Fix post-processing steps and add CMSO/ASMO ontologies

ontologies.Makefile

@@ -1637,6 +1637,28 @@ db/minsysont.owl: download/minsysont.owl
 	robot merge -i $< -o $@.tmp.owl && perl -npe 's/\[HSiO4\]/%5BHSiO4%5D/g' $@.tmp.owl > $@ && rm $@.tmp.owl
 
 
+download/cmso.owl: STAMP
+	curl -L -s https://raw.githubusercontent.com/OCDO/cmso/v0.0.1/cmso.owl > $@.tmp
+	sha256sum -b $@.tmp > $@.sha256
+	mv $@.tmp $@
+
+.PRECIOUS: download/cmso.owl
+
+db/cmso.owl: download/cmso.owl
+	cp $< $@
+
+
+download/asmo.owl: STAMP
+	curl -L -s https://raw.githubusercontent.com/OCDO/asmo/v0.3.0/asmo.owl > $@.tmp
+	sha256sum -b $@.tmp > $@.sha256
+	mv $@.tmp $@
+
+.PRECIOUS: download/asmo.owl
+
+db/asmo.owl: download/asmo.owl
+	cp $< $@
+
+
 download/sulo.owl: STAMP
 	curl -L -s https://w3id.org/sulo/sulo.ttl > $@.tmp
 	sha256sum -b $@.tmp > $@.sha256
@@ -1658,4 +1680,4 @@ download/%.owl: STAMP
 db/%.owl: download/%.owl
 	robot merge -i $< -o $@
 
-EXTRA_ONTOLOGIES = swo chiro pcl chemessence ogco ncit fma maxo foodon chebiplus msio chemrof deb matpo panet phenx pride sosa emi npc modl phenio comploinc hba mba dmba dhba pba bero aio reacto xsmo bcio sio icd10who icd11f ordo gard icd10cm omim mondo-ingest oeo envthes wifire taxslim goldterms sdgio kin metpo d3o biovoices omop comet cco occo iof upa go go-lego go-amigo neo bao orcid ror cpont biolink biopax enanomapper mlo ito chemont molgenie cso obiws biopragmatics-reactome reactome-hs reactome-mm efo hcao hpinternational edam chr sweetAll oboe-core oboe-standards lov schema-dot-org prov dtype vaem qudtunit quantitykind cellosaurus cosmo gist gistBFO fhkb dbpendiaont uberoncm co_324 ppeo interpro pfam hgnc.genegroup hgnc sgd gtdb eccode uniprot uniprot.ptm credit rhea swisslipid drugbank drugcentral complexportal wikipathways pathbank kegg.genome drugmechdb rxnorm vccf ontobiotope nando ecso enigma_context cbo ontie pain como ecosim bervo valuesets micront nmdc_schema mixs kgcl fibo bfo2020 bfo2020_core bfo2020_notime bfo2020_time saref4ener saref4bldg hhearvs sdoho pathgo brick minsysont sulo
+EXTRA_ONTOLOGIES = swo chiro pcl chemessence ogco ncit fma maxo foodon chebiplus msio chemrof deb matpo panet phenx pride sosa emi npc modl phenio comploinc hba mba dmba dhba pba bero aio reacto xsmo bcio sio icd10who icd11f ordo gard icd10cm omim mondo-ingest oeo envthes wifire taxslim goldterms sdgio kin metpo d3o biovoices omop comet cco occo iof upa go go-lego go-amigo neo bao orcid ror cpont biolink biopax enanomapper mlo ito chemont molgenie cso obiws biopragmatics-reactome reactome-hs reactome-mm efo hcao hpinternational edam chr sweetAll oboe-core oboe-standards lov schema-dot-org prov dtype vaem qudtunit quantitykind cellosaurus cosmo gist gistBFO fhkb dbpendiaont uberoncm co_324 ppeo interpro pfam hgnc.genegroup hgnc sgd gtdb eccode uniprot uniprot.ptm credit rhea swisslipid drugbank drugcentral complexportal wikipathways pathbank kegg.genome drugmechdb rxnorm vccf ontobiotope nando ecso enigma_context cbo ontie pain como ecosim bervo valuesets micront nmdc_schema mixs kgcl fibo bfo2020 bfo2020_core bfo2020_notime bfo2020_time saref4ener saref4bldg hhearvs sdoho pathgo brick minsysont cmso asmo sulo

src/semsql/builder/builder.py

@@ -133,9 +133,13 @@ def get_postprocessing_steps(
     registry = yaml_loader.load(
         str(registry_path), target_class=registry_schema.Registry
     )
-    # steps = [step.format(ont=ontology, db=db) for step in registry.ontologies.get(ontology, []).post_processing_steps]
-    steps = registry.ontologies.get(ontology, [])
-    return steps
+    ontology_entry = registry.ontologies.get(ontology)
+    if ontology_entry is None:
+        return []
+    return [
+        step.format(ont=ontology, ontology=ontology, db=db)
+        for step in ontology_entry.post_processing_steps
+    ]
 
 
 def compile_registry(registry_path: str, local_prefix_file: TextIO = None) -> str:

src/semsql/builder/cli.py

@@ -1,6 +1,7 @@
 import logging
 import subprocess
 from itertools import chain, combinations
+from pathlib import Path
 
 import click
 from linkml_runtime import SchemaView
@@ -24,6 +25,23 @@ def powerset(iterable):
     return chain.from_iterable(combinations(s, r) for r in range(len(s) + 1))
 
 
+def ontology_from_target_path(path: str):
+    """
+    Extract ontology ID from a standard db target path.
+
+    >>> ontology_from_target_path("db/cmso.db")
+    'cmso'
+    >>> ontology_from_target_path("db/reactome-mm.db")
+    'reactome-mm'
+    >>> ontology_from_target_path("/tmp/cmso.db") is None
+    True
+    """
+    target_path = Path(path)
+    if target_path.suffix != ".db" or target_path.parent.name != "db":
+        return None
+    return target_path.stem
+
+
 @click.group()
 @click.option("-v", "--verbose", count=True)
 @click.option("-q", "--quiet")
@@ -63,12 +81,8 @@ def make(path, docker, **kwargs):
     else:
         docker_config = None
     builder.make(path, docker_config=docker_config, **kwargs)
-    # check if path is db/{foo}.db using regular expression
-    import re
-
-    matches = re.match(r"db/(\w+).db", path)
-    if matches:
-        ontology = matches.group(1)
+    ontology = ontology_from_target_path(path)
+    if ontology:
         steps = builder.get_postprocessing_steps(ontology, path)
         for step in steps:
             print(f"RUNNING: {step}")

src/semsql/builder/prefixes/prefixes.csv

@@ -252,6 +252,10 @@ sh,http://www.w3.org/ns/shacl#
 vcard,http://www.w3.org/2006/vcard/ns#
 MinSysOnt,http://www.semanticweb.org/hbabaie/ontologies/2021/5/MinSysOnt#
 CMO.minerals,http://www.semanticweb.org/Davarpanah-Babaie/Ontologies/2022/CMO#
+CMSO,http://purls.helmholtz-metadaten.de/cmso/
+CDCO,http://purls.helmholtz-metadaten.de/cdos/cdco/
+ASMO,https://purls.helmholtz-metadaten.de/asmo/
+MDO.calculation,https://w3id.org/mdo/calculation/
 sulo,https://w3id.org/sulo/
 RBO,http://purl.obolibrary.org/obo/RBO_
 RBO,http://purl.obolibrary.org/obo/RBO_

src/semsql/builder/prefixes/prefixes_local.csv

@@ -189,4 +189,8 @@ sh,http://www.w3.org/ns/shacl#
 vcard,http://www.w3.org/2006/vcard/ns#
 MinSysOnt,http://www.semanticweb.org/hbabaie/ontologies/2021/5/MinSysOnt#
 CMO.minerals,http://www.semanticweb.org/Davarpanah-Babaie/Ontologies/2022/CMO#
+CMSO,http://purls.helmholtz-metadaten.de/cmso/
+CDCO,http://purls.helmholtz-metadaten.de/cdos/cdco/
+ASMO,https://purls.helmholtz-metadaten.de/asmo/
+MDO.calculation,https://w3id.org/mdo/calculation/
 sulo,https://w3id.org/sulo/

src/semsql/builder/registry/ontologies.yaml

@@ -764,6 +764,18 @@ ontologies:
     prefixmap:
       MinSysOnt: "http://www.semanticweb.org/hbabaie/ontologies/2021/5/MinSysOnt#"
       CMO.minerals: "http://www.semanticweb.org/Davarpanah-Babaie/Ontologies/2022/CMO#"
+  cmso:
+    description: Computational Materials Sample Ontology
+    url: https://raw.githubusercontent.com/OCDO/cmso/v0.0.1/cmso.owl
+    prefixmap:
+      CMSO: http://purls.helmholtz-metadaten.de/cmso/
+      CDCO: http://purls.helmholtz-metadaten.de/cdos/cdco/
+  asmo:
+    description: Atomistic Simulation Methods Ontology
+    url: https://raw.githubusercontent.com/OCDO/asmo/v0.3.0/asmo.owl
+    prefixmap:
+      ASMO: https://purls.helmholtz-metadaten.de/asmo/
+      MDO.calculation: https://w3id.org/mdo/calculation/
   sulo:
     description: Simplified Upper Level Ontology
     url: https://w3id.org/sulo/sulo.ttl

tests/test_builder/test_builder.py

@@ -1,4 +1,6 @@
 import os
+import tempfile
+import textwrap
 import unittest
 
 from semsql.builder import builder
@@ -20,3 +22,54 @@ def setUp(self) -> None:
     def test_builder(self):
         mkfile = builder.compile_registry(self.registry)
         print(mkfile)
+
+    def test_get_postprocessing_steps(self):
+        registry_text = textwrap.dedent(
+            """
+            id: test-registry
+            license: CC0
+            ontologies:
+              reactome-mm:
+                url: https://example.org/reactome-mm.owl
+                post_processing_steps:
+                  - "sqlite3 {db} < views/reactome.sql"
+                  - "echo {ont}"
+            """
+        ).strip()
+        with tempfile.TemporaryDirectory() as tmpdir:
+            registry_path = os.path.join(tmpdir, "registry.yaml")
+            with open(registry_path, "w") as stream:
+                stream.write(registry_text)
+            steps = builder.get_postprocessing_steps(
+                "reactome-mm",
+                "db/reactome-mm.db",
+                registry_path=registry_path,
+            )
+            self.assertEqual(
+                [
+                    "sqlite3 db/reactome-mm.db < views/reactome.sql",
+                    "echo reactome-mm",
+                ],
+                steps,
+            )
+
+    def test_get_postprocessing_steps_missing_ontology(self):
+        registry_text = textwrap.dedent(
+            """
+            id: test-registry
+            license: CC0
+            ontologies:
+              cmso:
+                url: https://example.org/cmso.owl
+            """
+        ).strip()
+        with tempfile.TemporaryDirectory() as tmpdir:
+            registry_path = os.path.join(tmpdir, "registry.yaml")
+            with open(registry_path, "w") as stream:
+                stream.write(registry_text)
+            steps = builder.get_postprocessing_steps(
+                "asmo",
+                "db/asmo.db",
+                registry_path=registry_path,
+            )
+            self.assertEqual([], steps)

tests/test_builder/test_cli.py

@@ -4,7 +4,7 @@
 
 from click.testing import CliRunner
 
-from semsql.builder.cli import main
+from semsql.builder.cli import main, ontology_from_target_path
 
 cwd = os.path.abspath(os.path.dirname(__file__))
 DB_DIR = os.path.join(cwd, "../inputs")
@@ -35,6 +35,17 @@ def test_main_help(self):
             0, self.runner.invoke(main, ["view2table", "--help"]).exit_code
         )
 
+    def test_ontology_from_target_path(self):
+        cases = [
+            ("db/cmso.db", "cmso"),
+            ("db/reactome-mm.db", "reactome-mm"),
+            ("./db/asmo.db", "asmo"),
+            ("/tmp/asmo.db", None),
+            ("db/asmo.owl", None),
+        ]
+        for path, expected in cases:
+            self.assertEqual(expected, ontology_from_target_path(path))
+
     def test_view2table(self):
         cases = [
             (["--no-index"], "CREATE TABLE deprecated_node AS SELECT"),