archive/zip: handle files with data descriptors

Fixes #1471.

R=rsc
CC=golang-dev
https://golang.org/cl/4183048

4 files changed, 57 insertions, 14 deletions

diff --git a/src/pkg/archive/zip/reader.go b/src/pkg/archive/zip/reader.go
index 579ba16029..d8d9bba60b 100644
--- a/src/pkg/archive/zip/reader.go
+++ b/src/pkg/archive/zip/reader.go
@@ -42,6 +42,10 @@ type File struct {
 	bodyOffset   int64
 }
 
+func (f *File) hasDataDescriptor() bool {
+	return f.Flags&0x8 != 0
+}
+
 // OpenReader will open the Zip file specified by name and return a Reader.
 func OpenReader(name string) (*Reader, os.Error) {
 	f, err := os.Open(name, os.O_RDONLY, 0644)
@@ -93,7 +97,16 @@ func (f *File) Open() (rc io.ReadCloser, err os.Error) {
 			return
 		}
 	}
-	r := io.NewSectionReader(f.zipr, off+f.bodyOffset, int64(f.CompressedSize))
+	size := int64(f.CompressedSize)
+	if f.hasDataDescriptor() {
+		if size == 0 {
+			// permit SectionReader to see the rest of the file
+			size = f.zipsize - (off + f.bodyOffset)
+		} else {
+			size += dataDescriptorLen
+		}
+	}
+	r := io.NewSectionReader(f.zipr, off+f.bodyOffset, size)
 	switch f.Method {
 	case 0: // store (no compression)
 		rc = nopCloser{r}
@@ -103,7 +116,7 @@ func (f *File) Open() (rc io.ReadCloser, err os.Error) {
 		err = UnsupportedMethod
 	}
 	if rc != nil {
-		rc = &checksumReader{rc, crc32.NewIEEE(), f.CRC32}
+		rc = &checksumReader{rc, crc32.NewIEEE(), f, r}
 	}
 	return
 }
@@ -111,7 +124,8 @@ func (f *File) Open() (rc io.ReadCloser, err os.Error) {
 type checksumReader struct {
 	rc   io.ReadCloser
 	hash hash.Hash32
-	sum  uint32
+	f    *File
+	zipr io.Reader // for reading the data descriptor
 }
 
 func (r *checksumReader) Read(b []byte) (n int, err os.Error) {
@@ -120,7 +134,12 @@ func (r *checksumReader) Read(b []byte) (n int, err os.Error) {
 	if err != os.EOF {
 		return
 	}
-	if r.hash.Sum32() != r.sum {
+	if r.f.hasDataDescriptor() {
+		if err = readDataDescriptor(r.zipr, r.f); err != nil {
+			return
+		}
+	}
+	if r.hash.Sum32() != r.f.CRC32 {
 		err = ChecksumError
 	}
 	return
@@ -205,6 +224,18 @@ func readDirectoryHeader(f *File, r io.Reader) (err os.Error) {
 	return
 }
 
+func readDataDescriptor(r io.Reader, f *File) (err os.Error) {
+	defer func() {
+		if rerr, ok := recover().(os.Error); ok {
+			err = rerr
+		}
+	}()
+	read(r, &f.CRC32)
+	read(r, &f.CompressedSize)
+	read(r, &f.UncompressedSize)
+	return
+}
+
 func readDirectoryEnd(r io.ReaderAt, size int64) (d *directoryEnd, err os.Error) {
 	// look for directoryEndSignature in the last 1k, then in the last 65k
 	var b []byte
diff --git a/src/pkg/archive/zip/reader_test.go b/src/pkg/archive/zip/reader_test.go
index 3c24f1467c..72e8cccfd4 100644
--- a/src/pkg/archive/zip/reader_test.go
+++ b/src/pkg/archive/zip/reader_test.go
@@ -52,6 +52,15 @@ var tests = []ZipTest{
 	},
 	{Name: "readme.zip"},
 	{Name: "readme.notzip", Error: FormatError},
+	{
+		Name: "dd.zip",
+		File: []ZipTestFile{
+			{
+				Name:    "filename",
+				Content: []byte("This is a test textfile.\n"),
+			},
+		},
+	},
 }
 
 func TestReader(t *testing.T) {
@@ -102,16 +111,18 @@ func readTestZip(t *testing.T, zt ZipTest) {
 	}
 
 	// test invalid checksum
-	z.File[0].CRC32++ // invalidate
-	r, err := z.File[0].Open()
-	if err != nil {
-		t.Error(err)
-		return
-	}
-	var b bytes.Buffer
-	_, err = io.Copy(&b, r)
-	if err != ChecksumError {
-		t.Errorf("%s: copy error=%v, want %v", z.File[0].Name, err, ChecksumError)
+	if !z.File[0].hasDataDescriptor() { // skip test when crc32 in dd
+		z.File[0].CRC32++ // invalidate
+		r, err := z.File[0].Open()
+		if err != nil {
+			t.Error(err)
+			return
+		}
+		var b bytes.Buffer
+		_, err = io.Copy(&b, r)
+		if err != ChecksumError {
+			t.Errorf("%s: copy error=%v, want %v", z.File[0].Name, err, ChecksumError)
+		}
 	}
 }
 
diff --git a/src/pkg/archive/zip/struct.go b/src/pkg/archive/zip/struct.go
index 8a8c727d47..bfe0aae2e9 100644
--- a/src/pkg/archive/zip/struct.go
+++ b/src/pkg/archive/zip/struct.go
@@ -4,6 +4,7 @@ const (
 	fileHeaderSignature      = 0x04034b50
 	directoryHeaderSignature = 0x02014b50
 	directoryEndSignature    = 0x06054b50
+	dataDescriptorLen        = 12
 )
 
 type FileHeader struct {
diff --git a/src/pkg/archive/zip/testdata/dd.zip b/src/pkg/archive/zip/testdata/dd.zip
new file mode 100644
index 0000000000..e53378b0b0
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/dd.zip