Merge pull request #428 from skalish/attribute-from-dataset

TC: Move tc.attribute.from_dataset_all to tc.dataset.attributes

CHANGELOG.md

@@ -4,6 +4,7 @@
   - Added function to get operation from resource ID
   - [#421](https://github.com/Datatamer/tamr-client/pull/421) Added functions for getting and replacing the transformations of a projects via `tc.transformations.get_all()` and `tc.transformations.replace_all()`
    - Added new dataclasses `Transformations` and `InputTransformations` to support these functions
+  - Moved function `tc.attribute.from_dataset_all` to `tc.dataset.attributes`
 
   **NEW FEATURES**
   - [#383](https://github.com/Datatamer/tamr-client/issues/383) Now able to create an Operation from Job resource id

docs/beta/attribute/attribute.rst

@@ -4,7 +4,6 @@ Attribute
 .. autoclass:: tamr_client.Attribute
 
 .. autofunction:: tamr_client.attribute.from_resource_id
-.. autofunction:: tamr_client.attribute.from_dataset_all
 .. autofunction:: tamr_client.attribute.to_json
 .. autofunction:: tamr_client.attribute.create
 .. autofunction:: tamr_client.attribute.update

docs/beta/dataset/dataset.rst

@@ -4,6 +4,7 @@ Dataset
 .. autoclass:: tamr_client.Dataset
 
 .. autofunction:: tamr_client.dataset.from_resource_id
+.. autofunction:: tamr_client.dataset.attributes
 
 Exceptions
 ----------

noxfile.py

@@ -22,6 +22,8 @@ def format(session):
     session.run("poetry", "install", external=True)
     if "--fix" in session.posargs:
         session.run("black", ".")
+    elif "--diff" in session.posargs:
+        session.run("black", ".", "--diff")
     else:
         session.run("black", ".", "--check")
 

tamr_client/attribute/__init__.py

@@ -5,7 +5,6 @@
     AlreadyExists,
     create,
     delete,
-    from_dataset_all,
     from_resource_id,
     NotFound,
     ReservedName,

tamr_client/attribute/_attribute.py

@@ -3,7 +3,7 @@
 """
 from copy import deepcopy
 from dataclasses import replace
-from typing import Optional, Tuple
+from typing import Optional
 
 from tamr_client import response
 from tamr_client._types import Attribute, AttributeType, Dataset, JsonDict, Session, URL
@@ -105,31 +105,6 @@ def _from_json(url: URL, data: JsonDict) -> Attribute:
     )
 
 
-def from_dataset_all(session: Session, dataset: Dataset) -> Tuple[Attribute, ...]:
-    """Get all attributes from a dataset
-
-    Args:
-        dataset: Dataset containing the desired attributes
-
-    Returns:
-        The attributes for the specified dataset
-
-    Raises:
-        requests.HTTPError: If an HTTP error is encountered.
-    """
-    attrs_url = replace(dataset.url, path=dataset.url.path + "/attributes")
-    r = session.get(str(attrs_url))
-    attrs_json = response.successful(r).json()
-
-    attrs = []
-    for attr_json in attrs_json:
-        id = attr_json["name"]
-        attr_url = replace(attrs_url, path=attrs_url.path + f"/{id}")
-        attr = _from_json(attr_url, attr_json)
-        attrs.append(attr)
-    return tuple(attrs)
-
-
 def to_json(attr: Attribute) -> JsonDict:
     """Serialize attribute into JSON
 

tamr_client/dataset/__init__.py

@@ -1,2 +1,2 @@
 from tamr_client.dataset import dataframe, record, unified
-from tamr_client.dataset._dataset import from_resource_id, NotFound
+from tamr_client.dataset._dataset import attributes, from_resource_id, NotFound

tamr_client/dataset/_dataset.py

@@ -2,9 +2,12 @@
 See https://docs.tamr.com/reference/dataset-models
 """
 from copy import deepcopy
+from dataclasses import replace
+from typing import Tuple
 
 from tamr_client import response
-from tamr_client._types import Dataset, Instance, JsonDict, Session, URL
+from tamr_client._types import Attribute, Dataset, Instance, JsonDict, Session, URL
+from tamr_client.attribute import _from_json as _attribute_from_json
 from tamr_client.exception import TamrClientException
 
 
@@ -74,3 +77,28 @@ def _from_json(url: URL, data: JsonDict) -> Dataset:
         description=cp.get("description"),
         key_attribute_names=tuple(cp["keyAttributeNames"]),
     )
+
+
+def attributes(session: Session, dataset: Dataset) -> Tuple[Attribute, ...]:
+    """Get all attributes from a dataset
+
+    Args:
+        dataset: Dataset containing the desired attributes
+
+    Returns:
+        The attributes for the specified dataset
+
+    Raises:
+        requests.HTTPError: If an HTTP error is encountered.
+    """
+    attrs_url = replace(dataset.url, path=dataset.url.path + "/attributes")
+    r = session.get(str(attrs_url))
+    attrs_json = response.successful(r).json()
+
+    attrs = []
+    for attr_json in attrs_json:
+        id = attr_json["name"]
+        attr_url = replace(attrs_url, path=attrs_url.path + f"/{id}")
+        attr = _attribute_from_json(attr_url, attr_json)
+        attrs.append(attr)
+    return tuple(attrs)

tests/tamr_client/attribute/test_attribute.py

@@ -1,7 +1,4 @@
-from dataclasses import replace
-
 import pytest
-import responses
 
 import tamr_client as tc
 from tests.tamr_client import fake, utils
@@ -30,7 +27,7 @@ def test_json():
         assert attr == tc.attribute._from_json(url, tc.attribute.to_json(attr))
 
 
-@responses.activate
+@fake.json
 def test_create():
     s = fake.session()
     dataset = fake.dataset()
@@ -46,10 +43,6 @@ def test_create():
         ]
     )
 
-    attrs_url = tc.URL(path=dataset.url.path + "/attributes")
-    url = replace(attrs_url, path=attrs_url.path + "/attr")
-    attr_json = utils.load_json("attribute.json")
-    responses.add(responses.POST, str(attrs_url), json=attr_json)
     attr = tc.attribute.create(
         s,
         dataset,
@@ -58,59 +51,61 @@ def test_create():
         type=tc.attribute.type.Record(attributes=attrs),
     )
 
-    assert attr == tc.attribute._from_json(url, attr_json)
+    assert attr.name == "attr"
+    assert not attr.is_nullable
+    assert isinstance(attr.type, tc.attribute.type.Record)
+    assert attr.type.attributes == attrs
 
 
-@responses.activate
+@fake.json
 def test_update():
     s = fake.session()
+    attr = fake.attribute()
 
-    url = tc.URL(path="datasets/1/attributes/RowNum")
-    attr_json = utils.load_json("attributes.json")[0]
-    attr = tc.attribute._from_json(url, attr_json)
-
-    updated_attr_json = utils.load_json("updated_attribute.json")
-    responses.add(responses.PUT, str(attr.url), json=updated_attr_json)
     updated_attr = tc.attribute.update(
-        s, attr, description=updated_attr_json["description"]
+        s, attr, description="Synthetic row number updated"
     )
 
-    assert updated_attr == replace(attr, description=updated_attr_json["description"])
+    assert updated_attr.description == "Synthetic row number updated"
 
 
-@responses.activate
+@fake.json
 def test_delete():
     s = fake.session()
+    attr = fake.attribute()
 
-    url = tc.URL(path="datasets/1/attributes/RowNum")
-    attr_json = utils.load_json("attributes.json")[0]
-    attr = tc.attribute._from_json(url, attr_json)
-
-    responses.add(responses.DELETE, str(attr.url), status=204)
     tc.attribute.delete(s, attr)
 
 
-@responses.activate
+@fake.json
 def test_from_resource_id():
     s = fake.session()
     dataset = fake.dataset()
 
-    url = tc.URL(path=dataset.url.path + "/attributes/attr")
-    attr_json = utils.load_json("attribute.json")
-    responses.add(responses.GET, str(url), json=attr_json)
+    attrs = tuple(
+        [
+            tc.SubAttribute(
+                name=str(i),
+                is_nullable=True,
+                type=tc.attribute.type.Array(tc.attribute.type.STRING),
+            )
+            for i in range(4)
+        ]
+    )
+
     attr = tc.attribute.from_resource_id(s, dataset, "attr")
 
-    assert attr == tc.attribute._from_json(url, attr_json)
+    assert attr.name == "attr"
+    assert not attr.is_nullable
+    assert isinstance(attr.type, tc.attribute.type.Record)
+    assert attr.type.attributes == attrs
 
 
-@responses.activate
+@fake.json
 def test_from_resource_id_attribute_not_found():
     s = fake.session()
     dataset = fake.dataset()
 
-    url = replace(dataset.url, path=dataset.url.path + "/attributes/attr")
-
-    responses.add(responses.GET, str(url), status=404)
     with pytest.raises(tc.attribute.NotFound):
         tc.attribute.from_resource_id(s, dataset, "attr")
 
@@ -123,58 +118,28 @@ def test_create_reserved_attribute_name():
         tc.attribute.create(s, dataset, name="clusterId", is_nullable=False)
 
 
-@responses.activate
-def test_from_dataset_all():
-    s = fake.session()
-    dataset = fake.dataset()
-
-    attrs_url = replace(dataset.url, path=dataset.url.path + "/attributes")
-    attrs_json = utils.load_json("attributes.json")
-    responses.add(responses.GET, str(attrs_url), json=attrs_json, status=204)
-
-    attrs = tc.attribute.from_dataset_all(s, dataset)
-
-    row_num = attrs[0]
-    assert row_num.name == "RowNum"
-    assert row_num.type == tc.attribute.type.STRING
-
-    geom = attrs[1]
-    assert geom.name == "geom"
-    assert isinstance(geom.type, tc.attribute.type.Record)
-
-
-@responses.activate
+@fake.json
 def test_create_attribute_exists():
     s = fake.session()
     dataset = fake.dataset()
 
-    url = replace(dataset.url, path=dataset.url.path + "/attributes")
-    responses.add(responses.POST, str(url), status=409)
     with pytest.raises(tc.attribute.AlreadyExists):
         tc.attribute.create(s, dataset, name="attr", is_nullable=False)
 
 
-@responses.activate
+@fake.json
 def test_update_attribute_not_found():
     s = fake.session()
+    attr = fake.attribute()
 
-    url = tc.URL(path="datasets/1/attributes/RowNum")
-    attr_json = utils.load_json("attributes.json")[0]
-    attr = tc.attribute._from_json(url, attr_json)
-
-    responses.add(responses.PUT, str(attr.url), status=404)
     with pytest.raises(tc.attribute.NotFound):
         tc.attribute.update(s, attr)
 
 
-@responses.activate
+@fake.json
 def test_delete_attribute_not_found():
     s = fake.session()
+    attr = fake.attribute()
 
-    url = tc.URL(path="datasets/1/attributes/RowNum")
-    attr_json = utils.load_json("attributes.json")[0]
-    attr = tc.attribute._from_json(url, attr_json)
-
-    responses.add(responses.PUT, str(attr.url), status=404)
     with pytest.raises(tc.attribute.NotFound):
-        attr = tc.attribute.update(s, attr)
+        tc.attribute.delete(s, attr)