archiver/archive.go

package archiver

import (
	"archive/tar"
	"bytes"
	"compress/gzip"
	"embed"
	"errors"
	"github.com/go-git/go-billy/v5"
	"io/fs"
	"os"
	"regexp"
	"sync"
)

type Type string

var (
	ErrArchiveTypeNotSupported = errors.New("archive type not supported")
	ErrArchiveFileNotFound     = errors.New("archive file not found")
)

const (
	TarGzip Type = "tar.gz"
	TarBzip      = "tar.bz2"
	TarXz        = "tar.xz"
	Tar          = "tar"
	Zip          = "zip"
)

type Archive struct {
	// Type of Archive
	Type Type
	// Path of Archive file
	Path        string
	reader      interface{}
	tarReader   *tar.Reader // Used for anything with .tar due to how tar.Reader cannot be reset.
	files       map[string]*File
	archiveFile *bytes.Reader
	mu          sync.Mutex
}

// Filesystem represents a standard interface for filesystems.
// Currently, supports fs.FS and billy.Filesystem.
type Filesystem struct {
	billyFS billy.Filesystem
	fs      fs.FS
	embed   *embed.FS
	file    bool
	path    string
}

// WrapBillyFS wraps billy.Filesystem to the Filesystem interface.
// Returns Filesystem.
func WrapBillyFS(filesystem billy.Filesystem) Filesystem {
	return Filesystem{
		billyFS: filesystem,
	}
}

// WrapFS wraps fs.FS to the Filesystem interface.
// Returns Filesystem.
func WrapFS(filesystem fs.FS) Filesystem {
	return Filesystem{
		fs: filesystem,
	}
}

// WrapPath wraps a path to the Filesystem interface.
// Returns Filesystem.
func WrapPath(path string) Filesystem {
	return Filesystem{
		path: path,
	}
}

// ExtractOptions represents options for extracting an Archive
type ExtractOptions struct {
	// Whether to overwrite files if they already exist.
	Overwrite bool
	// Folder to extract Archive to.
	Folder string
	// Whether to preserve the file structure in the Archive or not.
	NotPreserveFileStructure bool
	// Optional regex filter for specific files.
	Filter *regexp.Regexp
}

// ExtractFileOptions represents options for extracting an ArchiveFile
type ExtractFileOptions struct {
	// Whether to overwrite files if they already exist.
	Overwrite bool
	// Perms for output file, default is the permission in the Archive.
	Perms os.FileMode
	// Folder to extract Archive to.
	Folder string
	// Whether to preserve the file structure in the Archive or not.
	NotPreserveFileStructure bool
}

// File represents a file in an archive.
type File struct {
	// Name of file (e.g "test.txt")
	FileName string
	// Path of file in archive (e.g "folder/test.txt")
	Path    string
	archive *Archive
}

// GetFile gets a File from the Archive
// It takes the path of the file in the Archive as its parameter.
// The function returns ArchiveFile and an error, if any.
func (a *Archive) GetFile(path string) (*File, error) {
	a.mu.Lock()
	defer a.mu.Unlock()
	file, ok := a.files[path]
	if !ok {
		return nil, ErrArchiveFileNotFound
	}

	return file, nil
}

// FileCount gets the total file count in the Archive.
// The function returns an integer.
func (a *Archive) FileCount() int {
	return len(a.files)
}

// ExtractBillyFS extracts the specified Archive into billy.Filesystem.
// It takes billy.Filesystem and ExtractOptions as its parameters.
// The function returns an error, if any.
func (a *Archive) ExtractBillyFS(filesystem billy.Filesystem, opts ExtractOptions) error {
	return extract(WrapBillyFS(filesystem), opts, a)
}

// Extract extracts the specified Archive to a specified directory.
// It takes ExtractOptions as its parameter
// The function returns an error, if any.
func (a *Archive) Extract(opts ExtractOptions) error {
	return extract(Filesystem{file: true}, opts, a)
}

// ExtractFile extracts a specified file from an archive to a specified directory.
// It takes a file path, and ExtractFileOptions as its parameters.
// The function returns an error, if any.
func (a *Archive) ExtractFile(path string, opts ExtractFileOptions) error {
	file, err := a.GetFile(path)
	if err != nil {
		return err
	}
	return file.Extract(opts)
}

// Close closes the Archive.
// The function returns an error, if any.
func (a *Archive) Close() error {
	a.archiveFile = nil

	switch a.Type {
	case TarGzip:
		if err := a.reader.(*gzip.Reader).Close(); err != nil {
			return err
		}
	}

	return nil
}

// ExtractBillyFS extracts the specified File in the archive into billy.Filesystem.
// It takes billy.Filesystem and ExtractFileOptions as its parameters.
// The function returns an error, if any.
func (f *File) ExtractBillyFS(filesystem billy.Filesystem, opts ExtractFileOptions) error {
	return extractFile(WrapBillyFS(filesystem), opts, f)
}

// Extract extracts the specified File in the archive to a specified directory.
// It takes ExtractFileOptions as its parameter.
// The function returns an error, if any.
func (f *File) Extract(opts ExtractFileOptions) error {
	return extractFile(Filesystem{file: true}, opts, f)
}

// Read reads the specified File in the archive and returns the content.
// The function returns the content of the File, and an error, if any.
func (f *File) Read() ([]byte, error) {
	switch f.archive.Type {
	case Tar, TarGzip, TarBzip, TarXz:
		return tarRead(f)
	case Zip:
		return zipRead(f)
	default:
		return nil, ErrArchiveTypeNotSupported
	}
}