// We should parse the tags in the specified files.
// We should not parse other tag files
func TestVirtualBagTagFileOptions(t *testing.T) {
tarFilePath := testhelper.VbagGetPath("example.edu.tagsample_good.tar")
files := []string{}
vbag := models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, _ := vbag.Read()
assert.Equal(t, 0, len(obj.IngestTags))
files = []string{"bagit.txt"}
vbag = models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, _ = vbag.Read()
assert.Equal(t, 2, len(obj.IngestTags))
files = []string{"bagit.txt", "bag-info.txt"}
vbag = models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, _ = vbag.Read()
assert.Equal(t, 8, len(obj.IngestTags))
files = []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag = models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, _ = vbag.Read()
assert.Equal(t, 10, len(obj.IngestTags))
}
func TestVirtualBagRead_ChecksumOptions(t *testing.T) {
tempDir, bagPath, err := testhelper.UntarTestBag("example.edu.tagsample_good.tar")
if err != nil {
assert.Fail(t, err.Error())
}
if tempDir != "" {
defer os.RemoveAll(tempDir)
}
files := []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag := models.NewVirtualBag(bagPath, files, true, false)
assert.NotNil(t, vbag)
obj, _ := vbag.Read()
// Should calculate md5 only
for _, gf := range obj.GenericFiles {
assert.NotEmpty(t, gf.IngestMd5)
assert.Empty(t, gf.IngestSha256)
}
vbag = models.NewVirtualBag(bagPath, files, false, true)
assert.NotNil(t, vbag)
obj, _ = vbag.Read()
// Should calculate sha256 only
for _, gf := range obj.GenericFiles {
assert.Empty(t, gf.IngestMd5)
assert.NotEmpty(t, gf.IngestSha256)
}
}
// With md5 manifest only, sha256 only, and both
func TestVirtualBagRead_ManifestOptions(t *testing.T) {
tempDir, bagPath, err := testhelper.UntarTestBag("example.edu.tagsample_good.tar")
if err != nil {
assert.Fail(t, err.Error())
}
if tempDir != "" {
defer os.RemoveAll(tempDir)
}
// Delete the md5 manifest
os.Remove(filepath.Join(bagPath, "manifest-md5.txt"))
files := []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag := models.NewVirtualBag(bagPath, files, true, true)
assert.NotNil(t, vbag)
obj, _ := vbag.Read()
// Should have manifest values for sha256
for _, gf := range obj.GenericFiles {
if gf.IngestFileType == constants.PAYLOAD_FILE {
assert.Empty(t, gf.IngestManifestMd5)
assert.NotEmpty(t, gf.IngestManifestSha256)
}
}
tempDir, bagPath, err = testhelper.UntarTestBag("example.edu.tagsample_good.tar")
if err != nil {
assert.Fail(t, err.Error())
}
if tempDir != "" {
defer os.RemoveAll(tempDir)
}
// Delete the sha256 manifest
os.Remove(filepath.Join(bagPath, "manifest-sha256.txt"))
files = []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag = models.NewVirtualBag(bagPath, files, true, true)
assert.NotNil(t, vbag)
obj, _ = vbag.Read()
// Should have manifest values for sha256
for _, gf := range obj.GenericFiles {
if gf.IngestFileType == constants.PAYLOAD_FILE {
assert.NotEmpty(t, gf.IngestManifestMd5)
assert.Empty(t, gf.IngestManifestSha256)
}
}
}
// NewBagValidator creates a new BagValidator. Param pathToBag
// should be an absolute path to either the tarred bag (.tar file)
// or to the untarred bag (a directory). Param bagValidationConfig
// defines what we need to validate, in addition to the checksums in the
// manifests.
func NewBagValidator(pathToBag string, bagValidationConfig *BagValidationConfig) (*BagValidator, error) {
if !fileutil.FileExists(pathToBag) {
return nil, fmt.Errorf("Bag does not exist at %s", pathToBag)
}
if bagValidationConfig == nil {
return nil, fmt.Errorf("Param bagValidationConfig cannot be nil")
}
configErrors := bagValidationConfig.ValidateConfig()
if len(configErrors) > 0 {
errString := "BagValidationConfig has the following errors:"
for _, e := range configErrors {
errString += fmt.Sprintf("\n%s", e.Error())
}
return nil, fmt.Errorf(errString)
}
err := bagValidationConfig.CompileFileNameRegex()
if err != nil {
return nil, fmt.Errorf("Error in BagValidationConfig: %v", err)
}
calculateMd5 := util.StringListContains(bagValidationConfig.FixityAlgorithms, constants.AlgMd5)
calculateSha256 := util.StringListContains(bagValidationConfig.FixityAlgorithms, constants.AlgSha256)
tagFilesToParse := make([]string, 0)
for pathToFile, filespec := range bagValidationConfig.FileSpecs {
if filespec.ParseAsTagFile {
tagFilesToParse = append(tagFilesToParse, pathToFile)
}
}
bagValidator := &BagValidator{
PathToBag: pathToBag,
BagValidationConfig: bagValidationConfig,
virtualBag: models.NewVirtualBag(pathToBag, tagFilesToParse, calculateMd5, calculateSha256),
}
return bagValidator, nil
}
func TestVirtualBagRead_FromTarFile(t *testing.T) {
tarFilePath := testhelper.VbagGetPath("example.edu.tagsample_good.tar")
files := []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag := models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, summary := vbag.Read()
runAssertions(t, obj, summary, "TestVirtualBagRead_FromTarFile")
}
func TestVirtualBagRead_FromDirectory(t *testing.T) {
tempDir, bagPath, err := testhelper.UntarTestBag("example.edu.tagsample_good.tar")
if err != nil {
assert.Fail(t, err.Error())
}
if tempDir != "" {
defer os.RemoveAll(tempDir)
}
files := []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag := models.NewVirtualBag(bagPath, files, true, true)
assert.NotNil(t, vbag)
obj, summary := vbag.Read()
runAssertions(t, obj, summary, "TestVirtualBagRead_FromDirectory")
}
func TestVirtualBagReadReportsMissingFiles(t *testing.T) {
// This bad bag has a number of problems.
// Here, we're specifically testing to see if a missing file is reported.
tarFilePath := testhelper.VbagGetPath("example.edu.tagsample_bad.tar")
files := []string{"bagit.txt", "bag-info.txt", "aptrust-info.txt"}
vbag := models.NewVirtualBag(tarFilePath, files, true, true)
assert.NotNil(t, vbag)
obj, summary := vbag.Read()
// custom_tags/tag_file_xyz.pdf appears twice in missing files
// list because it's mentioned in manifest-md5 and manifest-sha256
assert.True(t, summary.HasErrors())
require.Equal(t, 3, len(obj.IngestMissingFiles))
assert.Equal(t, "data/file-not-in-bag", obj.IngestMissingFiles[0].FilePath)
assert.Equal(t, "custom_tags/tag_file_xyz.pdf", obj.IngestMissingFiles[1].FilePath)
assert.Equal(t, "custom_tags/tag_file_xyz.pdf", obj.IngestMissingFiles[2].FilePath)
}
func TestNewVirtualBag(t *testing.T) {
tarFilePath := testhelper.VbagGetPath("example.edu.tagsample_good.tar")
vbag := models.NewVirtualBag(tarFilePath, nil, false, false)
assert.NotNil(t, vbag)
}