def patch_field_auth(schema: DatasetSchema, table_id, field_id, *subfields, auth: list[str]):
"""Monkeypatch an Amsterdam Schema to set "auth" on a table."""
# This updates the low-level dict data so all high-level objects get it.
schema.get_table_by_id(table_id).get_field_by_id(field_id) # check existence
raw_table = next(t for t in schema["tables"] if t.default["id"] == table_id)
raw_field = next(
f for f_id, f in raw_table.default["schema"]["properties"].items() if f_id == field_id
)
# Allow to resolve sub fields too
for subfield in subfields:
# Auto jump over array, object or "array of objects"
if raw_field["type"] == "array":
raw_field = raw_field["items"]
if raw_field["type"] == "object":
raw_field = raw_field["properties"]
raw_field = raw_field[subfield]
raw_field["auth"] = auth
# Also patch the active model
model = apps.get_model(schema.id, table_id)
model_field = model._meta.get_field(to_snake_case(field_id))
for subfield in subfields:
model_field = model_field.related_model._meta.get_field(subfield)
model_field.field_schema["auth"] = auth
def patch_table_auth(schema: DatasetSchema, table_id, *, auth: list[str]):
"""Monkeypatch an Amsterdam Schema to set "auth" on a table."""
# This updates the low-level dict data so all high-level objects get it.
schema.get_table_by_id(table_id) # checks errors
raw_table = next(t for t in schema["tables"] if t.default["id"] == table_id)
raw_table.default["auth"] = auth
# Also patch the active model, as that's already loaded and has a copy of the table schema
model = apps.get_model(schema.id, table_id)
model.table_schema()["auth"] = auth
def import_schema(self, name: str,
schema: DatasetSchema) -> Optional[Dataset]:
"""Import a single dataset schema."""
try:
dataset = Dataset.objects.get(name=schema.id)
except Dataset.DoesNotExist:
dataset = Dataset.objects.create(name=schema.id,
schema_data=schema.json_data())
self.stdout.write(f" Created {name}")
return dataset
else:
dataset.schema_data = schema.json_data()
if dataset.schema_data_changed():
dataset.save()
self.stdout.write(f" Updated {name}")
return dataset
return None
def create_meta_table_data(engine, dataset_schema: DatasetSchema):
session = sessionmaker(bind=engine)()
ds_content = {
camel_case_to_snake(k): v
for k, v in dataset_schema.items() if k != "tables"
}
ds_content["contact_point"] = str(ds_content.get("contact_point", ""))
ds_transformer = transformer_factory(models.Dataset)
dataset = models.Dataset(**ds_transformer(ds_content))
session.add(dataset)
for table_data in dataset_schema["tables"]:
table_content = {
camel_case_to_snake(k): v
for k, v in table_data.items() if k != "schema"
}
table = models.Table(
**{
**table_content,
**{
f: table_data["schema"].get(f)
for f in ("required", "display")
},
})
table.dataset_id = dataset.id
session.add(table)
for field_name, field_value in table_data["schema"][
"properties"].items():
field_content = {
k.replace("$", ""): v
for k, v in field_value.items() if k not in {"$comment"}
}
field_content["name"] = field_name
try:
field = models.Field(**field_content)
except TypeError as e:
raise NotImplementedError(
f'Import failed: at "{field_name}": {field_value!r}:\n{e}'
) from e
field.table_id = table.id
field.dataset_id = dataset.id
session.add(field)
session.commit()
def _load_geojson(postgres_conn_id):
"""As airflow executes tasks at different hosts,
these tasks need to happen in a single call.
Otherwise, the (large) file is downloaded by one host,
and stored in the XCom table to be shared between tasks.
"""
tmp_dir = Path(f"/tmp/{dag_id}")
tmp_dir.mkdir(parents=True, exist_ok=True)
# 1. download files
files = {}
for route in ROUTES:
dest = f"{tmp_dir}/{route.name}.geojson"
logger.info("Downloading %s to %s", route.url, dest)
download_file(route.url, dest, http_conn_id=None)
files[route.name] = dest
# 2. generate schema ("schema introspect geojson *.geojson")
schema = introspect_geojson_files(
"gevaarlijke-routes", files=list(files.values())
)
schema = DatasetSchema.from_dict(schema) # TODO: move to schema-tools?
# XXX This is not running as one transaction atm, but autocommitting per chunk
# 3. import data
db_engine = get_engine()
importer = GeoJSONImporter(schema, db_engine, logger=logger)
for route in ROUTES:
geojson_path = files[route.name]
logger.info(
"Importing %s into %s", route.name, route.tmp_db_table_name
)
importer.generate_db_objects(
table_name=route.schema_table_name,
db_table_name=route.tmp_db_table_name,
truncate=True, # when reexecuting the same task
ind_tables=True,
ind_extra_index=False,
)
importer.load_file(
geojson_path,
)
if route.post_process:
hook = PostgresHook(postgres_conn_id=postgres_conn_id)
hook.run(route.post_process)
def execute(self, context=None):
"""Executes the 'generate_db_object' method from schema-tools.
Which leads to the creation of tables and/or an index on the identifier (as specified in the data JSON schema).
By default both tables and the identifier and 'many-to-many table' indexes are created.
By setting the boolean indicators in the method parameters, tables or an identifier index (per table) can be created.
"""
data_schema_url = f"{SCHEMA_URL.split('//')[0]}//{self.data_schema_env}{SCHEMA_URL.split('//')[1]}{self.data_schema_name}/{self.data_schema_name}"
data = schema_fetch_url_file(data_schema_url)
engine = _get_engine(self.db_conn)
parent_schema = SchemaType(data)
dataset_schema = DatasetSchema(parent_schema)
importer = BaseImporter(dataset_schema, engine)
for table in data["tables"]:
if (self.data_schema_name + "_" +
table["id"] == f"{self.data_table_name}"):
importer.generate_db_objects(
table["id"],
ind_tables=self.ind_table,
ind_extra_index=self.ind_extra_index,
)
else:
continue
def vestiging_schema(vestiging_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(vestiging_schema_json)
def parkeervakken_schema(parkeervakken_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(parkeervakken_schema_json)
def schema(self) -> DatasetSchema:
"""Provide access to the schema data"""
if not self.schema_data:
raise RuntimeError("Dataset.schema_data is empty")
return DatasetSchema.from_dict(self.schema_data)
def woningbouwplannen_schema(woningbouwplannen_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(woningbouwplannen_schema_json)
def gebieden_schema(gebieden_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(gebieden_schema_json)
def download_url_schema(download_url_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(download_url_schema_json)
def explosieven_schema(explosieven_schema_json, ) -> DatasetSchema:
return DatasetSchema.from_dict(explosieven_schema_json)
def afval_schema(afval_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(afval_schema_json)
def fietspaaltjes_schema(fietspaaltjes_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(fietspaaltjes_schema_json)
def fietspaaltjes_schema_no_display(
fietspaaltjes_schema_json_no_display, ) -> DatasetSchema:
return DatasetSchema.from_dict(fietspaaltjes_schema_json_no_display)
def afval_schema_backwards_embedded(
afval_schema_backwards_embedded_json, ) -> DatasetSchema:
return DatasetSchema.from_dict(afval_schema_backwards_embedded_json)
def indirect_self_ref_schema(indirect_self_ref_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(indirect_self_ref_schema_json)
def afval_schema_backwards_summary(
afval_schema_backwards_summary_json, ) -> DatasetSchema:
return DatasetSchema.from_dict(afval_schema_backwards_summary_json)
def meldingen_schema(meldingen_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(meldingen_schema_json)
def bommen_v2_schema(bommen_v2_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(bommen_v2_schema_json)
def bag_schema(bag_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(bag_schema_json)
def geometry_zoom_schema():
return DatasetSchema.from_dict(
json.loads((HERE / "files" / "geometry_zoom.json").read_text()))
TABLE_SCHEMA = {
"id": "mytable",
"type": "table",
"version": str(V1),
"schema": {
"$schema": "http://json-schema.org/draft-07/schema#",
},
}
REMOTE_SCHEMA = DatasetTableSchema(
TABLE_SCHEMA,
parent_schema=DatasetSchema(
{
"id": "adhoc",
"tables": [
TableVersions(
id=TABLE_SCHEMA["id"],
default_version_number=V1,
active=dict(V1=TABLE_SCHEMA),
)
],
}
),
)
@pytest.mark.parametrize(
"case",
[
("http://remote", "http://remote/foo?bar=baz"),
("http://remote/", "http://remote/foo?bar=baz"),
("http://remote/quux/{table_id}", "http://remote/quux/mytable/foo?bar=baz"),
("http://remote/quux/{table_id}/", "http://remote/quux/mytable/foo?bar=baz"),
def geometry_authdataset_schema(
geometry_authdataset_schema_json) -> DatasetSchema:
return DatasetSchema.from_dict(geometry_authdataset_schema_json)
