ = NULL;
    464   }
    465 
    466   // Suppress the referrer if we're redirecting out of https.
    467   if (GURL(referrer_).SchemeIsSecure() && !location.SchemeIsSecure())
    468     referrer_.clear();
    469 
    470   url_ = location;
    471   --redirect_limit_;
    472 
    473   if (strip_post_specific_headers) {
    474     // If being switched from POST to GET, must remove headers that were
    475     // specific to the POST and don't have meaning in GET. For example
    476     // the inclusion of a multipart Content-Type header in GET can cause
    477     // problems with some servers:
    478     // http://code.google.com/p/chromium/issues/detail?id=843
    479     //
    480     // TODO(eroman): It would be better if this data was structured into
    481     // specific fields/flags, rather than a stew of extra headers.
    482     extra_request_headers_ = StripPostSpecificHeaders(extra_request_headers_);
    483   }
    484 
    485   if (!final_upload_progress_)
    486     final_upload_progress_ = job_->GetUploadProgress();
    487 
    488   PrepareToRestart();
    489   Start();
    490   return net::OK;
    491 }
    492 
    493 URLRequestContext* URLRequest::context() {
    494   return context_.get();
    495 }
    496 
    497 void URLRequest::set_context(URLRequestContext* context) {
    498   scoped_refptr<URLRequestContext> prev_context = context_;
    499 
    500   context_ = context;
    501 
    502   // If the context this request belongs to has changed, update the tracker(s).
    503   if (prev_context != context) {
    504     if (prev_context)
    505       prev_context->url_request_tracker()->Remove(this);
    506     if (context) {
    507       if (!load_log_) {
    508         // Create the LoadLog -- we waited until now to create it so we know
    509         // what constraints the URLRequestContext is enforcing on log levels.
    510         load_log_ = context->url_request_tracker()->CreateLoadLog();
    511       }
    512 
    513       context->url_request_tracker()->Add(this);
    514     }
    515   }
    516 }
    517 
    518 int64 URLRequest::GetExpectedContentSize() const {
    519   int64 expected_content_size = -1;
    520   if (job_)
    521     expected_content_size = job_->expected_content_size();
    522 
    523   return expected_content_size;
    524 }
    525 
    526 URLRequest::UserData* URLRequest::GetUserData(const void* key) const {
    527   UserDataMap::const_iterator found = user_data_.find(key);
    528   if (found != user_data_.end())
    529     return found->second.get();
    530   return NULL;
    531 }
    532 
    533 void URLRequest::SetUserData(const void* key, UserData* data) {
    534   user_data_[key] = linked_ptr<UserData>(data);
    535 }
    536 
    537 void URLRequest::GetInfoForTracker(
    538     RequestTracker<URLRequest>::RecentRequestInfo* info) const {
    539   DCHECK(info);
    540   info->original_url = original_url_;
    541   info->load_log = load_log_;
    542 }
    543