Implement OpenRefine data extension API

.github/workflows/build.yml

@@ -149,6 +149,7 @@ jobs:
           context: .
           platforms: linux/amd64,linux/arm64
           push: true
+          pull: true
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           cache-from: type=gha

Makefile

@@ -23,4 +23,4 @@ test:
 typecheck:
 	mypy --strict yente
 
-check: typecheck integration-test unit-test
+check: typecheck test

tests/test_reconcile.py

@@ -12,9 +12,10 @@ def test_reconcile_metadata():
     assert data["identifierSpace"].startswith(url), data
     assert len(data["defaultTypes"]) > 3, data
     assert "suggest" in data, data
+    assert "extend" in data, data
 
 
-def test_reconcile_post():
+def test_reconcile_post_query():
     queries = {"mutti": {"query": "Yevgeny Popov"}}
     resp = client.post("/reconcile/default", data={"queries": json.dumps(queries)})
     assert resp.status_code == 200, resp.text
@@ -23,8 +24,22 @@ def test_reconcile_post():
     assert res[0]["id"] == "Q18634850", res
 
 
+def test_reconcile_post_extend():
+    query = {"ids": ["Q7747"], "properties": [{"id": "name"}, {"id": "birthDate"}]}
+    resp = client.post("/reconcile/default", data={"extend": json.dumps(query)})
+    assert resp.status_code == 200, resp.text
+    data = resp.json()
+    assert len(data["meta"]) == 2
+    assert data["meta"][0]["id"] == "name", data["meta"]
+    assert "Q7747" in data["rows"], data
+    assert "name" in data["rows"]["Q7747"], data
+    names = data["rows"]["Q7747"]["name"]
+    assert len(names) > 0, names
+    assert "putin" in "".join([n["str"] for n in names]).lower(), names
+
+
 def test_reconcile_invalid():
-    queries = {"mutti": {"query": 37473874}}
+    queries = {"mutti": {"type": "Banana"}}
     resp = client.post("/reconcile/default", data={"queries": json.dumps(queries)})
     assert resp.status_code == 400, resp.text
 
@@ -109,3 +124,22 @@ def test_reconcile_suggest_type_prefix_dummy():
     assert "result" in data
     res = data["result"]
     assert len(res) == 0, data
+
+
+def test_reconcile_extend_properties():
+    resp = client.get("/reconcile/default/extend/property?limit=5&type=LegalEntity")
+    assert resp.status_code == 200, resp.text
+    data = resp.json()
+    assert "type" in data
+    assert data["type"] == "LegalEntity", data
+    assert data["limit"] == 5, data
+    props = data["properties"]
+    assert len(props) == 5
+    ids = [p["id"] for p in props]
+    assert "name" in ids
+    assert "country" in ids
+
+
+def test_reconcile_extend_properties_invalid_type():
+    resp = client.get("/reconcile/default/extend/property?limit=5&type=Banana")
+    assert resp.status_code == 400, resp.text

yente/data/common.py

@@ -1,5 +1,5 @@
 from datetime import datetime
-from typing import Dict, List, Union, Optional
+from typing import Any, Dict, List, Union, Optional
 from pydantic import BaseModel, Field
 from nomenklatura.matching.types import MatchingResult, FeatureDocs
 
@@ -83,7 +83,7 @@ class SearchResponse(ResultsResponse):
 class EntityExample(BaseModel):
     id: Optional[str] = Field(None, examples=["my-entity-id"])
     schema_: str = Field(..., examples=["Person"], alias="schema")
-    properties: Dict[str, Union[str, List[str]]] = Field(
+    properties: Dict[str, Union[str, List[Any]]] = Field(
         ..., examples=[{"name": ["John Doe"]}]
     )
 

yente/data/entity.py

@@ -47,7 +47,7 @@ def from_example(cls, example: "EntityExample") -> "Entity":
                 log.warning(
                     "Invalid example property",
                     prop=prop_name,
-                    value=str(values),
+                    value=repr(values),
                 )
                 continue
             obj.add(prop_name, values, cleaned=False, fuzzy=True)

yente/data/freebase.py

@@ -1,4 +1,5 @@
-from typing import List, Optional
+from enum import Enum
+from typing import Any, Dict, List, Optional
 from pydantic import BaseModel, Field
 from pydantic.networks import AnyHttpUrl
 from followthemoney import model
@@ -88,6 +89,53 @@ class FreebasePropertySuggestResponse(FreebaseSuggestResponse):
     result: List[FreebaseProperty]
 
 
+class FreebaseExtendProperty(BaseModel):
+    id: str
+    name: str
+
+
+class FreebaseExtendPropertiesResponse(BaseModel):
+    limit: int
+    type: str
+    properties: List[FreebaseExtendProperty]
+
+
+class FreebaseRenderMethod(str, Enum):
+    raw = "raw"
+    caption = "caption"
+
+
+class FreebaseExtendQueryPropertySettings(BaseModel):
+    limit: int = 0
+    render: FreebaseRenderMethod = FreebaseRenderMethod.caption
+
+
+class FreebaseExtendQueryProperty(BaseModel):
+    id: str
+    settings: FreebaseExtendQueryPropertySettings = (
+        FreebaseExtendQueryPropertySettings()
+    )
+
+
+class FreebaseExtendQuery(BaseModel):
+    ids: List[str]
+    properties: List[FreebaseExtendQueryProperty]
+
+
+class FreebaseExtendResponseMeta(BaseModel):
+    id: str
+    name: str
+
+
+class FreebaseExtendResponseValue(BaseModel):
+    str: str
+
+
+class FreebaseExtendResponse(BaseModel):
+    meta: List[FreebaseExtendResponseMeta]
+    rows: Dict[str, Dict[str, List[FreebaseExtendResponseValue]]]
+
+
 class FreebaseManifestView(BaseModel):
     url: str
 
@@ -109,14 +157,41 @@ class FreebaseManifestSuggest(BaseModel):
     property: FreebaseManifestSuggestType
 
 
+class FreebaseManifestExtendProposeProperties(BaseModel):
+    service_url: AnyHttpUrl
+    service_path: str
+
+
+class FreebaseManifestExtendPropertySettingChoice(BaseModel):
+    id: str
+    name: str
+
+
+class FreebaseManifestExtendPropertySetting(BaseModel):
+    name: str
+    label: str
+    type: str
+    default: Any
+    help_text: str
+    choices: List[FreebaseManifestExtendPropertySettingChoice] = []
+
+
+class FreebaseManifestExtend(BaseModel):
+    propose_properties: FreebaseManifestExtendProposeProperties
+    propose_settings: List[FreebaseManifestExtendPropertySetting]
+
+
 class FreebaseManifest(BaseModel):
     versions: List[str] = Field(..., examples=[["0.2"]])
     name: str = Field(..., examples=[settings.TITLE])
     identifierSpace: AnyHttpUrl
     schemaSpace: AnyHttpUrl
+    documentation: AnyHttpUrl
+    batchSize: int
     view: FreebaseManifestView
     preview: FreebaseManifestPreview
     suggest: FreebaseManifestSuggest
+    extend: FreebaseManifestExtend
     defaultTypes: List[FreebaseType]