Further consolidate CURIE/URI parsing (#257)

- Remove custom parsing logic from relation and property parsing,
replace with Reference.from_curie_or_uri
- Add bioontologies URI upgrading into Reference.from_curie_or_uri
- Reuse bioregistry's identifier standardization code where possible

src/pyobo/identifier_utils.py

@@ -6,6 +6,7 @@
 from functools import wraps
 from typing import ClassVar
 
+import bioontologies.upgrade
 import bioregistry
 from curies import Reference, ReferenceTuple
 
@@ -74,7 +75,7 @@ def normalize_curie(
     strict: bool = True,
     ontology_prefix: str | None = None,
     node: Reference | None = None,
-) -> tuple[str, str] | tuple[None, None]:
+) -> ReferenceTuple | tuple[None, None]:
     """Parse a string that looks like a CURIE.
 
     :param curie: A compact uniform resource identifier (CURIE)
@@ -98,6 +99,9 @@ def normalize_curie(
     if curie_has_blacklisted_suffix(curie):
         return None, None
 
+    if reference_t := bioontologies.upgrade.upgrade(curie):
+        return reference_t
+
     if curie.startswith("http:") or curie.startswith("https:"):
         if reference := parse_iri(curie):
             return reference.pair
@@ -120,7 +124,7 @@ def normalize_curie(
 
     norm_node_prefix = bioregistry.normalize_prefix(prefix)
     if norm_node_prefix:
-        return norm_node_prefix, identifier
+        return ReferenceTuple(norm_node_prefix, identifier)
     elif strict:
         raise MissingPrefixError(curie=curie, ontology_prefix=ontology_prefix, node=node)
     else:

src/pyobo/reader.py

@@ -9,15 +9,13 @@
 from pathlib import Path
 from typing import Any
 
-import bioontologies.upgrade
 import bioregistry
 import networkx as nx
 from curies import ReferenceTuple
 from more_itertools import pairwise
 from tqdm.auto import tqdm
 
 from .constants import DATE_FORMAT, PROVENANCE_PREFIXES
-from .identifier_utils import normalize_curie
 from .registries import curie_has_blacklisted_prefix, curie_is_blacklisted, remap_prefix
 from .struct import (
     Obo,
@@ -42,8 +40,6 @@
 
 logger = logging.getLogger(__name__)
 
-RELATION_REMAPPINGS: Mapping[str, ReferenceTuple] = bioontologies.upgrade.load()
-
 
 def from_obo_path(
     path: str | Path,
@@ -363,7 +359,9 @@ def iterate_graph_synonym_typedefs(
             if reference is not None:
                 yield SynonymTypeDef(reference=reference)
             elif strict:
-                raise ValueError(f"Could not parse {sid}")
+                raise ValueError(
+                    f"[{ontology_prefix}] could not parse synonym type definition: {sid}"
+                )
             else:
                 continue
 
@@ -657,15 +655,7 @@ def _get_prop(
         if prop.startswith(sw):
             identifier = prop.removeprefix(sw)
             return default_reference(ontology_prefix, identifier)
-    if prop.startswith("http"):
-        # TODO upstream this into an omni-parser for references?
-        _pref, _id = bioregistry.parse_iri(prop)
-        if _pref and _id:
-            return Reference(prefix=_pref, identifier=_id)
-        else:
-            logger.warning("[%s] unable to handle property: %s", node.curie, prop)
-            return None
-    elif ":" not in prop:
+    if ":" not in prop:
         return default_reference(ontology_prefix, prop)
     else:
         return Reference.from_curie_or_uri(
@@ -713,23 +703,21 @@ def iterate_node_relationships(
     """Extract relationships from a :mod:`obonet` node's data."""
     for s in data.get("relationship", []):
         relation_curie, target_curie = s.split(" ")
-        relation_prefix: str | None
-        relation_identifier: str | None
-        if relation_curie in RELATION_REMAPPINGS:
-            relation_prefix, relation_identifier = RELATION_REMAPPINGS[relation_curie]
-        else:
-            relation_prefix, relation_identifier = normalize_curie(
+
+        if ":" in relation_curie:
+            relation = Reference.from_curie_or_uri(
                 relation_curie, strict=strict, ontology_prefix=ontology_prefix, node=node
             )
-        if relation_prefix is not None and relation_identifier is not None:
-            relation = Reference(prefix=relation_prefix, identifier=relation_identifier)
         else:
             relation = default_reference(ontology_prefix, relation_curie)
             logger.debug(
                 "unhandled relation: %s. Parsing as default relation: %s",
                 relation_curie,
                 relation.curie,
             )
+        if relation is None:
+            logger.warning("[%s] could not parse relation %s", node.curie, relation_curie)
+            continue
 
         target = Reference.from_curie_or_uri(
             target_curie, strict=strict, ontology_prefix=ontology_prefix, node=node

tests/test_reader.py

@@ -8,6 +8,7 @@
 from obonet import read_obo
 
 from pyobo import Obo, Reference, Term
+from pyobo.identifier_utils import UnparsableIRIError
 from pyobo.reader import from_obonet, get_first_nonescaped_quote
 from pyobo.struct import default_reference
 from pyobo.struct.struct import DEFAULT_SYNONYM_TYPE
@@ -416,13 +417,29 @@ def test_property_unparsable_object(self) -> None:
 
     def test_property_literal_url_unregistered(self) -> None:
         """Test using a full OBO PURL as the property."""
-        ontology = _read("""\
+        with self.assertRaises(UnparsableIRIError):
+            _read(
+                """\
+                ontology: chebi
+
+                [Term]
+                id: CHEBI:1234
+                property_value: https://example.com/nope/nope CHEBI:5678
+                """,
+                strict=True,
+            )
+
+        ontology = _read(
+            """\
             ontology: chebi
 
             [Term]
             id: CHEBI:1234
             property_value: https://example.com/nope/nope CHEBI:5678
-        """)
+            """,
+            strict=False,
+        )
+
         term = self.get_only_term(ontology)
         self.assertEqual(0, len(list(term.annotations_literal)))
         self.assertEqual(0, len(list(term.annotations_object)))