Pub + split query fix

.github/workflows/python-test.yml

@@ -7,10 +7,10 @@ on:
     branches: [ main, dev ]
   workflow_dispatch:  # Enables manual triggering
   schedule:
-    - cron: '0 0 2 * *'  # Runs at 00:00 UTC on the 2nd day of every month
+    - cron: '0 0 * * 0'  # Runs at 00:00 UTC every Sunday (weekly data-health audit)
 
 # One live run per branch. Every run drives the full correctness suite against
-# the production VFB backend (Neo4j / SOLR / Owlery), so a series of quick
+# the production data sources (VFB's Neo4j / SOLR / Owlery and FlyBase Chado), so a series of quick
 # pushes to a PR would otherwise stack several full suites against production
 # simultaneously. Superseded runs are cancelled — only the newest commit's
 # result is meaningful anyway.
@@ -75,9 +75,19 @@ jobs:
           # on today's GitHub-hosted ubuntu runner, so this is not a slowdown —
           # it just stops the load on VFB infra changing silently if the hosted
           # runner spec grows.
+          #
+          # data_health: the '*_live' term_info variants assert on the external
+          # indexer's built vfb_json content, not this repo's render code, so
+          # an incomplete/stale production document fails them. They must NOT gate
+          # code PRs — the PR-blocking versions are fixture-based and deterministic.
+          # So they run ONLY on the schedule (a data-health audit); on push /
+          # pull_request they are deselected. See TESTING.md.
+          MARK=(-m "not data_health")
+          [ "${{ github.event_name }}" = "schedule" ] && MARK=()
           pytest -v -ra -n 4 --dist loadscope \
             --ignore=src/test/test_query_performance.py \
             --ignore=src/test/test_example_queries.py \
+            "${MARK[@]}" \
             src/test tests 2>&1 | tee pytest_output.log
       - name: Flag skipped tests (backend unavailable)
         if: always()
@@ -105,7 +115,7 @@ jobs:
             } >> "$GITHUB_STEP_SUMMARY"
           fi
           if [ -n "$summary" ]; then
-            echo "::warning title=Tests skipped — VFB backend unreachable::${summary}. ${digest:+First failure: ${digest}. }These are NOT test failures and not a problem with this branch: the VFB backend (Neo4j / SOLR / Owlery) did not answer, so those queries went unverified this run. Treat a green check with skips as an incomplete run — re-run once the backend is healthy before relying on it. Full detail (every skipped test with its reason; failing URLs and probe verdicts for backend failures): the job summary and the PR comment."
+            echo "::warning title=Tests skipped — upstream data source unreachable::${summary}. ${digest:+First failure: ${digest}. }These are NOT test failures and not a problem with this branch: an upstream data source (VFB's Neo4j / SOLR / Owlery, or FlyBase's Chado) did not answer, so those queries went unverified this run. Treat a green check with skips as an incomplete run — re-run once the data sources are healthy before relying on it. Full detail (every skipped test with its reason; failing URLs and probe verdicts for backend failures): the job summary and the PR comment."
           else
             echo "No tests skipped."
           fi
@@ -151,7 +161,7 @@ jobs:
             // re-running it would do nothing.
             const runUrl = `${process.env.GITHUB_SERVER_URL}/${owner}/${repo}/actions/runs/${context.runId}`;
             const rerun = `To re-run: open [this workflow run](${runUrl}) and click ` +
-              `**Re-run all jobs** once the backend is healthy (re-running the ` +
+              `**Re-run all jobs** once the data sources are healthy (re-running the ` +
               `“Run completeness” check itself does nothing — it has no job behind it).`;
 
             // A neutral (grey) status check so the PR's checks box stops reading
@@ -168,16 +178,16 @@ jobs:
               details_url: runUrl,
               output: {
                 title: skipped > 0
-                  ? `${skipped} test(s) skipped — backend unreachable (incomplete run)`
+                  ? `${skipped} test(s) skipped — upstream data source unreachable (incomplete run)`
                   : 'All backend tests ran',
                 summary: skipped > 0
-                  ? (`**${skipped}** test(s) were skipped because the VFB backend ` +
-                     `(Neo4j / SOLR / Owlery) did not answer, so those queries went ` +
+                  ? (`**${skipped}** test(s) were skipped because an upstream data source ` +
+                     `(VFB's Neo4j / SOLR / Owlery, or FlyBase's Chado) did not answer, so those queries went ` +
                      `unverified. This is not a branch failure — but the run is ` +
                      `incomplete.\n\n${rerun}` +
                      (summary ? '\n\n```\n' + summary + '\n```' : '') +
                      (outage ? '\n\n' + outage : ''))
-                  : 'Every backend-dependent test reached the VFB backend and ran.',
+                  : 'Every backend-dependent test reached its data source and ran.',
               },
             });
             const comments = await github.paginate(github.rest.issues.listComments,
@@ -187,10 +197,10 @@ jobs:
             if (skipped > 0) {
               const body = [
                 marker,
-                `### ⚠️ ${skipped} test(s) skipped — VFB backend was unreachable`,
+                `### ⚠️ ${skipped} test(s) skipped — an upstream data source was unreachable`,
                 '',
-                `The full suite ran, but **${skipped}** test(s) were **skipped** because the ` +
-                `VFB backend (Neo4j / SOLR / Owlery) did not answer during this run.`,
+                `The full suite ran, but **${skipped}** test(s) were **skipped** because ` +
+                `an upstream data source (VFB's Neo4j / SOLR / Owlery, or FlyBase's Chado) did not answer during this run.`,
                 '',
                 'These are **not failures** and **not a problem with this branch** — but those ' +
                 'queries went **unverified**, so a green check here is an *incomplete* run.',

TESTING.md

@@ -4,8 +4,7 @@ Most VFBquery tests run **live queries against the production VFB backend**
 (SOLR, Neo4j, Owlery, FlyBase Chado). That makes them powerful — they catch real
 regressions in query results — but also easy to write badly: a test that never
 checks its query returned anything passes forever while the query is silently
-broken. A whole class of such tests was found and fixed in Aug 2026; this doc
-exists so they don't come back.
+broken. This doc exists so that class of test does not creep back in.
 
 Read this before adding or changing a test.
 
@@ -69,6 +68,50 @@ So an empty result is **never** an acceptable outcome for a known-populated
 term — it is a bug. A backend outage is handled *for you*; you do not need
 (and must not add) your own try/except to survive it.
 
+## Fixture vs live (`data_health`): what an empty result actually means
+
+The "empty ⇒ bug" rule assumes empty means **our** code is wrong. That holds
+when a test reads the **source of truth (pdb / Neo4j) through this repo's own
+query code** — connectivity, the neurotransmitter queries, and the like. There,
+an empty result *is* a defect here, so the test asserts real content and is
+**PR-blocking**.
+
+It does **not** hold for **term_info** tests. `get_term_info` / `process` /
+`term_info_parse_object` *render* a pre-built SOLR `vfb_json` document; that
+document is built by a **separate repository** (`VFB_json_schema_indexer`), not
+here. If that built document is stale or incomplete (e.g. a pub with no title, a
+split class missing its `Expression_pattern` type), our render code faithfully
+produces an empty/partial result — which is a problem with the **build**, not
+with this repo. Gating a code PR on it is a false signal.
+
+**The rule of thumb:** *if the build code is in this repo, test it here against
+the source (empty = fail, PR-blocking); if the build is in another repo, use a
+fixture here and test the build at its source.* Connectivity/NT queries build
+from pdb here → tested here. term_info documents are built by
+`VFB_json_schema_indexer` → **fixtures here, build tested there.**
+
+So term_info tests come in two variants:
+
+| Variant | Reads | Runs | A failure means |
+|---|---|---|---|
+| **fixture** (plain name) | a committed complete `vfb_json` doc in `src/test/fixtures/term_info/` | **every PR** (blocking) | this repo's **render code** regressed |
+| **live** (`<name>_live`, `@pytest.mark.data_health`) | the live SOLR `vfb_json` document | **schedule only** | the **built document / pipeline** (`VFB_json_schema_indexer`) is wrong — a data-health signal, not a code gate |
+
+Mechanics:
+
+* Fixtures are captured from the PDB (the source of truth) — i.e. what the
+  indexer *should* produce — and verified to render the asserted output. They
+  make the code assertion **deterministic and independent of the live SOLR
+  `vfb_json` document and the caches**, so code improvements are never blocked by
+  stale data.
+* The live variant carries `@pytest.mark.data_health`. PRs run
+  `-m 'not data_health'`; the weekly `schedule` in `python-test.yml` runs the
+  full set. A red scheduled `data_health` run means production `vfb_json` (or the
+  indexer that builds it) needs attention — it does **not** block merges.
+* Do **not** paper over an empty term_info doc with a skip guard — that violates
+  rule 2. Use the fixture for the code assertion; let the `_live` variant fail on
+  the schedule, where empty is exactly the signal you want.
+
 ## The rules
 
 ### 1. Assert content, not just shape

pyproject.toml

@@ -12,6 +12,7 @@ build-backend = "setuptools.build_meta"
 # blocks in performance.md — pure noise.
 markers = [
     "integration: integration tests that exercise the live VFB upstream (Neo4j, SOLR, owlery). Skip via -m 'not integration'.",
+    "data_health: live SOLR content-completeness checks (the '*_live' variants of the term_info tests). These assert on the external indexer's built 'vfb_json' documents, not this repo's render code, so an empty/incomplete production document fails them. Run on the schedule; deselected on PRs via -m 'not data_health'. See TESTING.md.",
 ]
 
 # Ceiling on any single test. The upstream Neo4j intermittently stops answering

src/test/fixtures/term_info/pub_FBrf0242477.json

@@ -0,0 +1,162 @@
+{
+ "term": {
+  "core": {
+   "iri": "http://flybase.org/reports/FBrf0242477",
+   "symbol": "",
+   "types": [
+    "Entity",
+    "Individual",
+    "pub"
+   ],
+   "short_form": "FBrf0242477",
+   "unique_facets": [
+    "pub"
+   ],
+   "label": "Dolan et al., 2019, eLife 8: e43079"
+  },
+  "description": [],
+  "comment": []
+ },
+ "query": "Get JSON for pub",
+ "version": "0dcd736",
+ "dataset_license": [
+  {
+   "license": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/VFBlicense_CC_BY_4_0",
+     "symbol": "CC_BY",
+     "types": [
+      "Entity",
+      "Individual",
+      "License"
+     ],
+     "short_form": "VFBlicense_CC_BY_4_0",
+     "unique_facets": [],
+     "label": "CC-BY_4.0"
+    },
+    "icon": "http://mirrors.creativecommons.org/presskit/buttons/88x31/png/by.png",
+    "link": "https://creativecommons.org/licenses/by/4.0/legalcode"
+   },
+   "dataset": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/LateralHorn2019",
+     "symbol": "",
+     "types": [
+      "Entity",
+      "Individual",
+      "DataSet"
+     ],
+     "short_form": "LateralHorn2019",
+     "unique_facets": [],
+     "label": "FlyLight split-GAL4 lines for Lateral Horn"
+    },
+    "link": ""
+   }
+  },
+  {
+   "license": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/VFBlicense_CC_BY_4_0",
+     "symbol": "CC_BY",
+     "types": [
+      "Entity",
+      "Individual",
+      "License"
+     ],
+     "short_form": "VFBlicense_CC_BY_4_0",
+     "unique_facets": [],
+     "label": "CC-BY_4.0"
+    },
+    "icon": "http://mirrors.creativecommons.org/presskit/buttons/88x31/png/by.png",
+    "link": "https://creativecommons.org/licenses/by/4.0/legalcode"
+   },
+   "dataset": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/Dolan2019",
+     "symbol": "",
+     "types": [
+      "Entity",
+      "Individual",
+      "DataSet"
+     ],
+     "short_form": "Dolan2019",
+     "unique_facets": [],
+     "label": "GAL4 Split expression patterns from Dolan et al. 2019"
+    },
+    "link": ""
+   }
+  },
+  {
+   "license": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/VFBlicense_CC_BY_4_0",
+     "symbol": "CC_BY",
+     "types": [
+      "Entity",
+      "Individual",
+      "License"
+     ],
+     "short_form": "VFBlicense_CC_BY_4_0",
+     "unique_facets": [],
+     "label": "CC-BY_4.0"
+    },
+    "icon": "http://mirrors.creativecommons.org/presskit/buttons/88x31/png/by.png",
+    "link": "https://creativecommons.org/licenses/by/4.0/legalcode"
+   },
+   "dataset": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/FlyLight2019LateralHorn2019",
+     "symbol": "",
+     "types": [
+      "Entity",
+      "Individual",
+      "DataSet"
+     ],
+     "short_form": "FlyLight2019LateralHorn2019",
+     "unique_facets": [],
+     "label": "FlyLight split-GAL4 lines for Lateral Horn"
+    },
+    "link": ""
+   }
+  },
+  {
+   "license": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/VFBlicense_CC_BY_4_0",
+     "symbol": "CC_BY",
+     "types": [
+      "Entity",
+      "Individual",
+      "License"
+     ],
+     "short_form": "VFBlicense_CC_BY_4_0",
+     "unique_facets": [],
+     "label": "CC-BY_4.0"
+    },
+    "icon": "http://mirrors.creativecommons.org/presskit/buttons/88x31/png/by.png",
+    "link": "https://creativecommons.org/licenses/by/4.0/legalcode"
+   },
+   "dataset": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/FafbDolan2019",
+     "symbol": "",
+     "types": [
+      "Entity",
+      "Individual",
+      "DataSet"
+     ],
+     "short_form": "FafbDolan2019",
+     "unique_facets": [],
+     "label": "EM FAFB Dolan et al. 2019"
+    },
+    "link": "https://fafb.catmaid.virtualflybrain.org"
+   }
+  }
+ ],
+ "pub_specific_content": {
+  "title": "Neurogenetic dissection of the Drosophila lateral horn reveals major outputs, diverse behavioural functions, and interactions with the mushroom body.",
+  "FlyBase": "FBrf0242477",
+  "PubMed": "31112130",
+  "DOI": "10.7554/eLife.43079"
+ }
+}

src/test/fixtures/term_info/pub_FBrf0243986.json

@@ -0,0 +1,63 @@
+{
+ "term": {
+  "core": {
+   "iri": "http://flybase.org/reports/FBrf0243986",
+   "symbol": "",
+   "types": [
+    "Entity",
+    "Individual",
+    "pub"
+   ],
+   "short_form": "FBrf0243986",
+   "unique_facets": [
+    "pub"
+   ],
+   "label": "Sayin et al., 2019, Neuron 104(3): 544--558.e6"
+  },
+  "description": [],
+  "comment": []
+ },
+ "query": "Get JSON for pub",
+ "version": "0dcd736",
+ "dataset_license": [
+  {
+   "license": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/VFBlicense_CC_BY_4_0",
+     "symbol": "CC_BY",
+     "types": [
+      "Entity",
+      "Individual",
+      "License"
+     ],
+     "short_form": "VFBlicense_CC_BY_4_0",
+     "unique_facets": [],
+     "label": "CC-BY_4.0"
+    },
+    "icon": "http://mirrors.creativecommons.org/presskit/buttons/88x31/png/by.png",
+    "link": "https://creativecommons.org/licenses/by/4.0/legalcode"
+   },
+   "dataset": {
+    "core": {
+     "iri": "http://virtualflybrain.org/reports/Sayin2019",
+     "symbol": "",
+     "types": [
+      "Entity",
+      "Individual",
+      "DataSet"
+     ],
+     "short_form": "Sayin2019",
+     "unique_facets": [],
+     "label": "EM FAFB Sayin et al 2019"
+    },
+    "link": "https://fafb.catmaid.virtualflybrain.org"
+   }
+  }
+ ],
+ "pub_specific_content": {
+  "title": "A Neural Circuit Arbitrates between Persistence and Withdrawal in Hungry Drosophila.",
+  "FlyBase": "FBrf0243986",
+  "PubMed": "31471123",
+  "DOI": "10.1016/j.neuron.2019.07.028"
+ }
+}