archiver/open.go

package archiver

import (
	"archive/tar"
	"archive/zip"
	"bytes"
	"compress/bzip2"
	"compress/gzip"
	"github.com/ulikunitz/xz"
	"io"
	"os"
	"path/filepath"
)

var magicByteLookup = map[Type][][]byte{
	Tar: {
		{0x75, 0x73, 0x74, 0x61, 0x72, 0x00, 0x30, 0x30},
		{0x75, 0x73, 0x74, 0x61, 0x72, 0x20, 0x20, 0x00},
	},
	TarGzip: {
		{0x1F, 0x8B},
	},
	TarXz: {
		{0xFD, 0x37, 0x7A, 0x58, 0x5A, 0x00},
	},
	TarBzip: {
		{0x42, 0x5A, 0x68},
	},
	Zip: {
		{0x50, 0x4B, 0x03, 0x04},
		{0x50, 0x4B, 0x05, 0x06},
		{0x50, 0x4B, 0x07, 0x08},
	},
}

// OpenFS opens an archive file from a given fs.FS and returns an Archive struct.
// It takes fs.FS, and a path to the archive file as its parameters.
// The function returns an Archive struct and an error, if any.
func OpenFS(filesystem Filesystem, path string) (*Archive, error) {
	return openArchive(filesystem, path)
}

// Open opens an archive file and returns an Archive struct.
// It takes a path to the archive file as its parameter.
// The function returns an Archive struct and an error, if any.
func Open(path string) (*Archive, error) {
	archiveFolderPath, err := filepath.Abs(filepath.Dir(path))
	if err != nil {
		return nil, err
	}
	return openArchive(WrapPath(archiveFolderPath), path)
}

// Extract opens an archive file and extracts the contents.
// It takes a path, and ExtractOptions for its parameters.
// The function returns an error, if any.
func Extract(path string, options ExtractOptions) error {
	archiveFile, err := Open(path)
	if err != nil {
		return err
	}
	defer archiveFile.Close()
	if err := archiveFile.Extract(options); err != nil {
		return err
	}
	if err := archiveFile.Close(); err != nil {
		return err
	}
	return nil
}

// ExtractFile opens an archive file and extracts a specific file within the archive.
// It takes a path, a file path, and ExtractFileOptions for its parameters.
// The function returns an error, if any.
func ExtractFile(path string, file string, options ExtractFileOptions) error {
	archive, err := Open(path)
	if err != nil {
		return err
	}
	defer archive.Close()

	archiveFile, err := archive.GetFile(file)
	if err != nil {
		return err
	}

	if err := archiveFile.Extract(options); err != nil {
		return err
	}
	if err := archive.Close(); err != nil {
		return err
	}
	return nil
}

func openArchive(filesystem Filesystem, path string) (*Archive, error) {
	archive := new(Archive)
	archive.files = make(map[string]*File)

	var file *bytes.Reader

	if filesystem.billyFS != nil {
		billyFS := filesystem.billyFS
		archiveFile, err := billyFS.Open(path)
		if err != nil {
			return nil, err
		}

		fileData, err := io.ReadAll(archiveFile)
		if err != nil {
			return nil, err
		}
		file = bytes.NewReader(fileData)
	} else if filesystem.fs != nil {
		genericFS := filesystem.fs
		archiveFile, err := genericFS.Open(path)
		if err != nil {
			return nil, err
		}
		fileData, err := io.ReadAll(archiveFile)
		if err != nil {
			return nil, err
		}
		file = bytes.NewReader(fileData)
	} else if len(filesystem.path) != 0 {
		archiveFile, err := os.Open(path)
		if err != nil {
			return nil, err
		}
		fileData, err := io.ReadAll(archiveFile)
		if err != nil {
			return nil, err
		}
		file = bytes.NewReader(fileData)
	}

	if file == nil {
		return nil, ErrArchiveFileNotFound
	}

	magicBytes := make([]byte, 512)
	if _, err := file.Read(magicBytes); err != nil {
		return nil, err
	}

	if _, err := file.Seek(0, io.SeekStart); err != nil {
		return nil, err
	}

	var foundArchiveType Type

	for archiveT, fileBytes := range magicByteLookup {
		if len(foundArchiveType) != 0 {
			break
		}
		for _, b := range fileBytes {
			var magicByteWithOffset []byte
			if archiveT == Tar {
				magicByteWithOffset = magicBytes[257 : 257+len(b)]
			} else {
				magicByteWithOffset = magicBytes[0:len(b)]
			}
			if bytes.Equal(b, magicByteWithOffset) {
				foundArchiveType = archiveT
				break
			}

			switch foundArchiveType {
			case TarGzip:
				gzipReader, err := gzip.NewReader(bytes.NewReader(b))
				if err != nil {
					return nil, err
				}
				fileMagicBytes := make([]byte, 512)
				if _, err := gzipReader.Read(fileMagicBytes); err != nil {
					return nil, err
				}
				if !bytes.Equal(b, fileMagicBytes[257:257+len(b)]) {
					return nil, ErrArchiveTypeNotSupported
				}
				if err := gzipReader.Close(); err != nil {
					return nil, err
				}
			case TarXz:
				xzReader, err := xz.NewReader(bytes.NewReader(b))
				if err != nil {
					return nil, err
				}
				fileMagicBytes := make([]byte, 512)
				if _, err := xzReader.Read(fileMagicBytes); err != nil {
					return nil, err
				}
				if !bytes.Equal(b, fileMagicBytes[257:257+len(b)]) {
					return nil, ErrArchiveTypeNotSupported
				}
			case TarBzip:
				bzReader := bzip2.NewReader(bytes.NewReader(b))

				fileMagicBytes := make([]byte, 512)
				if _, err := bzReader.Read(fileMagicBytes); err != nil {
					return nil, err
				}
				if !bytes.Equal(b, fileMagicBytes[257:257+len(b)]) {
					return nil, ErrArchiveTypeNotSupported
				}
			}
		}
	}

	if len(foundArchiveType) == 0 {
		return nil, ErrArchiveTypeNotSupported
	}

	switch foundArchiveType {
	case TarGzip:
		gzipReader, err := gzip.NewReader(file)
		if err != nil {
			return nil, err
		}
		tarReader := tar.NewReader(gzipReader)
		for {
			header, err := tarReader.Next()
			if err != nil && err != io.EOF {
				return nil, err
			} else if err == io.EOF {
				if _, err := file.Seek(0, io.SeekStart); err != nil {
					return nil, err
				}
				break
			}

			_, fileName := filepath.Split(header.Name)
			archive.files[header.Name] = &File{FileName: fileName, Path: header.Name, archive: archive}
		}
		if err := gzipReader.Reset(file); err != nil {
			return nil, err
		}
		archive.reader = gzipReader
		archive.tarReader = tar.NewReader(gzipReader)
	case TarBzip:
		bzipReader := bzip2.NewReader(file)
		tarReader := tar.NewReader(bzipReader)
		for {
			header, err := tarReader.Next()
			if err != nil && err != io.EOF {
				return nil, err
			} else if err == io.EOF {
				if _, err := file.Seek(0, io.SeekStart); err != nil {
					return nil, err
				}
				break
			}

			_, fileName := filepath.Split(header.Name)
			archive.files[header.Name] = &File{FileName: fileName, Path: header.Name, archive: archive}
		}
		bzipReader = bzip2.NewReader(file)
		archive.reader = bzipReader
		archive.tarReader = tar.NewReader(bzipReader)
	case TarXz:
		xzReader, err := xz.NewReader(file)
		if err != nil {
			return nil, err
		}
		tarReader := tar.NewReader(xzReader)
		for {
			header, err := tarReader.Next()
			if err != nil && err != io.EOF {
				return nil, err
			} else if err == io.EOF {
				if _, err := file.Seek(0, io.SeekStart); err != nil {
					return nil, err
				}
				break
			}

			_, fileName := filepath.Split(header.Name)
			archive.files[header.Name] = &File{FileName: fileName, Path: header.Name, archive: archive}
		}
		xzReader, err = xz.NewReader(file)
		if err != nil {
			return nil, err
		}
		archive.reader = xzReader
		archive.tarReader = tar.NewReader(xzReader)
	case Tar:
		tarReader := tar.NewReader(file)
		for {
			header, err := tarReader.Next()
			if err != nil && err != io.EOF {
				return nil, err
			} else if err == io.EOF {
				if _, err := file.Seek(0, io.SeekStart); err != nil {
					return nil, err
				}
				break
			}

			_, fileName := filepath.Split(header.Name)
			archive.files[header.Name] = &File{FileName: fileName, Path: header.Name, archive: archive}
		}
		archive.reader = file
		archive.tarReader = tar.NewReader(file)
	case Zip:
		reader, err := zip.NewReader(file, file.Size())
		if err != nil {
			return nil, err
		}

		archive.reader = reader
		for _, file := range reader.File {
			_, fileName := filepath.Split(file.Name)
			archive.files[file.Name] = &File{FileName: fileName, Path: file.Name, archive: archive}
		}
	default:
		return nil, ErrArchiveTypeNotSupported
	}

	archive.Type = foundArchiveType
	archive.Path = path
	archive.archiveFile = file

	return archive, nil
}