def _insert(app_name):
"""
This REST endpoint allows an application to append rows to their database by submitting a JSON
object in the `payload` field.
"""
payload = loads(request.params.get("payload"))
insert_payload(api.config['storage_dir'], app_name, payload)
return ""
def create_dummy_dataset(storage_dir):
from sklearn.datasets import load_wine
wine_dataset = load_wine()
wine_dataset.feature_names = [k.replace("/", "_") for k in wine_dataset.feature_names]
register_app(storage_dir, "dummy", {
"wine": {
"description": "Rows sampled from the Wine classification dataset.",
"columns": {x: {"type": "float", "description": x} for x in wine_dataset.feature_names}
}
})
i = randint(0, len(wine_dataset.data)-1)
insert_payload(storage_dir, "dummy", {
"wine": [
{k: v for k, v in zip(wine_dataset.feature_names, wine_dataset.data[i])}
]
})
def test_handle_task(tmpdir):
storage.initialize(tmpdir)
storage.register_app(
tmpdir, "example_app", {
"tableA": {
"description": "This table contains A.",
"columns": {
"some_var": {
"type": "float",
"description": "This column contains 1."
}
}
}
})
storage.insert_payload(
tmpdir, "example_app", {
"tableA": [
{
"some_var": 1.0
},
{
"some_var": 2.0
},
{
"some_var": 3.0
},
{
"some_var": 4.0
},
{
"some_var": 5.0
},
]
})
task = {
"type": "basic",
"epsilon": 1.0,
"featurizer": "SELECT sum(some_var) FROM example_app.tableA",
"aggregator": "mean"
}
result, err = handler.handle_task(tmpdir, task)
assert result[0] == 15.0
def test_insert_payload_1(tmpdir):
storage.initialize(tmpdir)
storage.register_app(
tmpdir, "example_app", {
"tableA": {
"description": "This table contains A.",
"columns": {
"some_var": {
"type": "float",
"description": "This column contains 1."
},
"other_var": {
"type": "float",
"description": "This column contains 2."
}
}
}
})
storage.insert_payload(
tmpdir, "example_app", {
"tableA": [
{
"some_var": 0.0,
"other_var": 1.0
},
{
"some_var": 10.0,
"other_var": 1.0
},
]
})
data = storage.execute_sql(tmpdir, "SELECT * FROM example_app.tableA")
assert len(data) == 2, "Expected two rows of data"
for row in data:
assert "some_var" in row
assert "other_var" in row
assert sum(row["some_var"] for row in data) == 10.0
assert sum(row["other_var"] for row in data) == 2.0
def create_synthetic_dataset(storage_dir):
register_app(storage_dir, "profile", {
"demographics": {
"description": "",
"columns": {
"age": {"type": "float", "description": ""},
"gender": {"type": "float", "description": ""},
"income": {"type": "float", "description": ""},
"city": {"type": "float", "description": ""},
"state": {"type": "float", "description": ""},
"zipcode": {"type": "float", "description": ""},
}
}
})
profile = sample_profile()
insert_payload(storage_dir, "profile", profile)
register_app(storage_dir, "browsing", {
"history": {
"description": "",
"columns": {
"timestamp": {"type": "float", "description": "When the website was opened"},
"domain": {"type": "float", "description": "The domain (i.e. everything up to `.com`, `.net`, etc.)"},
}
},
})
insert_payload(storage_dir, "browsing", sample_browsing(profile))
register_app(storage_dir, "screen_time", {
"events": {
"description": "",
"columns": {
"timestamp": {"type": "float", "description": "When the event occurred."},
"event_type": {"type": "string", "description": "Whether the application was opened or closed."},
"application_name": {"type": "string", "description": "Standardized name for the application."},
}
},
"types": {
"description": "",
"columns": {
"application_name": {"type": "string", "description": "Standardized name for the application."},
"application_type": {"type": "string", "description": "Whether the application is for web browsing, software development, or communicaation."},
}
},
})
insert_payload(storage_dir, "screen_time", sample_screen_time(profile))
def test_insert_payload_2(tmpdir):
storage.initialize(tmpdir)
storage.register_app(
tmpdir, "app1", {
"tableA": {
"description": "This table contains A.",
"columns": {
"some_var": {
"type": "float",
"description": "This column contains 1."
}
}
}
})
storage.insert_payload(
tmpdir, "app1", {
"tableA": [
{
"some_var": 1.0
},
{
"some_var": 2.0
},
{
"some_var": 3.0
},
{
"some_var": 4.0
},
{
"some_var": 5.0
},
]
})
storage.register_app(
tmpdir, "app2", {
"tableA": {
"description": "This table contains A.",
"columns": {
"some_var": {
"type": "float",
"description": "This column contains 1."
}
}
},
"tableB": {
"description": "This table contains B.",
"columns": {
"some_var": {
"type": "float",
"description": "This column contains 1."
},
"other_var": {
"type": "float",
"description": "This column contains 2."
}
}
},
})
storage.insert_payload(
tmpdir, "app2", {
"tableA": [
{
"some_var": 3.0
},
],
"tableB": [
{
"some_var": 0.0,
"other_var": 1.0
},
{
"some_var": 10.0,
"other_var": 1.0
},
]
})
data = storage.execute_sql(tmpdir, "SELECT * FROM app1.tableA")
assert len(data) == 5, "Expected five rows of data"
data = storage.execute_sql(tmpdir, "SELECT * FROM app2.tableA")
assert len(data) == 1, "Expected one row of data"
data = storage.execute_sql(tmpdir, "SELECT * FROM app2.tableB")
assert len(data) == 2, "Expected two rows of data"