def process_dataset(options,_dataset,processors,*args,**kw):
"""Given options, Information Set and processors filenames execute
processors over the information set """
for pr in processors:
if not exists(pr):
pr = join(options.process_path,pr)
if not exists(pr):
logger.error('E01#001: Processor config file not exists (%s)',pr)
raise ValueError, 'E:MKEL:001'
cfg = udict(cfg2hash(pr))
if 'PROCESSOR' in cfg:
pcfg = udict(cfg.xget('PROCESSOR'))
pname = pcfg.xget('NAME')
if not pname:
logger.error('E01#002: Key NAME not exists in processor config (%s)',pr)
continue
processor = getProcessor(pname,
_dataset,
config=pcfg)
if processor:
_dataset = processor.execute()
else:
logger.error('E01#003: Processor not exists (%s)',pname)
return _dataset
def _inline_processor(self,_ts,name,kw):
""" Inline processor """
if kw['PROC']:
# logger.debug( 'calling processor functions: %s on %s',kw['PROC'],name )
names = (name,) if name else _ts.keys()
for name in names:
proc = getProcessor('INLINE_PROCESSOR',
_ts,
config=kw )
_ts = proc.execute(name) # returns full results
return _ts
else:
if kw.has_key('NAME') and kw['NAME']:
self._missing.extend(kw['NAME'].split(','))
else:
self._missing.extend([ r['Instrument'] for r in reqs])
return
# ...if post-process
if kw['PROC']:
# logger.debug('calling inline processor functions: %s',kw['PROC'])
for name in _ts.keys():
logger.debug('calling processor functions: %s for %s',kw['PROC'],name)
# we re-take each time the processor to pass the Information Set
# as modified from the previous operation
proc = getProcessor('INLINE_PROCESSOR',
self._res ,
config=kw )
self._res = proc.execute(name) # returns full results
return self._res
class DatastreamProvider(WithSourceProvider):
"""
Dataprovider facade for Thomson DataStream (DWE) source
>>> dwe = DatastreamProvider()
>>> dwe.append('dstream://Datastream/AGINDEMIF')
>>> dwe.request()
<DataSet of len 1: AGINDEMIF(13)>
"""
