def test_metadata_split_to_dict():
"""Should split metadata into a dictionary."""
pc_axis = pyaxis.read(data_path + '14001.px', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
assert type(metadata) == dict
assert len(metadata) == 28
def test_metadata_split_dict_quotation_marks():
"""Should ignore character split if it is between quotation marks."""
metadata_elements = [
'VALUES("estrato de empleo")="Total",">=10 empleados";'
]
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
assert type(metadata) == dict
assert metadata['VALUES(estrato de empleo)'] == ['Total', '>=10 empleados']
def test_metadata_split_to_dict():
"""Should split metadata into a dictionary."""
pc_axis = pyaxis.read(
'https://www.ine.es/jaxi/files/_px/es/px/t20/e301/matri/a2000/l0/'
'14001.px?nocab=1', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
assert type(metadata) == dict
assert len(metadata) == 23
def test_get_codes():
"""Should return two lists (dimension names with codes and codes)."""
pc_axis = pyaxis.read(data_path + '14001.px', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
dimensions_with_codes, dimension_codes = pyaxis.get_codes(metadata)
assert len(dimension_codes) == 1
assert len(dimensions_with_codes) == 1
assert dimensions_with_codes[
0] == 'Comunidad Autónoma de residencia del matrimonio'
assert dimension_codes[0][6] == 'CA06'
assert dimension_codes[0][11] == 'CA11'
def test_get_dimensions():
"""Should return two lists (dimension names and members)."""
pc_axis = pyaxis.read(data_path + '14001.px', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
dimension_names, dimension_members = pyaxis.get_dimensions(metadata)
assert len(dimension_names) == 4
assert dimension_names[
0] == 'Comunidad Autónoma de residencia del matrimonio'
assert dimension_names[3] == 'estado civil anterior de los cónyuges'
assert len(dimension_members) == 4
assert dimension_members[0][0] == 'Total'
assert dimension_members[3][3] == 'Divorciados/as'
def test_get_dimensions():
"""Should return two lists (dimension names and members)."""
pc_axis = pyaxis.read(
'https://www.ine.es/jaxi/files/_px/es/px/t20/e301/matri/a2000/l0/'
'14001.px?nocab=1', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
dimension_names, dimension_members = pyaxis.get_dimensions(metadata)
assert len(dimension_names) == 4
assert dimension_names[
0] == 'Comunidad Autónoma de residencia de los cónyuges'
assert dimension_names[3] == 'Estado civil anterior de los cónyuges'
assert len(dimension_members) == 4
assert dimension_members[0][0] == 'Todas las comunidades'
assert dimension_members[3][3] == 'Divorciados/as'
def test_build_dataframe():
"""Should return a dataframe with n+1 columns (dimensions + data)."""
null_values = r'^"\."$'
sd_values = r'"\.\."'
pc_axis = pyaxis.read(data_path + '14001.px', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
dimension_names, dimension_members = pyaxis.get_dimensions(metadata)
data_values = Series(raw_data.split())
df = pyaxis.build_dataframe(dimension_names,
dimension_members,
data_values,
null_values=null_values,
sd_values=sd_values)
assert df.shape == (8064, 5)
assert df['DATA'][7] == '28138'
assert df['DATA'][159] == '422'
def test_build_dataframe():
"""Should return a dataframe with n+1 columns (dimensions + data)."""
null_values = r'^"\."$'
sd_values = r'"\.\."'
pc_axis = pyaxis.read(
'https://www.ine.es/jaxi/files/_px/es/px/t20/e301/matri/a2000/l0/'
'14001.px?nocab=1', 'iso-8859-15')
metadata_elements, raw_data = pyaxis.metadata_extract(pc_axis)
metadata = pyaxis.metadata_split_to_dict(metadata_elements)
dimension_names, dimension_members = pyaxis.get_dimensions(metadata)
data_values = Series(raw_data.split())
df = pyaxis.build_dataframe(dimension_names,
dimension_members,
data_values,
null_values=null_values,
sd_values=sd_values)
assert df.shape == (8064, 5)
assert df['DATA'][7] == '10624.0'
assert df['DATA'][159] == '534.0'