Merge branch 'master' into master

README.md

@@ -79,11 +79,11 @@ would.
 
 So if my application before had:
 ```go
-os.Open('/tmp/foo')
+os.Open("/tmp/foo")
 ```
 We would replace it with:
 ```go
-AppFs.Open('/tmp/foo')
+AppFs.Open("/tmp/foo")
 ```
 
 `AppFs` being the variable we defined above.
@@ -259,6 +259,18 @@ system using InMemoryFile.
 Afero has experimental support for secure file transfer protocol (sftp). Which can
 be used to perform file operations over a encrypted channel.
 
+### GCSFs
+
+Afero has experimental support for Google Cloud Storage (GCS). You can either set the
+`GOOGLE_APPLICATION_CREDENTIALS_JSON` env variable to your JSON credentials or use `opts` in
+`NewGcsFS` to configure access to your GCS bucket.
+
+Some known limitations of the existing implementation:
+* No Chmod support - The GCS ACL could probably be mapped to *nix style permissions but that would add another level of complexity and is ignored in this version.
+* No Chtimes support - Could be simulated with attributes (gcs a/m-times are set implicitly) but that's is left for another version.
+* Not thread safe - Also assumes all file operations are done through the same instance of the GcsFs. File operations between different GcsFs instances are not guaranteed to be consistent.
+
+
 ## Filtering Backends
 
 ### BasePathFs

gcsfs/errors.go

@@ -0,0 +1,31 @@
+// Copyright © 2021 Vasily Ovchinnikov <[email protected]>.
+//
+// The code in this file is derived from afero fork github.com/Zatte/afero by Mikael Rapp
+// licensed under Apache License 2.0.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package gcsfs
+
+import (
+	"errors"
+	"syscall"
+)
+
+var (
+	ErrNoBucketInName     = errors.New("no bucket name found in the name")
+	ErrFileClosed         = errors.New("file is closed")
+	ErrOutOfRange         = errors.New("out of range")
+	ErrObjectDoesNotExist = errors.New("storage: object doesn't exist")
+	ErrEmptyObjectName    = errors.New("storage: object name is empty")
+	ErrFileNotFound       = syscall.ENOENT
+)

gcsfs/file.go

@@ -0,0 +1,307 @@
+// Copyright © 2021 Vasily Ovchinnikov <[email protected]>.
+//
+// The code in this file is derived from afero fork github.com/Zatte/afero by Mikael Rapp
+// licensed under Apache License 2.0.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package gcsfs
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"log"
+	"os"
+	"path/filepath"
+	"sort"
+	"syscall"
+
+	"github.com/googleapis/google-cloud-go-testing/storage/stiface"
+
+	"cloud.google.com/go/storage"
+
+	"google.golang.org/api/iterator"
+)
+
+// GcsFs is the Afero version adapted for GCS
+type GcsFile struct {
+	openFlags int
+	fhOffset  int64 //File handle specific offset
+	closed    bool
+	ReadDirIt stiface.ObjectIterator
+	resource  *gcsFileResource
+}
+
+func NewGcsFile(
+	ctx context.Context,
+	fs *Fs,
+	obj stiface.ObjectHandle,
+	openFlags int,
+	// Unused: there is no use to the file mode in GCloud just yet - but we keep it here, just in case we need it
+	fileMode os.FileMode,
+	name string,
+) *GcsFile {
+	return &GcsFile{
+		openFlags: openFlags,
+		fhOffset:  0,
+		closed:    false,
+		ReadDirIt: nil,
+		resource: &gcsFileResource{
+			ctx: ctx,
+			fs:  fs,
+
+			obj:      obj,
+			name:     name,
+			fileMode: fileMode,
+
+			currentGcsSize: 0,
+
+			offset: 0,
+			reader: nil,
+			writer: nil,
+		},
+	}
+}
+
+func NewGcsFileFromOldFH(
+	openFlags int,
+	fileMode os.FileMode,
+	oldFile *gcsFileResource,
+) *GcsFile {
+	res := &GcsFile{
+		openFlags: openFlags,
+		fhOffset:  0,
+		closed:    false,
+		ReadDirIt: nil,
+
+		resource: oldFile,
+	}
+	res.resource.fileMode = fileMode
+
+	return res
+}
+
+func (o *GcsFile) Close() error {
+	if o.closed {
+		// the afero spec expects the call to Close on a closed file to return an error
+		return ErrFileClosed
+	}
+	o.closed = true
+	return o.resource.Close()
+}
+
+func (o *GcsFile) Seek(newOffset int64, whence int) (int64, error) {
+	if o.closed {
+		return 0, ErrFileClosed
+	}
+
+	//Since this is an expensive operation; let's make sure we need it
+	if (whence == 0 && newOffset == o.fhOffset) || (whence == 1 && newOffset == 0) {
+		return o.fhOffset, nil
+	}
+	log.Printf("WARNING: Seek beavhior triggered, highly inefficent. Offset before seek is at %d\n", o.fhOffset)
+
+	//Fore the reader/writers to be reopened (at correct offset)
+	err := o.Sync()
+	if err != nil {
+		return 0, err
+	}
+	stat, err := o.Stat()
+	if err != nil {
+		return 0, nil
+	}
+
+	switch whence {
+	case 0:
+		o.fhOffset = newOffset
+	case 1:
+		o.fhOffset += newOffset
+	case 2:
+		o.fhOffset = stat.Size() + newOffset
+	}
+	return o.fhOffset, nil
+}
+
+func (o *GcsFile) Read(p []byte) (n int, err error) {
+	return o.ReadAt(p, o.fhOffset)
+}
+
+func (o *GcsFile) ReadAt(p []byte, off int64) (n int, err error) {
+	if o.closed {
+		return 0, ErrFileClosed
+	}
+
+	read, err := o.resource.ReadAt(p, off)
+	o.fhOffset += int64(read)
+	return read, err
+}
+
+func (o *GcsFile) Write(p []byte) (n int, err error) {
+	return o.WriteAt(p, o.fhOffset)
+}
+
+func (o *GcsFile) WriteAt(b []byte, off int64) (n int, err error) {
+	if o.closed {
+		return 0, ErrFileClosed
+	}
+
+	if o.openFlags&os.O_RDONLY != 0 {
+		return 0, fmt.Errorf("file is opend as read only")
+	}
+
+	_, err = o.resource.obj.Attrs(o.resource.ctx)
+	if err != nil {
+		if err == storage.ErrObjectNotExist {
+			if o.openFlags&os.O_CREATE == 0 {
+				return 0, ErrFileNotFound
+			}
+		} else {
+			return 0, fmt.Errorf("error getting file attributes: %v", err)
+		}
+	}
+
+	written, err := o.resource.WriteAt(b, off)
+	o.fhOffset += int64(written)
+	return written, err
+}
+
+func (o *GcsFile) Name() string {
+	return filepath.FromSlash(o.resource.name)
+}
+
+func (o *GcsFile) readdirImpl(count int) ([]*FileInfo, error) {
+	err := o.Sync()
+	if err != nil {
+		return nil, err
+	}
+
+	var ownInfo os.FileInfo
+	ownInfo, err = o.Stat()
+	if err != nil {
+		return nil, err
+	}
+
+	if !ownInfo.IsDir() {
+		return nil, syscall.ENOTDIR
+	}
+
+	path := o.resource.fs.ensureTrailingSeparator(o.resource.name)
+	if o.ReadDirIt == nil {
+		//log.Printf("Querying path : %s\n", path)
+		bucketName, bucketPath := o.resource.fs.splitName(path)
+
+		o.ReadDirIt = o.resource.fs.client.Bucket(bucketName).Objects(
+			o.resource.ctx, &storage.Query{Delimiter: o.resource.fs.separator, Prefix: bucketPath, Versions: false})
+	}
+	var res []*FileInfo
+	for {
+		object, err := o.ReadDirIt.Next()
+		if err == iterator.Done {
+			// reset the iterator
+			o.ReadDirIt = nil
+
+			if len(res) > 0 || count <= 0 {
+				return res, nil
+			}
+
+			return res, io.EOF
+		}
+		if err != nil {
+			return res, err
+		}
+
+		tmp := newFileInfoFromAttrs(object, o.resource.fileMode)
+
+		if tmp.Name() == "" {
+			// neither object.Name, not object.Prefix were present - so let's skip this unknown thing
+			continue
+		}
+
+		if object.Name == "" && object.Prefix == "" {
+			continue
+		}
+
+		if tmp.Name() == ownInfo.Name() {
+			// Hmmm
+			continue
+		}
+
+		res = append(res, tmp)
+
+		// This would interrupt the iteration, once we reach the count.
+		// But it would then have files coming before folders - that's not what we want to have exactly,
+		// since it makes the results unpredictable. Hence, we iterate all the objects and then do
+		// the cut-off in a higher level method
+		//if count > 0 && len(res) >= count {
+		//	break
+		//}
+	}
+	//return res, nil
+}
+
+func (o *GcsFile) Readdir(count int) ([]os.FileInfo, error) {
+	fi, err := o.readdirImpl(count)
+	if len(fi) > 0 {
+		sort.Sort(ByName(fi))
+	}
+
+	if count > 0 {
+		fi = fi[:count]
+	}
+
+	var res []os.FileInfo
+	for _, f := range fi {
+		res = append(res, f)
+	}
+	return res, err
+}
+
+func (o *GcsFile) Readdirnames(n int) ([]string, error) {
+	fi, err := o.Readdir(n)
+	if err != nil && err != io.EOF {
+		return nil, err
+	}
+	names := make([]string, len(fi))
+
+	for i, f := range fi {
+		names[i] = f.Name()
+	}
+	return names, err
+}
+
+func (o *GcsFile) Stat() (os.FileInfo, error) {
+	err := o.Sync()
+	if err != nil {
+		return nil, err
+	}
+
+	return newFileInfo(o.resource.name, o.resource.fs, o.resource.fileMode)
+}
+
+func (o *GcsFile) Sync() error {
+	return o.resource.maybeCloseIo()
+}
+
+func (o *GcsFile) Truncate(wantedSize int64) error {
+	if o.closed {
+		return ErrFileClosed
+	}
+	if o.openFlags == os.O_RDONLY {
+		return fmt.Errorf("file was opened as read only")
+	}
+	return o.resource.Truncate(wantedSize)
+}
+
+func (o *GcsFile) WriteString(s string) (ret int, err error) {
+	return o.Write([]byte(s))
+}