// GenericFileSave saves a Generic File record to Pharos. If the Generic
// File's ID is zero, this performs a POST to create a new record.
// For non-zero IDs, this performs a PUT to update the existing record.
// Either way, the record must have an IntellectualObject ID. The response
// object will have a new copy of the GenericFile if the save was successful.
func (client *PharosClient) GenericFileSave(obj *models.GenericFile) *PharosResponse {
// Set up the response object
resp := NewPharosResponse(PharosGenericFile)
resp.files = make([]*models.GenericFile, 1)
// URL and method
relativeUrl := fmt.Sprintf("/api/%s/files/", client.apiVersion)
httpMethod := "POST"
if obj.Id > 0 {
// PUT URL looks like /api/v2/files/college.edu%2Fobject_name%2Ffile.xml
relativeUrl = fmt.Sprintf("%s%s", relativeUrl, escapeSlashes(obj.Identifier))
httpMethod = "PUT"
}
absoluteUrl := client.BuildUrl(relativeUrl)
// Prepare the JSON data
postData, err := obj.SerializeForPharos()
if err != nil {
resp.Error = err
}
// Run the request
client.DoRequest(resp, httpMethod, absoluteUrl, bytes.NewBuffer(postData))
if resp.Error != nil {
return resp
}
// Parse the JSON from the response body
gf := &models.GenericFile{}
resp.Error = json.Unmarshal(resp.data, gf)
if resp.Error == nil {
resp.files[0] = gf
}
return resp
}
// buildFile saves a data file from the tar archive to disk,
// then returns a struct with data we'll need to construct the
// GenericFile object in Fedora later.
func (reader *Reader) saveWithChecksums(gf *models.GenericFile) {
// Set up a MultiWriter to stream data ONCE to file,
// md5 and sha256. We don't want to process the stream
// three separate times.
err := os.MkdirAll(filepath.Dir(gf.IngestLocalPath), 0755)
if err != nil {
gf.IngestErrorMessage = err.Error()
return
}
outputWriter, err := os.OpenFile(gf.IngestLocalPath, os.O_CREATE|os.O_WRONLY, 0644)
if outputWriter != nil {
defer outputWriter.Close()
}
if err != nil {
gf.IngestErrorMessage = fmt.Sprintf("Error opening %s for writing: %v", gf.IngestLocalPath, err)
return
}
md5Hash := md5.New()
shaHash := sha256.New()
multiWriter := io.MultiWriter(md5Hash, shaHash, outputWriter)
io.Copy(multiWriter, reader.tarReader)
gf.IngestMd5 = fmt.Sprintf("%x", md5Hash.Sum(nil))
gf.IngestSha256 = fmt.Sprintf("%x", shaHash.Sum(nil))
gf.IngestSha256GeneratedAt = time.Now().UTC()
gf.FileFormat, _ = platform.GuessMimeType(gf.IngestLocalPath) // on err, defaults to application/binary
return
}
// Initializes the uploader object with connection data and metadata
// for this specific GenericFile.
func (storer *APTStorer) initUploader(ingestState *models.IngestState, gf *models.GenericFile, sendWhere string) *network.S3Upload {
var region string
var bucket string
if sendWhere == "s3" {
region = storer.Context.Config.APTrustS3Region
bucket = storer.Context.Config.PreservationBucket
} else if sendWhere == "glacier" {
region = storer.Context.Config.APTrustGlacierRegion
bucket = storer.Context.Config.ReplicationBucket
} else {
ingestState.IngestManifest.StoreResult.AddError("Cannot save %s to %s because "+
"storer doesn't know where %s is", gf.Identifier, sendWhere)
ingestState.IngestManifest.StoreResult.ErrorIsFatal = true
return nil
}
uploader := network.NewS3Upload(
region,
bucket,
gf.IngestUUID,
gf.FileFormat,
)
uploader.AddMetadata("institution", ingestState.IngestManifest.Object.Institution)
uploader.AddMetadata("bag", ingestState.IngestManifest.Object.Identifier)
uploader.AddMetadata("bagpath", gf.OriginalPath())
uploader.AddMetadata("md5", gf.IngestMd5)
uploader.AddMetadata("sha256", gf.IngestSha256)
return uploader
}
func TestInstitutionIdentifier(t *testing.T) {
genericFile := models.GenericFile{}
genericFile.Identifier = "uc.edu/cin.675812/data/object.properties"
instId, err := genericFile.InstitutionIdentifier()
if err != nil {
t.Errorf(err.Error())
return
}
assert.Equal(t, "uc.edu", instId)
}
func (storer *APTStorer) markFileAsStored(gf *models.GenericFile, sendWhere, storageUrl string) {
if sendWhere == "s3" {
gf.IngestStoredAt = time.Now().UTC()
gf.IngestStorageURL = storageUrl
gf.URI = storageUrl
events := gf.FindEventsByType(constants.EventIdentifierAssignment)
var event *models.PremisEvent
for i := range events {
existingEvent := events[i]
if strings.HasPrefix(existingEvent.OutcomeDetail, "http://") ||
strings.HasPrefix(existingEvent.OutcomeDetail, "https://") {
event = existingEvent
break
}
}
if event != nil {
event.DateTime = time.Now().UTC()
}
} else if sendWhere == "glacier" {
gf.IngestReplicatedAt = time.Now().UTC()
gf.IngestReplicationURL = storageUrl
events := gf.FindEventsByType(constants.EventReplication)
if events != nil && len(events) > 0 {
events[0].DateTime = time.Now().UTC()
}
}
}
// 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
}
func TestPreservationStorageFileName(t *testing.T) {
genericFile := models.GenericFile{}
genericFile.URI = ""
fileName, err := genericFile.PreservationStorageFileName()
if err == nil {
t.Errorf("PreservationStorageFileName() should have returned an error")
}
genericFile.URI = "https://s3.amazonaws.com/aptrust.test.preservation/a58a7c00-392f-11e4-916c-0800200c9a66"
fileName, err = genericFile.PreservationStorageFileName()
if err != nil {
t.Errorf("PreservationStorageFileName() returned an error: %v", err)
return
}
assert.Equal(t, "a58a7c00-392f-11e4-916c-0800200c9a66", fileName)
}
func TestOriginalPathWithBagName(t *testing.T) {
genericFile := models.GenericFile{}
genericFile.IntellectualObjectIdentifier = "uc.edu/cin.675812"
// Top-level custom tag file
genericFile.Identifier = "uc.edu/cin.675812/tagmanifest-sha256.txt"
origPath, err := genericFile.OriginalPathWithBagName()
require.Nil(t, err)
assert.Equal(t, "cin.675812/tagmanifest-sha256.txt", origPath)
// Payload file
genericFile.Identifier = "uc.edu/cin.675812/data/object.properties"
origPath, err = genericFile.OriginalPathWithBagName()
require.Nil(t, err)
assert.Equal(t, "cin.675812/data/object.properties", origPath)
// Nested custom tag file
genericFile.Identifier = "uc.edu/cin.675812/custom/tag/dir/special_info.xml"
origPath, err = genericFile.OriginalPathWithBagName()
require.Nil(t, err)
assert.Equal(t, "cin.675812/custom/tag/dir/special_info.xml", origPath)
}
func (storer *APTStorer) saveFile(ingestState *models.IngestState, gf *models.GenericFile) {
existingSha256, err := storer.getExistingSha256(gf.Identifier)
if err != nil {
storer.Context.MessageLog.Error(err.Error())
ingestState.IngestManifest.StoreResult.AddError(err.Error())
return
}
// Set this, for the record.
if existingSha256 != nil {
gf.IngestPreviousVersionExists = true
gf.Id = existingSha256.GenericFileId
uuid, err := storer.getUuidOfExistingFile(gf.Identifier)
if err != nil {
message := fmt.Sprintf("Cannot find existing UUID for %s: %v", gf.Identifier, err.Error())
ingestState.IngestManifest.StoreResult.AddError(message)
storer.Context.MessageLog.Error(message)
// Probably not fatal, but treat it as such for now,
// because we don't want leave orphan objects in S3,
// or have the GenericFile.URL not match the actual
// storage URL. This should only happen if a depositor
// deletes the existing version of a GenericFile while
// we are processing this ingest. The window for that
// to happen is usually between a few seconds and a few
// hours.
ingestState.IngestManifest.StoreResult.ErrorIsFatal = true
return
}
if uuid == "" {
message := fmt.Sprintf("Cannot find existing UUID for %s.", gf.Identifier)
ingestState.IngestManifest.StoreResult.AddError(message)
storer.Context.MessageLog.Error(message)
// Probably not fatal, but treat it as such for now.
// Same note as in previous if statement above.
ingestState.IngestManifest.StoreResult.ErrorIsFatal = true
return
} else {
// OK. Set the GenericFile's UUID to match the existing file's
// UUID, so that we overwrite the existing file, and so the
// GenericFile record in Pharos still has the correct URL.
message := fmt.Sprintf("Resetting UUID for '%s' to '%s' so we can overwrite "+
"the currently stored version of the file.",
gf.Identifier, uuid)
storer.Context.MessageLog.Info(message)
// TODO: Test this in integration post test.
gf.IngestUUID = uuid
}
if existingSha256.Digest != gf.IngestSha256 {
storer.Context.MessageLog.Info(
"GenericFile %s has same sha256. Does not need save.", gf.Identifier)
gf.IngestNeedsSave = false
}
}
// Now copy to storage only if the file has changed.
if gf.IngestNeedsSave {
storer.Context.MessageLog.Info("File %s needs save", gf.Identifier)
if gf.IngestStoredAt.IsZero() || gf.IngestStorageURL == "" {
storer.copyToLongTermStorage(ingestState, gf, "s3")
}
if gf.IngestReplicatedAt.IsZero() || gf.IngestReplicationURL == "" {
storer.copyToLongTermStorage(ingestState, gf, "glacier")
}
}
}
