def test_pandas_storage_constraints():
# Export/Import
source = Package("data/storage/constraints.json")
storage = source.to_pandas()
target = Package.from_pandas(dataframes=storage.dataframes)
# Assert metadata
assert target.get_resource("constraints").schema == {
"fields": [
{
"name": "required",
"type": "string"
}, # constraint removal
{
"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_pandas_storage_types():
# Export/Import
source = Package("data/storage/types.json")
storage = source.to_pandas()
target = Package.from_pandas(dataframes=storage.dataframes)
# Assert metadata
assert target.get_resource("types").schema == {
"fields": [
{
"name": "any",
"type": "string"
}, # type fallback
{
"name": "array",
"type": "array"
},
{
"name": "boolean",
"type": "boolean"
},
{
"name": "date",
"type": "date"
},
{
"name": "date_year",
"type": "date"
}, # format removal
{
"name": "datetime",
"type": "datetime"
},
{
"name": "duration",
"type": "duration"
},
{
"name": "geojson",
"type": "object"
},
{
"name": "geopoint",
"type": "array"
},
{
"name": "integer",
"type": "integer"
},
{
"name": "number",
"type": "number"
},
{
"name": "object",
"type": "object"
},
{
"name": "string",
"type": "string"
},
{
"name": "time",
"type": "time"
},
{
"name": "year",
"type": "integer"
}, # type downgrade
{
"name": "yearmonth",
"type": "array"
}, # type downgrade
],
}
# 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),
"datetime": datetime.datetime(2015, 1, 1, 3, 0),
"duration": isodate.parse_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, 1],
},
]
# Cleanup storage
storage.delete_package(target.resource_names)
def test_pandas_storage_integrity():
# Export/Import
source = Package("data/storage/integrity.json")
storage = source.to_pandas()
target = Package.from_pandas(dataframes=storage.dataframes)
# Assert metadata (main)
assert target.get_resource("integrity_main").schema == {
"fields": [
# added required
{
"name": "id",
"type": "integer",
"constraints": {
"required": True
}
},
{
"name": "parent",
"type": "number"
}, # type downgrade
{
"name": "description",
"type": "string"
},
],
"primaryKey": ["id"],
# foreign keys removal
}
# Assert metadata (link)
assert target.get_resource("integrity_link").schema == {
"fields": [
# added required
{
"name": "main_id",
"type": "integer",
"constraints": {
"required": True
}
},
# added required; removed unique
{
"name": "some_id",
"type": "integer",
"constraints": {
"required": True
}
},
# removed unique
{
"name": "description",
"type": "string"
},
],
"primaryKey": ["main_id", "some_id"],
# 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)