Merge pull request #276 from RDFLib/lawson/predicate-validation

lawson/predicate validation

prez/config.py

@@ -1,15 +1,14 @@
 from os import environ
 from pathlib import Path
-from typing import Optional, List, Tuple
-from typing import Union, Any, Dict
+from typing import Any, Dict, List, Optional, Tuple, Union
 
 import toml
 from pydantic import field_validator
 from pydantic_settings import BaseSettings
-from rdflib import URIRef, DCTERMS, RDFS, SDO
+from rdflib import DCTERMS, RDFS, SDO, URIRef
 from rdflib.namespace import SKOS
 
-from prez.reference_data.prez_ns import REG, EP
+from prez.reference_data.prez_ns import EP, REG
 
 
 class Settings(BaseSettings):
@@ -40,19 +39,25 @@ class Settings(BaseSettings):
     order_lists_by_label: bool = True
     listing_count_limit: int = 100
     search_count_limit: int = 10
-    label_predicates: Optional[List[URIRef]] = [
+    label_predicates: list = [
         SKOS.prefLabel,
         DCTERMS.title,
         RDFS.label,
         SDO.name,
     ]
-    description_predicates: Optional[List[URIRef]] = [
+    description_predicates: list = [
         SKOS.definition,
         DCTERMS.description,
         SDO.description,
     ]
-    provenance_predicates: Optional[List[URIRef]] = [DCTERMS.provenance]
-    other_predicates: Optional[List[URIRef]] = [SDO.color, REG.status]
+    provenance_predicates: list = [DCTERMS.provenance]
+    search_predicates: list = [
+        RDFS.label,
+        SKOS.prefLabel,
+        SDO.name,
+        DCTERMS.title,
+    ]
+    other_predicates: list = [SDO.color, REG.status]
     sparql_repo_type: str = "remote"
     sparql_timeout: int = 30
     log_level: str = "INFO"
@@ -66,12 +71,6 @@ class Settings(BaseSettings):
     prez_contact: Optional[Dict[str, Union[str, Any]]] = None
     disable_prefix_generation: bool = False
     default_language: str = "en"
-    default_search_predicates: Optional[List[URIRef]] = [
-        RDFS.label,
-        SKOS.prefLabel,
-        SDO.name,
-        DCTERMS.title,
-    ]
     local_rdf_dir: str = "rdf"
     endpoint_structure: Optional[Tuple[str, ...]] = ("catalogs", "collections", "items")
     system_endpoints: Optional[List[URIRef]] = [
@@ -114,5 +113,22 @@ def get_version(cls, v):
                     "PREZ_VERSION not set, and cannot find a pyproject.toml to extract the version."
                 )
 
+    @field_validator(
+        "label_predicates",
+        "description_predicates",
+        "provenance_predicates",
+        "search_predicates",
+        "other_predicates",
+    )
+    def validate_predicates(cls, v):
+        try:
+            v = [URIRef(predicate) for predicate in v]
+        except ValueError as e:
+            raise ValueError(
+                "Could not parse predicates. predicates must be valid URIs no prefixes allowed "
+                f"original message: {e}"
+            )
+        return v
+
 
 settings = Settings()

prez/services/query_generation/search.py

@@ -57,7 +57,7 @@ def __init__(
         limit += 1  # increase the limit by one so we know if there are further pages of results.
 
         if not predicates:
-            predicates = settings.default_search_predicates
+            predicates = settings.search_predicates
 
         sr_uri: Var = Var(value="focus_node")
         pred: Var = Var(value="pred")

pyproject.toml

@@ -19,7 +19,7 @@ include = [
 ]
 
 [tool.poetry.dependencies]
-python = "^3.11"
+python = "^3.12"
 uvicorn = {version = "^0.30.0", optional = true }
 httpx = "^0.27.0"
 rdflib = "^7.0.0"

tests/test_predicates.py

@@ -0,0 +1,88 @@
+import pytest
+
+from prez.config import Settings
+
+
+@pytest.mark.parametrize(
+    "label_predicates, error",
+    [
+        [["https://schema.org/name"], None],
+        [["1", "2", "3"], None],
+        [[1], TypeError],
+        ["not a list", ValueError],
+    ],
+)
+def test_label_predicates(label_predicates, error):
+    if error:
+        with pytest.raises(error):
+            assert Settings(label_predicates=label_predicates)
+    else:
+        assert Settings(label_predicates=label_predicates)
+
+
+@pytest.mark.parametrize(
+    "description_predicates, error",
+    [
+        [["https://schema.org/description"], None],
+        [["1", "2", "3"], None],
+        [[1], TypeError],
+        ["not a list", ValueError],
+    ],
+)
+def test_description_predicates(description_predicates, error):
+    if error:
+        with pytest.raises(error):
+            assert Settings(description_predicates=description_predicates)
+    else:
+        assert Settings(description_predicates=description_predicates)
+
+
+@pytest.mark.parametrize(
+    "provenance_predicates, error",
+    [
+        [["https://schema.org/provenance"], None],
+        [["1", "2", "3"], None],
+        [[1], TypeError],
+        ["not a list", ValueError],
+    ],
+)
+def test_provenance_predicates(provenance_predicates, error):
+    if error:
+        with pytest.raises(error):
+            assert Settings(provenance_predicates=provenance_predicates)
+    else:
+        assert Settings(provenance_predicates=provenance_predicates)
+
+
+@pytest.mark.parametrize(
+    "search_predicates, error",
+    [
+        [["https://schema.org/search"], None],
+        [["1", "2", "3"], None],
+        [[1], TypeError],
+        ["not a list", ValueError],
+    ],
+)
+def test_search_predicates(search_predicates, error):
+    if error:
+        with pytest.raises(error):
+            assert Settings(search_predicates=search_predicates)
+    else:
+        assert Settings(search_predicates=search_predicates)
+
+
+@pytest.mark.parametrize(
+    "other_predicates, error",
+    [
+        [["https://schema.org/other"], None],
+        [["1", "2", "3"], None],
+        [[1], TypeError],
+        ["not a list", ValueError],
+    ],
+)
+def test_other_predicates(other_predicates, error):
+    if error:
+        with pytest.raises(error):
+            assert Settings(other_predicates=other_predicates)
+    else:
+        assert Settings(other_predicates=other_predicates)