tstor/src/vfs/archive.go
2024-12-09 23:44:01 +03:00

442 lines
9.4 KiB
Go

package vfs
import (
"archive/zip"
"context"
"fmt"
"io"
"io/fs"
"path"
"strings"
"sync"
"time"
"git.kmsign.ru/royalcat/tstor/pkg/ioutils"
"github.com/bodgit/sevenzip"
"github.com/nwaples/rardecode/v2"
"github.com/royalcat/ctxio"
)
var ArchiveFactories = map[string]FsFactory{
".zip": func(ctx context.Context, sourcePath string, f File) (Filesystem, error) {
stat, err := f.Info()
if err != nil {
return nil, err
}
return NewArchive(ctx, sourcePath, stat.Name(), f, stat.Size(), ZipLoader)
},
".rar": func(ctx context.Context, sourcePath string, f File) (Filesystem, error) {
stat, err := f.Info()
if err != nil {
return nil, err
}
return NewArchive(ctx, sourcePath, stat.Name(), f, stat.Size(), RarLoader)
},
".7z": func(ctx context.Context, sourcePath string, f File) (Filesystem, error) {
stat, err := f.Info()
if err != nil {
return nil, err
}
return NewArchive(ctx, sourcePath, stat.Name(), f, stat.Size(), SevenZipLoader)
},
}
type archiveLoader func(ctx context.Context, archivePath string, r File, size int64) (map[string]fileEntry, error)
var _ Filesystem = &ArchiveFS{}
type fileEntry struct {
fs.FileInfo
open func(ctx context.Context) (File, error)
}
type ArchiveFS struct {
name string
size int64
files map[string]fileEntry
}
// Rename implements Filesystem.
func (a *ArchiveFS) Rename(ctx context.Context, oldpath string, newpath string) error {
return ErrNotImplemented
}
// ModTime implements Filesystem.
func (a *ArchiveFS) ModTime() time.Time {
return time.Time{}
}
// Mode implements Filesystem.
func (a *ArchiveFS) Mode() fs.FileMode {
return fs.ModeDir
}
// Size implements Filesystem.
func (a *ArchiveFS) Size() int64 {
return int64(a.size)
}
// Sys implements Filesystem.
func (a *ArchiveFS) Sys() any {
return nil
}
// FsName implements Filesystem.
func (a *ArchiveFS) FsName() string {
return "archivefs"
}
func NewArchive(ctx context.Context, archivePath, name string, f File, size int64, loader archiveLoader) (*ArchiveFS, error) {
archiveFiles, err := loader(ctx, archivePath, f, size)
if err != nil {
return nil, err
}
// TODO make optional
singleDir := true
for k := range archiveFiles {
if !strings.HasPrefix(k, "/"+name+"/") {
singleDir = false
break
}
}
files := make(map[string]fileEntry, len(archiveFiles))
for k, v := range archiveFiles {
// TODO make optional
if strings.Contains(k, "/__MACOSX/") {
continue
}
if singleDir {
k, _ = strings.CutPrefix(k, "/"+name)
}
files[k] = v
}
// FIXME configurable
files["/.forcegallery"] = fileEntry{
FileInfo: NewFileInfo("/.forcegallery", 0),
open: func(ctx context.Context) (File, error) {
return NewMemoryFile(".forcegallery", []byte{}), nil
},
}
return &ArchiveFS{
name: name,
size: size,
files: files,
}, nil
}
// Unlink implements Filesystem.
func (a *ArchiveFS) Unlink(ctx context.Context, filename string) error {
return ErrNotImplemented
}
func (a *ArchiveFS) Open(ctx context.Context, filename string) (File, error) {
if filename == Separator {
return NewDirFile(filename), nil
}
f, ok := a.files[filename]
if ok {
return f.open(ctx)
}
for p := range a.files {
if strings.HasPrefix(p, filename) {
return NewDirFile(filename), nil
}
}
return nil, ErrNotExist
}
func (a *ArchiveFS) ReadDir(ctx context.Context, path string) ([]fs.DirEntry, error) {
infos := make(map[string]fs.FileInfo, len(a.files))
for k, v := range a.files {
infos[k] = v
}
return ListDirFromInfo(infos, path)
}
// Stat implements Filesystem.
func (afs *ArchiveFS) Stat(ctx context.Context, filename string) (fs.FileInfo, error) {
if entry, ok := afs.files[filename]; ok {
return entry, nil
}
for p, _ := range afs.files {
if strings.HasPrefix(p, filename) {
return NewDirInfo(path.Base(filename)), nil
}
}
return nil, ErrNotExist
}
// Info implements Filesystem.
func (a *ArchiveFS) Info() (fs.FileInfo, error) {
return a, nil
}
// IsDir implements Filesystem.
func (a *ArchiveFS) IsDir() bool {
return true
}
// Name implements Filesystem.
func (a *ArchiveFS) Name() string {
return a.name
}
// Type implements Filesystem.
func (a *ArchiveFS) Type() fs.FileMode {
return fs.ModeDir
}
var _ File = (*archiveFile)(nil)
func newArchiveFile(name string, size int64, rr *randomReaderFromLinear) *archiveFile {
return &archiveFile{
name: name,
size: size,
rr: rr,
}
}
type archiveFile struct {
name string
size int64
m sync.Mutex
offset int64
rr *randomReaderFromLinear
}
// Seek implements File.
func (d *archiveFile) Seek(offset int64, whence int) (int64, error) {
switch whence {
case io.SeekStart:
d.offset = offset
case io.SeekCurrent:
d.offset += offset
case io.SeekEnd:
d.offset = d.size + offset
}
return d.offset, nil
}
// Name implements File.
func (d *archiveFile) Name() string {
return d.name
}
// Type implements File.
func (d *archiveFile) Type() fs.FileMode {
return ModeFileRO
}
func (d *archiveFile) Info() (fs.FileInfo, error) {
return NewFileInfo(d.name, d.size), nil
}
func (d *archiveFile) Size() int64 {
return d.size
}
func (d *archiveFile) IsDir() bool {
return false
}
func (d *archiveFile) Read(ctx context.Context, p []byte) (n int, err error) {
ctx, span := tracer.Start(ctx, "archive.File.Read")
defer span.End()
n, err = d.rr.ReadAt(ctx, p, d.offset)
d.offset += int64(n)
return n, err
}
func (d *archiveFile) ReadAt(ctx context.Context, p []byte, off int64) (n int, err error) {
d.m.Lock()
defer d.m.Unlock()
return d.rr.ReadAt(ctx, p, off)
}
func (d *archiveFile) Close(ctx context.Context) error {
// FIXME close should do nothing as archive fs currently reuse the same file instances
return nil
}
type archiveFileReaderFactory func(ctx context.Context) (ctxio.ReadCloser, error)
var _ archiveLoader = ZipLoader
func ZipLoader(ctx context.Context, archivePath string, f File, size int64) (map[string]fileEntry, error) {
hash, err := FileHash(ctx, f)
if err != nil {
return nil, err
}
reader := ctxio.IoReaderAt(ctx, f)
zr, err := zip.NewReader(reader, size)
if err != nil {
return nil, err
}
out := make(map[string]fileEntry)
for i := range zr.File {
zipFile := zr.File[i]
if zipFile.FileInfo().IsDir() {
continue
}
i := i
af := func(ctx context.Context) (ctxio.ReadCloser, error) {
reader := ctxio.IoReaderAt(ctx, f)
zr, err := zip.NewReader(reader, size)
if err != nil {
return nil, fmt.Errorf("failed to create zip reader: %w", err)
}
rc, err := zr.File[i].Open()
if err != nil {
return nil, fmt.Errorf("failed to open file in zip archive: %w", err)
}
return ctxio.WrapIoReadCloser(rc), nil
}
info := zipFile.FileInfo()
rr := newRandomReaderFromLinear(archiveFileIndex{archiveHash: hash, filename: zipFile.Name}, info.Size(), af)
out[AbsPath(zipFile.Name)] = fileEntry{
FileInfo: info,
open: func(ctx context.Context) (File, error) {
return newArchiveFile(info.Name(), info.Size(), rr), nil
},
}
}
return out, nil
}
var _ archiveLoader = SevenZipLoader
func SevenZipLoader(ctx context.Context, archivePath string, ctxreader File, size int64) (map[string]fileEntry, error) {
hash, err := FileHash(ctx, ctxreader)
if err != nil {
return nil, err
}
reader := ctxio.IoReaderAt(ctx, ctxreader)
r, err := sevenzip.NewReader(reader, size)
if err != nil {
return nil, err
}
out := make(map[string]fileEntry)
for i, f := range r.File {
f := f
if f.FileInfo().IsDir() {
continue
}
i := i
af := func(ctx context.Context) (ctxio.ReadCloser, error) {
reader := ctxio.IoReaderAt(ctx, ctxreader)
zr, err := sevenzip.NewReader(reader, size)
if err != nil {
return nil, err
}
rc, err := zr.File[i].Open()
if err != nil {
return nil, err
}
return ctxio.WrapIoReadCloser(rc), nil
}
info := f.FileInfo()
rr := newRandomReaderFromLinear(archiveFileIndex{archiveHash: hash, filename: f.Name}, info.Size(), af)
out[AbsPath(f.Name)] = fileEntry{
FileInfo: f.FileInfo(),
open: func(ctx context.Context) (File, error) {
return newArchiveFile(info.Name(), info.Size(), rr), nil
},
}
}
return out, nil
}
var _ archiveLoader = RarLoader
func RarLoader(ctx context.Context, archivePath string, f File, size int64) (map[string]fileEntry, error) {
hash, err := FileHash(ctx, f)
if err != nil {
return nil, err
}
reader := ioutils.WrapIoReadSeeker(ctx, f, size)
r, err := rardecode.NewReader(reader)
if err != nil {
return nil, err
}
out := make(map[string]fileEntry)
for {
header, err := r.Next()
if err == io.EOF {
break
}
if err != nil {
return nil, err
}
name := header.Name
af := func(ctx context.Context) (ctxio.ReadCloser, error) {
reader := ioutils.WrapIoReadSeeker(ctx, f, size)
r, err := rardecode.NewReader(reader)
if err != nil {
return nil, err
}
for header, err := r.Next(); err != io.EOF; header, err = r.Next() {
if err != nil {
return nil, err
}
if header.Name == name {
return ctxio.NopCloser(ctxio.WrapIoReader(r)), nil
}
}
return nil, fmt.Errorf("file with name '%s' not found", name)
}
rr := newRandomReaderFromLinear(archiveFileIndex{archiveHash: hash, filename: header.Name}, header.UnPackedSize, af)
out[AbsPath(header.Name)] = fileEntry{
FileInfo: NewFileInfo(header.Name, header.UnPackedSize),
open: func(ctx context.Context) (File, error) {
return newArchiveFile(header.Name, header.UnPackedSize, rr), nil
},
}
}
return out, nil
}