Break the fs package up into smaller parts.

The purpose of this is to make it easier to maintain and eventually to
allow the rclone backends to be re-used in other projects without
having to use the rclone configuration system.

The new code layout is documented in CONTRIBUTING.

lib/readers/repeatable.go

@@ -0,0 +1,96 @@
+package readers
+
+import (
+	"io"
+
+	"github.com/pkg/errors"
+)
+
+// A RepeatableReader implements the io.ReadSeeker it allow to seek cached data
+// back and forth within the reader but will only read data from the internal Reader as necessary
+// and will play nicely with the Account and io.LimitedReader to reflect current speed
+type RepeatableReader struct {
+	in io.Reader // Input reader
+	i  int64     // current reading index
+	b  []byte    // internal cache buffer
+}
+
+var _ io.ReadSeeker = (*RepeatableReader)(nil)
+
+// Seek implements the io.Seeker interface.
+// If seek position is passed the cache buffer length the function will return
+// the maximum offset that can be used and "fs.RepeatableReader.Seek: offset is unavailable" Error
+func (r *RepeatableReader) Seek(offset int64, whence int) (int64, error) {
+	var abs int64
+	cacheLen := int64(len(r.b))
+	switch whence {
+	case 0: //io.SeekStart
+		abs = offset
+	case 1: //io.SeekCurrent
+		abs = r.i + offset
+	case 2: //io.SeekEnd
+		abs = cacheLen + offset
+	default:
+		return 0, errors.New("fs.RepeatableReader.Seek: invalid whence")
+	}
+	if abs < 0 {
+		return 0, errors.New("fs.RepeatableReader.Seek: negative position")
+	}
+	if abs > cacheLen {
+		return offset - (abs - cacheLen), errors.New("fs.RepeatableReader.Seek: offset is unavailable")
+	}
+	r.i = abs
+	return abs, nil
+}
+
+// Read data from original Reader into bytes
+// Data is either served from the underlying Reader or from cache if was already read
+func (r *RepeatableReader) Read(b []byte) (n int, err error) {
+	cacheLen := int64(len(r.b))
+	if r.i == cacheLen {
+		n, err = r.in.Read(b)
+		if n > 0 {
+			r.b = append(r.b, b[:n]...)
+		}
+	} else {
+		n = copy(b, r.b[r.i:])
+	}
+	r.i += int64(n)
+	return n, err
+}
+
+// NewRepeatableReader create new repeatable reader from Reader r
+func NewRepeatableReader(r io.Reader) *RepeatableReader {
+	return &RepeatableReader{in: r}
+}
+
+// NewRepeatableReaderSized create new repeatable reader from Reader r
+// with an initial buffer of size.
+func NewRepeatableReaderSized(r io.Reader, size int) *RepeatableReader {
+	return &RepeatableReader{
+		in: r,
+		b:  make([]byte, 0, size),
+	}
+}
+
+// NewRepeatableLimitReader create new repeatable reader from Reader r
+// with an initial buffer of size wrapped in a io.LimitReader to read
+// only size.
+func NewRepeatableLimitReader(r io.Reader, size int) *RepeatableReader {
+	return NewRepeatableReaderSized(io.LimitReader(r, int64(size)), size)
+}
+
+// NewRepeatableReaderBuffer create new repeatable reader from Reader r
+// using the buffer passed in.
+func NewRepeatableReaderBuffer(r io.Reader, buf []byte) *RepeatableReader {
+	return &RepeatableReader{
+		in: r,
+		b:  buf[:0],
+	}
+}
+
+// NewRepeatableLimitReaderBuffer create new repeatable reader from
+// Reader r and buf wrapped in a io.LimitReader to read only size.
+func NewRepeatableLimitReaderBuffer(r io.Reader, buf []byte, size int64) *RepeatableReader {
+	return NewRepeatableReaderBuffer(io.LimitReader(r, int64(size)), buf)
+}