with open(query_file, encoding='utf-8') as file:
parsed = BeautifulSoup(file, "html.parser")
query_list = parsed.find_all("topic")
print("There are %s queries." % len(query_list))
combined_stats = Stats()
fileids = set()
## try:
query_list_m = list(map(get_query, query_list)) # whole batch for now
args = [(query_list_m, topk, math_index)]
for p in args: # single-process execution
(fileid, stats) = process_query_batch(p)
fileids.add(fileid)
combined_stats.add(stats)
## except Exception as err:
## reason = str(err)
## print("Failed to process queries: "+reason, file=sys.stderr)
cntl.store("query_fileids", str(fileids))
print("Done preparing query batch for %s" % (query_file))
combined_stats.dump()
cntl.dump() # output the revised cntl file
end = time.time()
elapsed = end - start
print("Elapsed time %s" % (elapsed))
with open(query_file, encoding='utf-8') as file:
parsed = BeautifulSoup(file, "lxml")
query_list = parsed.find_all("topic")
print("There are %s queries." % (len(query_list)), flush=True)
combined_stats = Stats()
fileids = set()
try:
query_list_m = list(map(get_query,query_list)) # whole batch for now
args = [(system, db, run_tag, query_list_m, topk, math_index, weighting_strategy)]
for p in args: # single-process execution
(fileid,stats) = process_query_batch(p)
fileids.add(fileid)
combined_stats.add(stats)
except Exception as err:
reason = str(err)
print("Failed to process document "+filename+": "+reason, file=sys.stderr)
combined_stats.problem_files[reason] = combined_stats.problem_files.get(reason, set())
combined_stats.problem_files[reason].add(filename)
cntl.store("query_fileids",str(fileids))
print("Done preparing query batch for %s against %s" % (query_file, db))
combined_stats.dump()
cntl.dump() # output the revised cntl file
end = time.time()
elapsed = end - start
