def main():
#pdb.set_trace()
args = Args()
args.parse_args()
RQ = RedisQueue('DI_ReadyQueue')
PDSinfoDICT = json.load(open(pds_info, 'r'))
try:
archiveID = PDSinfoDICT[args.archive]['archiveid']
except KeyError:
print("\nArchive '{}' not found in {}\n".format(
args.archive, pds_info))
print("The following archives are available:")
for k in PDSinfoDICT.keys():
print("\t{}".format(k))
exit()
# ********* Set up logging *************
logger = logging.getLogger('DI_Queueing.' + args.archive)
logger.setLevel(logging.INFO)
logFileHandle = logging.FileHandler(pds_log + 'DI.log')
#logFileHandle = logging.FileHandler('/usgs/cdev/PDS/logs/DI.log')
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s, %(message)s')
logFileHandle.setFormatter(formatter)
logger.addHandler(logFileHandle)
logger.info('Starting %s DI Queueing', args.archive)
if args.volume:
logger.info('Queueing %s Volume', args.volume)
try:
# @TODO switch back to prd
session, _ = db_connect(pds_db)
logger.info('DataBase Connecton: Success')
except:
logger.error('DataBase Connection: Error')
# ************* date stuff ***************
td = (datetime.datetime.now(pytz.utc) -
datetime.timedelta(days=30)).strftime("%Y-%m-%d %H:%M:%S")
testing_date = datetime.datetime.strptime(str(td), "%Y-%m-%d %H:%M:%S")
if args.volume:
volstr = '%' + args.volume + '%'
testcount = session.query(Files).filter(
Files.archiveid == archiveID, Files.filename.like(volstr)).filter(
or_(
cast(Files.di_date, Date) < testing_date,
cast(Files.di_date, Date) == None)).count()
# logger.info('Query Count %s', testcount)
testQ = session.query(Files).filter(
Files.archiveid == archiveID, Files.filename.like(volstr)).filter(
or_(
cast(Files.di_date, Date) < testing_date,
cast(Files.di_date, Date) == None))
else:
testcount = session.query(Files).filter(
Files.archiveid == archiveID).filter(
or_(
cast(Files.di_date, Date) < testing_date,
cast(Files.di_date, Date) == None)).count()
testQ = session.query(Files).filter(
Files.archiveid == archiveID).filter(
or_(
cast(Files.di_date, Date) < testing_date,
cast(Files.di_date, Date) == None))
addcount = 0
for element in testQ:
try:
RQ.QueueAdd((element.filename, args.archive))
addcount = addcount + 1
except:
logger.error('File %s Not Added to DI_ReadyQueue',
element.filename)
logger.info('Files Added to Queue %s', addcount)
logger.info('DI Queueing Complete')
def main():
args = Args()
args.parse_args()
override = args.override
# ********* Set up logging *************
logger = logging.getLogger('Ingest_Process')
logger.setLevel(logging.INFO)
logFileHandle = logging.FileHandler(pds_log + 'Ingest.log')
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s, %(message)s')
logFileHandle.setFormatter(formatter)
logger.addHandler(logFileHandle)
logger.info("Starting Process")
PDSinfoDICT = json.load(open(pds_info, 'r'))
RQ_main = RedisQueue('Ingest_ReadyQueue')
RQ_work = RedisQueue('Ingest_WorkQueue')
RQ_upc = RedisQueue('UPC_ReadyQueue')
RQ_thumb = RedisQueue('Thumbnail_ReadyQueue')
RQ_browse = RedisQueue('Browse_ReadyQueue')
#RQ_pilotB = RedisQueue('PilotB_ReadyQueue')
try:
session, _ = db_connect(pds_db)
logger.info('DataBase Connecton: Success')
except:
logger.error('DataBase Connection: Error')
index = 1
while int(RQ_main.QueueSize()) > 0:
item = literal_eval(RQ_main.QueueGet().decode("utf-8"))
inputfile = item[0]
archive = item[1]
RQ_work.QueueAdd(inputfile)
subfile = inputfile.replace(PDSinfoDICT[archive]['path'], '')
# Calculate checksum in chunks of 4096
f_hash = hashlib.md5()
with open(inputfile, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
f_hash.update(chunk)
filechecksum = f_hash.hexdigest()
QOBJ = session.query(Files).filter_by(filename=subfile).first()
runflag = False
if QOBJ is None:
runflag = True
elif filechecksum != QOBJ.checksum:
runflag = True
if runflag == True or override == True:
date = datetime.datetime.now(
pytz.utc).strftime("%Y-%m-%d %H:%M:%S")
fileURL = inputfile.replace(archive_base, web_base)
# If all upc requirements are in 'inputfile,' flag for upc
upcflag = all(x in inputfile for x in PDSinfoDICT[archive]['upc_reqs'])
filesize = os.path.getsize(inputfile)
try:
# If we found an existing file and want to overwrite the data
if QOBJ is not None and override == True:
testIN = QOBJ
# If the file was not found, create a new entry
else:
testIN = Files()
testIN.archiveid = PDSinfoDICT[archive]['archiveid']
testIN.filename = subfile
testIN.entry_date = date
testIN.checksum = filechecksum
testIN.upc_required = upcflag
testIN.validation_required = True
testIN.header_only = False
testIN.release_date = date
testIN.file_url = fileURL
testIN.file_size = filesize
testIN.di_pass = True
testIN.di_date = date
session.merge(testIN)
session.flush()
if upcflag == True:
RQ_upc.QueueAdd((inputfile, testIN.fileid, archive))
RQ_thumb.QueueAdd((inputfile, testIN.fileid, archive))
RQ_browse.QueueAdd((inputfile, testIN.fileid, archive))
#RQ_pilotB.QueueAdd((inputfile, testIN.fileid, archive))
RQ_work.QueueRemove(inputfile)
index = index + 1
except Exception as e:
print(e)
logger.error("Error During File Insert %s", subfile)
elif runflag == False:
RQ_work.QueueRemove(inputfile)
if index >= 250:
try:
session.commit()
logger.info("Commit 250 files to Database: Success")
index = 1
except:
session.rollback()
logger.error("Something Went Wrong During DB Insert")
else:
logger.info("No Files Found in Ingest Queue")
try:
session.commit()
logger.info("Commit to Database: Success")
except:
session.rollback()
if RQ_main.QueueSize() == 0 and RQ_work.QueueSize() == 0:
logger.info("Process Complete All Queues Empty")
elif RQ_main.QueueSize() == 0 and RQ_work.QueueSize() != 0:
logger.warning("Process Done Work Queue NOT Empty Contains %s Files", str(
RQ_work.QueueSize()))
logger.info("Ingest Complete")
def main():
# pdb.set_trace()
Key = sys.argv[-1]
workarea = '/scratch/pds_services/' + Key + '/'
RQ_file = RedisQueue(Key + '_FileQueue')
RQ_work = RedisQueue(Key + '_WorkQueue')
RQ_zip = RedisQueue(Key + '_ZIP')
RQ_loggy = RedisQueue(Key + '_loggy')
RQ_final = RedisQueue('FinalQueue')
RHash = RedisHash(Key + '_info')
RHerror = RedisHash(Key + '_error')
if int(RQ_file.QueueSize()) == 0:
print "No Files Found in Redis Queue"
else:
jobFile = RQ_file.Qfile2Qwork(RQ_file.getQueueName(),
RQ_work.getQueueName())
#******************** Setup system logging **********************
basename = os.path.splitext(os.path.basename(jobFile))[0]
logger = logging.getLogger(Key + '.' + basename)
logger.setLevel(logging.INFO)
logFileHandle = logging.FileHandler('/usgs/cdev/PDS/logs/Service.log')
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s, %(message)s')
logFileHandle.setFormatter(formatter)
logger.addHandler(logFileHandle)
logger.info('Starting MAP Processing')
loggyOBJ = Loggy(basename)
# *************** File Naming ***************
infile = workarea + \
os.path.splitext(os.path.basename(jobFile))[0] + '.input.cub'
outfile = workarea + \
os.path.splitext(os.path.basename(jobFile))[0] + '.output.cub'
# *********** Recipe Stuff ********************
RQ_recipe = RedisQueue(Key + '_recipe')
status = 'success'
for element in RQ_recipe.RecipeGet():
if status == 'error':
break
elif status == 'success':
processOBJ = Process()
process = processOBJ.JSON2Process(element)
if 'gdal_translate' not in processOBJ.getProcessName():
if 'cubeatt-band' in processOBJ.getProcessName():
if '+' in jobFile:
# bandSplit = jobFile.split('+')
# infileB = infile + '+' + bandSplit[1]
processOBJ.updateParameter('from_', jobFile)
processOBJ.updateParameter('to', outfile)
processOBJ.ChangeProcess('cubeatt')
else:
continue
elif 'map2map' in processOBJ.getProcessName():
if '+' in jobFile:
processOBJ.updateParameter('from_', infile)
else:
processOBJ.updateParameter('from_', jobFile)
processOBJ.updateParameter('to', outfile)
elif 'cubeatt-bit' in processOBJ.getProcessName():
if RHash.OutBit() == 'unsignedbyte':
temp_outfile = outfile + '+lsb+tile+attached+unsignedbyte+1:254'
elif RHash.OutBit() == 'signedword':
temp_outfile = outfile + '+lsb+tile+attached+signedword+-32765:32765'
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', temp_outfile)
processOBJ.ChangeProcess('cubeatt')
elif 'isis2pds' in processOBJ.getProcessName():
# finalfile = infile.replace('.input.cub', '_final.img')
finalfile = workarea + RHash.getMAPname() + '.img'
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', finalfile)
else:
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', outfile)
print processOBJ.getProcess()
for k, v in processOBJ.getProcess().items():
func = getattr(isis, k)
subloggyOBJ = SubLoggy(k)
try:
func(**v)
logger.info('Process %s :: Success', k)
subloggyOBJ.setStatus('SUCCESS')
subloggyOBJ.setCommand(processOBJ.LogCommandline())
subloggyOBJ.setHelpLink(processOBJ.LogHelpLink())
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
if os.path.isfile(outfile):
os.rename(outfile, infile)
status = 'success'
except ProcessError as e:
logger.error('Process %s :: Error', k)
logger.error(e)
status = 'error'
eSTR = 'Error Executing ' + k + \
' Standard Error: ' + str(e)
RHerror.addError(
os.path.splitext(os.path.basename(jobFile))[0],
eSTR)
subloggyOBJ.setStatus('ERROR')
subloggyOBJ.setCommand(processOBJ.LogCommandline())
subloggyOBJ.setHelpLink(processOBJ.LogHelpLink())
subloggyOBJ.errorOut(eSTR)
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
else:
GDALcmd = ""
for process, v, in processOBJ.getProcess().items():
subloggyOBJ = SubLoggy(process)
GDALcmd += process
for key, value in v.items():
GDALcmd += ' ' + key + ' ' + value
if RHash.Format() == 'GeoTiff-BigTiff':
fileext = 'tif'
elif RHash.Format() == 'GeoJPEG-2000':
fileext = 'jp2'
elif RHash.Format() == 'JPEG':
fileext = 'jpg'
elif RHash.Format() == 'PNG':
fileext = 'png'
elif RHash.Format() == 'GIF':
fileext = 'gif'
logGDALcmd = GDALcmd + ' ' + basename + '.input.cub ' + RHash.getMAPname() + \
'.' + fileext
finalfile = workarea + RHash.getMAPname() + '.' + fileext
# finalfile = infile.replace('.input.cub', '_final.' + fileext)
GDALcmd += ' ' + infile + ' ' + finalfile
print GDALcmd
try:
result = subprocess.call(GDALcmd, shell=True)
logger.info('Process GDAL translate :: Success')
status = 'success'
subloggyOBJ.setStatus('SUCCESS')
subloggyOBJ.setCommand(logGDALcmd)
subloggyOBJ.setHelpLink(
'www.gdal.org/gdal_translate.html')
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
os.remove(infile)
except OSError, e:
logger.error('Process GDAL translate :: Error')
logger.error(e.stderr)
status = 'error'
RHerror.addError(
os.path.splitext(os.path.basename(jobFile))[0],
'Process GDAL translate :: Error')
subloggyOBJ.setStatus('ERROR')
subloggyOBJ.setCommand(logGDALcmd)
subloggyOBJ.setHelpLink(
'http://www.gdal.org/gdal_translate.html')
subloggyOBJ.errorOut(e.stderr)
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
if status == 'success':
if RHash.Format() == 'ISIS3':
finalfile = workarea + RHash.getMAPname() + '.cub'
# finalfile = infile.replace('.input.cub', '_final.cub')
shutil.move(infile, finalfile)
if RHash.getStatus() != 'ERROR':
RHash.Status('SUCCESS')
try:
RQ_zip.QueueAdd(finalfile)
logger.info('File Added to ZIP Queue')
except:
logger.error('File NOT Added to ZIP Queue')
try:
RQ_loggy.QueueAdd(loggyOBJ.Loggy2json())
logger.info('JSON Added to Loggy Queue')
except:
logger.error('JSON NOT Added to Loggy Queue')
RQ_work.QueueRemove(jobFile)
elif status == 'error':
RHash.Status('ERROR')
if os.path.isfile(infile):
os.remove(infile)
if RQ_file.QueueSize() == 0 and RQ_work.QueueSize() == 0:
try:
RQ_final.QueueAdd(Key)
logger.info('Key %s Added to Final Queue: Success', Key)
logger.info('Job Complete')
except:
logger.error('Key NOT Added to Final Queue')
else:
logger.warning('Queues Not Empty: filequeue = %s work queue = %s',
str(RQ_file.QueueSize()), str(RQ_work.QueueSize()))
def main():
# pdb.set_trace()
args = Args()
args.parse_args()
RQ = RedisQueue('ChecksumUpdate_Queue')
archiveDICT = {
'cassiniISS': '/pds_san/PDS_Archive/Cassini/ISS',
'mroCTX': '/pds_san/PDS_Archive/Mars_Reconnaissance_Orbiter/CTX',
'mroHIRISE': '/pds_san/PDS_Archive/Mars_Reconnaissance_Orbiter/HiRISE',
'LROLRC_EDR':
'/pds_san/PDS_Archive/Lunar_Reconnaissance_Orbiter/LROC/EDR/'
}
archiveID = {
'cassiniISS': 'cassini_iss_edr',
'mroCTX': 16,
'mroHIRISE_EDR': '124',
'LROLRC_EDR': 74
}
# ********* Set up logging *************
logger = logging.getLogger('ChecksumUpdate_Queueing.' + args.archive)
logger.setLevel(logging.INFO)
logFileHandle = logging.FileHandler('/usgs/cdev/PDS/logs/DI.log')
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s, %(message)s')
logFileHandle.setFormatter(formatter)
logger.addHandler(logFileHandle)
logger.info('Starting %s Checksum update Queueing', args.archive)
if args.volume:
logger.info('Queueing %s Volume', args.volume)
try:
# Throws away engine information
session, _ = db_connect(pds_db)
logger.info('DataBase Connecton: Success')
except:
logger.error('DataBase Connection: Error')
return 1
if args.volume:
volstr = '%' + args.volume + '%'
QueryOBJ = session.query(Files).filter(
Files.archiveid == archiveID[args.archive],
Files.filename.like(volstr))
else:
QueryOBJ = session.query(Files).filter(
Files.archiveid == archiveID[args.archive])
addcount = 0
for element in QueryOBJ:
try:
RQ.QueueAdd(element.filename)
addcount = addcount + 1
except:
logger.error('File %s Not Added to DI_ReadyQueue',
element.filename)
logger.info('Files Added to Queue %s', addcount)
logger.info('DI Queueing Complete')