def _bundle_metadata(self, metadata):
"""
Bundles the metadata into a tarfile formatted bundle
@param metadata: The metadata string to bundle
"""
#print >> sys.stderr, "Bundle meta!"
#print >> sys.stderr, metadata
metadata_file = None
try:
metadata_file = tempfile.TemporaryFile()
except IOError:
task_error("Can't create metadata file in working directory")
metadata_file.write(metadata)
metadata_file.seek(0)
if self.empty_tar:
tarball = tarfile.TarFile(name=self.bundle_path, mode='w')
self.empty_tar = False
else:
tarball = tarfile.TarFile(name=self.bundle_path, mode='a')
tar_info = tarfile.TarInfo("metadata.txt")
tar_info.size = len(metadata)
tar_info.mtime = time.time()
tarball.addfile(tar_info, metadata_file)
metadata_file.close()
tarball.close()
def _bundle_file(self, file_paths):
"""
A 'Pure Virtual' function that will perform file bundling in a child class
:Parameters:
file_name
The name of the file to bundle
file_arcname
An alternative name to use for the file inside of the bundle
"""
task_error("Can't bundle a file with the base class")
def __init__(self, bundle_path):
"""
Initializes a Tar_Bundler
:Parameters:
bundle_path
The path to the target bundle file
"""
if bundle_path == '' or bundle_path is None:
task_error('no bundle path')
# Initialize the Base Bundler Class
FileBundler.__init__(self, bundle_path)
tarball = tarfile.TarFile(name=self.bundle_path, mode='w')
tarball.close()
self.empty_tar = True
print >> sys.stderr, "Successfully created tarfile bundle %s" % self.bundle_path
def bundle_metadata(self, metadata):
"""
Bundles the metadata into a tarfile formatted bundle
@param metadata: The metadata string to bundle
"""
metadata_file = None
try:
metadata_file = tempfile.NamedTemporaryFile(delete=False)
except IOError:
task_error('Cannot create metadata file in working directory')
metadata_file.write(metadata)
fname = metadata_file.name
metadata_file.close()
metadata_file = open(fname, mode='rb')
# metadata_file.seek(0)
if self.empty_tar:
tarball = tarfile.TarFile(name=self.bundle_path, mode='w')
self.empty_tar = False
else:
tarball = tarfile.TarFile(name=self.bundle_path, mode='a')
try:
tar_info = tarfile.TarInfo('metadata.txt')
tar_info.size = len(metadata)
tar_info.mtime = time.time()
tarball.addfile(tar_info, metadata_file)
metadata_file.close()
tarball.close()
os.remove(fname)
except Exception, ex:
print ex
traceback.print_exc(file=sys.stdout)
raise ex
def bundle_file(self, file_paths, bundle_size=0):
"""
Bundles files into a tarfile formatted bundle
:Parameters:
file_name
The name of the file to bundle
file_arcname
An alternative name to use for the file inside of the bundle
"""
if self.empty_tar:
tarball = tarfile.TarFile(name=self.bundle_path, mode='w')
self.empty_tar = False
else:
tarball = tarfile.TarFile(name=self.bundle_path, mode='a')
self.running_size = 0
self.last_percent = 0
self.percent_complete = 0
self.bundle_size = bundle_size
try:
self.report_percent_complete()
for (file_path, file_arcname) in file_paths:
# hash the file and store in hash_dict
# percent complete is reported only as read and hashed
# hopefully that being the slowest part and all we have
# access to for completion statistics
self.hash_file(file_path, file_arcname)
# for version 1.2, push files to a data/ directory
#to avoid collisions with metadata.txt in the root
modified_arc_name = os.path.join('data', file_arcname)
tarball.add(file_path, arcname=modified_arc_name, recursive=False)
except BundlerError, err:
task_error("Failed to bundle file: %s" % (err.msg))
def __init__(self,
bundle_path,
proposal_ID='',
instrument_name='',
instrument_ID='',
groups=None):
"""
Initializes a Tar_Bundler
:Parameters:
bundle_path
The path to the target bundle file
proposal_ID
An optional string describing the proposal associated with this bundle
instrument_name
The name of the instrument that produced the data packaged in the bundle
"""
if bundle_path == '' or bundle_path == None:
bundle_path = 'bundle.tar'
# Initialize the Base Bundler Class
FileBundler.__init__(self, bundle_path,
proposal_id=proposal_ID,
instrument_name=instrument_name,
instrument_id=instrument_ID,
groups=groups)
try:
tarball = tarfile.TarFile(name=self.bundle_path, mode='w')
tarball.close()
except:
task_error("Couldn't create bundle tarball: %s" % self.bundle_path)
self.empty_tar = True
print >> sys.stderr, "Successfully created tarfile bundle %s" % self.bundle_path
def progress(_download_t, _download_d, upload_t, upload_d):
"""
gets the progress of the current pycurl upload
"""
if upload_t > 0:
try:
percent = 100.0 * float(upload_d) / float(upload_t)
if percent - TrackPercent.percent > 5:
meta_dict={'Status': "upload percent complete: " + str(int(percent))}
task_state("PROGRESS", meta_dict)
TrackPercent.percent = percent
except Exception, e:
raise task_error("Error during callback: " + e.message)
def bundle(bundle_name='', file_list=None, bundle_size=0, meta_list=None):
"""
Bundles a list of files into a single aggregated bundle file
:Parameters:
bundle_name
The target bundle file in which to aggregate the file list
file_list
The list of files to bundle
bundle_size
total size of the files to be bundled
meta_list
list of metadata items. File metadata will be added to this list
"""
# validate parameters
if bundle_name is None or bundle_name == '':
task_error("Missing bundle name")
if file_list is None or len(file_list) == 0:
task_error("Missing file list")
# Set up the bundle file
bundle_path = os.path.abspath(bundle_name)
# Set up the bundler object
bundler = None
bundler = TarBundler(bundle_path)
bundler.bundle_file(file_list, bundle_size, meta_list)
meta_str = json.dumps(meta_list)
bundler.bundle_metadata(meta_str)
TaskComm.set_state('PROGRESS', 'Bundling complete')
def upload_files(ingest_server='',
bundle_name='',
file_list=None,
bundle_size=0,
meta_list=None,
auth={},
verify=True,
tartar=False):
"""
task created on a separate Celery process to bundle and upload in the background
status and errors are pushed by celery to the main server through RabbitMQ
"""
# one big-ass exception handler for upload.
try:
target_dir = os.path.dirname(bundle_name)
if not os.path.isdir(target_dir):
task_error('Bundle directory does not exist')
return
TaskComm.set_state("PROGRESS", "Cleaning previous uploads")
# initial state pushed through celery
TaskComm.set_state("PROGRESS", "Starting Bundle/Upload Process")
bundle(bundle_name=bundle_name,
file_list=file_list,
meta_list=meta_list,
bundle_size=bundle_size)
TaskComm.set_state("PROGRESS", "Completed Bundling")
if tartar:
# create the file tuple list of 1 file
fname = os.path.basename(bundle_name)
file_tuples = []
file_tuples.append((bundle_name, fname))
bundle_size = os.path.getsize(bundle_name)
# dual extension indicates tartar
bundle_name += '.tar'
bundle(bundle_name=bundle_name,
file_list=file_tuples,
meta_list=meta_list,
bundle_size=bundle_size)
TaskComm.set_state(
"PROGRESS", "Starting Uploady: " + str(bundle_name) + ": " +
ingest_server + ": " + str(auth))
uploader = Uploader(bundle_name, ingest_server, auth, verify)
TaskComm.set_state("PROGRESS", "Uploader Initialized")
result = uploader.upload_bundle()
TaskComm.set_state("PROGRESS", "Finished Upload")
try:
status = json.loads(result)
except Exception, e:
print 'Upload Error'
print result
raise Exception('Upload error: ' + result)
print 'End Upload Error'
# check for a valid job id. Ingest error should return -99
job_id = status['job_id']
if job_id < 0:
task_error(err)
raise Exception('Upload error: ' + bundle_name)
TaskComm.set_state("PROGRESS", "Rename Tar File")
try:
rename_tar_file(target_dir, bundle_name, job_id)
except Exception, ex:
raise Exception(ex.message + ': ' + bundle_name + ': ' +
str(job_id))
rename_tar_file(target_dir, bundle_name, job_id)
except Exception, ex:
raise Exception(ex.message + ': ' + bundle_name + ': ' +
str(job_id))
print status
if TaskComm.USE_CELERY:
# set job ID here
print 'exit with deliberate error'
print result
raise StandardError(result)
else:
TaskComm.set_state("DONE", result)
except StandardError, error:
raise error
except Exception, ex:
print >> sys.stderr, "Exception in upload_files:"
print >> sys.stderr, '-' * 60
traceback.print_exc(file=sys.stderr)
print >> sys.stderr, '-' * 60
err = 'Task exception: upload_files :' + str(
ex.message) + ': ' + result + ': ' + traceback.format_exc()
task_error(err)
print err
raise Exception(err)
def hash_file(self, file_path, file_arcname):
"""
Bundle in a file or directory
:Parameters:
file_path
The path to the file to bundle
file_arcname
An alternative name to use for the file inside of the bundle
"""
file_path = os.path.abspath(file_path)
# If the file_arcname argument is None use the base file name as the
# arc name
if file_arcname is None:
file_arcname = os.path.basename(file_path)
if not os.path.exists(file_path):
task_error("%s doesn't exist" % file_path)
if not os.access(file_path, os.R_OK):
task_error("Can't read from %s" % file_path)
file_mode = os.stat(file_path)[stat.ST_MODE]
if not stat.S_ISDIR(file_mode) and not stat.S_ISREG(file_mode):
task_error("Unknown file type for %s" % file_path)
file_in = None
try:
# open to read binary. This is important.
file_in = open(file_path, 'rb')
except IOError:
task_error("Couldn't read from file: %s" % file_path)
# hash file 1Mb at a time
hashval = hashlib.sha1()
while True:
data = file_in.read(1024 * 1024)
if not data:
break
hashval.update(data)
# update file bundle status
self.running_size += len(data)
self.percent_complete = 100.0 * self.running_size / self.bundle_size
# only update significant progress
if self.percent_complete - self.last_percent > 1:
self.report_percent_complete()
self.last_percent = self.percent_complete
file_hash = hashval.hexdigest()
# print 'hash: ' + file_hash
file_in.close()
modified_name = os.path.join('data', file_arcname)
(file_dir, file_name) = os.path.split(modified_name)
# linuxfy the directory
file_dir = file_dir.replace('\\', '/')
info = {}
info['size'] = os.path.getsize(file_path)
mime_type = mimetypes.guess_type(file_path, strict=True)[0]
info[
'mimetype'] = mime_type if mime_type is not None else 'application/octet-stream'
info['name'] = file_name
info['mtime'] = DT.datetime.utcfromtimestamp(
int(os.path.getmtime(file_path))).isoformat()
info['ctime'] = DT.datetime.utcfromtimestamp(
int(os.path.getctime(file_path))).isoformat()
info['destinationTable'] = 'Files'
info['subdir'] = file_dir
info['hashsum'] = file_hash
info['hashtype'] = 'sha1'
# todo make sure errors bubble up without crashing
if file_arcname in self.file_meta:
print file_arcname
task_error(
"Different file with the same arcname is already in the bundle"
)
return
return info
def bundle(bundle_name='',
instrument_name='',
proposal='',
file_list=None,
groups=None,
bundle_size=0):
"""
Bundles a list of files into a single aggregated bundle file
:Parameters:
bundle_name
The target bundle file in which to aggregate the file list
instrument_name
The name of the instrument that produced the data files that will be bundled
groups
The a hash of type/name groups to attach to
tarfile
If true, tarfile format is used to bundle. Otherwise zipfile format is used
proposal
An optional proposal ID to attach to the bundle
file_list
The list of files to bundle
"""
# validate parameters
if bundle_name == None or bundle_name == '':
task_error("Missing bundle name")
if instrument_name == None or instrument_name == '':
task_error("Missing instrument name")
if proposal == None or proposal == '':
task_error("Missing proposal")
if file_list == None or len(file_list) == 0:
task_error("Missing file list")
if groups == None or groups == '':
task_error("Missing groups")
#print >> sys.stderr, "Start bundling %s" % bundle_name
# Set up the bundle file
bundle_path = os.path.abspath(bundle_name)
#print >> sys.stderr, "Bundle file set to %s" % bundle_path
# Set up the bundler object
bundler = None
# dfh note we are setting the instrument name and ID to the same thing,
# which is being
# sent in as the instrument name but is actually the instrument ID. Fix
# this.
bundler = TarBundler(bundle_path, proposal_ID=proposal,
instrument_name=instrument_name,
instrument_ID=instrument_name,
groups=groups)
bundler.bundle_file(file_list, bundle_size)
bundler.bundle_metadata()
#print >> sys.stderr, "Finished bundling"
task_state('PROGRESS', "Bundling complete")
def _bundle_metadata(self, metadata):
"""
A 'Pure Virtual' function that will perform metadata bundling in a child class
@param metadata: The metadata string to bundle
"""
task_error("Can't bundle metadata with the base class")
def hash_file(self, file_path, file_arcname):
"""
Bundle in a file or directory
:Parameters:
file_path
The path to the file to bundle
file_arcname
An alternative name to use for the file inside of the bundle
"""
file_path = os.path.abspath(file_path)
# If the file_arcname argument is None use the base file name as the
# arc name
if file_arcname == None:
file_arcname = os.path.basename(file_path)
if not os.path.exists(file_path):
task_error("%s doesn't exist" % file_path)
if not os.access(file_path, os.R_OK):
task_error("Can't read from %s" % file_path)
file_mode = os.stat(file_path)[stat.ST_MODE]
if not stat.S_ISDIR(file_mode) and not stat.S_ISREG(file_mode):
task_error("Unknown file type for %s" % file_path)
file_in = None
try:
# open to read binary. This is important.
file_in = open(file_path, 'rb')
except IOError:
task_error("Couldn't read from file: %s" % file_path)
# hash file 1Mb at a time
hashval = hashlib.sha1()
while True:
data = file_in.read(1024 * 1024)
if not data:
break
hashval.update(data)
# update file bundle status
self.running_size += len(data)
self.percent_complete = 100.0 * self.running_size / self.bundle_size
# only update significant progress
if self.percent_complete - self.last_percent > 1:
self.report_percent_complete()
self.last_percent = self.percent_complete
file_hash = hashval.hexdigest()
# print 'hash: ' + file_hash
file_in.close()
#todo make sure errors bubble up without crashing
if file_arcname in self.hash_dict:
if hash != self.hash_dict[file_arcname]:
print file_arcname
task_error("Different file with the same arcname is already in the bundle")
return
self.hash_dict[file_arcname] = file_hash
