public async Task <bool> PrepareAsync(CookieAwareWebClient client, CancellationToken cancellation)
{
var url = await GetRedirectOverrideAsync(_url, client, cancellation);
if (url == null || cancellation.IsCancellationRequested)
{
return(false);
}
_url = url;
Logging.Write($"{GetType().Name} download link: {_url}");
_innerLoader = await FlexibleLoader.CreateLoaderAsync(_url, this, cancellation);
if (_innerLoader == null || cancellation.IsCancellationRequested)
{
return(false);
}
if (_innerLoader.GetType() == GetType())
{
throw new Exception(ToolsStrings.DirectLoader_RecursionDetected);
}
return(await _innerLoader.PrepareAsync(client, cancellation));
}
private async Task <string> DownloadResumeSupportAsync([NotNull] CookieAwareWebClient client,
[NotNull] FlexibleLoaderGetPreferredDestinationCallback getPreferredDestination,
[CanBeNull] FlexibleLoaderReportDestinationCallback reportDestination, [CanBeNull] Func <bool> checkIfPaused,
IProgress <long> progress, CancellationToken cancellation)
{
// Common variables
string filename = null, selectedDestination = null, actualFootprint = null;
Stream remoteData = null;
string loadedData = null;
var resumeSupported = ResumeSupported;
try {
// Read resume-related data and remove it to avoid conflicts
var resumeDestination = CacheStorage.Get <string>(_keyDestination);
var resumePartiallyLoadedFilename = CacheStorage.Get <string>(_keyPartiallyLoadedFilename);
var resumeLastWriteDate = CacheStorage.Get <DateTime?>(_keyLastWriteDate);
var resumePreviousFootprint = CacheStorage.Get <string>(_keyFootprint);
ClearResumeData();
// Collect known information for destination callback
var information = FlexibleLoaderMetaInformation.FromLoader(this);
// Opening stream to read…
var headRequest = HeadRequestSupported && resumeDestination != null;
using (headRequest ? client.SetMethod("HEAD") : null) {
Logging.Warning($"Initial request: {(headRequest ? "HEAD" : "GET")}");
remoteData = await client.OpenReadTaskAsync(Url);
}
cancellation.ThrowIfCancellationRequested();
// Maybe we’ll be lucky enough to load the most accurate data
if (client.ResponseHeaders != null)
{
if (long.TryParse(client.ResponseHeaders[HttpResponseHeader.ContentLength] ?? "",
NumberStyles.Any, CultureInfo.InvariantCulture, out var length))
{
if (client.ResponseHeaders[HttpResponseHeader.ContentType]?.StartsWith("text/html") == true &&
length < 256 * 1024)
{
Logging.Debug("HTML webpage detected, checking for redirect");
if (headRequest)
{
Logging.Warning("Re-open request to be GET");
remoteData.Dispose();
remoteData = await client.OpenReadTaskAsync(Url);
}
loadedData = remoteData.ReadAsStringAndDispose();
remoteData = null;
var doc = new HtmlDocument();
doc.LoadHtml(loadedData);
var link = doc.DocumentNode.SelectSingleNode(@"//meta[contains(@http-equiv, 'refresh')]")?.Attributes[@"content"]?.Value;
if (link == null)
{
Logging.Warning("Redirect is missing: " + loadedData);
}
else
{
var url = Regex.Match(link, @"\bhttp.+");
if (url.Success)
{
Logging.Debug("Redirect to " + url.Value);
var innerLoader = await FlexibleLoader.CreateLoaderAsync(url.Value, this, cancellation);
if (innerLoader != null)
{
return(await innerLoader.DownloadAsync(client, getPreferredDestination, reportDestination,
checkIfPaused, progress, cancellation));
}
}
}
}
TotalSize = information.TotalSize = length;
}
if (TryGetFileName(client.ResponseHeaders, out var fileName))
{
FileName = information.FileName = fileName;
}
// For example, Google Drive responds with “none” and yet allows to download file partially,
// so this header will only be checked if value is not defined.
if (resumeSupported == null && loadedData == null)
{
var accept = client.ResponseHeaders[HttpResponseHeader.AcceptRanges] ?? "";
if (accept.Contains("bytes"))
{
resumeSupported = true;
}
else if (accept.Contains("none"))
{
resumeSupported = false;
}
}
client.LogResponseHeaders();
}
// Was the file partially loaded before?
var partiallyLoaded = ResumeSupported != false && resumePartiallyLoadedFilename != null
? new FileInfo(FileUtils.EnsureFilenameIsValid(resumePartiallyLoadedFilename, true)) : null;
if (partiallyLoaded != null)
{
Logging.Warning("Not finished: " + partiallyLoaded);
}
// Does it still exist
if (partiallyLoaded?.Exists != true)
{
Logging.Warning($"Partially downloaded file “{partiallyLoaded?.FullName}” does not exist");
partiallyLoaded = null;
}
// If so, wasn’t it changed since the last time?
if (partiallyLoaded?.LastWriteTime > resumeLastWriteDate + TimeSpan.FromMinutes(5))
{
Logging.Warning($"Partially downloaded file is newer that it should be: {partiallyLoaded.LastWriteTime}, expected: {resumeLastWriteDate}");
partiallyLoaded = null;
}
// Looks like file is partially downloaded, but let’s ensure link still leads to the same content
actualFootprint = GetFootprint(information, client.ResponseHeaders);
if (partiallyLoaded != null && resumePreviousFootprint != actualFootprint)
{
Logging.Warning($"Footprints don’t match: {resumePreviousFootprint}≠{actualFootprint}");
partiallyLoaded = null;
}
// Let’s check where to load data, which is potentially the most actual data at this point
var destination = getPreferredDestination(Url, information);
selectedDestination = destination.Filename;
if (partiallyLoaded != null && (!destination.CanResumeDownload || !FileUtils.ArePathsEqual(selectedDestination, resumeDestination)))
{
Logging.Warning($"Different destination chosen: {selectedDestination} (before: {resumeDestination})");
partiallyLoaded = null;
}
// TODO: Check that header?
// Where to write?
// ReSharper disable once MergeConditionalExpression
filename = partiallyLoaded != null ? partiallyLoaded.FullName : FileUtils.EnsureUnique(true, destination.Filename);
reportDestination?.Invoke(filename);
// Set cancellation token
cancellation.Register(o => client.CancelAsync(), null);
// Open write stream
if (partiallyLoaded != null && loadedData == null)
{
var rangeFrom = partiallyLoaded.Length;
using (client.SetRange(new Tuple <long, long>(rangeFrom, -1))) {
Logging.Warning($"Trying to resume download from {rangeFrom} bytes…");
remoteData.Dispose();
remoteData = await client.OpenReadTaskAsync(Url);
cancellation.ThrowIfCancellationRequested();
// client.LogResponseHeaders();
// It’s unknown if resume is supported or not at this point
if (resumeSupported == null)
{
var bytes = new byte[16];
var firstBytes = await remoteData.ReadAsync(bytes, 0, bytes.Length);
cancellation.ThrowIfCancellationRequested();
if (CouldBeBeginningOfAFile(bytes))
{
using (var file = File.Create(filename)) {
Logging.Warning("File beginning found, restart download");
file.Write(bytes, 0, firstBytes);
await CopyToAsync(remoteData, file, checkIfPaused, progress, cancellation);
cancellation.ThrowIfCancellationRequested();
}
Logging.Write("Download finished");
return(filename);
}
rangeFrom += firstBytes;
}
using (var file = new FileStream(filename, FileMode.Append, FileAccess.Write)) {
await CopyToAsync(remoteData, file, checkIfPaused, new Progress <long>(v => { progress?.Report(v + rangeFrom); }), cancellation);
cancellation.ThrowIfCancellationRequested();
}
}
}
else if (loadedData != null)
{
File.WriteAllText(filename, loadedData);
}
else
{
if (headRequest)
{
Logging.Warning("Re-open request to be GET");
remoteData.Dispose();
remoteData = await client.OpenReadTaskAsync(Url);
}
using (var file = File.Create(filename)) {
Logging.Debug("Downloading the whole file…");
await CopyToAsync(remoteData, file, checkIfPaused, progress, cancellation);
cancellation.ThrowIfCancellationRequested();
}
}
Logging.Write("Download finished");
return(filename);
} catch (Exception e) when(e is WebException || e.IsCancelled())
{
Logging.Write("Download is interrupted! Saving details to resume later…");
var download = filename == null ? null : new FileInfo(filename);
if (download?.Exists == true && filename.Length > 0)
{
CacheStorage.Set(_keyDestination, selectedDestination);
CacheStorage.Set(_keyPartiallyLoadedFilename, filename);
CacheStorage.Set(_keyFootprint, actualFootprint);
CacheStorage.Set(_keyLastWriteDate, download.LastWriteTime);
}
else
{
ClearResumeData();
}
throw;
} finally {
if (remoteData != null)
{
remoteData.Dispose();
}
}
}