def screenRecords(
stream,
separatorRE,
idRE=None,
keep=False,
screen_set=None,
screenFile=None):
"""
uses recordIterator(strean, separatorRE, idRE) to parse input into records
uses screen_set (can be read from screenFile) to identify records
identified records are kept or skipped based on the value of keep
"""
if screen_set is None:
if screenFile is None:
raise Exception(
"Please supply a hash(Python map) or file of record keys")
else:
screen_set = parse_list_to_set(screenFile)
for (
recordId,
recordLines) in recordIterator(
stream,
separatorRE,
idRE=idRE):
screened = recordId in screen_set
if screened == keep:
for line in recordLines:
yield line
def get_group_set(groups, taxonomy):
"""
Return set of all indicated taxids
input may be a taxid, taxon name, or file listing taxids
"""
if groups is None or len(groups) == 0:
return None
taxidset = set()
for group in groups:
if os.path.isfile(group):
taxidset.update(parse_list_to_set(group, keyType=int))
else:
taxon = getTaxonFromArg(taxonomy, group)
taxidset.update(create_taxid_set(taxon))
return taxidset
def get_group_set(groups, taxonomy):
"""
Return set of all indicated taxids
input may be a taxid, taxon name, or file listing taxids
"""
if groups is None or len(groups) == 0:
return None
taxidset = set()
for group in groups:
if os.path.isfile(group):
taxidset.update(parse_list_to_set(group, keyType=int))
else:
taxon = getTaxonFromArg(taxonomy, group)
taxidset.update(create_taxid_set(taxon))
return taxidset
