archive/zip: new package for reading ZIP files

R=rsc
CC=golang-dev
https://golang.org/cl/2125042

10 files changed, 504 insertions, 0 deletions

diff --git a/src/pkg/Makefile b/src/pkg/Makefile
index 33194918b8..d7351c5993 100644
--- a/src/pkg/Makefile
+++ b/src/pkg/Makefile
@@ -15,6 +15,7 @@ all: install
 
 DIRS=\
 	archive/tar\
+	archive/zip\
 	asn1\
 	big\
 	bufio\
diff --git a/src/pkg/archive/zip/Makefile b/src/pkg/archive/zip/Makefile
new file mode 100644
index 0000000000..32a543133c
--- /dev/null
+++ b/src/pkg/archive/zip/Makefile
@@ -0,0 +1,12 @@
+# Copyright 2010 The Go Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style
+# license that can be found in the LICENSE file.
+
+include ../../../Make.inc
+
+TARG=archive/zip
+GOFILES=\
+	reader.go\
+	struct.go\
+
+include ../../../Make.pkg
diff --git a/src/pkg/archive/zip/reader.go b/src/pkg/archive/zip/reader.go
new file mode 100644
index 0000000000..579ba16029
--- /dev/null
+++ b/src/pkg/archive/zip/reader.go
@@ -0,0 +1,278 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+/*
+The zip package provides support for reading ZIP archives.
+
+See: http://www.pkware.com/documents/casestudies/APPNOTE.TXT
+
+This package does not support ZIP64 or disk spanning.
+*/
+package zip
+
+import (
+	"bufio"
+	"bytes"
+	"compress/flate"
+	"hash"
+	"hash/crc32"
+	"encoding/binary"
+	"io"
+	"os"
+)
+
+var (
+	FormatError       = os.NewError("not a valid zip file")
+	UnsupportedMethod = os.NewError("unsupported compression algorithm")
+	ChecksumError     = os.NewError("checksum error")
+)
+
+type Reader struct {
+	r       io.ReaderAt
+	File    []*File
+	Comment string
+}
+
+type File struct {
+	FileHeader
+	zipr         io.ReaderAt
+	zipsize      int64
+	headerOffset uint32
+	bodyOffset   int64
+}
+
+// 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)
+	if err != nil {
+		return nil, err
+	}
+	fi, err := f.Stat()
+	if err != nil {
+		return nil, err
+	}
+	return NewReader(f, fi.Size)
+}
+
+// NewReader returns a new Reader reading from r, which is assumed to
+// have the given size in bytes.
+func NewReader(r io.ReaderAt, size int64) (*Reader, os.Error) {
+	end, err := readDirectoryEnd(r, size)
+	if err != nil {
+		return nil, err
+	}
+	z := &Reader{
+		r:       r,
+		File:    make([]*File, end.directoryRecords),
+		Comment: end.comment,
+	}
+	rs := io.NewSectionReader(r, 0, size)
+	if _, err = rs.Seek(int64(end.directoryOffset), 0); err != nil {
+		return nil, err
+	}
+	buf := bufio.NewReader(rs)
+	for i := range z.File {
+		z.File[i] = &File{zipr: r, zipsize: size}
+		if err := readDirectoryHeader(z.File[i], buf); err != nil {
+			return nil, err
+		}
+	}
+	return z, nil
+}
+
+// Open returns a ReadCloser that provides access to the File's contents.
+func (f *File) Open() (rc io.ReadCloser, err os.Error) {
+	off := int64(f.headerOffset)
+	if f.bodyOffset == 0 {
+		r := io.NewSectionReader(f.zipr, off, f.zipsize-off)
+		if err = readFileHeader(f, r); err != nil {
+			return
+		}
+		if f.bodyOffset, err = r.Seek(0, 1); err != nil {
+			return
+		}
+	}
+	r := io.NewSectionReader(f.zipr, off+f.bodyOffset, int64(f.CompressedSize))
+	switch f.Method {
+	case 0: // store (no compression)
+		rc = nopCloser{r}
+	case 8: // DEFLATE
+		rc = flate.NewReader(r)
+	default:
+		err = UnsupportedMethod
+	}
+	if rc != nil {
+		rc = &checksumReader{rc, crc32.NewIEEE(), f.CRC32}
+	}
+	return
+}
+
+type checksumReader struct {
+	rc   io.ReadCloser
+	hash hash.Hash32
+	sum  uint32
+}
+
+func (r *checksumReader) Read(b []byte) (n int, err os.Error) {
+	n, err = r.rc.Read(b)
+	r.hash.Write(b[:n])
+	if err != os.EOF {
+		return
+	}
+	if r.hash.Sum32() != r.sum {
+		err = ChecksumError
+	}
+	return
+}
+
+func (r *checksumReader) Close() os.Error { return r.rc.Close() }
+
+type nopCloser struct {
+	io.Reader
+}
+
+func (f nopCloser) Close() os.Error { return nil }
+
+func readFileHeader(f *File, r io.Reader) (err os.Error) {
+	defer func() {
+		if rerr, ok := recover().(os.Error); ok {
+			err = rerr
+		}
+	}()
+	var (
+		signature      uint32
+		filenameLength uint16
+		extraLength    uint16
+	)
+	read(r, &signature)
+	if signature != fileHeaderSignature {
+		return FormatError
+	}
+	read(r, &f.ReaderVersion)
+	read(r, &f.Flags)
+	read(r, &f.Method)
+	read(r, &f.ModifiedTime)
+	read(r, &f.ModifiedDate)
+	read(r, &f.CRC32)
+	read(r, &f.CompressedSize)
+	read(r, &f.UncompressedSize)
+	read(r, &filenameLength)
+	read(r, &extraLength)
+	f.Name = string(readByteSlice(r, filenameLength))
+	f.Extra = readByteSlice(r, extraLength)
+	return
+}
+
+func readDirectoryHeader(f *File, r io.Reader) (err os.Error) {
+	defer func() {
+		if rerr, ok := recover().(os.Error); ok {
+			err = rerr
+		}
+	}()
+	var (
+		signature          uint32
+		filenameLength     uint16
+		extraLength        uint16
+		commentLength      uint16
+		startDiskNumber    uint16 // unused
+		internalAttributes uint16 // unused
+		externalAttributes uint32 // unused
+	)
+	read(r, &signature)
+	if signature != directoryHeaderSignature {
+		return FormatError
+	}
+	read(r, &f.CreatorVersion)
+	read(r, &f.ReaderVersion)
+	read(r, &f.Flags)
+	read(r, &f.Method)
+	read(r, &f.ModifiedTime)
+	read(r, &f.ModifiedDate)
+	read(r, &f.CRC32)
+	read(r, &f.CompressedSize)
+	read(r, &f.UncompressedSize)
+	read(r, &filenameLength)
+	read(r, &extraLength)
+	read(r, &commentLength)
+	read(r, &startDiskNumber)
+	read(r, &internalAttributes)
+	read(r, &externalAttributes)
+	read(r, &f.headerOffset)
+	f.Name = string(readByteSlice(r, filenameLength))
+	f.Extra = readByteSlice(r, extraLength)
+	f.Comment = string(readByteSlice(r, commentLength))
+	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
+	for i, bLen := range []int64{1024, 65 * 1024} {
+		if bLen > size {
+			bLen = size
+		}
+		b = make([]byte, int(bLen))
+		if _, err := r.ReadAt(b, size-bLen); err != nil && err != os.EOF {
+			return nil, err
+		}
+		if p := findSignatureInBlock(b); p >= 0 {
+			b = b[p:]
+			break
+		}
+		if i == 1 || bLen == size {
+			return nil, FormatError
+		}
+	}
+
+	// read header into struct
+	defer func() {
+		if rerr, ok := recover().(os.Error); ok {
+			err = rerr
+			d = nil
+		}
+	}()
+	br := bytes.NewBuffer(b[4:]) // skip over signature
+	d = new(directoryEnd)
+	read(br, &d.diskNbr)
+	read(br, &d.dirDiskNbr)
+	read(br, &d.dirRecordsThisDisk)
+	read(br, &d.directoryRecords)
+	read(br, &d.directorySize)
+	read(br, &d.directoryOffset)
+	read(br, &d.commentLen)
+	d.comment = string(readByteSlice(br, d.commentLen))
+	return d, nil
+}
+
+func findSignatureInBlock(b []byte) int {
+	const minSize = 4 + 2 + 2 + 2 + 2 + 4 + 4 + 2 // fixed part of header
+	for i := len(b) - minSize; i >= 0; i-- {
+		// defined from directoryEndSignature in struct.go
+		if b[i] == 'P' && b[i+1] == 'K' && b[i+2] == 0x05 && b[i+3] == 0x06 {
+			// n is length of comment
+			n := int(b[i+minSize-2]) | int(b[i+minSize-1])<<8
+			if n+minSize+i == len(b) {
+				return i
+			}
+		}
+	}
+	return -1
+}
+
+func read(r io.Reader, data interface{}) {
+	if err := binary.Read(r, binary.LittleEndian, data); err != nil {
+		panic(err)
+	}
+}
+
+func readByteSlice(r io.Reader, l uint16) []byte {
+	b := make([]byte, l)
+	if l == 0 {
+		return b
+	}
+	if _, err := io.ReadFull(r, b); err != nil {
+		panic(err)
+	}
+	return b
+}
diff --git a/src/pkg/archive/zip/reader_test.go b/src/pkg/archive/zip/reader_test.go
new file mode 100644
index 0000000000..36b925c4fc
--- /dev/null
+++ b/src/pkg/archive/zip/reader_test.go
@@ -0,0 +1,180 @@
+// Copyright 2010 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package zip
+
+import (
+	"bytes"
+	"encoding/binary"
+	"io"
+	"io/ioutil"
+	"os"
+	"testing"
+)
+
+type ZipTest struct {
+	Name    string
+	Comment string
+	File    []ZipTestFile
+	Error   os.Error // the error that Opening this file should return
+}
+
+type ZipTestFile struct {
+	Name    string
+	Content []byte // if blank, will attempt to compare against File
+	File    string // name of file to compare to (relative to testdata/)
+}
+
+var tests = []ZipTest{
+	ZipTest{
+		Name:    "test.zip",
+		Comment: "This is a zipfile comment.",
+		File: []ZipTestFile{
+			ZipTestFile{
+				Name:    "test.txt",
+				Content: []byte("This is a test text file.\n"),
+			},
+			ZipTestFile{
+				Name: "gophercolor16x16.png",
+				File: "gophercolor16x16.png",
+			},
+		},
+	},
+	ZipTest{
+		Name: "r.zip",
+		File: []ZipTestFile{
+			ZipTestFile{
+				Name: "r/r.zip",
+				File: "r.zip",
+			},
+		},
+	},
+	ZipTest{Name: "readme.zip"},
+	ZipTest{Name: "readme.notzip", Error: FormatError},
+}
+
+func TestReader(t *testing.T) {
+	for _, zt := range tests {
+		readTestZip(t, zt)
+	}
+}
+
+func readTestZip(t *testing.T, zt ZipTest) {
+	z, err := OpenReader("testdata/" + zt.Name)
+	if err != zt.Error {
+		t.Errorf("error=%v, want %v", err, zt.Error)
+		return
+	}
+
+	// bail here if no Files expected to be tested
+	// (there may actually be files in the zip, but we don't care)
+	if zt.File == nil {
+		return
+	}
+
+	if z.Comment != zt.Comment {
+		t.Errorf("%s: comment=%q, want %q", zt.Name, z.Comment, zt.Comment)
+	}
+	if len(z.File) != len(zt.File) {
+		t.Errorf("%s: file count=%d, want %d", zt.Name, len(z.File), len(zt.File))
+	}
+
+	// test read of each file
+	for i, ft := range zt.File {
+		readTestFile(t, ft, z.File[i])
+	}
+
+	// test simultaneous reads
+	n := 0
+	done := make(chan bool)
+	for i := 0; i < 5; i++ {
+		for j, ft := range zt.File {
+			go func() {
+				readTestFile(t, ft, z.File[j])
+				done <- true
+			}()
+			n++
+		}
+	}
+	for ; n > 0; n-- {
+		<-done
+	}
+
+	// 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", err, ChecksumError)
+	}
+}
+
+func readTestFile(t *testing.T, ft ZipTestFile, f *File) {
+	if f.Name != ft.Name {
+		t.Errorf("name=%q, want %q", f.Name, ft.Name)
+	}
+	var b bytes.Buffer
+	r, err := f.Open()
+	if err != nil {
+		t.Error(err)
+		return
+	}
+	_, err = io.Copy(&b, r)
+	if err != nil {
+		t.Error(err)
+		return
+	}
+	r.Close()
+	var c []byte
+	if len(ft.Content) != 0 {
+		c = ft.Content
+	} else if c, err = ioutil.ReadFile("testdata/" + ft.File); err != nil {
+		t.Error(err)
+		return
+	}
+	if b.Len() != len(c) {
+		t.Errorf("%s: len=%d, want %d", f.Name, b.Len(), len(c))
+		return
+	}
+	for i, b := range b.Bytes() {
+		if b != c[i] {
+			t.Errorf("%s: content[%d]=%q want %q", i, b, c[i])
+			return
+		}
+	}
+}
+
+func TestInvalidFiles(t *testing.T) {
+	const size = 1024 * 70 // 70kb
+	b := make([]byte, size)
+
+	// zeroes
+	_, err := NewReader(sliceReaderAt(b), size)
+	if err != FormatError {
+		t.Errorf("zeroes: error=%v, want %v", err, FormatError)
+	}
+
+	// repeated directoryEndSignatures
+	sig := make([]byte, 4)
+	binary.LittleEndian.PutUint32(sig, directoryEndSignature)
+	for i := 0; i < size-4; i += 4 {
+		copy(b[i:i+4], sig)
+	}
+	_, err = NewReader(sliceReaderAt(b), size)
+	if err != FormatError {
+		t.Errorf("sigs: error=%v, want %v", err, FormatError)
+	}
+}
+
+type sliceReaderAt []byte
+
+func (r sliceReaderAt) ReadAt(b []byte, off int64) (int, os.Error) {
+	copy(b, r[int(off):int(off)+len(b)])
+	return len(b), nil
+}
diff --git a/src/pkg/archive/zip/struct.go b/src/pkg/archive/zip/struct.go
new file mode 100644
index 0000000000..8a8c727d47
--- /dev/null
+++ b/src/pkg/archive/zip/struct.go
@@ -0,0 +1,33 @@
+package zip
+
+const (
+	fileHeaderSignature      = 0x04034b50
+	directoryHeaderSignature = 0x02014b50
+	directoryEndSignature    = 0x06054b50
+)
+
+type FileHeader struct {
+	Name             string
+	CreatorVersion   uint16
+	ReaderVersion    uint16
+	Flags            uint16
+	Method           uint16
+	ModifiedTime     uint16
+	ModifiedDate     uint16
+	CRC32            uint32
+	CompressedSize   uint32
+	UncompressedSize uint32
+	Extra            []byte
+	Comment          string
+}
+
+type directoryEnd struct {
+	diskNbr            uint16 // unused
+	dirDiskNbr         uint16 // unused
+	dirRecordsThisDisk uint16 // unused
+	directoryRecords   uint16
+	directorySize      uint32
+	directoryOffset    uint32 // relative to file
+	commentLen         uint16
+	comment            string
+}
diff --git a/src/pkg/archive/zip/testdata/gophercolor16x16.png b/src/pkg/archive/zip/testdata/gophercolor16x16.png
new file mode 100644
index 0000000000..48854ff3b7
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/gophercolor16x16.png
diff --git a/src/pkg/archive/zip/testdata/r.zip b/src/pkg/archive/zip/testdata/r.zip
new file mode 100644
index 0000000000..ea0fa2ffcc
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/r.zip
diff --git a/src/pkg/archive/zip/testdata/readme.notzip b/src/pkg/archive/zip/testdata/readme.notzip
new file mode 100644
index 0000000000..06668c4c1c
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/readme.notzip
diff --git a/src/pkg/archive/zip/testdata/readme.zip b/src/pkg/archive/zip/testdata/readme.zip
new file mode 100644
index 0000000000..db3bb900e4
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/readme.zip
diff --git a/src/pkg/archive/zip/testdata/test.zip b/src/pkg/archive/zip/testdata/test.zip
new file mode 100644
index 0000000000..03890c05d4
--- /dev/null
+++ b/src/pkg/archive/zip/testdata/test.zip