def main():
args = Args()
args.parse_args()
override = args.override
logger = logging.getLogger('Ingest_Process')
level = logging.getLevelName(args.log_level)
logger.setLevel(level)
logFileHandle = logging.FileHandler(pds_log + 'Ingest.log')
print("Log File: {}Ingest.log".format(pds_log))
formatter = logging.Formatter(
'%(asctime)s - %(name)s - %(levelname)s, %(message)s')
logFileHandle.setFormatter(formatter)
logger.addHandler(logFileHandle)
logger.info("Starting Ingest Process")
PDSinfoDICT = json.load(open(pds_info, 'r'))
RQ_main = RedisQueue('Ingest_ReadyQueue')
RQ_lock = RedisLock(lock_obj)
RQ_lock.add({RQ_main.id_name: '1'})
RQ_work = RedisQueue('Ingest_WorkQueue')
RQ_upc = RedisQueue('UPC_ReadyQueue')
RQ_thumb = RedisQueue('Thumbnail_ReadyQueue')
RQ_browse = RedisQueue('Browse_ReadyQueue')
logger.info("UPC Queue: %s", RQ_upc.id_name)
logger.info("Thumbnail Queue: %s", RQ_thumb.id_name)
logger.info("Browse Queue: %s", RQ_browse.id_name)
try:
session, engine = db_connect(pds_db)
logger.info('DataBase Connecton: Success')
except:
logger.error('DataBase Connection: Error')
return 1
index = 1
while int(RQ_main.QueueSize()) > 0 and RQ_lock.available(RQ_main.id_name):
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 or filechecksum != QOBJ.checksum:
runflag = True
if runflag or override:
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:
ingest_entry = QOBJ
# If the file was not found, create a new entry
else:
ingest_entry = Files()
ingest_entry.archiveid = PDSinfoDICT[archive]['archiveid']
ingest_entry.filename = subfile
ingest_entry.entry_date = date
ingest_entry.checksum = filechecksum
ingest_entry.upc_required = upcflag
ingest_entry.validation_required = True
ingest_entry.header_only = False
ingest_entry.release_date = date
ingest_entry.file_url = fileURL
ingest_entry.file_size = filesize
ingest_entry.di_pass = True
ingest_entry.di_date = date
session.merge(ingest_entry)
session.flush()
if upcflag:
RQ_upc.QueueAdd((inputfile, ingest_entry.fileid, archive))
RQ_thumb.QueueAdd(
(inputfile, ingest_entry.fileid, archive))
RQ_browse.QueueAdd(
(inputfile, ingest_entry.fileid, archive))
#RQ_pilotB.QueueAdd((inputfile,ingest_entry.fileid, archive))
RQ_work.QueueRemove(inputfile)
index = index + 1
except Exception as e:
logger.error("Error During File Insert %s : %s", str(subfile),
str(e))
elif not runflag and not override:
RQ_work.QueueRemove(inputfile)
logger.warn(
"Not running ingest: file %s already present"
" in database and no override flag supplied", inputfile)
if index >= 250:
try:
session.commit()
logger.info("Commit 250 files to Database: Success")
index = 1
except Exception as e:
session.rollback()
logger.warn("Unable to commit to database: %s", str(e))
else:
logger.info("No Files Found in Ingest Queue")
try:
session.commit()
logger.info("Commit to Database: Success")
except Exception as e:
logger.error("Unable to commit to database: %s", str(e))
session.rollback()
# Close connection to database
session.close()
engine.dispose()
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:
print RQ_file.getQueueName()
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 POW Processing')
# set up loggy
loggyOBJ = Loggy(basename)
# File Naming
if '+' in jobFile:
bandSplit = jobFile.split('+')
inputFile = bandSplit[0]
else:
inputFile = jobFile
infile = workarea + \
os.path.splitext(os.path.basename(jobFile))[0] + '.input.cub'
outfile = workarea + \
os.path.splitext(os.path.basename(jobFile))[0] + '.output.cub'
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():
print processOBJ.getProcessName()
if '2isis' in processOBJ.getProcessName():
processOBJ.updateParameter('from_', inputFile)
processOBJ.updateParameter('to', outfile)
elif 'cubeatt-band' in processOBJ.getProcessName():
if '+' in jobFile:
infileB = infile + '+' + bandSplit[1]
processOBJ.updateParameter('from_', infileB)
processOBJ.updateParameter('to', outfile)
processOBJ.ChangeProcess('cubeatt')
else:
continue
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 'spice' in processOBJ.getProcessName():
processOBJ.updateParameter('from_', infile)
elif 'ctxevenodd' in processOBJ.getProcessName():
label = pvl.load(infile)
SS = label['IsisCube']['Instrument']['SpatialSumming']
print SS
if SS != 1:
continue
else:
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', outfile)
elif 'mocevenodd' in processOBJ.getProcessName():
label = pvl.load(infile)
CTS = label['IsisCube']['Instrument']['CrosstrackSumming']
print CTS
if CTS != 1:
continue
else:
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', outfile)
elif 'mocnoise50' in processOBJ.getProcessName():
label = pvl.load(infile)
CTS = label['IsisCube']['Instrument']['CrosstrackSumming']
if CTS != 1:
continue
else:
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', outfile)
elif 'cam2map' in processOBJ.getProcessName():
processOBJ.updateParameter('from_', infile)
processOBJ.updateParameter('to', outfile)
if RHash.getGRtype() == 'smart' or RHash.getGRtype() == 'fill':
subloggyOBJ = SubLoggy('cam2map')
camrangeOUT = workarea + basename + '_camrange.txt'
isis.camrange(from_=infile,
to=camrangeOUT)
cam = pvl.load(camrangeOUT)
if cam['UniversalGroundRange']['MaximumLatitude'] < float(RHash.getMinLat()) or \
cam['UniversalGroundRange']['MinimumLatitude'] > float(RHash.getMaxLat()) or \
cam['UniversalGroundRange']['MaximumLongitude'] < float(RHash.getMinLon()) or \
cam['UniversalGroundRange']['MinimumLongitude'] > float(RHash.getMaxLon()):
status = 'error'
eSTR = "Error Ground Range Outside Extent Range"
RHerror.addError(os.path.splitext(
os.path.basename(jobFile))[0], eSTR)
subloggyOBJ.setStatus('ERROR')
subloggyOBJ.errorOut(eSTR)
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
break
elif RHash.getGRtype() == 'smart':
if cam['UniversalGroundRange']['MinimumLatitude'] > float(RHash.getMinLat()):
minlat = cam['UniversalGroundRange']['MinimumLatitude']
else:
minlat = RHash.getMinLat()
if cam['UniversalGroundRange']['MaximumLatitude'] < float(RHash.getMaxLat()):
maxlat = cam['UniversalGroundRange']['MaximumLatitude']
else:
maxlat = RHash.getMaxLat()
if cam['UniversalGroundRange']['MinimumLongitude'] > float(RHash.getMinLon()):
minlon = cam['UniversalGroundRange']['MinimumLongitude']
else:
minlon = RHash.getMinLon()
if cam['UniversalGroundRange']['MaximumLongitude'] < float(RHash.getMaxLon()):
maxlon = cam['UniversalGroundRange']['MaximumLongitude']
else:
maxlon = RHash.getMaxLon()
elif RHash.getGRtype() == 'fill':
minlat = RHash.getMinLat()
maxlat = RHash.getMaxLat()
minlon = RHash.getMinLon()
maxlon = RHash.getMaxLon()
processOBJ.AddParameter('minlat', minlat)
processOBJ.AddParameter('maxlat', maxlat)
processOBJ.AddParameter('minlon', minlon)
processOBJ.AddParameter('maxlon', maxlon)
os.remove(camrangeOUT)
elif 'isis2pds' in processOBJ.getProcessName():
finalfile = infile.replace('.input.cub', '_final.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 ' + basename + '_final.' + fileext
finalfile = infile.replace(
'.input.cub', '_final.' + fileext)
GDALcmd += ' ' + infile + ' ' + finalfile
print GDALcmd
result = subprocess.call(GDALcmd, shell=True)
if result == 0:
logger.info('Process GDAL translate :: Success')
status = 'success'
subloggyOBJ.setStatus('SUCCESS')
subloggyOBJ.setCommand(logGDALcmd)
subloggyOBJ.setHelpLink(
'http://www.gdal.org/gdal_translate.html')
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
os.remove(infile)
else:
errmsg = 'Error Executing GDAL translate: Error'
logger.error(errmsg)
status = 'error'
RHerror.addError(os.path.splitext(
os.path.basename(jobFile))[0], errmsg)
subloggyOBJ.setStatus('ERROR')
subloggyOBJ.setCommand(logGDALcmd)
subloggyOBJ.setHelpLink(
'http://www.gdal.org/gdal_translate.html')
subloggyOBJ.errorOut('Process GDAL translate :: Error')
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
if status == 'success':
if RHash.Format() == 'ISIS3':
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')
elif status == 'error':
RHash.Status('ERROR')
if os.path.isfile(infile):
os.remove(infile)
try:
RQ_loggy.QueueAdd(loggyOBJ.Loggy2json())
RQ_work.QueueRemove(jobFile)
logger.info('JSON Added to Loggy Queue')
except:
logger.error('JSON NOT Added to Loggy Queue')
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('Both Queues Empty: filequeue = %s work queue = %s', str(
RQ_file.QueueSize()), str(RQ_work.QueueSize()))
logger.info('JOB Complete')
except:
logger.error('Key NOT Added to Final Queue')
elif RQ_file.QueueSize() == 0 and RQ_work.QueueSize() != 0:
logger.warning('Work Queue Not Empty: filequeue = %s work queue = %s', str(
RQ_file.QueueSize()), str(RQ_work.QueueSize()))
def main():
args = Args()
args.parse_args()
key = args.key
namespace = args.namespace
if namespace is None:
namespace is default_namespace
workarea = scratch + args.key + '/'
RQ_file = RedisQueue(key + '_FileQueue', namespace)
RQ_work = RedisQueue(key + '_WorkQueue', namespace)
RQ_zip = RedisQueue(key + '_ZIP', namespace)
RQ_loggy = RedisQueue(key + '_loggy', namespace)
RQ_final = RedisQueue('FinalQueue', namespace)
RHash = RedisHash(key + '_info')
RHerror = RedisHash(key + '_error')
RQ_lock = RedisLock(lock_obj)
RQ_lock.add({'MAP':'1'})
if int(RQ_file.QueueSize()) == 0 and RQ_lock.available('MAP'):
print("No Files Found in Redis Queue")
else:
jobFile = RQ_file.Qfile2Qwork(
RQ_file.getQueueName(), RQ_work.getQueueName()).decode('utf-8')
# Setup system logging
basename = os.path.splitext(os.path.basename(jobFile))[0]
logger = logging.getLogger(key + '.' + basename)
logger.setLevel(logging.INFO)
logFileHandle = logging.FileHandler(pds_log + '/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:
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().decode('utf-8') == 'unsignedbyte':
temp_outfile = outfile + '+lsb+tile+attached+unsignedbyte+1:254'
elif RHash.OutBit().decode('utf-8') == '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().decode('utf-8') + '.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
img_format = RHash.Format().decode('utf-8')
if img_format == 'GeoTiff-BigTiff':
fileext = 'tif'
elif img_format == 'GeoJPEG-2000':
fileext = 'jp2'
elif img_format == 'JPEG':
fileext = 'jpg'
elif img_format == 'PNG':
fileext = 'png'
elif img_format == 'GIF':
fileext = 'gif'
logGDALcmd = GDALcmd + ' ' + basename + '.input.cub ' + RHash.getMAPname().decode('utf-8') + '.' + fileext
finalfile = workarea + RHash.getMAPname().decode('utf-8') + '.' + fileext
GDALcmd += ' ' + infile + ' ' + finalfile
print(GDALcmd)
try:
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 as e:
logger.error('Process GDAL translate :: Error')
logger.error(e)
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)
loggyOBJ.AddProcess(subloggyOBJ.getSLprocess())
if status == 'success':
if RHash.Format().decode('utf-8') == 'ISIS3':
finalfile = workarea + RHash.getMAPname().decode('utf-8') + '.cub'
shutil.move(infile, finalfile)
if RHash.getStatus() != b'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()))