Ingest BioPortal PURLs from prefixmaps

src/bioregistry/curation/ingest_prefixmaps_bioportal.py

@@ -0,0 +1,65 @@
+"""Intest manually curated BioPortal PURLs from :mod:`prefixmaps`, curated by Harry Caufield."""
+
+import requests
+import yaml
+import bioregistry
+from bioregistry.external.bioportal import get_bioportal
+
+URL = "https://raw.githubusercontent.com/linkml/prefixmaps/main/src/prefixmaps/data/bioportal.curated.yaml"
+#: A mapping from BioPortal prefixes to lists of URI prefixes to skip
+BLACKLIST = {"BFO": ["http://www.ifomis.org/bfo/1.1/snap#"]}
+
+
+def main():
+    count = 0
+    max_count = 10
+    bioportal = get_bioportal(force_download=False)
+
+    bioportal_to_bioregistry = bioregistry.get_registry_invmap("bioportal")
+
+    res = requests.get(URL)
+    data = yaml.safe_load(res.text)["prefixes"]
+    for bioportal_prefix, uri_prefixes in data.items():
+        if bioportal_prefix not in bioportal:
+            # these are nonsense
+            continue
+
+        bioregistry_prefix = bioportal_to_bioregistry.get(bioportal_prefix)
+        if bioregistry_prefix is None:
+            # these might be relevant, but are not currently in the Bioregistry.
+            # note that there's no quality filter on BioPortal content, and it's not
+            # clear if there's a quality filter on the curation here, so we skip them
+            continue
+
+        resource = bioregistry.get_resource(bioregistry_prefix)
+        bioregistry_uri_prefixes = resource.get_uri_prefixes()
+        if isinstance(uri_prefixes, str):
+            uri_prefixes = [uri_prefixes]
+        for uri_prefix in uri_prefixes:
+            if uri_prefix in BLACKLIST.get(bioportal_prefix, []):
+                continue
+            if uri_prefix.startswith("OBO:"):
+                uri_prefix = "http://purl.obolibrary.org/obo/" + uri_prefix[len("OBO:") :]
+            if uri_prefix in bioregistry_uri_prefixes:
+                continue
+            print(bioregistry_prefix, uri_prefix)
+
+            if count > max_count:
+                continue
+            p = bioregistry.Provider(
+                code="",
+                name="",
+                homepage="",
+                description="",
+                uri_format=uri_prefix + "$1",
+            )
+            if resource.providers is None:
+                resource.providers = []
+            resource.providers.append(p)
+            count += 1
+
+    bioregistry.manager.write_registry()
+
+
+if __name__ == "__main__":
+    main()

src/bioregistry/data/bioregistry.json

@@ -1014,6 +1014,15 @@
       "prefix": "ADO"
     },
     "pattern": "^\\d{7}$",
+    "providers": [
+      {
+        "code": "legacy",
+        "description": "Legacy URI used before switching to OBO PURLs",
+        "homepage": "http://scai.fraunhofer.de",
+        "name": "Legacy",
+        "uri_format": "http://scai.fraunhofer.de/AlzheimerOntology#$1"
+      }
+    ],
     "publications": [
       {
         "doi": "10.1016/j.jalz.2013.02.009",
@@ -2717,10 +2726,10 @@
     "pattern": "^ACTRN\\d+$",
     "providers": [
       {
-        "code": "",
-        "description": "",
-        "homepage": "",
-        "name": "",
+        "code": "redirect",
+        "description": "Australian New Zealand Clinical Trials Registry (redirect)",
+        "homepage": "https://anzctr.org.au",
+        "name": "Australian New Zealand Clinical Trials Registry (redirect)",
         "uri_format": "https://anzctr.org.au/ACTRN$1.aspx"
       }
     ],
@@ -4799,6 +4808,15 @@
       ],
       "uri_format": "http://www.whocc.no/atc_ddd_index/?code=$1"
     },
+    "providers": [
+      {
+        "code": "bioportal.purl",
+        "description": "BioPortal assigned additional PURLs for ATC codes",
+        "homepage": "http://purl.bioontology.org/ontology/ATC",
+        "name": "BioPortal",
+        "uri_format": "http://purl.bioontology.org/ontology/ATC/$1"
+      }
+    ],
     "publications": [
       {
         "pubmed": "7368387",
@@ -9539,6 +9557,15 @@
     },
     "name": "Biomedical Informatics Research Network Lexicon",
     "pattern": "^\\d+$",
+    "providers": [
+      {
+        "code": "bioportal.purl",
+        "description": "Pseudo-BioPortal PURL assigned to BirnLex",
+        "homepage": "http://bioontology.org/projects/ontologies/birnlex",
+        "name": "BioPortal",
+        "uri_format": "http://bioontology.org/projects/ontologies/birnlex#$1"
+      }
+    ],
     "uri_format": "http://uri.neuinfo.org/nif/nifstd/birnlex_$1"
   },
   "biro": {
@@ -26859,10 +26886,17 @@
     "providers": [
       {
         "code": "purl",
-        "description": "Legacy PURLs found in OAE",
+        "description": "Legacy PURLs for DOID, including the redundant DOID_",
         "homepage": "http://purl.org/obo/owl/",
         "name": "Legacy PURL",
         "uri_format": "http://purl.org/obo/owl/DOID#DOID_$1"
+      },
+      {
+        "code": "purl2",
+        "description": "Legacy PURLs for DOID, not including the redundant DOID_",
+        "homepage": "http://purl.org/obo/owl/",
+        "name": "Legacy PURL",
+        "uri_format": "http://purl.org/obo/owl/DOID#$1"
       }
     ],
     "publications": [
@@ -34809,6 +34843,15 @@
       "prefix": "EPIO"
     },
     "pattern": "^\\d{7}$",
+    "providers": [
+      {
+        "code": "legacy",
+        "description": "Legacy internal URL before switching to OBO PURL",
+        "homepage": "https://bio.scai.fraunhofer.de/ontolog",
+        "name": "Legacy",
+        "uri_format": "https://bio.scai.fraunhofer.de/ontology/epilepsy#$1"
+      }
+    ],
     "uri_format": "http://purl.obolibrary.org/obo/EPIO_$1"
   },
   "epo": {
@@ -39673,7 +39716,7 @@
         "description": "Access funder data through a DOI for crossref funders.",
         "homepage": "https://doi.org",
         "name": "DOI",
-        "uri_format": "https://dx.doi.org/10.13039/501100000995"
+        "uri_format": "https://dx.doi.org/10.13039/$1"
       }
     ],
     "synonyms": [
@@ -42715,6 +42758,15 @@
       "fairsharing": "FAIRsharing.175hsz"
     },
     "name": "General Formal Ontology",
+    "providers": [
+      {
+        "code": "alt1",
+        "description": "Alternate identifier using gfo-basic.owl instead of gfo.owl",
+        "homepage": "http://www.onto-med.de/ontologies",
+        "name": "Alternate 1",
+        "uri_format": "http://www.onto-med.de/ontologies/gfo-basic.owl#$1"
+      }
+    ],
     "uri_format": "http://www.onto-med.de/ontologies/gfo.owl#$1"
   },
   "ghr": {
@@ -57210,10 +57262,10 @@
     "pattern": "^jRCT\\w?\\d+$",
     "providers": [
       {
-        "code": "",
-        "description": "",
-        "homepage": "",
-        "name": "",
+        "code": "detail",
+        "description": "Japan Registry of Clinical Trials (Details)",
+        "homepage": "https://jrct.niph.go.jp",
+        "name": "Japan Registry of Clinical Trials (Details)",
         "uri_format": "https://jrct.niph.go.jp/latest-detail/$1"
       }
     ],

tests/test_data.py

@@ -672,6 +672,10 @@ def test_providers(self):
             for provider in resource.providers:
                 with self.subTest(prefix=prefix, code=provider.code):
                     self.assertNotEqual(provider.code, prefix)
+                    self.assertNotEqual(provider.code, "", msg="code should not be an empty string")
+                    self.assertNotEqual(provider.homepage, "", msg="homepage should not be an empty string")
+                    self.assertNotEqual(provider.description, "", msg="desc. should not be an empty string")
+                    self.assertNotEqual(provider.name, "", msg="name should not be an empty string")
                     self.assertNotIn(
                         provider.code,
                         set(self.metaregistry),
@@ -683,7 +687,7 @@ def test_providers(self):
                         provider.code,
                         msg="Provider codes must be lowercase. Ideally, they should be simple and memorable",
                     )
-                    # self.assertIn("$1", provider.uri_format)
+                    self.assertIn("$1", provider.uri_format)
                     self.assertNotIn(
                         "$2",
                         provider.uri_format,