Compare commits

..
Author SHA1 Message Date
CC WorkerandClaude Opus 4.8 41614b78ef FX-3: project image-only papers to Neo4j after auto-map
Only the born-digital fast path enqueued project_template_safe; the async OCR
job (_run_auto_map_job) — where image-only papers, R3's primary target, are
routed — never projected, so those papers never reached the graph after
auto-map. Project at the end of the async job too.

Adds test_auto_map_ocr_path_projects_to_neo4j.

Co-Authored-By: Claude Opus 4.8 <[email protected]>
2026-07-02 21:22:54 +00:00
3 changed files with 43 additions and 62 deletions
+3
View File
@@ -634,6 +634,9 @@ def _run_auto_map_job(job_id: str, ctx: ExamContext, template_id: str, pdf_bytes
_set_auto_map_status(job_id, {"status": "running", "template_id": template_id})
try:
rows = _run_auto_map_merge(ctx, template_id, pdf_bytes, source_label)
# Project to Neo4j like the born-digital fast path does — otherwise image-only papers (R3's
# primary target, routed here because they need OCR) never reach the graph after auto-map.
project_template_safe(template_id)
_set_auto_map_status(job_id, {"status": "completed", "template_id": template_id, "counts": {k: len(v) for k, v in rows.items()}})
except Exception as exc:
logger.exception(f"auto-map job failed for template {template_id}: {exc}")
+28 -62
View File
@@ -1,19 +1,17 @@
"""
init_exam_graph.py — Initialise the cc.public.exams Neo4j knowledge graph.
Creates the shared, public exam database, its uniqueness constraints, and seeds the AQA exam board
+ the 6 current test specifications (GCSE & A-level Physics/Chemistry/Biology) with their top-level
topic SpecPoints (44 in total). Idempotent (CREATE DATABASE IF NOT EXISTS / CREATE CONSTRAINT IF NOT
EXISTS / MERGE).
Creates the shared, public exam database, its uniqueness constraints, and seeds the AQA exam
board + AQA GCSE Physics (8463) specification with its 8 top-level topic SpecPoints. Idempotent
(CREATE DATABASE IF NOT EXISTS / CREATE CONSTRAINT IF NOT EXISTS / MERGE).
Run inside the ccapi container:
python3 -c "from run.initialization.init_exam_graph import init; import json; print(json.dumps(init()))"
NOTE: only *top-level* topics are seeded (the granularity a teacher plans against). The full sub-point
breakdown (e.g. 4.1.1.1 ...) is a later data-population task (sourceable from the AQA spec PDF via
Docling). Seeding all 6 specs means a template's spec_ref finds a matching SpecPoint so
(:Part)-[:ASSESSES]->(:SpecPoint) fires beyond GCSE Physics; spec_code (e.g. AQA-PHYS-8463) must match
the eb_exams/eb_specifications seed (card S4-3) and the app's deriveSpecCode.
NOTE: the 8 SpecPoints seeded here are the real AQA GCSE Physics *top-level* topics. The full
sub-point breakdown (e.g. 4.1.1.1 ...) is a later data-population task (sourceable from the AQA
spec PDF via Docling). spec_code AQA-PHYS-8463 is the standalone GCSE Physics code that matches
"AQA Physics Paper 1H"; the eb_exams/eb_specifications seed (card S4-3) must use the same code.
"""
import uuid
from typing import Dict, Any
@@ -43,37 +41,6 @@ SPEC_POINTS = [
("4.8", "Space physics"),
]
# Full AQA catalogue for the current test specs (top-level topics; ref = topic number). Seeding all of
# them means a template's spec_ref finds a matching SpecPoint so (:Part)-[:ASSESSES]->(:SpecPoint) fires
# beyond AQA GCSE Physics. Sub-point granularity (e.g. 4.1.1.1) remains a later data-population task.
SPECIFICATIONS = [
{**SPEC, "topics": SPEC_POINTS},
{"spec_code": "AQA-CHEM-8462", "exam_board_code": "AQA", "subject_code": "CHEM", "award_code": "GCSE",
"title": "AQA GCSE Chemistry (8462)", "topics": [
("4.1", "Atomic structure and the periodic table"), ("4.2", "Bonding, structure, and the properties of matter"),
("4.3", "Quantitative chemistry"), ("4.4", "Chemical changes"), ("4.5", "Energy changes"),
("4.6", "The rate and extent of chemical change"), ("4.7", "Organic chemistry"), ("4.8", "Chemical analysis"),
("4.9", "Chemistry of the atmosphere"), ("4.10", "Using resources")]},
{"spec_code": "AQA-BIOL-8461", "exam_board_code": "AQA", "subject_code": "BIOL", "award_code": "GCSE",
"title": "AQA GCSE Biology (8461)", "topics": [
("4.1", "Cell biology"), ("4.2", "Organisation"), ("4.3", "Infection and response"), ("4.4", "Bioenergetics"),
("4.5", "Homeostasis and response"), ("4.6", "Inheritance, variation and evolution"), ("4.7", "Ecology")]},
{"spec_code": "AQA-PHYS-7408", "exam_board_code": "AQA", "subject_code": "PHYS", "award_code": "A-level",
"title": "AQA A-level Physics (7408)", "topics": [
("3.1", "Measurements and their errors"), ("3.2", "Particles and radiation"), ("3.3", "Waves"),
("3.4", "Mechanics and materials"), ("3.5", "Electricity"), ("3.6", "Further mechanics and thermal physics"),
("3.7", "Fields and their consequences"), ("3.8", "Nuclear physics")]},
{"spec_code": "AQA-CHEM-7405", "exam_board_code": "AQA", "subject_code": "CHEM", "award_code": "A-level",
"title": "AQA A-level Chemistry (7405)", "topics": [
("3.1", "Physical chemistry"), ("3.2", "Inorganic chemistry"), ("3.3", "Organic chemistry")]},
{"spec_code": "AQA-BIOL-7402", "exam_board_code": "AQA", "subject_code": "BIOL", "award_code": "A-level",
"title": "AQA A-level Biology (7402)", "topics": [
("3.1", "Biological molecules"), ("3.2", "Cells"), ("3.3", "Organisms exchange substances with their environment"),
("3.4", "Genetic information, variation and relationships between organisms"),
("3.5", "Energy transfers in and between organisms"), ("3.6", "Organisms respond to changes"),
("3.7", "Genetics, populations, evolution and ecosystems"), ("3.8", "The control of gene expression")]},
]
CONSTRAINTS = [
"CREATE CONSTRAINT exam_board_uid IF NOT EXISTS FOR (n:ExamBoard) REQUIRE n.uuid_string IS UNIQUE",
"CREATE CONSTRAINT spec_uid IF NOT EXISTS FOR (n:Specification) REQUIRE n.uuid_string IS UNIQUE",
@@ -114,38 +81,37 @@ def init() -> Dict[str, Any]:
s.run(c).consume()
result["constraints"] += 1
# 3. board (once)
# 3. board + spec
board_uid = _uid("ExamBoard", BOARD["code"])
spec_uid = _uid("Specification", SPEC["spec_code"])
s.run(
"MERGE (b:ExamBoard {uuid_string:$uid}) "
"SET b.code=$code, b.name=$name, b.node_storage_path=$nsp",
uid=board_uid, code=BOARD["code"], name=BOARD["name"],
nsp=f"{EXAM_DB}/ExamBoard/{BOARD['code']}",
).consume()
s.run(
"MERGE (sp:Specification {uuid_string:$uid}) "
"SET sp.spec_code=$sc, sp.exam_board_code=$ebc, sp.subject_code=$subj, "
" sp.award_code=$award, sp.title=$title, sp.node_storage_path=$nsp "
"WITH sp MATCH (b:ExamBoard {code:$ebc}) MERGE (b)-[:PUBLISHES]->(sp)",
uid=spec_uid, sc=SPEC["spec_code"], ebc=SPEC["exam_board_code"],
subj=SPEC["subject_code"], award=SPEC["award_code"], title=SPEC["title"],
nsp=f"{EXAM_DB}/Specification/{SPEC['spec_code']}",
).consume()
# 4. each specification + its top-level spec points (idempotent MERGE)
for spec in SPECIFICATIONS:
spec_uid = _uid("Specification", spec["spec_code"])
# 4. spec points
for ref, desc in SPEC_POINTS:
sp_uid = _uid("SpecPoint", SPEC["spec_code"], ref)
s.run(
"MERGE (sp:Specification {uuid_string:$uid}) "
"SET sp.spec_code=$sc, sp.exam_board_code=$ebc, sp.subject_code=$subj, "
" sp.award_code=$award, sp.title=$title, sp.node_storage_path=$nsp "
"WITH sp MATCH (b:ExamBoard {code:$ebc}) MERGE (b)-[:PUBLISHES]->(sp)",
uid=spec_uid, sc=spec["spec_code"], ebc=spec["exam_board_code"],
subj=spec["subject_code"], award=spec["award_code"], title=spec["title"],
nsp=f"{EXAM_DB}/Specification/{spec['spec_code']}",
"MERGE (p:SpecPoint {uuid_string:$uid}) "
"SET p.ref=$ref, p.description=$desc, p.spec_code=$sc, "
" p.exam_board_code=$ebc, p.node_storage_path=$nsp "
"WITH p MATCH (s:Specification {spec_code:$sc}) MERGE (s)-[:HAS_SPEC_POINT]->(p)",
uid=sp_uid, ref=ref, desc=desc, sc=SPEC["spec_code"],
ebc=SPEC["exam_board_code"], nsp=f"{EXAM_DB}/SpecPoint/{SPEC['spec_code']}/{ref}",
).consume()
for ref, desc in spec["topics"]:
sp_uid = _uid("SpecPoint", spec["spec_code"], ref)
s.run(
"MERGE (p:SpecPoint {uuid_string:$uid}) "
"SET p.ref=$ref, p.description=$desc, p.spec_code=$sc, "
" p.exam_board_code=$ebc, p.node_storage_path=$nsp "
"WITH p MATCH (s:Specification {spec_code:$sc}) MERGE (s)-[:HAS_SPEC_POINT]->(p)",
uid=sp_uid, ref=ref, desc=desc, sc=spec["spec_code"],
ebc=spec["exam_board_code"], nsp=f"{EXAM_DB}/SpecPoint/{spec['spec_code']}/{ref}",
).consume()
result["spec_points"] += 1
result["spec_points"] += 1
counts = s.run(
"MATCH (b:ExamBoard) WITH count(b) AS boards "
+12
View File
@@ -718,4 +718,16 @@ def test_auto_map_ocr_returns_job_id_and_status_completes(monkeypatch):
body = status.json()
assert body["status"] == "completed"
assert body["counts"]["questions"] >= 2
def test_auto_map_ocr_path_projects_to_neo4j(monkeypatch, _stub_projection):
# Image-only papers route through the async OCR job; regression: that path must project to Neo4j
# like the born-digital fast path, or the graph is never built for the primary target.
store = _template_with_source()
client, store = make_client(store=store)
_patch_auto_map(monkeypatch, store, fast=False)
resp = client.post("/api/exam/templates/t1/auto-map")
assert resp.status_code == 202
# the BackgroundTask runs after the response under TestClient
assert "t1" in _stub_projection
assert body["template"]["layout"]