Compressed resource optimization

cmd/rwp/cmd/serve/api.go

@@ -17,7 +17,9 @@ import (
 	httprange "github.com/gotd/contrib/http_range"
 	"github.com/pkg/errors"
 	"github.com/readium/go-toolkit/cmd/rwp/cmd/serve/cache"
+	"github.com/readium/go-toolkit/pkg/archive"
 	"github.com/readium/go-toolkit/pkg/asset"
+	"github.com/readium/go-toolkit/pkg/fetcher"
 	"github.com/readium/go-toolkit/pkg/manifest"
 	"github.com/readium/go-toolkit/pkg/pub"
 	"github.com/readium/go-toolkit/pkg/streamer"
@@ -243,8 +245,17 @@ func (s *Server) getAsset(w http.ResponseWriter, r *http.Request) {
 		w.WriteHeader(http.StatusPartialContent)
 	}
 
-	// Stream the asset
-	_, rerr = res.Stream(w, start, end)
+	cres, ok := res.(fetcher.CompressedResource)
+	if ok && cres.CompressedAs(archive.CompressionMethodDeflate) && start == 0 && end == 0 && supportsDeflate(r) {
+		// Stream the asset in compressed format
+		w.Header().Set("content-encoding", "deflate")
+		w.Header().Set("content-length", strconv.FormatInt(cres.CompressedLength(), 10))
+		_, err = cres.StreamCompressed(w)
+	} else {
+		// Stream the asset
+		_, rerr = res.Stream(w, start, end)
+	}
+
 	if rerr != nil {
 		if errors.Is(err, syscall.EPIPE) || errors.Is(err, syscall.ECONNRESET) {
 			// Ignore client errors

cmd/rwp/cmd/serve/helpers.go

@@ -1,6 +1,7 @@
 package serve
 
 import (
+	"net/http"
 	"strings"
 
 	"github.com/readium/go-toolkit/pkg/manifest"
@@ -62,3 +63,28 @@ func conformsToAsMimetype(conformsTo manifest.Profiles) string {
 	}
 	return mime
 }
+
+func supportsDeflate(r *http.Request) bool {
+	vv := r.Header.Values("Accept-Encoding")
+	for _, v := range vv {
+		for _, sv := range strings.Split(v, ",") {
+			coding := parseCoding(sv)
+			if coding == "" {
+				continue
+			}
+			if coding == "deflate" {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func parseCoding(s string) (coding string) {
+	p := strings.IndexRune(s, ';')
+	if p == -1 {
+		p = len(s)
+	}
+	coding = strings.ToLower(strings.TrimSpace(s[:p]))
+	return
+}

pkg/archive/archive.go

@@ -55,9 +55,10 @@ type Entry interface {
 	Path() string                                              // Absolute path to the entry in the archive.
 	Length() uint64                                            // Uncompressed data length.
 	CompressedLength() uint64                                  // Compressed data length.
+	CompressedAs(compressionMethod CompressionMethod) bool     // Whether the entry is compressed using the given method.
 	Read(start int64, end int64) ([]byte, error)               // Reads the whole content of this entry, or a portion when [start] or [end] are specified.
 	Stream(w io.Writer, start int64, end int64) (int64, error) // Streams the whole content of this entry to a writer, or a portion when [start] or [end] are specified.
-	// Close()
+	StreamCompressed(w io.Writer) (int64, error)               // Streams the compressed content of this entry to a writer.
 }
 
 // Represents an immutable archive.

pkg/archive/archive_exploded.go

@@ -26,6 +26,10 @@ func (e explodedArchiveEntry) CompressedLength() uint64 {
 	return 0
 }
 
+func (e explodedArchiveEntry) CompressedAs(compressionMethod CompressionMethod) bool {
+	return false
+}
+
 func (e explodedArchiveEntry) Read(start int64, end int64) ([]byte, error) {
 	if end < start {
 		return nil, errors.New("range not satisfiable")
@@ -82,6 +86,10 @@ func (e explodedArchiveEntry) Stream(w io.Writer, start int64, end int64) (int64
 	return n, nil
 }
 
+func (e explodedArchiveEntry) StreamCompressed(w io.Writer) (int64, error) {
+	return -1, errors.New("entry is not compressed")
+}
+
 // An archive exploded on the file system as a directory.
 type explodedArchive struct {
 	directory string // Directory, already cleaned!

pkg/archive/archive_zip.go

@@ -31,6 +31,13 @@ func (e gozipArchiveEntry) CompressedLength() uint64 {
 	return e.file.CompressedSize64
 }
 
+func (e gozipArchiveEntry) CompressedAs(compressionMethod CompressionMethod) bool {
+	if compressionMethod != CompressionMethodDeflate {
+		return false
+	}
+	return e.file.Method == zip.Deflate
+}
+
 // This is a special mode to minimize the number of reads from the underlying reader.
 // It's especially useful when trying to stream the ZIP from a remote file, e.g.
 // cloud storage. It's only enabled when trying to read the entire file and compression
@@ -145,6 +152,18 @@ func (e gozipArchiveEntry) Stream(w io.Writer, start int64, end int64) (int64, e
 	return n, nil
 }
 
+func (e gozipArchiveEntry) StreamCompressed(w io.Writer) (int64, error) {
+	if e.file.Method != zip.Deflate {
+		return -1, errors.New("not a compressed resource")
+	}
+	f, err := e.file.OpenRaw()
+	if err != nil {
+		return -1, err
+	}
+
+	return io.Copy(w, f)
+}
+
 // An archive from a zip file using go's stdlib
 type gozipArchive struct {
 	zip           *zip.Reader

pkg/archive/compression.go

@@ -0,0 +1,10 @@
+package archive
+
+import "archive/zip"
+
+type CompressionMethod uint16
+
+const (
+	CompressionMethodStore   CompressionMethod = CompressionMethod(zip.Store)
+	CompressionMethodDeflate CompressionMethod = CompressionMethod(zip.Deflate)
+)

pkg/fetcher/fetcher_archive.go

@@ -152,6 +152,25 @@ func (r *entryResource) Stream(w io.Writer, start int64, end int64) (int64, *Res
 	return -1, Other(err)
 }
 
+// CompressedAs implements CompressedResource
+func (r *entryResource) CompressedAs(compressionMethod archive.CompressionMethod) bool {
+	return r.entry.CompressedAs(compressionMethod)
+}
+
+// CompressedLength implements CompressedResource
+func (r *entryResource) CompressedLength() int64 {
+	return int64(r.entry.CompressedLength())
+}
+
+// StreamCompressed implements CompressedResource
+func (r *entryResource) StreamCompressed(w io.Writer) (int64, *ResourceError) {
+	i, err := r.entry.StreamCompressed(w)
+	if err == nil {
+		return i, nil
+	}
+	return -1, Other(err)
+}
+
 // Length implements Resource
 func (r *entryResource) Length() (int64, *ResourceError) {
 	return int64(r.entry.Length()), nil

pkg/fetcher/resource.go

@@ -3,12 +3,14 @@ package fetcher
 import (
 	"encoding/json"
 	"encoding/xml"
+	"errors"
 	"fmt"
 	"io"
 	"net/http"
 	"os"
 	"strings"
 
+	"github.com/readium/go-toolkit/pkg/archive"
 	"github.com/readium/go-toolkit/pkg/manifest"
 	"github.com/readium/xmlquery"
 	"golang.org/x/text/encoding/unicode"
@@ -365,6 +367,33 @@ func (r ProxyResource) ReadAsXML(prefixes map[string]string) (*xmlquery.Node, *R
 	return r.Res.ReadAsXML(prefixes)
 }
 
+// CompressedAs implements CompressedResource
+func (r ProxyResource) CompressedAs(compressionMethod archive.CompressionMethod) bool {
+	cres, ok := r.Res.(CompressedResource)
+	if !ok {
+		return false
+	}
+	return cres.CompressedAs(compressionMethod)
+}
+
+// CompressedLength implements CompressedResource
+func (r ProxyResource) CompressedLength() int64 {
+	cres, ok := r.Res.(CompressedResource)
+	if !ok {
+		return -1
+	}
+	return cres.CompressedLength()
+}
+
+// StreamCompressed implements CompressedResource
+func (r ProxyResource) StreamCompressed(w io.Writer) (int64, *ResourceError) {
+	cres, ok := r.Res.(CompressedResource)
+	if !ok {
+		return -1, Other(errors.New("resource is not compressed"))
+	}
+	return cres.StreamCompressed(w)
+}
+
 /**
  * Transforms the bytes of [resource] on-the-fly.
  *

pkg/fetcher/traits.go

@@ -0,0 +1,13 @@
+package fetcher
+
+import (
+	"io"
+
+	"github.com/readium/go-toolkit/pkg/archive"
+)
+
+type CompressedResource interface {
+	CompressedAs(compressionMethod archive.CompressionMethod) bool
+	CompressedLength() int64
+	StreamCompressed(w io.Writer) (int64, *ResourceError)
+}

pkg/parser/epub/deobfuscator.go

@@ -7,6 +7,7 @@ import (
 	"strings"
 
 	"github.com/pkg/errors"
+	"github.com/readium/go-toolkit/pkg/archive"
 	"github.com/readium/go-toolkit/pkg/fetcher"
 )
 
@@ -32,15 +33,23 @@ type DeobfuscatingResource struct {
 	identifier string
 }
 
-func (d DeobfuscatingResource) Read(start, end int64) ([]byte, *fetcher.ResourceError) {
+func (d DeobfuscatingResource) obfuscation() (string, int64) {
 	algorithm := ""
 	penc := d.Res.Link().Properties.Encryption()
 	if penc != nil {
 		algorithm = penc.Algorithm
 	}
 
 	v, ok := algorithm2length[algorithm]
-	if ok {
+	if !ok {
+		return algorithm, 0
+	}
+	return algorithm, v
+}
+
+func (d DeobfuscatingResource) Read(start, end int64) ([]byte, *fetcher.ResourceError) {
+	algorithm, v := d.obfuscation()
+	if v > 0 {
 		data, err := d.ProxyResource.Read(start, end)
 		if err != nil {
 			return nil, err
@@ -62,14 +71,8 @@ func (d DeobfuscatingResource) Read(start, end int64) ([]byte, *fetcher.Resource
 }
 
 func (d DeobfuscatingResource) Stream(w io.Writer, start int64, end int64) (int64, *fetcher.ResourceError) {
-	algorithm := ""
-	penc := d.Res.Link().Properties.Encryption()
-	if penc != nil {
-		algorithm = penc.Algorithm
-	}
-
-	v, ok := algorithm2length[algorithm]
-	if ok {
+	algorithm, v := d.obfuscation()
+	if v > 0 {
 		if start >= v {
 			// We're past the obfuscated part, just proxy it
 			return d.ProxyResource.Stream(w, start, end)
@@ -141,6 +144,36 @@ func (d DeobfuscatingResource) Stream(w io.Writer, start int64, end int64) (int6
 	return d.ProxyResource.Stream(w, start, end)
 }
 
+// CompressedAs implements CompressedResource
+func (d DeobfuscatingResource) CompressedAs(compressionMethod archive.CompressionMethod) bool {
+	_, v := d.obfuscation()
+	if v > 0 {
+		return false
+	}
+
+	return d.ProxyResource.CompressedAs(compressionMethod)
+}
+
+// CompressedLength implements CompressedResource
+func (d DeobfuscatingResource) CompressedLength() int64 {
+	_, v := d.obfuscation()
+	if v > 0 {
+		return -1
+	}
+
+	return d.ProxyResource.CompressedLength()
+}
+
+// StreamCompressed implements CompressedResource
+func (d DeobfuscatingResource) StreamCompressed(w io.Writer) (int64, *fetcher.ResourceError) {
+	_, v := d.obfuscation()
+	if v > 0 {
+		return 0, fetcher.Other(errors.New("cannot stream compressed resource when obfuscated"))
+	}
+
+	return d.ProxyResource.StreamCompressed(w)
+}
+
 func (d DeobfuscatingResource) getHashKeyAdobe() []byte {
 	hexbytes, _ := hex.DecodeString(
 		strings.Replace(