def get_table(datarows, types, titles):
try:
table = journalism.Table(datarows, types, titles)
return table
except:
logging.exception('problem loading table')
return None
def test_create_table(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
self.assertEqual(len(table.rows), 3)
self.assertEqual(table.rows[0], [1, 4, 'a'])
self.assertEqual(table.rows[1], [2, 3, 'b'])
self.assertEqual(table.rows[2], [None, 2, 'c'])
def test_reject(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
new_table = table.reject('one', [2, None])
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.rows), 1)
self.assertEqual(new_table.rows[0], [1, 4, 'a'])
def test_filter(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
new_table = table.filter('one', [2, None])
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.rows), 2)
self.assertEqual(new_table.rows[0], [2, 3, 'b'])
def test_sort_by_reverse(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
new_table = table.sort_by('two', reverse=True)
self.assertEqual(len(new_table.rows), 3)
self.assertEqual(new_table.rows[0], [1, 4, 'a'])
self.assertEqual(new_table.rows[1], [2, 3, 'b'])
self.assertEqual(new_table.rows[2], [None, 2, 'c'])
def test_create_table_header(self):
rows = [['one', 'two', 'three']]
rows.extend(self.rows)
table = journalism.Table(rows, self.column_types)
self.assertEqual(table.rows[0], [1, 4, 'a'])
self.assertEqual(table.rows[1], [2, 3, 'b'])
self.assertEqual(table.rows[2], [None, 2, 'c'])
def test_where(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
new_table = table.where('one', lambda x: x is not None)
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.rows), 2)
self.assertEqual(new_table.rows[0], [1, 4, 'a'])
self.assertEqual(new_table.rows[1], [2, 3, 'b'])
def setUp(self):
self.rows = [[1, 2, 'a'], [1, 1, 3, 'b'], [None, 4, 'c'], [2, 1, 'c']]
self.column_names = ['one', 'two', 'three']
self.column_types = [
journalism.IntColumn, journalism.IntColumn, journalism.TextColumn
]
self.table = journalism.Table(self.rows, self.column_types,
self.column_names)
def test_count(self):
rows = self.rows
rows.append(rows[0])
rows.append(rows[0])
table = journalism.Table(rows, self.column_types, self.column_names)
self.assertEqual(table.columns['one'].count(1), 3)
self.assertEqual(table.columns['one'].count(4), 0)
self.assertEqual(table.columns['one'].count(None), 1)
def test_aggregate_sum_two_columns(self):
table = journalism.Table(self.rows, self.column_types)
new_table = table.aggregate('one', [('two', 'sum'), ('four', 'sum')])
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.rows), 3)
self.assertEqual(new_table._column_names, ['one', 'two', 'four'])
self.assertEqual(new_table.rows[0], ['a', 4, 4])
self.assertEqual(new_table.rows[1], [None, 3, 0])
self.assertEqual(new_table.rows[2], ['b', 3, 0])
def load_data():
"""
Load the initial table data from a CSV.
"""
with open('simple.csv') as f:
reader = csv.reader(f)
next(reader)
table = journalism.Table(reader, COLUMN_TYPES, COLUMN_NAMES)
return table
def setUp(self):
self.rows = [[1.1, 2.19, 'a'], [2.7, 3.42, 'b'], [None, 4.1, 'c'],
[2.7, 1, 'c']]
self.column_names = ['one', 'two', 'three']
self.column_types = [
journalism.FloatColumn, journalism.FloatColumn,
journalism.TextColumn
]
self.table = journalism.Table(self.rows, self.column_types,
self.column_names)
def setUp(self):
self.rows = [
['one', 'two', 'three', 'four'],
['a', 2, 3, 4],
[None, 3, 5, None],
['a', 2, 4, None],
['b', 3, 4, None]
]
self.column_types = [journalism.TextColumn, journalism.IntColumn, journalism.IntColumn, journalism.IntColumn]
self.table = journalism.Table(self.rows, self.column_types)
def test_sort_by_cmp(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
def func(a, b):
return -cmp(a, b)
new_table = table.sort_by('two', cmp=func)
self.assertEqual(len(new_table.rows), 3)
self.assertEqual(new_table.rows[0], [1, 4, 'a'])
self.assertEqual(new_table.rows[1], [2, 3, 'b'])
self.assertEqual(new_table.rows[2], [None, 2, 'c'])
def test_sort_by(self):
table = journalism.Table(self.rows, self.column_types, self.column_names)
new_table = table.sort_by('two')
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.rows), 3)
self.assertEqual(new_table.rows[0], [None, 2, 'c'])
self.assertEqual(new_table.rows[1], [2, 3, 'b'])
self.assertEqual(new_table.rows[2], [1, 4, 'a'])
# Verify old table not changed
self.assertEqual(table.rows[0], [1, 4, 'a'])
self.assertEqual(table.rows[1], [2, 3, 'b'])
self.assertEqual(table.rows[2], [None, 2, 'c'])
def test_counts(self):
rows = self.rows
rows.append(rows[0])
rows.append(rows[0])
table = journalism.Table(rows, self.column_types, self.column_names)
new_table = table.columns['one'].counts()
self.assertIsNot(new_table, table)
self.assertEqual(len(new_table.columns), 2)
self.assertEqual(len(new_table.rows), 3)
self.assertEqual(new_table.rows[0], [1, 3])
self.assertEqual(new_table.rows[1], [2, 1])
self.assertEqual(new_table.rows[2], [None, 1])
self.assertEqual(new_table.columns['one'], [1, 2, None])
self.assertEqual(new_table.columns['count'], [3, 1, 1])
def test_validate_table_fails(self):
column_types = [journalism.IntColumn, journalism.IntColumn, journalism.IntColumn]
with self.assertRaises(journalism.ColumnValidationError):
journalism.Table(self.rows, column_types, self.column_names, validate=True)
def get_table(new_arr, types, titles):
try:
table = journalism.Table(new_arr, types, titles)
return table
except Exception as e:
print e
for v in example_row:
value_type = ctype_text[v.ctype]
if value_type == 'text':
types.append(text_type)
elif value_type == 'number':
types.append(number_type)
elif value_type == 'xldate':
types.append(date_type)
else:
types.append(text_type)
columns = zip(titles, types)
columns
try:
table = journalism.Table(country_rows, types, titles)
#Throws error for conversion
except:
pass
cleaned_rows = []
def float_to_str(val):
if isinstance(val, float):
return str(val)
elif isinstance(val, (str, unicode)):
print 'unicode is', val.encode('utf-8')
return val.encode('ascii', errors='replace').strip()
return val
def test_aggregate_sum_invalid(self):
table = journalism.Table(self.rows, self.column_types)
with self.assertRaises(journalism.UnsupportedOperationError):
table.aggregate('two', [('one', 'sum')])
def test_validate_table(self):
journalism.Table(self.rows, self.column_types, self.column_names, validate=True)
