# Default values - the order in which we attempt to determine data types is critical
# because some formats are much more flexibly defined than others.
if len(self.sniff_order) < 1:
self.sniff_order = [
binary.Bam(),
binary.Sff(),
xml.GenericXml(),
sequence.Maf(),
sequence.Lav(),
sequence.csFasta(),
qualityscore.QualityScoreSOLiD(),
qualityscore.QualityScore454(),
sequence.Fasta(),
sequence.Fastq(),
interval.Wiggle(),
images.Html(),
sequence.Axt(),
interval.Bed(),
interval.CustomTrack(),
interval.Gtf(),
interval.Gff(),
interval.Gff3(),
tabular.Pileup(),
interval.Interval(),
tabular.Sam(),
tabular.Eland()
]
def append_to_sniff_order():
# Just in case any supported data types are not included in the config's sniff_order section.
for ext in self.datatypes_by_extension:
datatype = self.datatypes_by_extension[ext]
'data': data.Data(),
'bed': interval.Bed(),
'txt': data.Text(),
'text': data.Text(),
'interval': interval.Interval(),
'tabular': interval.Tabular(),
'png': images.Image(),
'pdf': images.Image(),
'fasta': sequence.Fasta(),
'maf': sequence.Maf(),
'axt': sequence.Axt(),
'gff': interval.Gff(),
'gmaj.zip': images.Gmaj(),
'laj': images.Laj(),
'lav': sequence.Lav(),
'html': images.Html()
}
def get_datatype_by_extension(ext):
"""
Returns a datatype based on an extension
"""
try:
builder = datatypes_by_extension[ext]
except KeyError:
builder = data.Text()
log.warning('unkown extension in data factory %s' % ext)
return builder