def detect_empty_columns(sample):
empty_col = []
if len(sample) > 0:
columns = [[] for _ in sample[0]]
for row in sample:
for j, c in enumerate(row):
columns[j].append(c)
for i, col in enumerate(columns):
if all(["EMPTY" in simple_type_detection.detectType(c) for c in col]):
empty_col.append(i)
return empty_col
def guess_headers(sample, empty_columns=None):
# first store types of first x rows
types = []
for i, row in enumerate(sample):
if i >= HEADER_CONFIDENCE:
break
row_types = []
for c in row:
t = simple_type_detection.detectType(c)
row_types.append(t)
types.append(row_types)
# now analyse types
first = types[0]
if empty_columns:
first = [i for j, i in enumerate(first) if j not in empty_columns]
first_is_alpha = all(["ALPHA" in t for t in first])
if first_is_alpha:
return sample[0]
return []
