def train(db_path, out_path, **kwargs):
db = sqlite3.connect(db_path)
db.create_function("LC", 1, linecount)
# auto-detect whether it's a GitHub repo
kwargs['gh'] = dbutil.is_github(db)
ret = create_corpus(db, out_path, **kwargs)
if ret:
sys.exit(ret)
def explore(db_path, graph=False):
locale.setlocale(locale.LC_ALL, 'en_GB.utf-8')
db = sqlite3.connect(db_path)
if dbutil.is_github(db):
db.close()
explore_gh(db_path)
return
if graph and not os.path.exists(img_dir):
os.makedirs(img_dir)
# Worker process pool
pool, jobs = Pool(processes=4), []
if graph:
jobs.append(pool.apply_async(graph_ocl_lc, (db_path,)))
# TODO: If GH dataset:
# jobs.append(pool.apply_async(graph_ocl_stars, (db_path,)))
future_stats = pool.apply_async(stats_worker, (db_path,))
# Wait for jobs to finish
[job.wait() for job in jobs]
# Print stats
print()
stats = future_stats.get()
maxlen = max([len(x[0]) for x in stats])
for stat in stats:
k,v = stat
if k:
print(k, ':', ' ' * (maxlen - len(k) + 2), v, sep='')
elif v == '':
print(k)
else:
print()
def test_is_github(self):
self.assertFalse(dbutil.is_github(tests.db('empty')))
self.assertTrue(dbutil.is_github(tests.db('empty-gh')))