def test_ContentConstraint(dsd, dks):
""":class:`.ContentConstraint` can be written to XML."""
sdmx.to_xml(
m.ContentConstraint(
role=m.ConstraintRole(role=m.ConstraintRoleType.allowable),
content=[dsd],
data_content_keys=dks,
))
def test_ds(dsd, obs):
# Write DataSet with Observations not in Series
ds = DataSet(structured_by=dsd)
ds.obs.append(obs)
result = sdmx.to_xml(ds, pretty_print=True)
print(result.decode())
def test_obs(obs):
# Generate <gen:ObsKey> element for 2+-dimensional Observation.dimension
exp = (
'<gen:ObsKey><gen:Value id="FOO" value="1"/>'
'<gen:Value id="BAR" value="2"/></gen:ObsKey>'
)
assert exp in sdmx.to_xml(obs).decode()
def update_dsd():
"""Generate the iTEM SDMX data structures.
The file item/data/structure.xml is updated.
"""
import sdmx
from item.structure import generate
with open(Path(__file__).parent / "data" / "structure.xml", "wb") as f:
f.write(sdmx.to_xml(generate(), pretty_print=True))
def test_structuremessage(tmp_path, structuremessage):
result = sdmx.to_xml(structuremessage, pretty_print=True)
print(result.decode())
# Message can be round-tripped to/from file
path = tmp_path / "output.xml"
path.write_bytes(result)
msg = sdmx.read_sdmx(path)
# Contents match the original object
assert (msg.codelist["CL_COLLECTION"]["A"].name["en"] ==
structuremessage.codelist["CL_COLLECTION"]["A"].name["en"])
# False because `structuremessage` lacks URNs, which are constructed automatically
# by `to_xml`
assert not msg.compare(structuremessage, strict=True)
# Compares equal when allowing this difference
assert msg.compare(structuremessage, strict=False)
def test_structure_roundtrip(pytestconfig, specimen_id, strict, tmp_path):
"""Test that SDMX-ML StructureMessages can be 'round-tripped'."""
# Read a specimen file
with specimen(specimen_id) as f:
msg0 = sdmx.read_sdmx(f)
# Write to file
path = tmp_path / "output.xml"
path.write_bytes(sdmx.to_xml(msg0, pretty_print=True))
# Read again
msg1 = sdmx.read_sdmx(path)
# Contents are identical
assert msg0.compare(msg1, strict), (
path.read_text() if pytestconfig.getoption("verbose") else path
)
def test_sdmx_roundtrip(tmp_path):
path = tmp_path / "structure.xml"
# Structure can be written
with open(path, "wb") as f:
f.write(sdmx.to_xml(generate(), pretty_print=True))
# Structure can be read
sm = sdmx.read_sdmx(path)
# One CubeRegion
assert 1 == len(sm.constraint["PRICE_FUEL"].data_content_region)
# One dimension with a MemberSelection
cr = sm.constraint["PRICE_FUEL"].data_content_region[0]
assert {"FUEL"} == set(d.id for d in cr.member.keys())
# 3 values in the MemberSelection
assert 3 == len(cr.member["FUEL"].values)
def test_data_roundtrip(pytestconfig, data_id, structure_id, tmp_path):
"""Test that SDMX-ML DataMessages can be 'round-tripped'."""
# Read structure from file
with specimen(structure_id) as f:
dsd = sdmx.read_sdmx(f).structure[0]
# Read data from file, using the DSD
with specimen(data_id) as f:
msg0 = sdmx.read_sdmx(f, dsd=dsd)
# Write to file
path = tmp_path / "output.xml"
path.write_bytes(sdmx.to_xml(msg0, pretty_print=True))
# Read again, using the same DSD
msg1 = sdmx.read_sdmx(path, dsd=dsd)
# Contents are identical
assert msg0.compare(msg1, strict=True), (
path.read_text() if pytestconfig.getoption("verbose") else path
)
def build(self, language=None):
"""Write the SDMX output. Overrides parent."""
status = True
# The SDMX output is a special case, and does not need to be
# translated separately for each language. So we only continue
# if this is an untranslated or language-agnostic build.
if language is not None and language != 'untranslated':
return status
all_serieses = {}
all_metadata_serieses = []
metadata_template = Template(self.get_metadata_template())
dfd = DataflowDefinition(id="OPEN_SDG_DFD", structure=self.dsd)
time_period = next(dim for dim in self.dsd.dimensions if dim.id == 'TIME_PERIOD')
header_info = self.get_header_info()
header = self.create_header(header_info)
metadata_base_vars = header_info.copy()
for indicator_id in self.get_indicator_ids():
indicator = self.get_indicator_by_id(indicator_id)
data = indicator.data.copy()
# Some hardcoded dataframe changes.
data = data.rename(columns={
'Value': 'OBS_VALUE',
'Units': 'UNIT_MEASURE',
'Series': 'SERIES',
'Year': 'TIME_PERIOD',
})
# Any user-specified dataframe changes.
self.apply_column_map(data)
self.apply_code_map(data)
if self.constrain_data:
before = len(data.index)
data = indicator.get_data_matching_schema(self.data_schema, data=data)
after = len(data.index)
message = '{indicator_id} - Removed {difference} rows while constraining data (out of {total}).'
difference = str(before - after)
self.warn(message, indicator_id=indicator_id, difference=difference, total=before)
data = data.replace(np.nan, '', regex=True)
if not data.empty:
serieses = {}
for _, row in data.iterrows():
series_key = self.dsd.make_key(SeriesKey, self.get_dimension_values(row, indicator))
series_key.attrib = self.get_series_attribute_values(row, indicator)
attributes = self.get_observation_attribute_values(row, indicator)
dimension_key = self.dsd.make_key(Key, values={
'TIME_PERIOD': str(row['TIME_PERIOD']),
})
observation = Observation(
series_key=series_key,
dimension=dimension_key,
attached_attribute=attributes,
value_for=self.dsd.measures[0],
value=row[self.dsd.measures[0].id],
)
if series_key not in serieses:
serieses[series_key] = []
serieses[series_key].append(observation)
dataset = self.create_dataset(serieses)
msg = DataMessage(data=[dataset], dataflow=dfd, header=header, observation_dimension=time_period)
sdmx_path = os.path.join(self.sdmx_folder, indicator_id + '.xml')
with open(sdmx_path, 'wb') as f:
status = status & f.write(sdmx.to_xml(msg))
all_serieses.update(serieses)
concepts = indicator.meta
if self.constrain_meta and self.msd is not None:
concepts = indicator.get_meta_matching_schema(self.msd)
reporting_type = self.meta_reporting_type
if reporting_type is None and 'REPORTING_TYPE' in data.columns and len(data) > 0:
reporting_type = self.get_first_value_from_data_column(data, 'REPORTING_TYPE')
ref_area = self.meta_ref_area
if ref_area is None and 'REF_AREA' in data.columns and len(data) > 0:
ref_area = self.get_first_value_from_data_column(data, 'REF_AREA')
if concepts and ref_area is not None and reporting_type is not None:
series_codes = helpers.sdmx.get_all_series_codes_from_indicator_id(indicator_id,
dsd_path=self.dsd_path,
request_params=self.request_params,
)
# Make sure the indicator is fully translated.
for language in self.all_languages:
indicator.translate(language, self.translation_helper)
concept_items = []
for key in concepts:
translation_items = []
for language in self.all_languages:
translated_value = indicator.language(language).get_meta_field_value(key)
translation_items.append({
'language': language,
'value': translated_value,
})
concept_items.append({
'key': key,
'translations': translation_items,
})
metadata_serieses = []
for code in series_codes:
metadata_series = {
'set_id': uuid.uuid4(),
'series': code,
'reporting_type': reporting_type,
'ref_area': ref_area,
'concepts': concept_items,
}
metadata_serieses.append(metadata_series)
metadata = metadata_base_vars.copy()
metadata['serieses'] = metadata_serieses
metadata_sdmx = metadata_template.render(metadata)
meta_path = os.path.join(self.meta_folder, indicator_id + '.xml')
with open(meta_path, 'w') as f:
status = status & f.write(metadata_sdmx)
all_metadata_serieses = all_metadata_serieses + metadata_serieses
dataset = self.create_dataset(all_serieses)
msg = DataMessage(data=[dataset], dataflow=dfd, header=header, observation_dimension=time_period)
all_sdmx_path = os.path.join(self.sdmx_folder, 'all.xml')
with open(all_sdmx_path, 'wb') as f:
status = status & f.write(sdmx.to_xml(msg))
metadata = metadata_base_vars.copy()
metadata['serieses'] = all_metadata_serieses
metadata_sdmx = metadata_template.render(metadata)
meta_path = os.path.join(self.meta_folder, 'all.xml')
with open(meta_path, 'w') as f:
status = status & f.write(metadata_sdmx)
return status
def test_DataKeySet(dks):
""":class:`.DataKeySet` can be written to XML."""
sdmx.to_xml(dks)
def test_codelist(tmp_path, codelist):
result = sdmx.to_xml(codelist, pretty_print=True)
print(result.decode())
