Begin using curies.Converter in more places

.gitignore

@@ -158,3 +158,4 @@ schema/sssom_datamodel.py
 sssom/internal_context.py
 sssom/sssom_datamodel.py
 *.rej
+poetry.lock

pyproject.toml

@@ -6,7 +6,7 @@ authors = [
     "Chris Mungall <cjmungall@lbl.gov>",
     "Nicolas Matentzoglu <nicolas.matentzoglu@gmail.com>",
     "Harshad Hegde <hhegde@lbl.gov>"
-    ]
+]
 license = "MIT"
 readme = "README.md"
 
@@ -19,6 +19,7 @@ bioregistry = ">=0.9.43"
 deprecation = ">=2.1.0"
 linkml-runtime = ">=1.5.3"
 networkx = ">=3.1"
+curies = ">=0.5.7"
 pandas = ">=2.0.2"
 pansql = "^0.0.1"
 pyyaml = ">=6.0"

src/sssom/cli.py

@@ -356,11 +356,8 @@ def sparql(
         endpoint.limit = limit
     if object_labels is not None:
         endpoint.include_object_labels = object_labels
-    if prefix is not None:
-        if endpoint.prefix_map is None:
-            endpoint.prefix_map = {}
-        for k, v in prefix:
-            endpoint.prefix_map[k] = v
+    for k, v in prefix or []:
+        endpoint.prefix_map[k] = v
     msdf = query_mappings(endpoint)
     write_table(msdf, output)
 

src/sssom/context.py

@@ -44,7 +44,7 @@ def get_extended_prefix_map():
     :return: Prefix map.
     """
     converter = Converter.from_extended_prefix_map(EXTENDED_PREFIX_MAP)
-    return converter.prefix_map
+    return {record.prefix: record.uri_prefix for record in converter.records}
 
 
 def get_built_in_prefix_map() -> PrefixMap:
@@ -86,9 +86,23 @@ def add_built_in_prefixes_to_prefix_map(
                 raise ValueError(
                     f"Built-in prefix {k} is specified ({prefix_map[k]}) but differs from default ({builtinmap[k]})"
                 )
+    clean_dc(prefix_map)
     return prefix_map
 
 
+def clean_dc(prefix_map):
+    """Remove a common issue with prefix maps in-place for DC/DCTERMS.
+
+    This happens when both DC and DCTERMS are set to the new URI prefix
+    for DCTERMS. DC has historically been used to point to DC elements.
+    If this happens, then the prefix map doesn't follow the bijectivity
+    rule. This function deletes the DC annotation in favor of the keeping
+    the DCTERMS one. This should be fixed upstream in the SSSOM Schema.
+    """
+    if "dc" in prefix_map and "dcterms" in prefix_map and prefix_map["dc"] == prefix_map["dcterms"]:
+        del prefix_map["dc"]
+
+
 def get_default_metadata() -> Metadata:
     """Get @context property value from the sssom_context variable in the auto-generated 'internal_context.py' file.
 
@@ -108,8 +122,12 @@ def get_default_metadata() -> Metadata:
             if "@id" in v and "@prefix" in v:
                 if v["@prefix"]:
                     prefix_map[key] = v["@id"]
+    del prefix_map["@vocab"]
 
     prefix_map.update({(k, v) for k, v in contxt_external.items() if k not in prefix_map})
+    clean_dc(prefix_map)
+    # Tests if the prefix map is a valid bijective map
+    Converter.from_prefix_map(prefix_map)
 
     metadata = Metadata(prefix_map=prefix_map, metadata=metadata_dict)
     metadata.metadata["mapping_set_id"] = DEFAULT_MAPPING_SET_ID

src/sssom/parsers.py

@@ -15,6 +15,7 @@
 import pandas as pd
 import requests
 import yaml
+from curies import Converter
 from deprecation import deprecated
 from linkml_runtime.loaders.json_loader import JSONLoader
 from pandas.errors import EmptyDataError
@@ -64,11 +65,10 @@
     SSSOM_DEFAULT_RDF_SERIALISATION,
     URI_SSSOM_MAPPINGS,
     MappingSetDataFrame,
-    NoCURIEException,
-    curie_from_uri,
     get_file_extension,
     is_multivalued_slot,
     raise_for_bad_path,
+    safe_compress,
     to_mapping_set_dataframe,
 )
 
@@ -506,6 +506,7 @@ def from_sssom_rdf(
     :return: MappingSetDataFrame object
     """
     prefix_map = _ensure_prefix_map(prefix_map)
+    converter = Converter.from_prefix_map(prefix_map)
 
     ms = _init_mapping_set(meta)
     mlist: List[Mapping] = []
@@ -515,7 +516,7 @@ def from_sssom_rdf(
         for _s, p, o in g.triples((ox, None, None)):
             if isinstance(p, URIRef):
                 try:
-                    p_id = curie_from_uri(p, prefix_map)
+                    p_id = safe_compress(p, converter)
                     k = None
 
                     if p_id.startswith("sssom:"):
@@ -529,14 +530,14 @@ def from_sssom_rdf(
 
                     if isinstance(o, URIRef):
                         v: Any
-                        v = curie_from_uri(o, prefix_map)
+                        v = safe_compress(o, converter)
                     else:
                         v = o.toPython()
                     if k:
                         v = _address_multivalued_slot(k, v)
                         mdict[k] = v
 
-                except NoCURIEException as e:
+                except ValueError as e:
                     logging.warning(e)
         if mdict:
             m = _prepare_mapping(Mapping(**mdict))
@@ -596,6 +597,7 @@ def from_alignment_minidom(
     """
     # FIXME: should be prefix_map =  _check_prefix_map(prefix_map)
     _ensure_prefix_map(prefix_map)
+    converter = Converter.from_prefix_map(prefix_map)
     ms = _init_mapping_set(meta)
     mlist: List[Mapping] = []
     # bad_attrs = {}
@@ -612,7 +614,7 @@ def from_alignment_minidom(
                     cell = e.getElementsByTagName("Cell")
                     for c_node in cell:
                         mdict = _cell_element_values(
-                            c_node, prefix_map, mapping_predicates=mapping_predicates
+                            c_node, converter, mapping_predicates=mapping_predicates
                         )
                         if mdict:
                             m = _prepare_mapping(mdict)
@@ -665,6 +667,7 @@ def from_obographs(
     :return: An SSSOM data frame (MappingSetDataFrame)
     """
     _ensure_prefix_map(prefix_map)
+    converter = Converter.from_prefix_map(prefix_map)
     ms = _init_mapping_set(meta)
     mlist: List[Mapping] = []
     # bad_attrs = {}
@@ -705,13 +708,13 @@ def from_obographs(
                                 xref_id = xref["val"]
                                 mdict: Dict[str, Any] = {}
                                 try:
-                                    mdict[SUBJECT_ID] = curie_from_uri(nid, prefix_map)
-                                    mdict[OBJECT_ID] = curie_from_uri(xref_id, prefix_map)
+                                    mdict[SUBJECT_ID] = safe_compress(nid, converter)
+                                    mdict[OBJECT_ID] = safe_compress(xref_id, converter)
                                     mdict[SUBJECT_LABEL] = label
                                     mdict[PREDICATE_ID] = "oboInOwl:hasDbXref"
                                     mdict[MAPPING_JUSTIFICATION] = MAPPING_JUSTIFICATION_UNSPECIFIED
                                     mlist.append(Mapping(**mdict))
-                                except NoCURIEException as e:
+                                except ValueError as e:
                                     # FIXME this will cause all sorts of ragged Mappings
                                     logging.warning(e)
                         if "basicPropertyValues" in n["meta"]:
@@ -721,15 +724,15 @@ def from_obographs(
                                     xref_id = value["val"]
                                     mdict = {}
                                     try:
-                                        mdict[SUBJECT_ID] = curie_from_uri(nid, prefix_map)
-                                        mdict[OBJECT_ID] = curie_from_uri(xref_id, prefix_map)
+                                        mdict[SUBJECT_ID] = safe_compress(nid, converter)
+                                        mdict[OBJECT_ID] = safe_compress(xref_id, converter)
                                         mdict[SUBJECT_LABEL] = label
-                                        mdict[PREDICATE_ID] = curie_from_uri(pred, prefix_map)
+                                        mdict[PREDICATE_ID] = safe_compress(pred, converter)
                                         mdict[
                                             MAPPING_JUSTIFICATION
                                         ] = MAPPING_JUSTIFICATION_UNSPECIFIED
                                         mlist.append(Mapping(**mdict))
-                                    except NoCURIEException as e:
+                                    except ValueError as e:
                                         # FIXME this will cause ragged mappings
                                         logging.warning(e)
             if "edges" in g:
@@ -739,15 +742,15 @@ def from_obographs(
                     predicate_id = _get_obographs_predicate_id(edge["pred"])
                     object_id = edge["obj"]
                     if predicate_id in mapping_predicates:
-                        mdict[SUBJECT_ID] = curie_from_uri(subject_id, prefix_map)
-                        mdict[OBJECT_ID] = curie_from_uri(object_id, prefix_map)
+                        mdict[SUBJECT_ID] = safe_compress(subject_id, converter)
+                        mdict[OBJECT_ID] = safe_compress(object_id, converter)
                         mdict[SUBJECT_LABEL] = (
                             labels[subject_id] if subject_id in labels.keys() else ""
                         )
                         mdict[OBJECT_LABEL] = (
                             labels[object_id] if object_id in labels.keys() else ""
                         )
-                        mdict[PREDICATE_ID] = curie_from_uri(predicate_id, prefix_map)
+                        mdict[PREDICATE_ID] = safe_compress(predicate_id, converter)
                         mdict[MAPPING_JUSTIFICATION] = MAPPING_JUSTIFICATION_UNSPECIFIED
                         mlist.append(Mapping(**mdict))
             if "equivalentNodesSets" in g and OWL_EQUIV_CLASS_URI in mapping_predicates:
@@ -757,10 +760,10 @@ def from_obographs(
                             for ec2 in equivalents["nodeIds"]:
                                 if ec1 != ec2:
                                     mdict = {}
-                                    mdict[SUBJECT_ID] = curie_from_uri(ec1, prefix_map)
-                                    mdict[OBJECT_ID] = curie_from_uri(ec2, prefix_map)
-                                    mdict[PREDICATE_ID] = curie_from_uri(
-                                        OWL_EQUIV_CLASS_URI, prefix_map
+                                    mdict[SUBJECT_ID] = safe_compress(ec1, converter)
+                                    mdict[OBJECT_ID] = safe_compress(ec2, converter)
+                                    mdict[PREDICATE_ID] = safe_compress(
+                                        OWL_EQUIV_CLASS_URI, converter
                                     )
                                     mdict[MAPPING_JUSTIFICATION] = MAPPING_JUSTIFICATION_UNSPECIFIED
                                     mdict[SUBJECT_LABEL] = (
@@ -868,19 +871,15 @@ def _set_metadata_in_mapping_set(
                 mapping_set[k] = v
 
 
-def _cell_element_values(cell_node, prefix_map: PrefixMap, mapping_predicates) -> Optional[Mapping]:
+def _cell_element_values(cell_node, converter: Converter, mapping_predicates) -> Optional[Mapping]:
     mdict: Dict[str, Any] = {}
     for child in cell_node.childNodes:
         if child.nodeType == Node.ELEMENT_NODE:
             try:
                 if child.nodeName == "entity1":
-                    mdict[SUBJECT_ID] = curie_from_uri(
-                        child.getAttribute("rdf:resource"), prefix_map
-                    )
+                    mdict[SUBJECT_ID] = safe_compress(child.getAttribute("rdf:resource"), converter)
                 elif child.nodeName == "entity2":
-                    mdict[OBJECT_ID] = curie_from_uri(
-                        child.getAttribute("rdf:resource"), prefix_map
-                    )
+                    mdict[OBJECT_ID] = safe_compress(child.getAttribute("rdf:resource"), converter)
                 elif child.nodeName == "measure":
                     mdict[CONFIDENCE] = child.firstChild.nodeValue
                 elif child.nodeName == "relation":
@@ -902,7 +901,7 @@ def _cell_element_values(cell_node, prefix_map: PrefixMap, mapping_predicates) -
                         logging.warning(f"{relation} not a recognised relation type.")
                 else:
                     logging.warning(f"Unsupported alignment api element: {child.nodeName}")
-            except NoCURIEException as e:
+            except ValueError as e:
                 logging.warning(e)
 
     mdict[MAPPING_JUSTIFICATION] = MAPPING_JUSTIFICATION_UNSPECIFIED

src/sssom/rdf_util.py

@@ -3,10 +3,9 @@
 import logging
 from typing import Any, Dict, List, Optional
 
+from curies import Converter
 from linkml_runtime.utils.metamodelcore import URIorCURIE
 from rdflib import Graph, URIRef
-
-# from .sssom_datamodel import EntityReference, Mapping
 from sssom_schema import EntityReference, Mapping
 
 from .parsers import to_mapping_set_document
@@ -24,17 +23,12 @@ def rewire_graph(
     precedence: Optional[List[str]] = None,
 ) -> int:
     """Rewire an RDF Graph replacing using equivalence mappings."""
-    pm = mset.prefix_map
     mdoc = to_mapping_set_document(mset)
-    rewire_map: Dict[URIorCURIE, URIorCURIE] = {}
-
-    def expand_curie(curie: str) -> URIRef:
-        """Expand CURIE into URIRef."""
-        pfx, local = curie.split(":")
-        return URIRef(f"{pm[pfx]}{local}")
-
     if mdoc.mapping_set.mappings is None:
         raise TypeError
+
+    converter = Converter.from_prefix_map(mdoc.prefix_map)
+    rewire_map: Dict[URIorCURIE, URIorCURIE] = {}
     for m in mdoc.mapping_set.mappings:
         if not isinstance(m, Mapping):
             continue
@@ -49,8 +43,8 @@ def expand_curie(curie: str) -> URIRef:
                 curr_tgt = rewire_map[src]
                 logging.info(f"Ambiguous: {src} -> {tgt} vs {curr_tgt}")
                 if precedence:
-                    curr_pfx, _ = curr_tgt.split(":")
-                    tgt_pfx, _ = tgt.split(":")
+                    curr_pfx, _ = converter.parse_curie(curr_tgt)
+                    tgt_pfx, _ = converter.parse_curie(tgt)
                     if tgt_pfx in precedence:
                         if curr_pfx not in precedence or precedence.index(
                             tgt_pfx
@@ -63,7 +57,8 @@ def expand_curie(curie: str) -> URIRef:
                 rewire_map[src] = tgt
 
     uri_ref_rewire_map: Dict[URIRef, URIRef] = {
-        expand_curie(k): expand_curie(v) for k, v in rewire_map.items()
+        URIRef(converter.expand_strict(k)): URIRef(converter.expand_strict(v))
+        for k, v in rewire_map.items()
     }
 
     def rewire_node(n: Any):