public static async ValueTask <ExtendedHashes> FromFile(IExtractedFile file)
{
var hashes = new ExtendedHashes();
await using var stream = await file.OpenRead();
hashes.SHA256 = System.Security.Cryptography.SHA256.Create().ComputeHash(stream).ToHex();
stream.Position = 0;
hashes.SHA1 = System.Security.Cryptography.SHA1.Create().ComputeHash(stream).ToHex();
stream.Position = 0;
hashes.MD5 = System.Security.Cryptography.MD5.Create().ComputeHash(stream).ToHex();
stream.Position = 0;
var bytes = new byte[1024 * 8];
var crc = new Crc32();
while (true)
{
var read = stream.Read(bytes, 0, bytes.Length);
if (read == 0)
{
break;
}
crc.Update(bytes, 0, read);
}
hashes.CRC = crc.DigestBytes().ToHex();
return(hashes);
}
public static async Task <VirtualFile> Analyze(Context context, VirtualFile parent, IExtractedFile extractedFile,
IPath relPath, int depth = 0)
{
var hash = await extractedFile.HashAsync();
if (!context.UseExtendedHashes && FileExtractor.MightBeArchive(relPath.FileName.Extension))
{
var result = await TryGetContentsFromServer(hash);
if (result != null)
{
Utils.Log($"Downloaded VFS data for {relPath.FileName}");
return(ConvertFromIndexedFile(context, result, relPath, parent, extractedFile));
}
}
if (TryGetFromCache(context, parent, relPath, extractedFile, hash, out var vself))
{
return(vself);
}
var self = new VirtualFile
{
Context = context,
Name = relPath,
Parent = parent,
Size = extractedFile.Size,
LastModified = extractedFile.LastModifiedUtc.AsUnixTime(),
LastAnalyzed = DateTime.Now.AsUnixTime(),
Hash = hash
};
self.FillFullPath(depth);
if (context.UseExtendedHashes)
{
self.ExtendedHashes = await ExtendedHashes.FromFile(extractedFile);
}
if (!await extractedFile.CanExtract())
{
return(self);
}
try
{
await using var extracted = await extractedFile.ExtractAll(context.Queue);
var list = await extracted
.PMap(context.Queue,
file => Analyze(context, self, file.Value, file.Key, depth + 1));
self.Children = list.ToImmutableList();
}
catch (Exception ex)
{
Utils.Log($"Error while examining the contents of {relPath.FileName}");
throw;
}
await using var ms = new MemoryStream();
self.ToIndexedVirtualFile().ToJson(ms);
_vfsCache.Put(self.Hash.ToArray(), ms.ToArray());
return(self);
}
private static VirtualFile ConvertFromIndexedFile(Context context, IndexedVirtualFile file, IPath path, VirtualFile vparent, IExtractedFile extractedFile)
{
var vself = new VirtualFile
{
Context = context,
Name = path,
Parent = vparent,
Size = file.Size,
LastModified = extractedFile.LastModifiedUtc.AsUnixTime(),
LastAnalyzed = DateTime.Now.AsUnixTime(),
Hash = file.Hash
};
vself.FillFullPath();
vself.Children = file.Children.Select(f => ConvertFromIndexedFile(context, f, f.Name, vself, extractedFile)).ToImmutableList();
return(vself);
}
private static bool TryGetFromCache(Context context, VirtualFile parent, IPath path, IExtractedFile extractedFile, Hash hash, out VirtualFile found)
{
var result = _vfsCache.Get(hash.ToArray());
if (result == null)
{
found = null;
return(false);
}
var data = new MemoryStream(result).FromJson <IndexedVirtualFile>();
found = ConvertFromIndexedFile(context, data, path, parent, extractedFile);
return(true);
}
public static async Task <VirtualFile> Analyze(Context context, VirtualFile parent, IExtractedFile extractedFile,
IPath relPath, int depth = 0)
{
var hash = await extractedFile.HashAsync();
if (!context.UseExtendedHashes && FileExtractor.MightBeArchive(relPath.FileName.Extension))
{
var result = await TryGetContentsFromServer(hash);
if (result != null)
{
Utils.Log($"Downloaded VFS data for {relPath.FileName}");
VirtualFile Convert(IndexedVirtualFile file, IPath path, VirtualFile vparent)
{
var vself = new VirtualFile
{
Context = context,
Name = path,
Parent = vparent,
Size = file.Size,
LastModified = extractedFile.LastModifiedUtc.AsUnixTime(),
LastAnalyzed = DateTime.Now.AsUnixTime(),
Hash = file.Hash
};
vself.FillFullPath();
vself.Children = file.Children.Select(f => Convert(f, f.Name, vself)).ToImmutableList();
return(vself);
}
return(Convert(result, relPath, parent));
}
}
var self = new VirtualFile
{
Context = context,
Name = relPath,
Parent = parent,
Size = extractedFile.Size,
LastModified = extractedFile.LastModifiedUtc.AsUnixTime(),
LastAnalyzed = DateTime.Now.AsUnixTime(),
Hash = hash
};
self.FillFullPath(depth);
if (context.UseExtendedHashes)
{
self.ExtendedHashes = ExtendedHashes.FromFile(extractedFile);
}
if (!await extractedFile.CanExtract())
{
return(self);
}
try
{
await using var extracted = await extractedFile.ExtractAll(context.Queue);
var list = await extracted
.PMap(context.Queue,
file => Analyze(context, self, file.Value, file.Key, depth + 1));
self.Children = list.ToImmutableList();
}
catch (Exception ex)
{
Utils.Log($"Error while examining the contents of {relPath.FileName}");
throw;
}
return(self);
}
