def parse(self):
global_run_id = None
history = set()
results = Results()
with open(self.filename) as f:
for line in f:
line_components = line.strip().split()
if len(line_components) != 6:
raise ResultsParseError(
'lines in results file should have exactly 6 columns')
query_id, _, doc_id, rank, score, run_id = line_components
rank = int(rank)
score = float(score)
if global_run_id is None:
global_run_id = run_id
elif global_run_id != run_id:
raise ResultsParseError(
'Mismatching runIDs in results file')
key = query_id + doc_id
if key in history:
raise ResultsParseError(
'Duplicate query_id, doc_id in results file')
history.add(key)
results.add_result(query_id, Result(doc_id, score, rank))
return global_run_id, results
def parse(self):
global_run_id = None
history = set()
results = Results()
with open(self.filename) as f:
for line in f:
line_components = line.strip().split()
if len(line_components) != 6:
raise ResultsParser.ResultsParseError(
'lines in %s do not have exactly 6 columns' %
self.filename)
query_id, _, doc_id, rank, score, run_id = line_components
try:
rank = int(rank)
score = float(score)
except ValueError:
raise ResultsParser.ResultsParseError(
'Error parsing rank or score in %s' % self.filename)
if doc_id not in self.docno_list:
raise ResultsParser.ResultsParseError(
'Docno %s does not exist for query %s in %s' %
(doc_id, query_id, self.filename))
if global_run_id is None:
global_run_id = run_id
elif global_run_id != run_id:
raise ResultsParser.ResultsParseError(
'Mismatching runIDs in %s' % self.filename)
key = query_id + doc_id
if key in history:
raise ResultsParser.ResultsParseError(
'Duplicate query_id, doc_id in %s' % self.filename)
history.add(key)
results.add_result(query_id, Result(doc_id, score, rank))
return global_run_id, results
