def guess_column_types(path, dialect, sample_size, encoding='utf-8'):
"""
Guess column types based on a sample of data.
"""
book = xlrd.open_workbook(path, on_demand=True)
sheet = book.sheet_by_index(0)
column_types = []
for i in range(sheet.ncols):
values = sheet.col_values(i)[1:sample_size + 1]
types = sheet.col_types(i)[1:sample_size + 1]
nominal_type = determine_column_type(types)
if nominal_type == xlrd.biffh.XL_CELL_EMPTY:
column_types.append(None)
elif nominal_type == xlrd.biffh.XL_CELL_TEXT:
column_types.append(unicode)
elif nominal_type == xlrd.biffh.XL_CELL_NUMBER:
column_types.append(determine_number_type(values))
elif nominal_type == xlrd.biffh.XL_CELL_DATE:
column_types.append(determine_date_type(values, datemode=book.datemode))
elif nominal_type == xlrd.biffh.XL_CELL_BOOLEAN:
column_types.append(bool)
elif nominal_type == xlrd.biffh.XL_CELL_ERROR:
column_types.append(unicode)
else:
raise TypeInferenceError(_('Unknown column type found in xls file: %s') % nominal_type)
return [t.__name__ if t else None for t in column_types]
def guess_column_types(path, dialect, sample_size, encoding='utf-8'):
"""
Guess column types based on a sample of data.
"""
book = xlrd.open_workbook(path, on_demand=True)
sheet = book.sheet_by_index(0)
column_types = []
for i in range(sheet.ncols):
values = sheet.col_values(i)[1:sample_size + 1]
types = sheet.col_types(i)[1:sample_size + 1]
nominal_type = determine_column_type(types)
if nominal_type == xlrd.biffh.XL_CELL_EMPTY:
column_types.append(None)
elif nominal_type == xlrd.biffh.XL_CELL_TEXT:
column_types.append(unicode)
elif nominal_type == xlrd.biffh.XL_CELL_NUMBER:
column_types.append(determine_number_type(values))
elif nominal_type == xlrd.biffh.XL_CELL_DATE:
column_types.append(
determine_date_type(values, datemode=book.datemode))
elif nominal_type == xlrd.biffh.XL_CELL_BOOLEAN:
column_types.append(bool)
elif nominal_type == xlrd.biffh.XL_CELL_ERROR:
column_types.append(unicode)
else:
raise TypeInferenceError(
'Unknown column type found in xls file: %s' % nominal_type)
return [t.__name__ if t else None for t in column_types]
def test_determine_column_type_empty(self):
column_type = xls.determine_column_type([xlrd.biffh.XL_CELL_EMPTY, xlrd.biffh.XL_CELL_EMPTY, xlrd.biffh.XL_CELL_EMPTY])
self.assertEquals(column_type, xlrd.biffh.XL_CELL_EMPTY)
def test_determine_column_type_multiple(self):
column_type = xls.determine_column_type([xlrd.biffh.XL_CELL_NUMBER, xlrd.biffh.XL_CELL_TEXT, xlrd.biffh.XL_CELL_EMPTY])
self.assertEquals(column_type, xlrd.biffh.XL_CELL_TEXT)
def test_determine_column_type_single(self):
column_type = xls.determine_column_type([
xlrd.biffh.XL_CELL_NUMBER, xlrd.biffh.XL_CELL_NUMBER,
xlrd.biffh.XL_CELL_EMPTY
])
self.assertEquals(column_type, xlrd.biffh.XL_CELL_NUMBER)
def test_determine_column_type_empty(self):
column_type = xls.determine_column_type([
xlrd.biffh.XL_CELL_EMPTY, xlrd.biffh.XL_CELL_EMPTY,
xlrd.biffh.XL_CELL_EMPTY
])
self.assertEquals(column_type, xlrd.biffh.XL_CELL_EMPTY)
