test: empty _OWED_BOTH_ROOTS — all eleven gates now sweep both roots (#689)

src/communitymech/paths.py

@@ -56,6 +56,31 @@ def default_record_roots() -> list[Path]:
     return [KB_COMMUNITIES, DATA_ISOLATES]
 
 
+def record_files() -> list[Path]:
+    """Every `MicrobialCommunity` record file, across every root (#689).
+
+    `default_record_roots()` answers "which directories"; almost every caller
+    then writes the same `for root in ...: root.glob("*.yaml")` line. Eleven test
+    modules skipped that step entirely and globbed `kb/communities` alone --
+    correct-looking, and blind to 4 records carrying 66 snippets, 3 interactions
+    and 3 GTDB groundings.
+
+    One function so the loop is written once, and so a module that wants both
+    roots cannot get half of them by writing slightly less code than the module
+    next to it.
+    """
+    return [path for root in default_record_roots() for path in sorted(root.glob("*.yaml"))]
+
+
+def record_path(name: str) -> Path | None:
+    """Resolve a record filename against every root, for named fixtures."""
+    for root in default_record_roots():
+        candidate = root / name
+        if candidate.is_file():
+            return candidate
+    return None
+
+
 def taxon_descriptor_roots() -> list[Path]:
     """Every directory whose records can carry a `TaxonDescriptor` (#656).
 

tests/test_community_level_connectivity_credit.py

@@ -35,7 +35,13 @@
 import pytest
 import yaml
 
+from communitymech.paths import record_files
+
 REPO = pathlib.Path(__file__).parent.parent
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 COMMUNITIES = REPO / "kb/communities"
 
 
@@ -54,7 +60,7 @@ def _names(block: dict) -> set[str]:
 def _survey() -> dict[str, int]:
     records = with_cl = mixed = cl_only = 0
     taxa = solely = 0
-    for path in sorted(COMMUNITIES.glob("*.yaml")):
+    for path in record_files():
         document = yaml.safe_load(path.read_text()) or {}
         records += 1
         interactions = [

tests/test_cultivation_units_are_constrained.py

@@ -38,9 +38,14 @@ class attributes, which is why `test_the_generated_datamodel_carries_them`
 import pytest
 import yaml
 
+from communitymech.paths import record_files
+
 REPO = pathlib.Path(__file__).parent.parent
 SCHEMA = REPO / "src/communitymech/schema/communitymech.yaml"
-COMMUNITIES = REPO / "kb/communities"
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 
 # slot -> the enum it must be ranged to.
 CONSTRAINED = {
@@ -80,7 +85,7 @@ def test_every_value_in_the_corpus_is_permissible(schema):
     """
     enums = schema["enums"]
     offenders = []
-    for path in sorted(COMMUNITIES.glob("*.yaml")):
+    for path in record_files():
         for entry in (yaml.safe_load(path.read_text()) or {}).get("cultivation_setup") or []:
             for slot, enum_name in CONSTRAINED.items():
                 value = entry.get(slot)
@@ -95,7 +100,7 @@ def test_the_corpus_actually_uses_these_slots():
     """Guard: at zero populated slots the test above passes on nothing."""
     populated = sum(
         1
-        for path in COMMUNITIES.glob("*.yaml")
+        for path in record_files()
         for entry in (yaml.safe_load(path.read_text()) or {}).get("cultivation_setup") or []
         for slot in CONSTRAINED
         if entry.get(slot) is not None
@@ -127,11 +132,7 @@ def test_a_wrong_unit_is_actually_rejected(tmp_path):
     real record, writes the exact string this issue was filed about, and
     requires a non-zero exit.
     """
-    source = next(
-        p
-        for p in sorted(COMMUNITIES.glob("*.yaml"))
-        if "operating_temperature_unit: °C" in p.read_text()
-    )
+    source = next(p for p in record_files() if "operating_temperature_unit: °C" in p.read_text())
     broken = tmp_path / "broken.yaml"
     broken.write_text(
         source.read_text().replace(

tests/test_gtdb_near_tie_marker.py

@@ -21,15 +21,28 @@
 from __future__ import annotations
 
 import importlib.util
+import pathlib
 import subprocess
 from pathlib import Path
 
 import pytest
 import yaml
 
+from communitymech.paths import taxon_descriptor_roots
+from communitymech.taxon_blocks import iter_taxon_descriptors
+
 REPO = Path(__file__).parent.parent
 
 
+# A GTDB grounding can live on any `TaxonDescriptor`, and the schema hangs that
+# class off two roots: MicrobialCommunity.taxonomy[].taxon_term and
+# CommonTaxon.taxon_term. So this needs BOTH the wider directory list and the
+# shared walker -- iterating `document["taxonomy"]` over kb/taxa finds nothing,
+# because a CommonTaxon has no `taxonomy` key at all (#656, #689).
+def _record_paths() -> list[pathlib.Path]:
+    return [p for root in taxon_descriptor_roots() for p in sorted(root.glob("*.yaml"))]
+
+
 def _module():
     spec = importlib.util.spec_from_file_location("_gtdb", REPO / "scripts/gtdb_ground.py")
     module = importlib.util.module_from_spec(spec)
@@ -85,9 +98,9 @@ def test_the_bound_is_where_the_population_gap_is():
 
 
 def _grounded():
-    for path in sorted((REPO / "kb/communities").glob("*.yaml")):
-        for entry in (yaml.safe_load(path.read_text()) or {}).get("taxonomy") or []:
-            block = (entry.get("taxon_term") or {}).get("gtdb_classification")
+    for path in _record_paths():
+        for descriptor in iter_taxon_descriptors(yaml.safe_load(path.read_text())):
+            block = descriptor.get("gtdb_classification")
             if block:
                 yield path.name, block
 

tests/test_gtdb_status_writer.py

@@ -22,6 +22,9 @@
 import pytest
 import yaml
 
+from communitymech.paths import taxon_descriptor_roots
+from communitymech.taxon_blocks import iter_taxon_descriptors
+
 REPO = Path(__file__).parent.parent
 
 
@@ -435,14 +438,14 @@ def test_every_curated_pin_carries_a_note_and_a_value():
     recorded, so a silent change to a pinned grounding fails.
     """
     pins = {}
-    for path in sorted((REPO / "kb/communities").glob("*.yaml")):
-        for entry in yaml.safe_load(path.read_text()).get("taxonomy") or []:
-            term_block = entry.get("taxon_term") or {}
-            block = term_block.get("gtdb_classification") or {}
-            if block.get("curated"):
-                key = (path.name, (term_block.get("term") or {}).get("id"))
-                pins[key] = block.get("gtdb_id")
-                assert block.get("curation_note"), f"{path.name}: curated with no note"
+    for root in taxon_descriptor_roots():
+        for path in sorted(root.glob("*.yaml")):
+            for term_block in iter_taxon_descriptors(yaml.safe_load(path.read_text())):
+                block = term_block.get("gtdb_classification") or {}
+                if block.get("curated"):
+                    key = (path.name, (term_block.get("term") or {}).get("id"))
+                    pins[key] = block.get("gtdb_id")
+                    assert block.get("curation_note"), f"{path.name}: curated with no note"
 
     assert pins, "no curated pins found; the flag protects nothing"
     # The two known pins are value-pinned, so a mapping build that makes either

tests/test_interaction_participants_outside_taxonomy.py

@@ -47,9 +47,13 @@
 import pytest
 
 from communitymech.network.auditor import IssueType, NetworkIntegrityAuditor
+from communitymech.paths import record_files
 
 REPO = pathlib.Path(__file__).parent.parent
-COMMUNITIES = REPO / "kb/communities"
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 
 # (record, participant) for every interaction endpoint absent from `taxonomy`,
 # grouped by why it is absent. Sourced from the auditor itself, not re-derived:
@@ -121,9 +125,9 @@
 
 def _outside_taxonomy() -> set[tuple[str, str]]:
     """Every (record, participant) the auditor reports as not a member."""
-    auditor = NetworkIntegrityAuditor(COMMUNITIES)
+    auditor = NetworkIntegrityAuditor()
     found = set()
-    for path in sorted(COMMUNITIES.glob("*.yaml")):
+    for path in record_files():
         for issue in auditor.audit_community(path) or []:
             if issue["type"] in (IssueType.UNKNOWN_SOURCE, IssueType.UNKNOWN_TARGET):
                 found.add((path.name, issue.get("taxon")))
@@ -181,9 +185,9 @@ def test_all_of_them_are_warnings_not_errors(outside):
     interaction turns that participant into an error and reddens the build on a
     record nobody changed the biology of.
     """
-    auditor = NetworkIntegrityAuditor(COMMUNITIES)
+    auditor = NetworkIntegrityAuditor()
     errors = []
-    for path in sorted(COMMUNITIES.glob("*.yaml")):
+    for path in record_files():
         for issue in auditor.audit_community(path) or []:
             if (
                 issue["type"] in (IssueType.UNKNOWN_SOURCE, IssueType.UNKNOWN_TARGET)

tests/test_ncbi_domain_scope.py

@@ -23,11 +23,22 @@
 import pytest
 import yaml
 
+from communitymech.paths import taxon_descriptor_roots
+from communitymech.taxon_blocks import iter_taxon_descriptors
 from communitymech.validators.ncbi_domain import BACTERIA, EUKARYOTA, domain_of, outside_gtdb_scope
 
 REPO = pathlib.Path(__file__).parent.parent
 
 
+# A GTDB grounding can live on any `TaxonDescriptor`, and the schema hangs that
+# class off two roots: MicrobialCommunity.taxonomy[].taxon_term and
+# CommonTaxon.taxon_term. So this needs BOTH the wider directory list and the
+# shared walker -- iterating `document["taxonomy"]` over kb/taxa finds nothing,
+# because a CommonTaxon has no `taxonomy` key at all (#656, #689).
+def _record_paths() -> list[pathlib.Path]:
+    return [p for root in taxon_descriptor_roots() for p in sorted(root.glob("*.yaml"))]
+
+
 @pytest.mark.parametrize(
     ("curie", "expected"),
     [
@@ -79,9 +90,8 @@ def test_an_unavailable_adapter_degrades_rather_than_guesses(monkeypatch):
 def _statuses():
     counts = collections.Counter()
     offenders = []
-    for path in sorted((REPO / "kb/communities").glob("*.yaml")):
-        for entry in (yaml.safe_load(path.read_text()) or {}).get("taxonomy") or []:
-            term_block = entry.get("taxon_term") or {}
+    for path in _record_paths():
+        for term_block in iter_taxon_descriptors(yaml.safe_load(path.read_text())):
             status = term_block.get("gtdb_grounding_status")
             if not status:
                 continue
@@ -121,7 +131,7 @@ def test_sulcia_is_a_bacterium_not_a_spider():
     The id↔label gate cannot catch this class: `NCBITaxon:2716471` really is
     labelled "Sulcia" (#292).
     """
-    for path in sorted((REPO / "kb/communities").glob("*.yaml")):
+    for path in _record_paths():
         text = path.read_text()
         assert "NCBITaxon:2716471" not in text, f"{path.name} still uses the spider id"
 

tests/test_network_auditor.py

@@ -18,6 +18,7 @@
     issue_severity,
     severity_of,
 )
+from communitymech.paths import record_files
 
 
 @pytest.fixture
@@ -1268,10 +1269,16 @@ def test_the_kb_has_no_duplicate_taxon_names():
     error — so the test audited zero records and passed vacuously (#334). It is
     the sole support for gating on this finding, so an empty sweep must fail.
     """
-    communities = Path(__file__).parent.parent / "kb/communities"
-    assert len(list(communities.glob("*.yaml"))) > 100, "audited an empty or wrong directory"
-
-    auditor = NetworkIntegrityAuditor(communities_dir=communities)
+    # No `communities_dir=` override. Passing one silently replaced the
+    # auditor's own `default_record_roots()` default, so this swept
+    # kb/communities alone -- the shape #350 fixed IN the auditor and left
+    # standing in its test. The auditor's default is the thing under test here
+    # as much as the corpus is (#689).
+    records = record_files()
+    assert len(records) > 100, "audited an empty or wrong directory"
+    assert any(p.parent.name == "isolates" for p in records), "isolates are not being audited"
+
+    auditor = NetworkIntegrityAuditor()
     auditor.audit_all(quiet=True)
 
     offenders = sorted(

tests/test_no_duplicate_yaml_keys.py

@@ -26,6 +26,11 @@
 import pytest
 import yaml
 
+from communitymech.paths import record_files
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 COMMUNITIES = Path(__file__).parent.parent / "kb/communities"
 
 # Empty since #289 was fixed: both records that needed a curator decision have
@@ -58,7 +63,7 @@ def construct_mapping(loader, node, deep=False):
 
 
 def _community_files() -> list[Path]:
-    return sorted(COMMUNITIES.glob("*.yaml"))
+    return record_files()
 
 
 def test_there_are_community_files_to_check():

tests/test_no_vacuous_go_annotations.py

@@ -35,8 +35,13 @@
 import pytest
 import yaml
 
+from communitymech.paths import record_files
+
 REPO = pathlib.Path(__file__).parent.parent
-COMMUNITIES = REPO / "kb/communities"
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 
 # GO terms so close to the root of the biological-process branch that asserting
 # them of a microbial community conveys nothing. Each needs a reason, so that
@@ -66,7 +71,7 @@ def _walk(node, filename):
 
 def _descriptors(corpus: pathlib.Path | None = None):
     """Every (file, preferred_term, id, label) biological-process descriptor."""
-    for path in sorted((corpus or COMMUNITIES).glob("*.yaml")):
+    for path in sorted(corpus.glob("*.yaml")) if corpus else record_files():
         document = yaml.safe_load(path.read_text(encoding="utf-8")) or {}
         yield from _walk(document, path.name)
 

tests/test_participating_taxa.py

@@ -36,8 +36,13 @@
 import yaml
 
 from communitymech.network.auditor import IssueType, NetworkIntegrityAuditor
+from communitymech.paths import record_files
 
 REPO = pathlib.Path(__file__).parent.parent
+
+# Both record roots, not kb/communities alone. `data/isolates` holds the same
+# root class -- 4 records with 66 snippets, 3 ecological_interactions and 3
+# gtdb_classification blocks -- and this module could not see any of it (#689).
 COMMUNITIES = REPO / "kb/communities"
 # #312's illustration: 28 taxa, every interaction COMMUNITY_LEVEL.
 EXAMPLE = COMMUNITIES / "GLBRC_Populus_Variovorax_SynCom28.yaml"
@@ -157,7 +162,7 @@ def test_the_corpus_is_unchanged_by_this_feature():
     """
     users = [
         path.name
-        for path in sorted(COMMUNITIES.glob("*.yaml"))
+        for path in record_files()
         for interaction in (yaml.safe_load(path.read_text()) or {}).get("ecological_interactions")
         or []
         if isinstance(interaction, dict) and interaction.get("participating_taxa")