def test_storage_constraints(options):
# Export/Import
source = Package("data/storage/constraints.json")
storage = source.to_bigquery(force=True, **options)
target = Package.from_bigquery(**options)
# Assert metadata
assert target.get_resource("constraints").schema == {
"fields": [
{"name": "required", "type": "string", "constraints": {"required": True}},
{"name": "minLength", "type": "string"}, # constraint removal
{"name": "maxLength", "type": "string"}, # constraint removal
{"name": "pattern", "type": "string"}, # constraint removal
{"name": "enum", "type": "string"}, # constraint removal
{"name": "minimum", "type": "integer"}, # constraint removal
{"name": "maximum", "type": "integer"}, # constraint removal
],
}
# Assert data
assert target.get_resource("constraints").read_rows() == [
{
"required": "passing",
"minLength": "passing",
"maxLength": "passing",
"pattern": "passing",
"enum": "passing",
"minimum": 5,
"maximum": 5,
},
]
# Cleanup storage
storage.delete_package(target.resource_names)
def test_bigquery_storage_constraints(options):
prefix = options.pop("prefix")
service = options.pop("service")
dialect = BigqueryDialect(table=prefix, **options)
source = Package("data/storage/constraints.json")
storage = source.to_bigquery(service, dialect=dialect)
target = Package.from_bigquery(service, dialect=dialect)
# Assert metadata
assert target.get_resource("constraints").schema == {
"fields": [
{
"name": "required",
"type": "string",
"constraints": {
"required": True
}
},
{
"name": "minLength",
"type": "string"
}, # constraint removal
{
"name": "maxLength",
"type": "string"
}, # constraint removal
{
"name": "pattern",
"type": "string"
}, # constraint removal
{
"name": "enum",
"type": "string"
}, # constraint removal
{
"name": "minimum",
"type": "integer"
}, # constraint removal
{
"name": "maximum",
"type": "integer"
}, # constraint removal
],
}
# Assert data
assert target.get_resource("constraints").read_rows() == [
{
"required": "passing",
"minLength": "passing",
"maxLength": "passing",
"pattern": "passing",
"enum": "passing",
"minimum": 5,
"maximum": 5,
},
]
# Cleanup storage
storage.delete_package(target.resource_names)
def test_storage_types(options):
# Export/Import
source = Package("data/storage/types.json")
storage = source.to_bigquery(force=True, **options)
target = Package.from_bigquery(**options)
# Assert metadata
assert target.get_resource("types").schema == {
"fields": [
{"name": "any", "type": "string"}, # type fallback
{"name": "array", "type": "string"}, # type fallback
{"name": "boolean", "type": "boolean"},
{"name": "date", "type": "date"},
{"name": "date_year", "type": "date"}, # format removal
{"name": "datetime", "type": "datetime"},
{"name": "duration", "type": "string"}, # type fallback
{"name": "geojson", "type": "string"}, # type fallback
{"name": "geopoint", "type": "string"}, # type fallback
{"name": "integer", "type": "integer"},
{"name": "number", "type": "number"},
{"name": "object", "type": "string"}, # type fallback
{"name": "string", "type": "string"},
{"name": "time", "type": "time"},
{"name": "year", "type": "integer"}, # type downgrade
{"name": "yearmonth", "type": "string"}, # type fallback
],
}
# Assert data
assert target.get_resource("types").read_rows() == [
{
"any": "中国人",
"array": '["Mike", "John"]',
"boolean": True,
"date": datetime.date(2015, 1, 1),
"date_year": datetime.date(2015, 1, 1),
# converted into UTC
"datetime": datetime.datetime(2015, 1, 1, 3, 0),
"duration": "P1Y1M",
"geojson": '{"type": "Point", "coordinates": [33, 33.33]}',
"geopoint": "30,70",
"integer": 1,
"number": 7,
"object": '{"chars": 560}',
"string": "english",
"time": datetime.time(3, 0),
"year": 2015,
"yearmonth": "2015-01",
},
]
# Cleanup storage
storage.delete_package(target.resource_names)
def test_storage_integrity(options):
# Export/Import
source = Package("data/storage/integrity.json")
storage = source.to_bigquery(force=True, **options)
target = Package.from_bigquery(**options)
# Assert metadata (main)
assert target.get_resource("integrity_main").schema == {
"fields": [
# added required
{"name": "id", "type": "integer"},
{"name": "parent", "type": "integer"},
{"name": "description", "type": "string"},
],
# primary key removal
# foreign keys removal
}
# Assert metadata (link)
assert target.get_resource("integrity_link").schema == {
"fields": [
{"name": "main_id", "type": "integer"},
{"name": "some_id", "type": "integer"}, # constraint removal
{"name": "description", "type": "string"}, # constraint removal
],
# primary key removal
# foreign keys removal
}
# Assert data (main)
assert target.get_resource("integrity_main").read_rows() == [
{"id": 1, "parent": None, "description": "english"},
{"id": 2, "parent": 1, "description": "中国人"},
]
# Assert data (link)
assert target.get_resource("integrity_link").read_rows() == [
{"main_id": 1, "some_id": 1, "description": "note1"},
{"main_id": 2, "some_id": 2, "description": "note2"},
]
# Cleanup storage
storage.delete_package(target.resource_names)
def test_bigquery_storage_integrity(options):
prefix = options.pop("prefix")
service = options.pop("service")
dialect = BigqueryDialect(table=prefix, **options)
source = Package("data/storage/integrity.json")
storage = source.to_bigquery(service, dialect=dialect)
target = Package.from_bigquery(service, dialect=dialect)
# Assert metadata (main)
assert target.get_resource("integrity_main").schema == {
"fields": [
# added required
{
"name": "id",
"type": "integer"
},
{
"name": "parent",
"type": "integer"
},
{
"name": "description",
"type": "string"
},
],
# primary key removal
# foreign keys removal
}
# Assert metadata (link)
assert target.get_resource("integrity_link").schema == {
"fields": [
{
"name": "main_id",
"type": "integer"
},
{
"name": "some_id",
"type": "integer"
}, # constraint removal
{
"name": "description",
"type": "string"
}, # constraint removal
],
# primary key removal
# foreign keys removal
}
# Assert data (main)
assert target.get_resource("integrity_main").read_rows() == [
{
"id": 1,
"parent": None,
"description": "english"
},
{
"id": 2,
"parent": 1,
"description": "中国人"
},
]
# Assert data (link)
assert target.get_resource("integrity_link").read_rows() == [
{
"main_id": 1,
"some_id": 1,
"description": "note1"
},
{
"main_id": 2,
"some_id": 2,
"description": "note2"
},
]
# Cleanup storage
storage.delete_package(target.resource_names)