.NetCore 解析Zip - 源码解读 ZipFile

摘要

在之前写的内容中涉及到的内容比较高层次简单，本篇文章更深一步分析ZipFile如何解析zip压缩包中具体文件，然后进行读取解压。

1、.NET Core 文件路径解决方法，统一Linux Window
2、 ZipFIle解压原理

解析主流程

0 获取压缩包对象文档

在上篇文章中说到首先是读取压缩包文件，然后讲压缩包转为对象

ZipArchive

using (ZipArchive archive = ZipFile.Open(sourceArchiveFileName, ZipArchiveMode.Read, entryNameEncoding))
{
archive.ExtractToDirectoryExtension(destinationDirectoryName, overwriteFiles);
}

1 初始化压缩文档的处理模式

ZipFile.Open(string archiveFileName, ZipArchiveMode mode, Encoding? entryNameEncoding) 方法中

1. 声明文件操作模式为打开（打开、创建、截取等）
2. 声明对文件的访问方式为读操作（读、写、读并写）
3. 共享权限
   详细查看读取文件流
   

   FileStream(String, FileMode, FileAccess, FileShare, Int32, Boolean)

   
   使用指定的路径、创建模式、读/写和共享权限、缓冲区大小和同步或异步状态初始化 FileStream 类的新实例。
   参考资料：https://docs.microsoft.com/zh-cn/dotnet/api/system.io.filestream?view=netcore-3.1

2 初始化压缩包对象的数据

接下来就是创建ZipArchive对象

public ZipArchive(Stream stream, ZipArchiveMode mode, bool leaveOpen, Encoding? entryNameEncoding)

1. 打开压缩文件为文件流后转为二进制流，供解析Zip数据量用
2. 初始化压缩文档集合变量，供获取压缩条目用于储存到本地
3. 初始化压缩文档字典，根据文档名称获取压缩文档
4. 其他初始化工作
5. 读取压缩包核心目录结束标记，用于计算压缩头目录开始位置

public ZipArchive(Stream stream, ZipArchiveMode mode, bool leaveOpen, Encoding? entryNameEncoding)
{
if (stream == null)
throw new ArgumentNullException(nameof(stream));

EntryNameEncoding = entryNameEncoding;
Stream? extraTempStream = null;

try
{
//省略代码
//初始化 ZipArchive 数据
//省略代码
_entriesCollection = new ReadOnlyCollection<ZipArchiveEntry>(_entries);

switch (mode)
{
//省略代码
//。。。
//省略代码
case ZipArchiveMode.Read:
//读取中央目录结束标记
ReadEndOfCentralDirectory();
break;
//省略代码
//。。。
//省略代码
}
}
catch
{
if (extraTempStream != null)
extraTempStream.Dispose();

throw;
}
}

方法

ReadEndOfCentralDirectory

private void ReadEndOfCentralDirectory()
{
// This seeks backwards almost to the beginning of the EOCD, one byte after where the signature would be
// located if the EOCD had the minimum possible size (no file zip comment)
// 向后查找到EOCD的开头，即签名所在位置之后的一个字节
// 查找EOCD具有可能的最小值(没有文件zip注释)
_archiveStream.Seek(-ZipEndOfCentralDirectoryBlock.SizeOfBlockWithoutSignature, SeekOrigin.End);

// If the EOCD has the minimum possible size (no zip file comment), then exactly the previous 4 bytes will contain the signature
// But if the EOCD has max possible size, the signature should be found somewhere in the previous 64K + 4 bytes
//如果EOCD具有尽可能小的大小(没有zip文件注释)，那么前面的4个字节将恰好包含签名
//但是如果EOCD有最大的可能大小，签名应该在前面的64K + 4字节中找到
if (!ZipHelper.SeekBackwardsToSignature(_archiveStream,
ZipEndOfCentralDirectoryBlock.SignatureConstant,
ZipEndOfCentralDirectoryBlock.ZipFileCommentMaxLength + ZipEndOfCentralDirectoryBlock.SignatureSize))
throw new InvalidDataException();

//省略代码
//。。。
//省略代码
}

ZipHelper.SeekBackwardsToSignature

internal static bool SeekBackwardsToSignature(Stream stream, uint signatureToFind, int maxBytesToRead)
{
int bufferPointer = 0;
bool signatureFound = false;

//省略代码
//计算核心目录头的偏移量
//省略代码

if (!signatureFound)
{
return false;
}
else
{
//将数据流设置到当前流中‘核心目录头’的位置。
stream.Seek(bufferPointer, SeekOrigin.Current);
return true;
}
}

3 读取压缩包中文档集合

循环读取压缩包中文档

foreach (ZipArchiveEntry entry in source.Entries)
{
entry.ExtractRelativeToDirectoryExtension(destinationDirectoryName, overwriteFiles);
}

读取ZipArchive属性

Entries

public ReadOnlyCollection<ZipArchiveEntry> Entries
{
get
{
if (_mode == ZipArchiveMode.Create)
throw new NotSupportedException();

ThrowIfDisposed();

//确保已经读取过核心目录头，核心目录头：Zip数据协议格式中项，后面补上
//
EnsureCentralDirectoryRead();
return _entriesCollection;
}
}

private void EnsureCentralDirectoryRead()
{
//是否未过读取压缩文档
if (!_readEntries)
{
//没有读取过则开始读取，并转换未ZipArchiveEntry集合
ReadCentralDirectory();
_readEntries = true;
}
}
private void ReadCentralDirectory()
{
try
{
// assume ReadEndOfCentralDirectory has been called and has populated _centralDirectoryStart
//设置当前文件流到核心目录开始的位置
_archiveStream.Seek(_centralDirectoryStart, SeekOrigin.Begin);

long numberOfEntries = 0;

Debug.Assert(_archiveReader != null);
//read the central directory
ZipCentralDirectoryFileHeader currentHeader;
bool saveExtraFieldsAndComments = Mode == ZipArchiveMode.Update;
//循环读取文件核心头
while (ZipCentralDirectoryFileHeader.TryReadBlock(_archiveReader,
saveExtraFieldsAndComments, out currentHeader))
{
//转为ZipArchiveEntry 并添加到集合中
AddEntry(new ZipArchiveEntry(this, currentHeader));
numberOfEntries++;
}

if (numberOfEntries != _expectedNumberOfEntries)
throw new InvalidDataException();
}
catch (EndOfStreamException ex)
{
throw new InvalidDataException();
}
}

3.1 读取压缩目录文件头

public static bool TryReadBlock(BinaryReader reader, bool saveExtraFieldsAndComments, out ZipCentralDirectoryFileHeader header)
{
header = default;

if (reader.ReadUInt32() != SignatureConstant)
return false;
header.VersionMadeBySpecification = reader.ReadByte();
header.VersionMadeByCompatibility = reader.ReadByte();
header.VersionNeededToExtract = reader.ReadUInt16();
header.GeneralPurposeBitFlag = reader.ReadUInt16();
header.CompressionMethod = reader.ReadUInt16();
header.LastModified = reader.ReadUInt32();
header.Crc32 = reader.ReadUInt32();
uint compressedSizeSmall = reader.ReadUInt32();
uint uncompressedSizeSmall = reader.ReadUInt32();
header.FilenameLength = reader.ReadUInt16();
header.ExtraFieldLength = reader.ReadUInt16();
header.FileCommentLength = reader.ReadUInt16();
ushort diskNumberStartSmall = reader.ReadUInt16();
header.InternalFileAttributes = reader.ReadUInt16();
header.ExternalFileAttributes = reader.ReadUInt32();
uint relativeOffsetOfLocalHeaderSmall = reader.ReadUInt32();

//代码省略，读取其他信息

return true;
}
}

文件核心目录头

ZipCentralDirectoryFileHeader

相对于本地头的偏移量，用于后面计算读取文件内容的起始位置

3.2 转为压缩文档

while (ZipCentralDirectoryFileHeader.TryReadBlock(_archiveReader,
saveExtraFieldsAndComments, out currentHeader))
{
//转为ZipArchiveEntry 并添加到集合中
AddEntry(new ZipArchiveEntry(this, currentHeader));
numberOfEntries++;
}
->
internal ZipArchiveEntry(ZipArchive archive, ZipCentralDirectoryFileHeader cd)
{
_archive = archive;

_originallyInArchive = true;

_diskNumberStart = cd.DiskNumberStart;
_versionMadeByPlatform = (ZipVersionMadeByPlatform)cd.VersionMadeByCompatibility;
_versionMadeBySpecification = (ZipVersionNeededValues)cd.VersionMadeBySpecification;
_versionToExtract = (ZipVersionNeededValues)cd.VersionNeededToExtract;
//省略代码
//初始化 ZipArchiveEntry
}

4 返回文档集合

前面步骤处理完成后返回文档集合

5 遍历文档条目集合

public static void ExtractToFileExtension(this ZipArchiveEntry source, string destinationFileName, bool overwrite)
{
//省略代码
using (Stream fs = new FileStream(destinationFileName, fMode, FileAccess.Write, FileShare.None, bufferSize: 0x1000, useAsync: false))
{
//source.Open 根据文档条目目录文件头信息读取数据
using (Stream es = source.Open())
es.CopyTo(fs);  //保存
}

//设置最后修改时间
File.SetLastWriteTime(destinationFileName, source.LastWriteTime.DateTime);
}

6 读取数据

public Stream Open()
{
ThrowIfInvalidArchive();

switch (_archive.Mode)
{
case ZipArchiveMode.Read:
//我们只看这行
return OpenInReadMode(checkOpenable: true);
case ZipArchiveMode.Create:
return OpenInWriteMode();
case ZipArchiveMode.Update:
default:
Debug.Assert(_archive.Mode == ZipArchiveMode.Update);
return OpenInUpdateMode();
}
}
private Stream OpenInReadMode(bool checkOpenable)
{
//省略代码

//_archive.ArchiveStream 压缩包二进制流
//OffsetOfCompressedData 当前文件目录相对于头的偏移量，也就是文件内容读取的开始位置
//_compressedSize 文档内容长度，这个长度是压缩后的长度
Stream compressedStream = new SubReadStream(_archive.ArchiveStream, OffsetOfCompressedData, _compressedSize);
return GetDataDecompressor(compressedStream);
}

//开始读取
private Stream GetDataDecompressor(Stream compressedStreamToRead)
{
Stream? uncompressedStream = null;
//CompressionMethod 压缩方式，根据不同的压缩算法获取文件内容
// Stored = 0x0, Deflate = 0x8, Deflate64 = 0x9, BZip2 = 0xC, LZMA = 0xE
switch (CompressionMethod)
{
case CompressionMethodValues.Deflate:
uncompressedStream = new DeflateStream(compressedStreamToRead, CompressionMode.Decompress, _uncompressedSize);
break;
case CompressionMethodValues.Deflate64:
uncompressedStream = new DeflateManagedStream(compressedStreamToRead, CompressionMethodValues.Deflate64, _uncompressedSize);
break;
case CompressionMethodValues.Stored:
default:
// we can assume that only deflate/deflate64/stored are allowed because we assume that
// IsOpenable is checked before this function is called
Debug.Assert(CompressionMethod == CompressionMethodValues.Stored);

uncompressedStream = compressedStreamToRead;
break;
}

return uncompressedStream;
}

读取到文件内容后则可以进行保存工作

总结

整个解析Zip过程比较清晰，但是内容较多需要慢慢消化，而

Zip

补充
查看Zip数据协议格式