def test(self, infile):
_f = anyfile(infile)
ff = rec_handler(_f)
gd = []
err_cnt = 0
for rec in ff:
if not rec.startswith('rs'):
continue
lines = rec.strip().split('\n')
self._parse_rsline(lines)
d = self._parse_GMAF(lines)
if not d:
err_cnt += 1
gd.append(d)
print(err_cnt)
return gd
def load_data(input_file):
# the first two line of clinvar_xml is not useful information
cv_data = rec_handler(input_file, block_end='</ClinVarSet>\n',
skip=2, include_block_end=True)
print input_file
for record in cv_data:
# some exceptions
if record.startswith('\n</ReleaseSet>'):
continue
try:
record_parsed = clinvar.parseString(record, silence=1)
except:
print(record)
raise
for record_mapped in _map_line_to_json(record_parsed):
yield record_mapped
def load_data(input_file):
# the first two line of clinvar_xml is not useful information
cv_data = rec_handler(input_file,
block_end='</ClinVarSet>\n',
skip=2,
include_block_end=True)
print input_file
for record in cv_data:
# some exceptions
if record.startswith('\n</ReleaseSet>'):
continue
try:
record_parsed = clinvar.parseString(record, silence=1)
except:
print(record)
raise
for record_mapped in _map_line_to_json(record_parsed):
yield record_mapped
def parse(self, infile):
print(os.path.split(infile)[1])
cnt = 0
err_d = {}
_f = anyfile(infile)
ff = rec_handler(_f)
for rec in ff:
if not rec.startswith('rs'):
continue
doc = self.parse_one_record(rec)
if isinstance(doc, dict):
cnt += 1
yield doc
else:
if doc in err_d:
err_d[doc] += 1
else:
err_d[doc] = 1
print(cnt, err_d)