func TestTFIFind(t *testing.T) {
_, filename, _, _ := runtime.Caller(0)
tarFilePath, _ := filepath.Abs(path.Join(filepath.Dir(filename),
"..", "..", "testdata", "unit_test_bags", "example.edu.tagsample_good.tar"))
tfi, err := fileutil.NewTarFileIterator(tarFilePath)
if tfi != nil {
defer tfi.Close()
}
assert.NotNil(t, tfi)
require.Nil(t, err)
readCloser, err := tfi.Find("example.edu.tagsample_good/junk_file.txt")
assert.Nil(t, err)
assert.NotNil(t, readCloser)
readCloser.Close()
readCloser, err = tfi.Find("example.edu.tagsample_good/tagmanifest-sha256.txt")
assert.Nil(t, err)
assert.NotNil(t, readCloser)
readCloser.Close()
readCloser, err = tfi.Find("this-file-does-not-exist")
assert.NotNil(t, err)
assert.Nil(t, readCloser)
}
func TestNewTarFileIterator(t *testing.T) {
_, filename, _, _ := runtime.Caller(0)
tarFilePath, _ := filepath.Abs(path.Join(filepath.Dir(filename),
"..", "..", "testdata", "unit_test_bags", "example.edu.tagsample_good.tar"))
tfi, err := fileutil.NewTarFileIterator(tarFilePath)
assert.NotNil(t, tfi)
assert.Nil(t, err)
}
// Read() reads the bag and returns an IntellectualObject and a WorkSummary.
// The WorkSummary will include a list of errors, if there were any.
// The list of files contained in IntellectualObject.GenericFiles will include
// ALL files found in the bag, even some we may not want to save, such as
// those beginning with dots and dashes. If you don't want to preserve those
// files you can delete them from the IntellectualObject manually later.
func (vbag *VirtualBag) Read() (*IntellectualObject, *WorkSummary) {
vbag.summary = NewWorkSummary()
vbag.summary.Start()
vbag.obj = NewIntellectualObject()
vbag.obj.Identifier = util.CleanBagName(path.Base(vbag.pathToBag))
if strings.HasSuffix(vbag.pathToBag, ".tar") {
vbag.obj.IngestTarFilePath = vbag.pathToBag
} else {
vbag.obj.IngestUntarredPath = vbag.pathToBag
}
// Compile a list of the bag's contents (GenericFiles),
// and calculate checksums for everything in the bag.
var err error
if vbag.obj.IngestTarFilePath != "" {
vbag.readIterator, err = fileutil.NewTarFileIterator(vbag.obj.IngestTarFilePath)
} else {
vbag.readIterator, err = fileutil.NewFileSystemIterator(vbag.obj.IngestUntarredPath)
}
if err != nil {
vbag.summary.AddError("Could not read bag: %v", err)
vbag.summary.Finish()
return vbag.obj, vbag.summary
} else {
vbag.addGenericFiles()
}
vbag.obj.IngestTopLevelDirNames = vbag.readIterator.GetTopLevelDirNames()
// Golang's tar file reader is forward-only, so we need to
// open a new iterator to read through a handful of tag files,
// manifests and tag manifests.
vbag.readIterator = nil
if vbag.obj.IngestTarFilePath != "" {
vbag.readIterator, err = fileutil.NewTarFileIterator(vbag.obj.IngestTarFilePath)
} else {
vbag.readIterator, err = fileutil.NewFileSystemIterator(vbag.obj.IngestUntarredPath)
}
if err != nil {
vbag.summary.AddError("Could not read bag: %v", err)
} else {
vbag.parseManifestsTagFilesAndMimeTypes()
}
vbag.summary.Finish()
return vbag.obj, vbag.summary
}
// Should be able to close repeatedly without panic.
func TestTarFileIteratorClose(t *testing.T) {
_, filename, _, _ := runtime.Caller(0)
tarFilePath, _ := filepath.Abs(path.Join(filepath.Dir(filename),
"..", "..", "testdata", "unit_test_bags", "example.edu.tagsample_good.tar"))
tfi, _ := fileutil.NewTarFileIterator(tarFilePath)
if tfi == nil {
assert.Fail(t, "Could not get TarFileIterator")
}
assert.NotPanics(t, tfi.Close, "TarFileIterator.Close() freaked out")
assert.NotPanics(t, tfi.Close, "TarFileIterator.Close() freaked out")
}
func TestTFINext(t *testing.T) {
_, filename, _, _ := runtime.Caller(0)
tarFilePath, _ := filepath.Abs(path.Join(filepath.Dir(filename),
"..", "..", "testdata", "unit_test_bags", "example.edu.tagsample_good.tar"))
tfi, err := fileutil.NewTarFileIterator(tarFilePath)
if tfi != nil {
defer tfi.Close()
}
assert.NotNil(t, tfi)
assert.Nil(t, err)
for {
reader, fileSummary, err := tfi.Next()
if err == io.EOF {
break
}
if reader == nil {
assert.Fail(t, "Reader is nil")
}
if fileSummary == nil {
assert.Fail(t, "FileSummary is nil")
}
if fileSummary.IsDir {
continue
}
assert.NotEmpty(t, fileSummary.RelPath)
assert.False(t, strings.HasPrefix(fileSummary.RelPath, string(os.PathSeparator)))
// On Windows, where separator is '\', tar files may still use '/'
assert.False(t, strings.HasPrefix(fileSummary.RelPath, "/"))
assert.Empty(t, fileSummary.AbsPath)
assert.NotNil(t, fileSummary.Mode)
if fileSummary.IsRegularFile {
assert.True(t, fileSummary.Size > int64(0))
}
assert.False(t, fileSummary.ModTime.IsZero())
buf := make([]byte, 1024)
_, err = reader.Read(buf)
if err != nil {
assert.Equal(t, io.EOF, err)
}
}
}
func TestTFIGetTopLevelDirNames(t *testing.T) {
_, filename, _, _ := runtime.Caller(0)
tarFilePath, _ := filepath.Abs(path.Join(filepath.Dir(filename),
"..", "..", "testdata", "unit_test_bags", "example.edu.tagsample_good.tar"))
tfi, _ := fileutil.NewTarFileIterator(tarFilePath)
if tfi == nil {
assert.Fail(t, "Could not get TarFileIterator")
}
// Read the entire tar file, so we know the reader
// has looked at all directories.
for {
_, _, err := tfi.Next()
if err != nil {
break
}
}
topLevelDirs := tfi.GetTopLevelDirNames()
require.NotEmpty(t, topLevelDirs)
assert.Equal(t, 1, len(topLevelDirs))
assert.Equal(t, "example.edu.tagsample_good", topLevelDirs[0])
}
// calculateTagManifestDigest calculates the sha256 digest of the bag's
// tagmanifest-sha256.txt file.
func (copier *DPNCopier) calculateTagManifestDigest(manifest *models.ReplicationManifest) {
tarFileIterator, err := fileutil.NewTarFileIterator(manifest.LocalPath)
if err != nil {
manifest.CopySummary.AddError("Can't get TarFileIterator for %s: %v",
manifest.LocalPath, err.Error)
return
}
// DPN BagIt spec says that the top-level dir inside the bag should
// have the same name as the bag itself (a UUID).
// https://wiki.duraspace.org/display/DPN/BagIt+Specification#BagItSpecification-DPNBagitStructure
tagManifestPath := filepath.Join(manifest.ReplicationTransfer.Bag, "tagmanifest-sha256.txt")
readCloser, err := tarFileIterator.Find(tagManifestPath)
if readCloser != nil {
defer readCloser.Close()
}
if err != nil {
manifest.CopySummary.AddError("Can't get tagmanifest from bag: %v",
err.Error)
return
}
nonce := ""
if manifest.ReplicationTransfer.FixityNonce != nil && *manifest.ReplicationTransfer.FixityNonce != "" {
nonce = *manifest.ReplicationTransfer.FixityNonce
copier.Context.MessageLog.Info("FixityNonce for replication %s is %s",
manifest.ReplicationTransfer.ReplicationId, nonce)
} else {
copier.Context.MessageLog.Info("No FixityNonce for replication %s",
manifest.ReplicationTransfer.ReplicationId)
}
digest, err := copier.calculateSha256(readCloser, nonce)
if err != nil {
manifest.CopySummary.AddError("Error calculating tagmanifest digest: %v",
err.Error)
return
}
manifest.ReplicationTransfer.FixityValue = digest
copier.Context.MessageLog.Info("Xfer %s has digest %s",
manifest.ReplicationTransfer.ReplicationId,
*manifest.ReplicationTransfer.FixityValue)
}
// Returns a reader that can read the file from within the tar archive.
// The S3 uploader uses this reader to stream data to S3 and Glacier.
func (storer *APTStorer) getReadCloser(ingestState *models.IngestState, gf *models.GenericFile) (*fileutil.TarFileIterator, io.ReadCloser) {
tarFilePath := ingestState.IngestManifest.Object.IngestTarFilePath
tfi, err := fileutil.NewTarFileIterator(tarFilePath)
if err != nil {
msg := fmt.Sprintf("Can't get TarFileIterator for %s: %v", tarFilePath, err)
ingestState.IngestManifest.StoreResult.AddError(msg)
return nil, nil
}
origPathWithBagName, err := gf.OriginalPathWithBagName()
if err != nil {
ingestState.IngestManifest.StoreResult.AddError(err.Error())
return nil, nil
}
readCloser, err := tfi.Find(origPathWithBagName)
if err != nil {
msg := fmt.Sprintf("Can't get reader for %s: %v", gf.Identifier, err)
ingestState.IngestManifest.StoreResult.AddError(msg)
if readCloser != nil {
readCloser.Close()
}
return nil, nil
}
return tfi, readCloser
}
