Compare commits

..
Author SHA1 Message Date
tudor 404ba95275 Merge branch 'main' into fix/gias-blank-name-codes
PR Checks / Frontend Typecheck + Tests (pull_request) Successful in 1m3s
PR Checks / Backend Smoke (pull_request) Successful in 7s
PR Checks / Build Backend (no push) (pull_request) Successful in 10s
PR Checks / Build Frontend (no push) (pull_request) Successful in 50s
PR Checks / Build Pipeline (no push) (pull_request) Successful in 10s
PR Checks / AI Code Review (Claude) (pull_request) Successful in 33s
2026-07-16 09:14:42 +00:00
Tudor 493ea39c29 Updating the number of schools
PR Checks / Frontend Typecheck + Tests (pull_request) Successful in 9m44s
PR Checks / Backend Smoke (pull_request) Successful in 7s
PR Checks / Build Backend (no push) (pull_request) Successful in 16s
PR Checks / Build Frontend (no push) (pull_request) Successful in 49s
PR Checks / Build Pipeline (no push) (pull_request) Successful in 38s
PR Checks / AI Code Review (Claude) (pull_request) Failing after 27s
2026-07-09 22:53:42 +01:00
50 changed files with 171 additions and 2329 deletions
-5
View File
@@ -1,7 +1,2 @@
venv
__pycache__/
# dbt local build artifacts (embed absolute paths + anonymous-usage UUID)
pipeline/transform/target/
pipeline/transform/logs/
pipeline/transform/.user.yml
+23 -35
View File
@@ -677,7 +677,6 @@ async def compare_schools(
"deprivation": None,
}
supplementary_by_urn: dict = {}
census_benchmarks = None
db = None
try:
db = database.SessionLocal()
@@ -689,31 +688,6 @@ async def compare_schools(
key: supp.get(key, default)
for key, default in _EMPTY_SUPPLEMENTARY.items()
}
# Import-time census context benchmarks (fact_census_benchmarks);
# absent mart → None, and compute_benchmarks leaves those fields null.
try:
from .models import CensusBenchmark
rows = db.query(CensusBenchmark).all()
by_phase = {
r.phase: {
"year": r.year,
"fsm_pct": r.fsm_pct,
"eal_pct": r.eal_pct,
"median_pupils": r.median_pupils,
}
for r in rows
if getattr(r, "phase", None) in ("primary", "secondary")
}
if by_phase:
census_benchmarks = by_phase
except Exception:
# Missing mart (or a stubbed session in tests) must never break
# the compare payload — and not every session has rollback().
try:
db.rollback()
except Exception:
pass
except Exception:
supplementary_by_urn = {}
finally:
@@ -740,9 +714,6 @@ async def compare_schools(
"religious_denomination": convert_to_native(latest.get("religious_denomination")),
"age_range": convert_to_native(latest.get("age_range")),
"gender": convert_to_native(latest.get("gender")),
# Needed by the admissions "What this means" copy: selective
# schools get entrance-test framing, never the distance template.
"admissions_policy": convert_to_native(latest.get("admissions_policy")),
"has_sixth_form": convert_to_native(latest.get("has_sixth_form")),
"capacity": convert_to_native(latest.get("capacity")),
"gias_total_pupils": convert_to_native(latest.get("gias_total_pupils")),
@@ -757,7 +728,7 @@ async def compare_schools(
# Official DfE anchors + computed state-school benchmarks so the
# compare UI can label provenance correctly (spec §8.6).
"national_averages": _national_averages_payload(df),
"benchmarks": compute_benchmarks(df, census_benchmarks=census_benchmarks),
"benchmarks": compute_benchmarks(df),
}
@@ -826,11 +797,11 @@ def _national_averages_payload(df: pd.DataFrame) -> dict:
/api/compare.
Both series are persisted marts computed at import time: official DfE
KS2 figures (fact_ks2_national_averages) and official DfE KS4 figures
(fact_ks4_national_averages) — the API never aggregates the performance
dataframe per request. If the KS4 mart hasn't been built yet, the
secondary series is empty — never a computed stand-in, because the UI
labels these figures as official DfE data.
KS2 figures (fact_ks2_national_averages) and dataset-computed KS4
averages (fact_ks4_national_averages) — the API never aggregates the
performance dataframe per request. If the KS4 mart hasn't been built
yet (deploy lands before the next DAG run), fall back to computing the
latest year only — a single-year scan, never the historical loop.
"""
if df.empty:
return {"primary": {}, "secondary": {}}
@@ -870,6 +841,23 @@ def _national_averages_payload(df: pd.DataFrame) -> dict:
primary_by_year = {r.year: _row_metrics(r, _KS2_NATIONAL_METRICS) for r in ks2_rows}
secondary_by_year = {r.year: _row_metrics(r, _KS4_NATIONAL_METRICS) for r in ks4_rows}
if not any(secondary_by_year.values()):
# KS4 mart missing/empty: compute the latest year only.
df_latest = df[df["year"] == latest_year]
sec = (
df_latest[df_latest["attainment_8_score"].notna()]
if "attainment_8_score" in df_latest.columns
else df_latest.iloc[0:0]
)
vals = {}
for col in _KS4_NATIONAL_METRICS:
if col in sec.columns:
v = sec[col].dropna()
if len(v) > 0:
vals[col] = round(float(v.mean()), 2)
if vals:
secondary_by_year[latest_year] = vals
all_years = sorted(set(primary_by_year) | set(secondary_by_year))
by_year = [
{
+14 -22
View File
@@ -525,20 +525,13 @@ def get_data_info(db: Session = None) -> dict:
# SUPPLEMENTARY DATA — per-school detail page
# =============================================================================
def compute_benchmarks(df: pd.DataFrame, census_benchmarks: dict | None = None) -> dict:
def compute_benchmarks(df: pd.DataFrame) -> dict:
"""State-school benchmarks computed from our dataset (spec §5/§8.6).
NOT official DfE figures — consumers must label them
"state-school average (computed from our dataset)". The disadvantaged
attainment average is weighted by cohort size (eligible_pupils) so
small schools don't dominate.
Context measures (FSM/EAL/pupil counts) come from `census_benchmarks`
(the fact_census_benchmarks mart, pupil-weighted, keyed by phase): the
performance df has no fsm_pct at all, and its eal/disadvantaged columns
are KS2-only — medianing them for "secondary" produced junk anchors
from the handful of all-through schools. When the mart is unavailable
these are None; never fall back across measure definitions.
small schools don't dominate; context measures are medians.
"""
if df.empty or "year" not in df.columns:
return {}
@@ -574,14 +567,18 @@ def compute_benchmarks(df: pd.DataFrame, census_benchmarks: dict | None = None)
)
return round(float(w), 1)
def _block(sub, phase, with_disadvantaged):
census = (census_benchmarks or {}).get(phase) or {}
def _block(sub, with_disadvantaged):
median_pupils = None
if "total_pupils" in sub.columns:
mp = sub["total_pupils"].median()
if pd.notna(mp):
median_pupils = int(mp)
block = {
"eal_pct": census.get("eal_pct"),
"eal_pct": _median(sub, "eal_pct"),
"sen_support_pct": _median(sub, "sen_support_pct"),
"disadvantaged_pct": _median(sub, "disadvantaged_pct") if with_disadvantaged else None,
"fsm_pct": census.get("fsm_pct"),
"median_pupils": census.get("median_pupils"),
"disadvantaged_pct": _median(sub, "disadvantaged_pct"),
"fsm_pct": _median(sub, "fsm_pct"),
"median_pupils": median_pupils,
}
if with_disadvantaged:
block["disadvantaged_rwm_expected_pct"] = _weighted_disadvantaged(sub)
@@ -590,8 +587,8 @@ def compute_benchmarks(df: pd.DataFrame, census_benchmarks: dict | None = None)
return {
"source": "state-school average (computed from our dataset)",
"year": int(latest_year),
"primary": _block(prim, "primary", with_disadvantaged=True),
"secondary": _block(sec, "secondary", with_disadvantaged=False),
"primary": _block(prim, with_disadvantaged=True),
"secondary": _block(sec, with_disadvantaged=False),
}
@@ -620,11 +617,6 @@ def _ofsted_block(o, urn: int) -> dict:
block = {
"framework": o.framework,
"inspection_date": o.inspection_date.isoformat() if o.inspection_date else None,
"rc_inspection_date": (
o.rc_inspection_date.isoformat()
if getattr(o, "rc_inspection_date", None)
else None
),
"inspection_type": o.inspection_type,
"overall_effectiveness": overall,
"grade_source": grade_source,
+1 -23
View File
@@ -156,9 +156,6 @@ class FactOfstedInspection(Base):
rc_leadership_governance = Column(Integer)
rc_early_years = Column(Integer)
rc_sixth_form = Column(Integer)
# Start date of the report-card inspection itself (renewed framework,
# Nov 2025+). Null for rows without report-card grades.
rc_inspection_date = Column(Date)
report_url = Column(Text)
@@ -234,27 +231,8 @@ class FactFinance(Base):
premises_cost_pct = Column(Float)
class CensusBenchmark(Base):
"""State-school context benchmarks from the pupil census — one row per phase.
fsm_pct / eal_pct are pupil-weighted means. Computed at import time;
consumers label them "state-school average (computed from our dataset)".
"""
__tablename__ = "fact_census_benchmarks"
__table_args__ = MARTS
phase = Column(String(20), primary_key=True)
year = Column(Integer)
fsm_pct = Column(Float)
eal_pct = Column(Float)
median_pupils = Column(Integer)
class Ks4NationalAverage(Base):
"""Official DfE KS4 national headline averages — one row per academic year.
gcse_grade_91_pct has no official national series and is always NULL.
"""
"""Computed national KS4 averages (from our dataset) — one row per year."""
__tablename__ = "fact_ks4_national_averages"
__table_args__ = MARTS
+8 -29
View File
@@ -57,40 +57,19 @@ def test_weighted_disadvantaged_average():
def test_medians_ignore_nan_and_older_years():
b = compute_benchmarks(_df())
assert b["year"] == LATEST
# sen medians over [10,14,18,20,22] = 18 — the only context measure still
# sourced from the performance df (the rest come from the census mart).
assert b["primary"]["sen_support_pct"] == 18.0
# disadvantaged_pct medians over [20,24,30,40,44] = 30
assert b["primary"]["disadvantaged_pct"] == 30.0
def test_benchmarks_use_census_mart_for_context():
census = {
"primary": {"year": LATEST, "fsm_pct": 25.3, "eal_pct": 21.8, "median_pupils": 240},
"secondary": {"year": LATEST, "fsm_pct": 24.1, "eal_pct": 18.9, "median_pupils": 980},
}
b = compute_benchmarks(_df(), census_benchmarks=census)
assert b["primary"]["fsm_pct"] == 25.3
assert b["primary"]["eal_pct"] == 21.8
assert b["secondary"]["eal_pct"] == 18.9
assert b["secondary"]["median_pupils"] == 980
def test_benchmarks_context_none_when_mart_missing():
# The performance df has no fsm_pct and its eal/disadvantaged columns are
# KS2-only — never silently fall back to medianing them for context.
b = compute_benchmarks(_df(), census_benchmarks=None)
assert b["primary"]["fsm_pct"] is None
assert b["primary"]["eal_pct"] is None
assert b["primary"]["median_pupils"] is None
# eal medians over [10,20,30,40,50] = 30
assert b["primary"]["eal_pct"] == 30.0
# fsm medians over [15,17,19,21,23] = 19
assert b["primary"]["fsm_pct"] == 19.0
# median pupils over [200,280,300,350,400] = 300
assert b["primary"]["median_pupils"] == 300
def test_secondary_block_has_no_disadvantaged_rwm():
b = compute_benchmarks(_df())
assert "disadvantaged_rwm_expected_pct" not in b["secondary"]
# KS2-only columns must not produce a fake secondary disadvantaged anchor
# (the old median over all-through schools' KS2 rows produced 50%).
assert b["secondary"]["disadvantaged_pct"] is None
assert b["secondary"]["fsm_pct"] == 13.0
assert b["secondary"]["median_pupils"] == 1100
def test_provenance_string():
+9 -10
View File
@@ -1,7 +1,7 @@
"""_national_averages_payload reads persisted marts (computed at import
time) — it must never aggregate the dataframe. Both marts hold OFFICIAL
DfE figures, so a missing KS4 mart yields an empty secondary series —
never a computed stand-in the UI would mislabel as official."""
time) — it must never loop the dataframe per year. The only dataframe work
allowed is the single-latest-year KS4 fallback for the window between a
deploy and the next DAG run."""
import numpy as np
import pandas as pd
@@ -84,11 +84,10 @@ def test_ks4_averages_come_from_the_mart_not_the_dataframe(payload):
assert body["by_year"][-1]["secondary"]["progress_8_score"] == -0.02
def test_ks4_secondary_empty_when_mart_missing(payload):
# No computed stand-in: the UI labels national figures as official DfE
# data, so an empty mart must yield an empty secondary series.
def test_missing_ks4_mart_falls_back_to_latest_year_only(payload):
body = payload(_Ks4MissingSession)
assert body["secondary"] == {}
assert all(not e["secondary"] for e in body["by_year"])
# The KS2 series is unaffected.
assert body["primary"]["rwm_expected_pct"] == 62.1
# Fallback computes the latest year from the df: mean(50, 30) = 40.0
assert body["secondary"]["attainment_8_score"] == 40.0
# ...and only the latest year — no historical KS4 loop
ks4_years = [e["year"] for e in body["by_year"] if e["secondary"]]
assert ks4_years == [LATEST]
@@ -3,7 +3,6 @@ labels, provider-page URL, graded-vs-carried-forward provenance, and the
admissions preference/cross-LA detail promoted in the data-foundation PR."""
import types
from datetime import date
from backend.data_loader import _admissions_row_dict, _ofsted_block
@@ -41,25 +40,6 @@ def test_grade_source_graded_vs_carried_forward():
assert _ofsted_block(_row(), urn=1)["grade_source"] is None
def test_ofsted_block_carries_rc_inspection_date():
o = _row(
ungraded_grade=2,
rc_achievement=1,
rc_inspection_date=date(2026, 2, 3),
inspection_date=date(2021, 10, 7),
)
block = _ofsted_block(o, urn=138690)
assert block["rc_inspection_date"] == "2026-02-03"
# The legacy inspection date is still present, unchanged.
assert block["inspection_date"] == "2021-10-07"
def test_ofsted_block_rc_inspection_date_none_when_absent():
o = _row(overall_effectiveness=1, inspection_date=date(2021, 10, 13))
block = _ofsted_block(o, urn=136276)
assert block["rc_inspection_date"] is None
def test_ofsted_block_keeps_existing_keys():
block = _ofsted_block(_row(overall_effectiveness=2, quality_of_education=2), urn=1)
for key in ("framework", "inspection_date", "overall_effectiveness",
@@ -1,931 +0,0 @@
# Compare Screen Must-Fix (Final Expert Review) Implementation Plan
> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
**Goal:** Fix the five promotion-blocking findings from the expert's final staging review: (1) blank all-secondary compare view, (2) report cards dated with pre-Nov-2025 legacy inspection dates, (3) FSM chip benchmarked against the wrong measure, (4) KS4 "national averages" that are dataset means presented as official DfE figures, (5) factually wrong "DfE didn't publish 2021/22" footnote.
**Architecture:** One branch/PR touching all three layers. Pipeline: a new tap field carries the report-card inspection's own date; a new EES stream ingests official KS4 national headlines; a new census-benchmarks mart replaces junk KS2-derived context medians. Backend: serialize the new fields, stop mislabelling computed KS4 means as official. Frontend: fix the phase-detection effect that leaves all-secondary comparisons stuck on an empty "primary" tab, date report cards correctly, drop the FSM→disadvantaged fallback, fix the footnote copy.
**Tech Stack:** Meltano/Singer taps (Python), dbt-postgres, FastAPI/SQLAlchemy/pandas, Next.js app router + Jest, Playwright e2e.
## Global Constraints
- Never push to `main`; work on branch `fix/compare-final-review-mustfix`, open a PR. Never trigger the "Promote to Production (manual)" workflow — promotion is exclusively the human's call.
- User-facing behaviour changes must extend the `e2e/` journeys in the same PR (they gate staging fitness and promotability).
- All user-facing copy on the compare screen comes verbatim from `docs/superpowers/specs/mockups/compare-desktop.html` / `compare-mobile.html` — except where this plan explicitly changes copy to fix a factual error (Task 6); the spec/mockup gets the same wording in the same commit.
- Benchmark provenance house style: official figures = "England average"; computed figures = "state-school average (computed from our dataset)".
- A report card must NEVER be displayed with a pre-November-2025 date. Report cards exist only from November 2025.
- Backend tests: `uv run --with-requirements requirements.txt --with pytest --with "httpx==0.27.0" python -m pytest backend/tests -q` (repo root; there is no local pytest).
- dbt: `cd pipeline/transform && uv run --with dbt-postgres python -m dbt.cli.main parse --profiles-dir .` (never bare `dbt` — the Fusion binary shadows dbt-postgres).
- Frontend: `cd nextjs-app && npx tsc --noEmit && npm test` (run tsc un-piped so exit codes are not masked).
- Do NOT start a local server to test the application (CLAUDE.md).
- Commits end with the Claude Code `Co-Authored-By` + `Claude-Session` trailers used on this branch's history.
## Root-Cause Evidence (verified 2026-07-16, do not re-derive)
- **Finding 1:** `nextjs-app/components/ComparisonView.tsx:164-176` — the auto-phase effect returns early when `selectedSchools.length === 0` (basket hydrates a beat after mount) and its dep array is only `[comparisonData]`, so it never re-fires; `comparePhase` stays `'primary'`, `activeSchools` is empty, the page renders "No primary schools in your comparison" (a11y snapshot confirmed). No console errors — not a crash.
- **Finding 2:** In the Ofsted MI CSV (`Management_information_-_state-funded_schools_-_latest_inspections_as_at_31_May_2026.csv`) the report-card grade columns (cols 3855, "Safeguarding standards", "Inclusion", …) belong to the **latest full inspection** block whose date is col 30 "Inspection start date" (Barclay 138690: `03/02/2026`). The tap's `inspection_date` COLUMN_PRIORITY matches col 60 "Inspection start date of latest OEIF graded inspection" first (the *legacy* date; NULL for Barclay, so stg coalesces to the 2021 *ungraded* date). The rc data is **real Ofsted data, not fabricated** — it is mis-dated. Also `discover_csv_url()` returns `matches[0]` = the oldest (2017) link on the GOV.UK page; staging works only because `mi_url` is set in the environment. Staging raw is stale for at least Watford Grammar 136276 (staging shows rc grades; the current MI file has all rc columns NULL for it) — a fresh extract fixes that via upsert on `(urn, inspection_date)`.
- **Finding 3:** `nextjs-app/components/compare/CompareCommunity.tsx:36``bench?.fsm_pct ?? bench?.disadvantaged_pct` falls back across definitions. `benchmarks.primary.fsm_pct` is null because `compute_benchmarks` (backend/data_loader.py:528) medians the *performance* df, which has no `fsm_pct` (school FSM comes from `census.fsm_pct` = `fact_pupil_characteristics`). `disadvantaged_pct` / `eal_pct` are KS2-only columns, so the "secondary" medians (50.0 / 10.0) are computed over the few all-through schools' KS2 rows — junk.
- **Finding 4:** `fact_ks4_national_averages.sql` computes unweighted school means (A8 38.94 vs official 46.0; national P8 0.27, impossible). Official series exists on EES: data-set `1b649e16-01e8-435b-a814-56be2faf9054` ("National characteristics summary data", KS4 performance publication), CSV endpoint same pattern as the KS2 national stream, national level, 2018/19→2024/25, `establishment_type_group = 'All state-funded'`, `breakdown_topic = 'Total'`, `breakdown = 'Total'`. Verified values: 2024/25 A8 46.0, P8 `z` (not published — no KS2 baseline for that cohort), EM 9-5 45.4%, EBacc entry 40.5%. It has **no** `gcse_91_percent` column.
- **Finding 5:** `ComparisonChart.tsx:245-246` claims "DfE didn't publish school-level figures for 2021/22". False — DfE published school-level KS2 for 2021/22 in Dec 2022; spec §8.1 itself lists loading it as a pipeline task. The honest claim is that the figures aren't in our dataset.
---
### Task 1: All-secondary comparison renders (phase-detection fix)
**Files:**
- Modify: `nextjs-app/components/ComparisonView.tsx:176`
- Create: `nextjs-app/__tests__/components/ComparisonView.phase.test.tsx`
- Modify: `e2e/tests/journeys.spec.ts` (add helper + journey after the existing `twoPrimaryUrns` helper / primary compare journey)
**Interfaces:**
- Consumes: existing `ComparisonView` props (`initialData`, `initialUrns`, `metrics`, `selectedMetric`), `ComparisonProvider`.
- Produces: no API changes; the auto-phase effect re-runs when the basket hydrates.
- [ ] **Step 1: Write the failing Jest test**
Create `nextjs-app/__tests__/components/ComparisonView.phase.test.tsx` (mirrors the mock setup of `ComparisonView.refresh.test.tsx`):
```tsx
/**
* Regression: an all-secondary comparison must render the secondary sections.
*
* The basket hydrates from the URL a beat after mount, so the auto-phase
* effect must re-run once selectedSchools arrives — with deps of only
* [comparisonData] it fired once against an empty basket, bailed, and the
* page stayed on an empty "primary" tab ("No primary schools in your
* comparison") even though all schools were secondary.
*/
import { render, screen, waitFor } from '@testing-library/react';
import { ComparisonView } from '@/components/ComparisonView';
import { ComparisonProvider } from '@/context/ComparisonProvider';
import type { ComparisonData, School } from '@/lib/types';
const fetchComparison = jest.fn();
jest.mock('@/lib/api', () => ({
fetchComparison: (...args: unknown[]) => fetchComparison(...args),
}));
jest.mock('@/lib/analytics', () => ({ track: jest.fn() }));
function secondarySchool(urn: number, name: string): School {
return {
urn,
school_name: name,
local_authority: 'Testshire',
school_type: 'Academy converter',
attainment_8_score: 55,
phase: 'Secondary',
} as School;
}
function data(urn: number, name: string): ComparisonData {
return {
school_info: secondarySchool(urn, name),
yearly_data: [{ year: 202425, attainment_8_score: 55 }] as ComparisonData['yearly_data'],
ofsted: null,
census: null,
admissions: null,
admissions_history: [],
deprivation: null,
};
}
const INITIAL_DATA = {
'300': data(300, 'Gamma High'),
'400': data(400, 'Delta Academy'),
};
test('an all-secondary comparison renders the sections, not an empty primary tab', async () => {
render(
<ComparisonProvider>
<ComparisonView
initialData={INITIAL_DATA}
initialNationalAverages={{
year: 202425,
primary: {},
secondary: { attainment_8_score: 46 },
by_year: [],
}}
initialBenchmarks={undefined}
initialUrns={[300, 400]}
metrics={[]}
selectedMetric="attainment_8_score"
/>
</ComparisonProvider>,
);
await waitFor(() => {
expect(screen.getByRole('heading', { name: 'At a glance' })).toBeInTheDocument();
});
expect(screen.getAllByText('Gamma High').length).toBeGreaterThan(0);
expect(screen.queryByText(/No primary schools in your comparison/)).toBeNull();
expect(fetchComparison).not.toHaveBeenCalled();
});
```
- [ ] **Step 2: Run it to verify it fails**
Run: `cd nextjs-app && npx jest __tests__/components/ComparisonView.phase.test.tsx`
Expected: FAIL — "No primary schools in your comparison" is rendered / "At a glance" never appears.
- [ ] **Step 3: Fix the effect dependencies**
In `nextjs-app/components/ComparisonView.tsx`, the auto-phase effect currently ends:
```tsx
}, [comparisonData]); // eslint-disable-line react-hooks/exhaustive-deps
```
Change to:
```tsx
// selectedSchools is a dep because the basket hydrates after mount: the
// first run sees an empty basket and bails, so it must re-fire when the
// schools arrive. primarySchools/secondarySchools/metrics/selectedMetric
// are intentionally omitted (derived or would cause loops).
}, [comparisonData, selectedSchools]); // eslint-disable-line react-hooks/exhaustive-deps
```
(`phaseLockedByUser` still suppresses re-detection after a manual tab click; re-running with unchanged inputs sets the same state, which React treats as a no-op.)
- [ ] **Step 4: Run the new test and the existing suite**
Run: `cd nextjs-app && npx tsc --noEmit && npm test`
Expected: PASS, including `ComparisonView.refresh.test.tsx` (the refresh regression must stay green).
- [ ] **Step 5: Add the e2e secondary journey**
In `e2e/tests/journeys.spec.ts`, add below `twoPrimaryUrns`:
```ts
async function twoSecondaryUrns(page: Page): Promise<[string, string]> {
const res = await page.request.get('/api/schools?search=school&per_page=100');
expect(res.ok()).toBeTruthy();
const body = await res.json();
const urns: string[] = (body.schools ?? [])
.filter((s: { phase?: string; attainment_8_score?: number | null }) =>
s.phase === 'Secondary' && s.attainment_8_score != null,
)
.map((s: { urn: number }) => String(s.urn));
expect(urns.length).toBeGreaterThanOrEqual(2);
return [urns[0], urns[1]];
}
```
(If `/api/schools` list rows lack `attainment_8_score`, filter on `s.phase === 'Secondary'` only — check the response first.) Then add a journey test next to the primary compare journey:
```ts
test('comparing two secondary schools renders the secondary sections', async ({ page }) => {
const [urn0, urn1] = await twoSecondaryUrns(page);
await page.goto(`/compare?urns=${urn0},${urn1}`);
await expect(page.locator(`a[href*="${urn0}"]`).first()).toBeVisible({ timeout: 15_000 });
// The parent-first sections must render — this page was completely blank
// for all-secondary baskets (expert review must-fix #1).
await expect(page.getByRole('heading', { name: 'At a glance' })).toBeVisible();
await expect(page.getByRole('heading', { name: 'Ofsted inspection' })).toBeVisible();
// A KS4 measure proves the secondary academics variant rendered.
await expect(page.getByText(/Attainment 8/i).first()).toBeVisible();
await expect(page.getByText(/No primary schools in your comparison/)).toHaveCount(0);
});
```
- [ ] **Step 6: Commit**
```bash
git add nextjs-app/components/ComparisonView.tsx nextjs-app/__tests__/components/ComparisonView.phase.test.tsx e2e/tests/journeys.spec.ts
git commit -m "fix(compare): render all-secondary comparisons — re-run phase detection after basket hydration"
```
---
### Task 2: Report-card inspection date through the pipeline
**Files:**
- Modify: `pipeline/plugins/extractors/tap-uk-ofsted/tap_uk_ofsted/tap.py` (COLUMN_PRIORITY, schema, `discover_csv_url`)
- Modify: `pipeline/transform/models/staging/stg_ofsted_inspections.sql`
- Modify: `pipeline/transform/models/intermediate/int_ofsted_latest.sql`
- Modify: `pipeline/transform/models/marts/fact_ofsted_inspection.sql`
- Modify: `pipeline/transform/models/marts/_marts_schema.yml` (add column doc if other fact_ofsted columns are documented there)
**Interfaces:**
- Consumes: MI CSV column `Inspection start date` (the latest **full** inspection = the report-card inspection in the renewed framework; NULL when a school's only inspections are legacy OEIF/ungraded — verified for Watford Grammar).
- Produces: `marts.fact_ofsted_inspection.rc_inspection_date` (DATE, null unless the row carries report-card grades). Task 3 depends on this exact column name.
- [ ] **Step 1: Add the tap field**
In `tap.py` COLUMN_PRIORITY, after the `rc_sixth_form` entry, add:
```python
# Date of the latest FULL inspection — in the renewed framework this is
# the report-card inspection's own start date (col "Inspection start
# date"), distinct from the legacy OEIF graded/ungraded dates above.
"rc_inspection_date": ["Inspection start date"],
```
and in the stream schema, next to the other rc properties:
```python
th.Property("rc_inspection_date", th.StringType),
```
Note: `inspection_date`'s own priority list also contains `"Inspection start date"` as a lower-priority candidate — that stays; in renewed-framework files the higher-priority OEIF column exists so they map to different columns, and in legacy files both map to the same column but rc grades are absent, and staging nulls `rc_inspection_date` in that case (Step 3).
- [ ] **Step 2: Fix `discover_csv_url` to pick the newest file, not `matches[0]`**
The GOV.UK page lists 2017 files first; `matches[0]` is a 2017 CSV. Replace the body of `discover_csv_url()` to date-sort the `latest_inspections_as_at` links, mirroring `discover_independent_csv_url`:
```python
def discover_csv_url() -> str | None:
"""Scrape GOV.UK page to find the latest MI CSV download link.
The page lists a decade of monthly files, oldest first — take the
newest 'latest inspections as at <date>' link by parsing its date,
never matches[0].
"""
resp = requests.get(GOV_UK_PAGE, timeout=30)
resp.raise_for_status()
csv_links = re.findall(
r'href="(https://assets\.publishing\.service\.gov\.uk/[^"]+\.csv)"',
resp.text,
)
months = {
'january': 1, 'february': 2, 'march': 3, 'april': 4, 'may': 5, 'june': 6,
'july': 7, 'august': 8, 'september': 9, 'october': 10, 'november': 11, 'december': 12,
'jan': 1, 'feb': 2, 'mar': 3, 'apr': 4, 'jun': 6,
'jul': 7, 'aug': 8, 'sep': 9, 'oct': 10, 'nov': 11, 'dec': 12,
}
parsed_links = []
for link in csv_links:
normalized = link.lower().replace('-', '_')
if 'latest_inspections_as_at' not in normalized:
continue
match = re.search(r'as_at_(\d{1,2})_([a-z]+)_(\d{4})', normalized)
if match:
day, month_str, year = match.groups()
month = months.get(month_str)
if month:
try:
parsed_links.append((datetime(int(year), month, int(day)), link))
except ValueError:
continue
parsed_links.sort(reverse=True)
if parsed_links:
return parsed_links[0][1]
if csv_links:
return csv_links[-1]
matches = re.findall(
r'href="(https://assets\.publishing\.service\.gov\.uk/[^"]+\.ods)"',
resp.text,
)
return matches[0] if matches else None
```
(`mi_url` config still wins when set — `self.config.get("mi_url") or discover_csv_url()` is unchanged.)
- [ ] **Step 3: Parse and guard the date in staging**
In `stg_ofsted_inspections.sql`, inside the `renamed` CTE after the `rc_sixth_form` line, add:
```sql
-- Start date of the latest FULL inspection (the report-card
-- inspection in the renewed framework). Guarded below: only kept
-- when the row actually carries report-card grades, because in
-- legacy-format files this column is the legacy inspection date.
to_date(nullif(trim(rc_inspection_date), 'NULL'), 'DD/MM/YYYY') as rc_inspection_date_raw,
```
and replace the final select:
```sql
select
*,
case
when rc_safeguarding_met is not null
or rc_inclusion is not null
or rc_curriculum_teaching is not null
or rc_achievement is not null
or rc_attendance_behaviour is not null
or rc_personal_development is not null
or rc_leadership_governance is not null
then rc_inspection_date_raw
end as rc_inspection_date
from renamed
where inspection_date is not null
```
- [ ] **Step 4: Propagate through int + mart**
Add `rc_inspection_date,` to the explicit column lists of `int_ofsted_latest.sql` and `fact_ofsted_inspection.sql` (after `rc_sixth_form`). Do NOT propagate `rc_inspection_date_raw`.
- [ ] **Step 5: Parse-check dbt**
Run: `cd pipeline/transform && uv run --with dbt-postgres python -m dbt.cli.main parse --profiles-dir .`
Expected: parse OK, no compilation errors.
- [ ] **Step 6: Commit**
```bash
git add pipeline/plugins/extractors/tap-uk-ofsted pipeline/transform/models
git commit -m "feat(pipeline): carry the report-card inspection's own date; pick newest MI file in discovery"
```
---
### Task 3: Report-card date in the API and UI
**Files:**
- Modify: `backend/models.py` (FactOfstedInspection)
- Modify: `backend/data_loader.py` (`_ofsted_block`)
- Test: `backend/tests/test_supplementary_enrichment.py` (extend the existing `_ofsted_block` tests)
- Modify: `nextjs-app/lib/types.ts` (OfstedInspection)
- Modify: `nextjs-app/components/compare/CompareOfsted.tsx` ("Inspected" measure)
- Test: `nextjs-app/__tests__/components/CompareOfsted.test.tsx`
**Interfaces:**
- Consumes: `marts.fact_ofsted_inspection.rc_inspection_date` (Task 2).
- Produces: API `ofsted.rc_inspection_date: string | null` (ISO date). UI rule: report-card displays are dated with `rc_inspection_date` only; when null, show "—" (never the legacy date).
- [ ] **Step 1: Failing backend test**
In `backend/tests/test_supplementary_enrichment.py`, alongside the existing `_ofsted_block` tests, add (reuse the file's existing fake-row helper/style):
```python
def test_ofsted_block_carries_rc_inspection_date():
o = _fake_ofsted_row( # use this file's existing fake/stub construction
overall_effectiveness=None,
ungraded_grade=2,
rc_achievement=1,
rc_inspection_date=date(2026, 2, 3),
inspection_date=date(2021, 10, 7),
)
block = _ofsted_block(o, 138690)
assert block["rc_inspection_date"] == "2026-02-03"
# The legacy inspection date is still present, unchanged.
assert block["inspection_date"] == "2021-10-07"
def test_ofsted_block_rc_inspection_date_none_when_absent():
o = _fake_ofsted_row(overall_effectiveness=1, inspection_date=date(2021, 10, 13))
block = _ofsted_block(o, 136276)
assert block["rc_inspection_date"] is None
```
Run: `uv run --with-requirements requirements.txt --with pytest --with "httpx==0.27.0" python -m pytest backend/tests/test_supplementary_enrichment.py -q`
Expected: FAIL (KeyError / AttributeError on `rc_inspection_date`).
- [ ] **Step 2: Backend implementation**
`backend/models.py`, in `FactOfstedInspection` after `rc_sixth_form`:
```python
# Start date of the report-card inspection itself (renewed framework,
# Nov 2025+). Null for rows without report-card grades.
rc_inspection_date = Column(Date)
```
`backend/data_loader.py` `_ofsted_block`, after the `"inspection_date"` entry:
```python
"rc_inspection_date": (
o.rc_inspection_date.isoformat()
if getattr(o, "rc_inspection_date", None)
else None
),
```
(`getattr` default keeps old test stubs working.) Run the backend suite; expected: PASS.
- [ ] **Step 3: Failing frontend test**
`nextjs-app/lib/types.ts`, in `OfstedInspection`, after `inspection_date`:
```ts
/** Start date of the report-card inspection itself (Nov 2025+); null otherwise. */
rc_inspection_date?: string | null;
```
In `nextjs-app/__tests__/components/CompareOfsted.test.tsx`, add to the existing suite (reusing its fixture style):
```tsx
it('dates a report card with the report-card inspection date, never the legacy date', () => {
const ofsted = reportCardOfsted({
inspection_date: '2021-10-07',
rc_inspection_date: '2026-02-03',
});
render(<CompareOfsted schools={[schoolFixture]} data={{ [String(schoolFixture.urn)]: { ...dataFixture, ofsted } }} />);
expect(screen.getByText(/3 Feb 2026/)).toBeInTheDocument();
expect(screen.queryByText(/7 Oct 2021/)).toBeNull();
expect(screen.queryByText('4+ years ago')).toBeNull();
});
it('shows an em dash when a report card has no rc_inspection_date yet', () => {
const ofsted = reportCardOfsted({ inspection_date: '2021-10-07', rc_inspection_date: null });
render(<CompareOfsted schools={[schoolFixture]} data={{ [String(schoolFixture.urn)]: { ...dataFixture, ofsted } }} />);
expect(screen.getByText('—')).toBeInTheDocument();
expect(screen.queryByText(/7 Oct 2021/)).toBeNull();
});
```
(`reportCardOfsted` = the file's existing report-card fixture builder, or build inline matching its other tests.) Run just this file; expected: FAIL.
- [ ] **Step 4: Frontend implementation**
In `CompareOfsted.tsx`, replace the body of the "Inspected" measure's map:
```tsx
{schools.map((school, i) => {
const ofsted = data[String(school.urn)]?.ofsted;
// A report card is dated by its OWN inspection date. The legacy
// inspection_date belongs to an older inspection and must never
// be shown against a report card (report cards exist only from
// Nov 2025).
const dateIso =
displays[i].kind === 'report_card'
? ofsted?.rc_inspection_date ?? null
: ofsted?.inspection_date ?? null;
const age = yearsSince(dateIso);
return (
<Cell key={school.urn} school={school} index={i}>
{formatInspectionDate(dateIso)}{' '}
{age != null && age > 4 && <Chip tone="neutral">4+ years ago</Chip>}
</Cell>
);
})}
```
- [ ] **Step 5: Run frontend checks**
Run: `cd nextjs-app && npx tsc --noEmit && npm test`
Expected: PASS.
- [ ] **Step 6: Commit**
```bash
git add backend/models.py backend/data_loader.py backend/tests nextjs-app/lib/types.ts nextjs-app/components/compare/CompareOfsted.tsx nextjs-app/__tests__/components/CompareOfsted.test.tsx
git commit -m "fix(compare): date report cards with their own inspection date, never the legacy one"
```
---
### Task 4: Census-based context benchmarks; kill the FSM fallback
**Files:**
- Create: `pipeline/transform/models/marts/fact_census_benchmarks.sql`
- Modify: `pipeline/transform/models/marts/_marts_schema.yml`
- Modify: `backend/models.py` (new `CensusBenchmark` model)
- Modify: `backend/data_loader.py` (`compute_benchmarks`)
- Modify: `backend/app.py` (compare endpoint call site, only if the signature change requires it)
- Test: `backend/tests/test_benchmarks.py`
- Modify: `nextjs-app/components/compare/CompareCommunity.tsx:36`
- Test: `nextjs-app/__tests__/lib/compareLogic.test.ts` or the community section's existing test home (add a fallback-removal test where the FSM chip logic is tested today)
**Interfaces:**
- Consumes: `marts.fact_pupil_characteristics` (urn, year, phase_type_grouping, total_pupils, fsm_pct, eal_pct).
- Produces: `marts.fact_census_benchmarks` — one row per phase (`'primary'`/`'secondary'`), columns `phase, year, fsm_pct, eal_pct, median_pupils`. `fsm_pct`/`eal_pct` are **pupil-weighted means** (so they approximate the national pupil-level rate, answering the expert's objection to school-median anchors). API `benchmarks.{primary,secondary}` keeps its existing keys; `fsm_pct`/`eal_pct`/`median_pupils` now come from this mart; `disadvantaged_pct` becomes primary-only (the KS2-column median was junk for secondary).
- [ ] **Step 1: dbt mart**
Create `pipeline/transform/models/marts/fact_census_benchmarks.sql`:
```sql
{{ config(materialized='table') }}
-- Mart: state-school context benchmarks from the pupil census — one row per
-- phase, latest census year. Computed at import time (never per request).
-- fsm_pct / eal_pct are pupil-weighted means, i.e. "what % of pupils", not
-- "the median school" — this matches how DfE quotes national FSM/EAL rates.
-- Consumers must label these "state-school average (computed from our
-- dataset)" (spec §8.6), never "England average".
with latest as (
select max(year) as year from {{ ref('fact_pupil_characteristics') }}
),
classified as (
select
case
when p.phase_type_grouping ilike '%primary%' then 'primary'
when p.phase_type_grouping ilike '%secondary%' then 'secondary'
end as phase,
p.total_pupils,
p.fsm_pct,
p.eal_pct,
l.year
from {{ ref('fact_pupil_characteristics') }} p
join latest l on p.year = l.year
where p.total_pupils is not null and p.total_pupils > 0
)
select
phase,
max(year) as year,
round((sum(fsm_pct * total_pupils) filter (where fsm_pct is not null)
/ nullif(sum(total_pupils) filter (where fsm_pct is not null), 0))::numeric, 1) as fsm_pct,
round((sum(eal_pct * total_pupils) filter (where eal_pct is not null)
/ nullif(sum(total_pupils) filter (where eal_pct is not null), 0))::numeric, 1) as eal_pct,
round(percentile_cont(0.5) within group (order by total_pupils))::integer as median_pupils
from classified
where phase is not null
group by phase
```
Add a `fact_census_benchmarks` entry to `_marts_schema.yml` in the file's existing style (name + description; column tests only if sibling marts have them).
Run: `cd pipeline/transform && uv run --with dbt-postgres python -m dbt.cli.main parse --profiles-dir .` — expected PASS.
- [ ] **Step 2: Failing backend test**
In `backend/tests/test_benchmarks.py` add:
```python
def test_benchmarks_use_census_mart_for_context(monkeypatch):
census = {
"primary": {"year": 202425, "fsm_pct": 25.3, "eal_pct": 21.8, "median_pupils": 240},
"secondary": {"year": 202425, "fsm_pct": 24.1, "eal_pct": 18.9, "median_pupils": 980},
}
result = compute_benchmarks(_sample_df(), census_benchmarks=census)
assert result["primary"]["fsm_pct"] == 25.3
assert result["secondary"]["eal_pct"] == 18.9
assert result["secondary"]["median_pupils"] == 980
# KS2-only columns must not produce a fake secondary disadvantaged anchor.
assert result["secondary"]["disadvantaged_pct"] is None
def test_benchmarks_context_none_when_mart_missing():
result = compute_benchmarks(_sample_df(), census_benchmarks=None)
assert result["primary"]["fsm_pct"] is None # never silently fall back
```
(`_sample_df()` = this file's existing dataframe fixture.) Run the file; expected: FAIL (unexpected keyword `census_benchmarks`).
- [ ] **Step 3: Backend implementation**
`backend/models.py` (next to the national-average models):
```python
class CensusBenchmark(Base):
"""State-school context benchmarks from the pupil census — one row per phase."""
__tablename__ = "fact_census_benchmarks"
__table_args__ = MARTS
phase = Column(String(20), primary_key=True)
year = Column(Integer)
fsm_pct = Column(Float) # pupil-weighted mean
eal_pct = Column(Float) # pupil-weighted mean
median_pupils = Column(Integer)
```
`backend/data_loader.py` — change the signature and `_block`:
```python
def compute_benchmarks(df: pd.DataFrame, census_benchmarks: dict | None = None) -> dict:
```
Inside, keep `_median` and `_weighted_disadvantaged` as-is, and replace `_block` with:
```python
def _block(sub, phase, with_disadvantaged):
census = (census_benchmarks or {}).get(phase) or {}
block = {
# Context measures come from the census mart (pupil-weighted):
# the performance df has no fsm_pct, and its eal/disadvantaged
# columns are KS2-only — medianing them for "secondary" produced
# junk anchors from the handful of all-through schools.
"eal_pct": census.get("eal_pct"),
"sen_support_pct": _median(sub, "sen_support_pct"),
"disadvantaged_pct": _median(sub, "disadvantaged_pct") if with_disadvantaged else None,
"fsm_pct": census.get("fsm_pct"),
"median_pupils": census.get("median_pupils"),
}
if with_disadvantaged:
block["disadvantaged_rwm_expected_pct"] = _weighted_disadvantaged(sub)
return block
```
and the return:
```python
return {
"source": "state-school average (computed from our dataset)",
"year": int(latest_year),
"primary": _block(prim, "primary", with_disadvantaged=True),
"secondary": _block(sec, "secondary", with_disadvantaged=False),
}
```
In `backend/app.py`'s compare endpoint, load the mart and pass it (same defensive style as the national-averages queries):
```python
census_benchmarks = None
try:
rows = db.query(CensusBenchmark).all()
if rows:
census_benchmarks = {
r.phase: {
"year": r.year,
"fsm_pct": r.fsm_pct,
"eal_pct": r.eal_pct,
"median_pupils": r.median_pupils,
}
for r in rows
}
except Exception:
db.rollback()
...
"benchmarks": compute_benchmarks(df, census_benchmarks=census_benchmarks),
```
(Import `CensusBenchmark`; use the endpoint's existing db session pattern.) Run the backend suite; expected: PASS (update any existing benchmark tests that asserted the old median-sourced fsm/eal values).
- [ ] **Step 4: Frontend — remove the cross-definition fallback**
`nextjs-app/components/compare/CompareCommunity.tsx:36`:
```tsx
const anchor = bench?.fsm_pct ?? null;
```
If the FSM chip has unit coverage, update/add the case: `anchor` null ⇒ no verdict chip rendered (bare value only). Run `cd nextjs-app && npx tsc --noEmit && npm test` — expected PASS.
- [ ] **Step 5: Commit**
```bash
git add pipeline/transform/models/marts backend/models.py backend/data_loader.py backend/app.py backend/tests/test_benchmarks.py nextjs-app/components/compare/CompareCommunity.tsx nextjs-app/__tests__
git commit -m "fix(compare): census-sourced FSM/EAL benchmarks; never fall back across measure definitions"
```
---
### Task 5: Official KS4 national averages
**Files:**
- Modify: `pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py` (new stream, registered in `discover_streams`)
- Create: `pipeline/transform/models/staging/stg_ees_ks4_national.sql`
- Modify: `pipeline/transform/models/staging/_stg_sources.yml` (add raw table `ees_ks4_national`)
- Modify: `pipeline/transform/models/marts/fact_ks4_national_averages.sql` (rewrite)
- Modify: `backend/models.py` (Ks4NationalAverage docstring), `backend/app.py` (`_national_averages_payload` — remove the computed fallback)
- Test: `backend/tests/test_national_averages_marts.py`
**Interfaces:**
- Consumes: EES data-catalogue CSV `https://explore-education-statistics.service.gov.uk/data-catalogue/data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv` (columns verified: `time_period, geographic_level, establishment_type_group, breakdown_topic, breakdown, attainment8_average, progress8_average, engmath_95_percent, engmath_94_percent, ebacc_entering_percent, ebacc_95_percent, ebacc_94_percent, ebacc_aps_average, …`).
- Produces: `marts.fact_ks4_national_averages` with the SAME columns as today (so `Ks4NationalAverage` needs no schema change), now holding official DfE figures; `gcse_grade_91_pct` is NULL (not in the official series — the England anchor for that measure disappears, which is correct: it was noise).
- [ ] **Step 1: Tap stream**
In `tap.py`, after the KS2 national stream, add:
```python
# ── KS4 National Headlines (national level only — one row per year) ──────────
# Dataset: "National characteristics summary data" (Key stage 4 performance).
# Official England state-funded headline measures, 2018/19 → latest.
# Suppressed values ('z', 'x') → NULL downstream. Progress 8 is legitimately
# absent in years with no KS2 baseline (e.g. 2024/25) — that is DfE policy,
# not missing data.
_KS4_NATIONAL_CSV_URL = (
"https://explore-education-statistics.service.gov.uk/data-catalogue/"
"data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv"
)
_KS4_NATIONAL_COL_MAP = {
"attainment8_average": "attainment_8_score",
"progress8_average": "progress_8_score",
"engmath_94_percent": "english_maths_standard_pass_pct",
"engmath_95_percent": "english_maths_strong_pass_pct",
"ebacc_entering_percent": "ebacc_entry_pct",
"ebacc_94_percent": "ebacc_standard_pass_pct",
"ebacc_95_percent": "ebacc_strong_pass_pct",
"ebacc_aps_average": "ebacc_avg_score",
}
class EESKs4NationalStream(Stream):
"""National KS4 headline averages — one row per academic year.
Filters to geographic_level == 'National', establishment_type_group ==
'All state-funded', breakdown_topic == 'Total', breakdown == 'Total'
so only the England-wide all-pupils row per year is emitted.
"""
name = "ees_ks4_national"
primary_keys = ["time_period"]
replication_key = None
schema = th.PropertiesList(
th.Property("time_period", th.StringType, required=True),
*[th.Property(out, th.StringType) for out in _KS4_NATIONAL_COL_MAP.values()],
).to_dict()
def get_records(self, context):
import pandas as pd
self.logger.info("Downloading KS4 national headlines: %s", _KS4_NATIONAL_CSV_URL)
resp = requests.get(_KS4_NATIONAL_CSV_URL, timeout=60)
resp.raise_for_status()
df = pd.read_csv(io.BytesIO(resp.content), dtype=str, keep_default_na=False)
df.columns = [c.strip().lower() for c in df.columns]
for col, want in [
("geographic_level", "national"),
("establishment_type_group", "all state-funded"),
("breakdown_topic", "total"),
("breakdown", "total"),
]:
if col in df.columns:
df = df[df[col].str.strip().str.lower() == want]
self.logger.info("Emitting %d national KS4 rows", len(df))
for _, row in df.iterrows():
record = {"time_period": row.get("time_period", "").strip()}
for src, out in _KS4_NATIONAL_COL_MAP.items():
record[out] = row.get(src, "")
yield record
```
Register `EESKs4NationalStream(self)` in `discover_streams` next to the KS2 national stream.
- [ ] **Step 2: Raw source + staging model**
Add to `_stg_sources.yml` under the raw source, matching the `ees_ks2_national` entry's style:
```yaml
- name: ees_ks4_national
description: Official DfE KS4 national headline averages (EES data catalogue)
```
Create `pipeline/transform/models/staging/stg_ees_ks4_national.sql`:
```sql
{{ config(materialized='table') }}
-- Staging model: official DfE KS4 national headline averages — one row per
-- academic year (England, all state-funded, all pupils). Source: EES data
-- catalogue "National characteristics summary data". Suppressed values
-- ('z', 'x') are coerced to NULL by safe_numeric — Progress 8 is 'z' in
-- years with no KS2 baseline (e.g. 2024/25): legitimately unpublished.
select
cast(trim(time_period) as integer) as year,
{{ safe_numeric('attainment_8_score') }} as attainment_8_score,
{{ safe_numeric('progress_8_score') }} as progress_8_score,
{{ safe_numeric('english_maths_standard_pass_pct') }} as english_maths_standard_pass_pct,
{{ safe_numeric('english_maths_strong_pass_pct') }} as english_maths_strong_pass_pct,
{{ safe_numeric('ebacc_entry_pct') }} as ebacc_entry_pct,
{{ safe_numeric('ebacc_standard_pass_pct') }} as ebacc_standard_pass_pct,
{{ safe_numeric('ebacc_strong_pass_pct') }} as ebacc_strong_pass_pct,
{{ safe_numeric('ebacc_avg_score') }} as ebacc_avg_score
from {{ source('raw', 'ees_ks4_national') }}
where time_period ~ '^[0-9]+$'
```
- [ ] **Step 3: Rewrite the mart**
Replace the entire body of `fact_ks4_national_averages.sql`:
```sql
{{ config(materialized='table') }}
-- Mart: OFFICIAL DfE KS4 national headline averages — one row per academic
-- year (England, state-funded, all pupils), from the EES national dataset.
-- Replaces the previous unweighted school-level means, which were 715
-- points off every headline measure and produced an arithmetically
-- impossible national Progress 8. gcse_grade_91_pct has no official
-- national series and is NULL (schema kept for the API model).
select
year,
attainment_8_score,
progress_8_score,
english_maths_standard_pass_pct,
english_maths_strong_pass_pct,
ebacc_entry_pct,
ebacc_standard_pass_pct,
ebacc_strong_pass_pct,
ebacc_avg_score,
cast(null as double precision) as gcse_grade_91_pct
from {{ ref('stg_ees_ks4_national') }}
order by year
```
Run: `cd pipeline/transform && uv run --with dbt-postgres python -m dbt.cli.main parse --profiles-dir .` — expected PASS.
- [ ] **Step 4: Backend — official provenance, no computed fallback**
`backend/models.py`: change the `Ks4NationalAverage` docstring to `"""Official DfE KS4 national headline averages — one row per academic year."""`.
`backend/app.py` `_national_averages_payload`: delete the entire `if not any(secondary_by_year.values()):` fallback block (it computes dataset means that the UI footnote then labels official). Update the function docstring's KS4 sentence to: `official DfE KS4 figures (fact_ks4_national_averages). If the KS4 mart hasn't been built yet, the secondary series is empty — never a computed stand-in, because the UI labels these figures as official.`
Update `backend/tests/test_national_averages_marts.py`: the test that exercised the fallback now asserts the opposite —
```python
def test_ks4_secondary_empty_when_mart_missing(...):
# No computed stand-in: the UI labels national figures as official DfE
# data, so an empty mart must yield an empty secondary series.
payload = _national_averages_payload(df)
assert all(not e["secondary"] for e in payload["by_year"])
```
(adapt to the file's existing fixtures/monkeypatching). Run the backend suite — expected PASS.
- [ ] **Step 5: Commit**
```bash
git add pipeline/plugins/extractors/tap-uk-ees pipeline/transform backend
git commit -m "fix(data): official DfE KS4 national headline averages; drop mislabelled computed means"
```
---
### Task 6: Honest 2021/22 footnote
**Files:**
- Modify: `nextjs-app/components/ComparisonChart.tsx:243-247`
- Modify: `nextjs-app/lib/compareChartData.ts` (comment lines 7, 5355 — comments only, no logic)
- Modify: `nextjs-app/__tests__/lib/compareChartData.test.ts` (test name/comment wording only)
- Modify: `docs/superpowers/specs/2026-07-11-compare-screen-redesign-design.md` §8.1
**Interfaces:** none — copy and docs only. This is the one place the plan changes reviewed copy, because the reviewed copy is factually wrong (Global Constraints exception).
- [ ] **Step 1: Fix the user-facing copy**
In `ComparisonChart.tsx` replace the note:
```tsx
{built.showUnpublished202122Note && (
<p className={styles.chartNote}>
No national tests were held in 2019/20 and 2020/21 (COVID), and our dataset doesn&apos;t
yet include school-level figures for 2021/22 the England average is shown for that
year.
</p>
)}
```
- [ ] **Step 2: Fix the lying comments**
In `compareChartData.ts`, update the header comment (line 7) and the `showUnpublished202122Note` doc comment (lines 5355) to say the 2021/22 school-level figures are *absent from our dataset* (DfE published them in Dec 2022; ingesting them is a backlog pipeline task), not "unpublished". Rename nothing (the flag name stays — pure rename churn). In `compareChartData.test.ts`, adjust the test description/comment wording the same way.
- [ ] **Step 3: Correct spec §8.1**
In the spec's §8.1, replace any wording that calls 2021/22 school-level KS2 a "permanent DfE gap" with: DfE published school-level KS2 results for 2021/22 in December 2022 (with comparability caveats); they are not yet ingested — loading them remains an open pipeline task, and the chart footnote says "our dataset doesn't yet include" accordingly.
- [ ] **Step 4: Verify + commit**
Run: `cd nextjs-app && npx tsc --noEmit && npm test` — expected PASS.
```bash
git add nextjs-app docs/superpowers/specs/2026-07-11-compare-screen-redesign-design.md
git commit -m "fix(compare): stop attributing the missing 2021/22 school-level year to DfE"
```
---
### Task 7: Full verification, PR, and post-deploy checklist
**Files:** none new (verification + PR).
- [ ] **Step 1: Run everything**
```bash
uv run --with-requirements requirements.txt --with pytest --with "httpx==0.27.0" python -m pytest backend/tests -q
cd nextjs-app && npx tsc --noEmit && npm test && cd ..
cd pipeline/transform && uv run --with dbt-postgres python -m dbt.cli.main parse --profiles-dir . && cd ../..
```
Expected: all PASS.
- [ ] **Step 2: Open the PR**
Push `fix/compare-final-review-mustfix`; open a PR via the Gitea API using `git credential fill` basic auth (token-header auth 401s). PR body: summarize the five findings and fixes, link the expert review, end with the standard Claude Code attribution + session URL. Note in the body that findings 2 and 4 also need a **DAG run after the staging deploy** before the UI shows corrected data.
- [ ] **Step 3: Post-merge staging verification (after the user merges and the daily DAG runs — record results, do not promote)**
```bash
# Report card dated by its own inspection (Barclay): expect 2026-02-03
curl -sk "https://stx.schoolcompare.co.uk/api/compare?urns=138690" | python3 -c "import json,sys; o=json.load(sys.stdin)['comparison']['138690']['ofsted']; print(o['rc_inspection_date'], o['inspection_date'])"
# Stale Watford rc grades cleared by the fresh extract: expect report_card == {}
curl -sk "https://stx.schoolcompare.co.uk/api/compare?urns=136276" | python3 -c "import json,sys; print(json.load(sys.stdin)['comparison']['136276']['ofsted']['report_card'])"
# Official KS4 nationals: expect A8 46.0 for 202425, progress_8_score absent
curl -sk "https://stx.schoolcompare.co.uk/api/national-averages" | python3 -c "import json,sys; print(json.load(sys.stdin)['secondary'])"
# FSM benchmark real (~24-26), secondary disadvantaged_pct gone
curl -sk "https://stx.schoolcompare.co.uk/api/compare?urns=138690,136276" | python3 -c "import json,sys; print(json.load(sys.stdin)['benchmarks'])"
```
Then re-screenshot both phase views (desktop + mobile, "More measures" expanded, Watford Grammar in the secondary set) and hand them to the Ofsted expert agent for the sign-off pass it said it expects. Production promotion remains the human's manual call.
---
## Out of Scope (expert should-fix/minor — separate follow-ups)
- 137086-style interim state (subgrades without an overall from an RI reinspection) rendering treatment (finding 6).
- Disadvantaged cohort sizes on the attainment row (finding 7, spec §8.5).
- SEN/EAL "typical school" labelling and secondary SEN benchmark (finding 8) — note Task 4 already upgrades EAL to a pupil-weighted census figure.
- Selective-school admissions copy variant (finding 9).
- Removing/relabelling `gcse_grade_91_pct` as a compare measure (finding 10) — Task 5 already removes its false England anchor.
- Palette deviation (11), trends picker label (12), "More measures" expanded-state verification (13).
- Actually ingesting the 2021/22 school-level KS2 release (the copy in Task 6 says "doesn't *yet* include").
@@ -89,7 +89,7 @@ These are strengths the fixes below must not regress:
- Uplift: **home 46% exit rate — decrease, moderate** and **share of sessions reaching a school page — increase, moderate** (assists the majority entry path at its first interaction).
- **P1.7 — The mobile hero omits the value proposition entirely** *(J1-F6)*
- Evidence: desktop shows the "UPDATED WITH 2026/2027 ADMISSIONS RESULTS" trust badge and the "24,000+ schools… side by side, in one place" subheading; mobile renders only the poetic H1 ("Every school in England, *compared.*") and a bare search box (`j1-home-desktop-fold.png` vs `j1-home-mobile-fold.png`).
- Evidence: desktop shows the "UPDATED WITH 2026/2027 ADMISSIONS RESULTS" trust badge and the "27,000+ schools… side by side, in one place" subheading; mobile renders only the poetic H1 ("Every school in England, *compared.*") and a bare search box (`j1-home-desktop-fold.png` vs `j1-home-mobile-fold.png`).
- Criterion: mobile content parity; Nielsen #1 — first-visit orientation ("what is this, why trust it") absent on the primary viewport.
- Argument: 63% of entries land here and 56% of traffic is mobile; a first-time visitor gets no statement of coverage, data source, or freshness above the fold. Weak value proposition at first glance is a classic bounce driver and plausibly a material slice of the 46% exit rate.
- Recommendation: restore a compact version of the badge + one-line value prop under the mobile H1 (one text block; the fold has room above the deadline rail).
@@ -256,12 +256,7 @@ implementation, beyond what the mockups can show:
DfE stated it would not publish KS2 2021/22 in performance tables
(verified 2026-07-12 against EES, the CSP download service, and
DfE release notes; see `# TASK 6 VERIFICATION` in
`pipeline/scripts/diagnose_compare_gaps.py`; re-verified 2026-07-16
after an expert-review challenge — the GOV.UK statistics announcement
"Primary school performance tables: 2022" is marked CANCELLED with
"will not be published in key stage 2 performance tables in academic
year 2021/22", so the footnote's "DfE didn't publish" claim stands
and must not be softened to "not in our dataset"). The chart's England-
`pipeline/scripts/diagnose_compare_gaps.py`). The chart's England-
only 2021/22 point with broken school lines is therefore the
correct permanent rendering; copy should say "DfE didn't publish
school-level figures for 2021/22", not "not in our dataset yet".
-70
View File
@@ -40,19 +40,6 @@ async function twoPrimaryUrns(page: Page): Promise<[string, string]> {
return [urns[0], urns[1]];
}
async function twoSecondaryUrns(page: Page): Promise<[string, string]> {
const res = await page.request.get('/api/schools?search=school&per_page=100');
expect(res.ok()).toBeTruthy();
const body = await res.json();
const urns: string[] = (body.schools ?? [])
.filter((s: { phase?: string; attainment_8_score?: number | null }) =>
s.phase === 'Secondary' && s.attainment_8_score != null,
)
.map((s: { urn: number }) => String(s.urn));
expect(urns.length).toBeGreaterThanOrEqual(2);
return [urns[0], urns[1]];
}
test('home page loads with hero search', async ({ page }) => {
await page.goto('/');
await expect(page.locator('h1').first()).toBeVisible();
@@ -197,16 +184,6 @@ test('comparing two schools shows the parent-first sections side by side', async
// Every number gets an anchor: at least one England-average tick or label
await expect(page.getByText(/England \d+/).first()).toBeVisible();
// Desktop: the sticky school bar shares the sections' grid template
// (200px label rail + one column per school) so chips align with the
// columns they label.
const barTemplate = await page
.locator('[aria-label="Schools in this comparison"]')
.evaluate((el) => getComputedStyle(el).gridTemplateColumns);
expect(barTemplate).toMatch(/^200px /);
// ...and its label rail carries the comparison caption.
await expect(page.getByText(/^\d+ (primary|secondary) schools?$/)).toBeVisible();
// Ofsted linkout goes to the school's provider page, never a report deep-link
const ofstedLink = page.getByRole('link', { name: /Ofsted page/i }).first();
await expect(ofstedLink).toBeVisible();
@@ -222,53 +199,6 @@ test('comparing two schools shows the parent-first sections side by side', async
}
});
test('comparing two secondary schools renders the secondary sections', async ({ page }) => {
const [urn0, urn1] = await twoSecondaryUrns(page);
await page.goto(`/compare?urns=${urn0},${urn1}`);
await expect(page.locator(`a[href*="${urn0}"]`).first()).toBeVisible({ timeout: 15_000 });
// The parent-first sections must render — this page was completely blank
// for all-secondary baskets (expert review must-fix #1).
await expect(page.getByRole('heading', { name: 'At a glance' }).first()).toBeVisible({
timeout: 15_000,
});
await expect(page.getByRole('heading', { name: 'Ofsted inspection' }).first()).toBeVisible();
// A KS4 measure proves the secondary academics variant rendered.
await expect(page.getByText(/Attainment 8/i).first()).toBeVisible();
await expect(page.getByText(/No primary schools in your comparison/)).toHaveCount(0);
// The admissions template must be phase-aware: the primaries' distance
// copy ("non-faith primaries") must never appear on a secondary comparison
// (expert sign-off must-fix M3).
await expect(page.getByText(/non-faith primaries/)).toHaveCount(0);
});
test('opening a different compare link after a previous comparison still renders', async ({ page }) => {
// Regression: the first visit stores a basket in localStorage; opening a
// link for a DIFFERENT school set then raced a stale fetch for the stored
// basket against the new SSR data, blanking every section (including the
// trends chart) until a hard refresh.
const [s0, s1] = await twoSecondaryUrns(page);
const [p0, p1] = await twoPrimaryUrns(page);
await page.goto(`/compare?urns=${s0},${s1}`);
await expect(page.getByRole('heading', { name: 'At a glance' }).first()).toBeVisible({
timeout: 15_000,
});
await page.goto(`/compare?urns=${p0},${p1}`);
await expect(page.getByRole('heading', { name: 'At a glance' }).first()).toBeVisible({
timeout: 15_000,
});
// Give any straggling stale response time to land, then confirm the new
// comparison is still on screen.
await page.waitForTimeout(1500);
await expect(page.getByRole('heading', { name: 'At a glance' }).first()).toBeVisible();
await expect(page.getByRole('heading', { name: 'Explore trends' }).first()).toBeVisible();
await expect(page.locator(`a[href*="${p0}"]`).first()).toBeVisible();
});
test('compare chart on mobile shows school chips with tap-to-focus', async ({ page }) => {
await page.setViewportSize({ width: 390, height: 844 });
@@ -1,65 +0,0 @@
/**
* Secondary academics: every headline number carries its England anchor and a
* verdict chip (expert sign-off SF1 — the grade-5 and EBacc rows previously
* rendered as bare numbers, breaking the "anchored against England" promise).
*/
import { render, screen, within } from '@testing-library/react';
import { CompareAcademics } from '@/components/compare/CompareAcademics';
import type { ComparisonData, NationalAverages, School } from '@/lib/types';
function school(urn: number, name: string): School {
return { urn, school_name: name, attainment_8_score: 58.7 } as School;
}
function data(urn: number): ComparisonData {
return {
school_info: school(urn, 'Test High'),
yearly_data: [
{
year: 202425,
attainment_8_score: 58.7,
english_maths_strong_pass_pct: 30,
ebacc_entry_pct: 10,
},
] as ComparisonData['yearly_data'],
ofsted: null,
census: null,
admissions: null,
admissions_history: [],
deprivation: null,
};
}
const NATIONAL: NationalAverages = {
year: 202425,
primary: {},
secondary: {
attainment_8_score: 46.0,
english_maths_strong_pass_pct: 45.4,
ebacc_entry_pct: 40.5,
},
by_year: [],
};
test('grade-5 and EBacc rows show the England anchor and a Below chip when under it', () => {
const s = school(137086, 'Bishop Stopford School');
render(
<CompareAcademics
schools={[s]}
data={{ '137086': data(137086) }}
nationalAverages={NATIONAL}
isSecondary
/>,
);
// The official anchors appear (45.4% and 40.5%), not just the school numbers.
expect(screen.getByText(/England average 45%/)).toBeInTheDocument();
expect(screen.getByText(/England average 41%/)).toBeInTheDocument();
// 30% grade-5 and 10% EBacc are both well below their anchors → Below chips.
// Attainment 8 (58.7 vs 46.0) is above → at least one "Above" chip too.
expect(screen.getAllByText(/Below England average/).length).toBeGreaterThanOrEqual(2);
expect(screen.getAllByText(/Above England average/).length).toBeGreaterThanOrEqual(1);
});
@@ -1,140 +0,0 @@
/**
* Getting a place — phase and school-type correctness (expert sign-off
* must-fixes M1/M3):
* - an all-through school's Year 7 round must never render on the primary
* tab as if it were Reception odds;
* - selective schools get entrance-test framing, and the secondary tab
* never shows the primaries' distance template.
*/
import { render, screen } from '@testing-library/react';
import { CompareAdmissions } from '@/components/compare/CompareAdmissions';
import type { ComparisonData, School, SchoolAdmissions } from '@/lib/types';
function school(urn: number, name: string, extra: Partial<School> = {}): School {
return { urn, school_name: name, ...extra } as School;
}
function admissions(partial: Partial<SchoolAdmissions>): SchoolAdmissions {
return {
year: 202627,
school_phase: 'Secondary',
places_offered: 173,
total_applications: 433,
first_preference_offer_pct: 83,
oversubscribed: true,
...partial,
} as SchoolAdmissions;
}
function entry(info: School, a: SchoolAdmissions | null): ComparisonData {
return {
school_info: info,
yearly_data: [],
ofsted: null,
census: null,
admissions: a,
admissions_history: a ? [a] : [],
deprivation: null,
};
}
describe('CompareAdmissions', () => {
it("does not show an all-through school's Year 7 round on the primary tab", () => {
// The real M1 scenario: an all-through school (Year 7 round only) beside
// a primary with a Reception round.
const allThrough = school(137306, 'Hessle High and Penshurst Primary');
const primary = school(138690, 'Barclay Primary School');
const data = {
'137306': entry(allThrough, admissions({ school_phase: 'Secondary' })),
'138690': entry(
primary,
admissions({
school_phase: 'Primary',
total_applications: 300,
places_offered: 120,
first_preference_offer_pct: 96,
}),
),
};
render(<CompareAdmissions schools={[allThrough, primary]} data={data} isSecondary={false} />);
// Hessle's Year 7 figures must not appear…
expect(screen.queryByText('433')).toBeNull();
expect(
screen.getByText(/We don't hold Reception admissions data for this school/),
).toBeInTheDocument();
// …while Barclay's Reception round renders normally.
expect(screen.getByText('300')).toBeInTheDocument();
});
it('phase-labels the section empty state when no matching round exists at all', () => {
const allThrough = school(137306, 'Hessle High and Penshurst Primary');
const data = { '137306': entry(allThrough, admissions({ school_phase: 'Secondary' })) };
render(<CompareAdmissions schools={[allThrough]} data={data} isSecondary={false} />);
expect(
screen.getByText(/No Reception admissions data is available for these schools yet/),
).toBeInTheDocument();
expect(screen.queryByText('433')).toBeNull();
});
it('shows the Year 7 round on the secondary tab', () => {
const allThrough = school(137306, 'Hessle High and Penshurst Primary');
const data = { '137306': entry(allThrough, admissions({ school_phase: 'Secondary' })) };
render(<CompareAdmissions schools={[allThrough]} data={data} isSecondary={true} />);
expect(screen.getByText('433')).toBeInTheDocument();
expect(screen.getByText('173')).toBeInTheDocument();
});
it('gives selective schools entrance-test framing, never the distance template', () => {
const grammar = school(136276, 'Watford Grammar School for Boys', {
admissions_policy: 'Selective',
religious_denomination: 'Church of England',
});
const data = {
'136276': entry(grammar, admissions({ first_preference_offer_pct: 43.7 })),
};
render(<CompareAdmissions schools={[grammar]} data={data} isSecondary={true} />);
expect(
screen.getByText(/Entry is by entrance test — the school is selective/),
).toBeInTheDocument();
expect(screen.queryByText(/non-faith primaries/)).toBeNull();
});
it('secondary faith school gets faith-aware copy, not the primaries template', () => {
const faithSchool = school(102052, "Bishop Stopford's School", {
admissions_policy: 'Non-selective',
religious_denomination: 'Church of England',
});
const data = {
'102052': entry(faithSchool, admissions({ first_preference_offer_pct: 68 })),
};
render(<CompareAdmissions schools={[faithSchool]} data={data} isSecondary={true} />);
expect(screen.getByText(/faith-based criteria may apply/)).toBeInTheDocument();
expect(screen.queryByText(/non-faith primaries/)).toBeNull();
});
it('keeps the reviewed distance copy for oversubscribed non-faith primaries', () => {
const primary = school(100140, 'Plumcroft Primary School');
const data = {
'100140': entry(
primary,
admissions({ school_phase: 'Primary', first_preference_offer_pct: 73.4 }),
),
};
render(<CompareAdmissions schools={[primary]} data={data} isSecondary={false} />);
expect(screen.getByText(/for most non-faith primaries, distance decides/)).toBeInTheDocument();
});
});
@@ -96,70 +96,6 @@ describe('CompareOfsted', () => {
expect(links[0]).toHaveAttribute('href', 'https://reports.ofsted.gov.uk/provider/21/1');
});
it('never renders Ofsted sentinel codes (9 = not applicable) as judgement chips', () => {
const sentinelSchool = school(6, 'Sentinel School');
const sentinelData: Record<string, ComparisonData> = {
'6': {
school_info: sentinelSchool,
yearly_data: [],
ofsted: ofsted({
overall_effectiveness: 2,
grade_source: 'graded',
quality_of_education: 1,
early_years_provision: 9,
sixth_form_provision: 2,
}),
},
};
render(<CompareOfsted schools={[sentinelSchool]} data={sentinelData} />);
// Real grades render…
expect(screen.getByText('Quality of education')).toBeInTheDocument();
// …the applicable sixth-form judgement renders (was previously dropped)…
expect(screen.getByText('Sixth form provision')).toBeInTheDocument();
// …and the not-applicable sentinel never appears, neither as area nor code.
expect(screen.queryByText('Early years provision')).toBeNull();
expect(screen.queryByText('9')).toBeNull();
});
it('dates a report card with the report-card inspection date, never the legacy date', () => {
const cardSchool = school(4, 'Dated Card School');
const cardData: Record<string, ComparisonData> = {
'4': {
school_info: cardSchool,
yearly_data: [],
ofsted: ofsted({
inspection_date: '2021-10-07',
rc_inspection_date: '2026-02-03',
rc_safeguarding_met: true,
report_card: { rc_achievement: { code: 1, label: 'Exceptional' } },
}),
},
};
render(<CompareOfsted schools={[cardSchool]} data={cardData} />);
expect(screen.getByText(/3 Feb 2026/)).toBeInTheDocument();
expect(screen.queryByText(/7 Oct 2021/)).toBeNull();
expect(screen.queryByText('4+ years ago')).toBeNull();
});
it('shows an em dash when a report card has no rc_inspection_date yet', () => {
const cardSchool = school(5, 'Undated Card School');
const cardData: Record<string, ComparisonData> = {
'5': {
school_info: cardSchool,
yearly_data: [],
ofsted: ofsted({
inspection_date: '2021-10-07',
rc_inspection_date: null,
rc_safeguarding_met: true,
report_card: { rc_achievement: { code: 1, label: 'Exceptional' } },
}),
},
};
render(<CompareOfsted schools={[cardSchool]} data={cardData} />);
expect(screen.getByText('—')).toBeInTheDocument();
expect(screen.queryByText(/7 Oct 2021/)).toBeNull();
});
it('renders a per-measure mobile tag with the short school name', () => {
render(<CompareOfsted schools={schools} data={data} />);
// Each measure repeats the schools, so the short name ("Graded" from
@@ -1,78 +0,0 @@
/**
* Regression: an all-secondary comparison must render the secondary sections.
*
* The basket hydrates from the URL a beat after mount, so the auto-phase
* effect must re-run once selectedSchools arrives — with deps of only
* [comparisonData] it fired once against an empty basket, bailed, and the
* page stayed on an empty "primary" tab ("No primary schools in your
* comparison") even though all schools were secondary.
*/
import { render, screen, waitFor } from '@testing-library/react';
import { ComparisonView } from '@/components/ComparisonView';
import { ComparisonProvider } from '@/context/ComparisonProvider';
import type { ComparisonData, School } from '@/lib/types';
const fetchComparison = jest.fn();
jest.mock('@/lib/api', () => ({
fetchComparison: (...args: unknown[]) => fetchComparison(...args),
}));
jest.mock('@/lib/analytics', () => ({ track: jest.fn() }));
function secondarySchool(urn: number, name: string): School {
return {
urn,
school_name: name,
local_authority: 'Testshire',
school_type: 'Academy converter',
attainment_8_score: 55,
phase: 'Secondary',
} as School;
}
function data(urn: number, name: string): ComparisonData {
return {
school_info: secondarySchool(urn, name),
yearly_data: [{ year: 202425, attainment_8_score: 55 }] as ComparisonData['yearly_data'],
ofsted: null,
census: null,
admissions: null,
admissions_history: [],
deprivation: null,
};
}
const INITIAL_DATA = {
'300': data(300, 'Gamma High'),
'400': data(400, 'Delta Academy'),
};
test('an all-secondary comparison renders the sections, not an empty primary tab', async () => {
render(
<ComparisonProvider>
<ComparisonView
initialData={INITIAL_DATA}
initialNationalAverages={{
year: 202425,
primary: {},
secondary: { attainment_8_score: 46 },
by_year: [],
}}
initialBenchmarks={undefined}
initialUrns={[300, 400]}
metrics={[]}
selectedMetric="attainment_8_score"
/>
</ComparisonProvider>,
);
await waitFor(() => {
expect(screen.getByRole('heading', { name: 'At a glance' })).toBeInTheDocument();
});
expect(screen.getAllByText('Gamma High').length).toBeGreaterThan(0);
expect(screen.queryByText(/No primary schools in your comparison/)).toBeNull();
// The sticky bar's rail caption reflects the active phase and count.
expect(screen.getByText('2 secondary schools')).toBeInTheDocument();
expect(fetchComparison).not.toHaveBeenCalled();
});
@@ -1,107 +0,0 @@
/**
* Regression: opening a compare link while a DIFFERENT basket is stored must
* not blank the page.
*
* The basket hydrates from localStorage first, which can fire a fetch for the
* OLD school set; the URL-seed effect then replaces the basket with the URL's
* schools (already covered by SSR data, so no new fetch). When the stale
* response for the old set finally lands, it must not clobber the fresh SSR
* data — that left every section (including the trends chart) empty until a
* hard refresh.
*/
import { act, render, screen, waitFor } from '@testing-library/react';
import { ComparisonView } from '@/components/ComparisonView';
import { ComparisonProvider } from '@/context/ComparisonProvider';
import type { ComparisonData, School } from '@/lib/types';
const fetchComparison = jest.fn();
jest.mock('@/lib/api', () => ({
fetchComparison: (...args: unknown[]) => fetchComparison(...args),
}));
jest.mock('@/lib/analytics', () => ({ track: jest.fn() }));
function school(urn: number, name: string): School {
return {
urn,
school_name: name,
local_authority: 'Testshire',
school_type: 'Community school',
rwm_expected_pct: 80,
phase: 'Primary',
} as School;
}
function data(urn: number, name: string): ComparisonData {
return {
school_info: school(urn, name),
yearly_data: [{ year: 202425, rwm_expected_pct: 80 }] as ComparisonData['yearly_data'],
ofsted: null,
census: null,
admissions: null,
admissions_history: [],
deprivation: null,
};
}
// The visitor's previously stored basket (a different school entirely).
const STORED_SCHOOL = school(900, 'Old Stored School');
// The comparison the URL (and SSR) actually asked for.
const URL_DATA = {
'100': data(100, 'Alpha Primary'),
'200': data(200, 'Beta Primary'),
};
beforeEach(() => {
fetchComparison.mockReset();
localStorage.clear();
});
test('a stale fetch for the previously stored basket does not clobber the URL comparison', async () => {
localStorage.setItem('selectedSchools', JSON.stringify([STORED_SCHOOL]));
const pending: Array<(v: unknown) => void> = [];
fetchComparison.mockImplementation(() => new Promise((resolve) => pending.push(resolve)));
render(
<ComparisonProvider>
<ComparisonView
initialData={URL_DATA}
initialNationalAverages={{
year: 202425,
primary: { rwm_expected_pct: 62 },
secondary: {},
by_year: [],
}}
initialBenchmarks={undefined}
initialUrns={[100, 200]}
metrics={[]}
selectedMetric="rwm_expected_pct"
/>
</ComparisonProvider>,
);
// The URL's schools render from SSR data once the basket is reseeded.
await waitFor(() => {
expect(screen.getByRole('heading', { name: 'At a glance' })).toBeInTheDocument();
});
expect(screen.getAllByText('Alpha Primary').length).toBeGreaterThan(0);
// The transient stored-basket fetch (for school 900) resolves LATE, after
// the basket has moved on to the URL's schools.
await act(async () => {
for (const resolve of pending) {
resolve({
comparison: { '900': data(900, 'Old Stored School') },
national_averages: { year: 202425, primary: {}, secondary: {}, by_year: [] },
benchmarks: undefined,
});
}
});
// The page must still show the URL comparison — not go blank.
expect(screen.getByRole('heading', { name: 'At a glance' })).toBeInTheDocument();
expect(screen.getAllByText('Alpha Primary').length).toBeGreaterThan(0);
});
@@ -73,15 +73,6 @@ describe('buildCompareChart', () => {
expect(eng.data[chart.years.indexOf(202122)]).toBe(58.7);
});
it('lists England-only years so the component can caption dashed-only stretches', () => {
const chart = buildCompareChart(THREE_SCHOOLS, SCHOOL_LIST, 'rwm_expected_pct', {
202122: 58.7,
});
expect(chart.englandOnlyYears).toEqual([202122]);
const none = buildCompareChart(THREE_SCHOOLS, SCHOOL_LIST, 'rwm_expected_pct');
expect(none.englandOnlyYears).toEqual([]);
});
it('flags the unpublished 2021/22 school-level year when England has data but schools do not', () => {
const withNational = buildCompareChart(THREE_SCHOOLS, SCHOOL_LIST, 'rwm_expected_pct', {
202122: 58.7,
@@ -7,7 +7,6 @@
import {
OFSTED_LEGACY_GRADES,
admissionsForPhase,
ofstedDisplay,
progressBand,
rcAreaLabel,
@@ -177,16 +176,6 @@ describe('summariseAdmissions', () => {
expect(s.chip).toEqual({ tone: 'warn', text: 'Over 1 in 4 first choices missed out' });
});
it('60% → "About 1 in 3 first choices missed out"', () => {
const s = summariseAdmissions(admissions({ first_preference_offer_pct: 60 }));
expect(s.chip).toEqual({ tone: 'warn', text: 'About 1 in 3 first choices missed out' });
});
it('44% (selective-scale demand) → "More than half of first choices missed out"', () => {
const s = summariseAdmissions(admissions({ first_preference_offer_pct: 43.69 }));
expect(s.chip).toEqual({ tone: 'warn', text: 'More than half of first choices missed out' });
});
it('100% → "All first choices offered"', () => {
const s = summariseAdmissions(admissions({ first_preference_offer_pct: 100 }));
expect(s.chip).toEqual({ tone: 'good', text: 'All first choices offered' });
@@ -199,44 +188,6 @@ describe('summariseAdmissions', () => {
});
});
describe('admissionsForPhase', () => {
const row = (year: number, school_phase: string | null): SchoolAdmissions =>
({ year, school_phase, places_offered: 100, total_applications: 200, first_preference_offer_pct: 80 }) as SchoolAdmissions;
it('returns the latest round matching the active phase', () => {
const data = {
admissions: row(202627, 'Secondary'),
admissions_history: [row(202526, 'Secondary'), row(202526, 'Primary'), row(202425, 'Primary')],
};
expect(admissionsForPhase(data, true)?.year).toBe(202627);
expect(admissionsForPhase(data, false)?.year).toBe(202526);
expect(admissionsForPhase(data, false)?.school_phase).toBe('Primary');
});
it("never substitutes the other phase's round (all-through with Year 7 data only)", () => {
const data = {
admissions: row(202627, 'Secondary'),
admissions_history: [row(202526, 'Secondary')],
};
expect(admissionsForPhase(data, false)).toBeNull();
expect(admissionsForPhase(data, true)?.year).toBe(202627);
});
it('uses untagged legacy rows only when no row carries a phase', () => {
const untagged = { admissions: row(202627, null), admissions_history: [row(202526, null)] };
expect(admissionsForPhase(untagged, false)?.year).toBe(202627);
expect(admissionsForPhase(untagged, true)?.year).toBe(202627);
const mixed = { admissions: row(202627, 'Secondary'), admissions_history: [row(202526, null)] };
expect(admissionsForPhase(mixed, false)).toBeNull();
});
it('handles missing data', () => {
expect(admissionsForPhase(null, false)).toBeNull();
expect(admissionsForPhase({ admissions: null, admissions_history: [] }, true)).toBeNull();
});
});
describe('progressBand', () => {
it('CI entirely above zero → above', () => {
expect(progressBand(1.2, 0.4, 2.0)).toBe('above');
+3 -8
View File
@@ -183,15 +183,10 @@ body {
}
/* Secondary: teal outline — supporting actions (+ Compare) */
/* NOTE: a duplicate `.btn` block further down this file sets `border: none`,
which wins over the base `.btn`'s `1px solid transparent`. The outline
variants below therefore declare the full `border` shorthand explicitly so
they don't depend on the base border-width — otherwise `border-color` alone
has no width and the outline never renders (buttons read as plain text). */
.btn-secondary {
background: transparent;
color: var(--accent-teal);
border: 1px solid var(--accent-teal);
border-color: var(--accent-teal);
}
.btn-secondary:hover:not(:disabled) {
background: var(--accent-teal-bg);
@@ -201,7 +196,7 @@ body {
.btn-tertiary {
background: var(--bg-secondary);
color: var(--text-secondary);
border: 1px solid var(--border-color);
border-color: var(--border-color);
}
.btn-tertiary:hover:not(:disabled) {
background: var(--border-color);
@@ -212,7 +207,7 @@ body {
.btn-active {
background: var(--accent-teal-bg);
color: var(--accent-teal);
border: 1px solid var(--accent-teal);
border-color: var(--accent-teal);
}
.btn-active:hover:not(:disabled) {
background: transparent;
@@ -1,23 +1,16 @@
/* Chart wrapper: chips (mobile) above, then the canvas, then the gap note.
The canvas has its OWN definite height (Chart.js needs one for
maintainAspectRatio: false); the chips and the note flow at their natural
size around it rather than competing with it for a fixed outer height —
so a longer note (e.g. the KS4 gap caption) or a two-row chip legend can
never squash the chart. */
/* Chart wrapper: chips (mobile) above, canvas filling the rest of the
parent .chartContainer, whose fixed height drives Chart.js sizing via
maintainAspectRatio: false. */
.wrapper {
display: flex;
flex-direction: column;
height: 100%;
}
.canvasBox {
position: relative;
height: 380px;
}
@media (max-width: 640px) {
.canvasBox {
height: 280px;
}
flex: 1 1 auto;
min-height: 0;
}
/* School chips: mobile-only legend + tap-to-focus control. Desktop keeps
+5 -19
View File
@@ -38,15 +38,13 @@ interface ComparisonChartProps {
/** Official England figure per academic year for this metric — renders a
* dashed grey reference line when provided. */
nationalByYear?: Record<number, number | null | undefined>;
/** KS4 metrics get a different (honest) gap caption than KS2. */
isSecondary?: boolean;
}
// One shape per basket slot (MAX_SCHOOLS = 5) — secondary encoding so
// converging lines stay tellable apart without relying on hue alone.
const POINT_STYLES: PointStyle[] = ['circle', 'triangle', 'rect', 'rectRot', 'star'];
export function ComparisonChart({ comparisonData, schools, metric, metricLabel, nationalByYear, isSecondary = false }: ComparisonChartProps) {
export function ComparisonChart({ comparisonData, schools, metric, metricLabel, nationalByYear }: ComparisonChartProps) {
const isMobile = useIsMobile();
const [focusedUrn, setFocusedUrn] = useState<number | null>(null);
@@ -170,7 +168,7 @@ export function ComparisonChart({ comparisonData, schools, metric, metricLabel,
display: true,
title: {
display: !isMobile,
text: kind === 'percentage' ? 'Percentage (%)' : kind === 'progress' ? 'Progress Score' : 'Score',
text: kind === 'percentage' ? 'Percentage (%)' : kind === 'progress' ? 'Progress Score' : 'Value',
font: {
size: 12,
weight: 'bold',
@@ -242,23 +240,11 @@ export function ComparisonChart({ comparisonData, schools, metric, metricLabel,
<div className={styles.canvasBox}>
<Line data={chartData} options={options} aria-label={`${metricLabel} comparison chart`} />
</div>
{isSecondary && built.englandOnlyYears.length > 0 ? (
// KS4's honest story differs from KS2's: 2019/202020/21 school-level
// GCSE results weren't published (COVID grading); later years WERE
// published by DfE but aren't in our dataset yet.
{built.showUnpublished202122Note && (
<p className={styles.chartNote}>
School-level GCSE figures for 2019/20 and 2020/21 weren&apos;t published (COVID
grading), and more recent years aren&apos;t in our dataset yet where lines break the
England average is shown where available.
No national tests were held in 2019/20 and 2020/21 (COVID), and DfE didn&apos;t publish
school-level figures for 2021/22 the England average is shown for that year.
</p>
) : (
!isSecondary &&
built.showUnpublished202122Note && (
<p className={styles.chartNote}>
No national tests were held in 2019/20 and 2020/21 (COVID), and DfE didn&apos;t publish
school-level figures for 2021/22 the England average is shown for that year.
</p>
)
)}
</div>
);
@@ -148,54 +148,6 @@
text-overflow: ellipsis;
}
/* Caption filling the label rail on desktop ("Comparing / 3 primary
schools"). Hidden on mobile, where the bar is a row of compact pills. */
.barCaption {
display: none;
}
/* Desktop (matches the sections' 761px breakpoint): the bar adopts the same
grid template as compareSections' .grid — a 200px row-label rail plus one
column per school — so each chip sits exactly over the column it labels.
The caption occupies the rail; chips flow into the school columns. */
@media (min-width: 761px) {
.schoolBar {
display: grid;
grid-template-columns: 200px repeat(var(--school-count, 3), 1fr);
gap: 0 0.75rem;
overflow-x: visible;
}
.schoolChip {
min-width: 0;
}
.barCaption {
grid-column: 1;
display: flex;
flex-direction: column;
justify-content: center;
gap: 0.1rem;
padding-right: 0.5rem;
min-width: 0;
}
.barCaptionEyebrow {
font-size: 0.72rem;
font-weight: 600;
letter-spacing: 0.06em;
text-transform: uppercase;
color: var(--text-muted, #6d685f);
}
.barCaptionCount {
font-size: 0.95rem;
font-weight: 600;
line-height: 1.3;
color: var(--text-primary, #1a1612);
}
}
.chipRemove {
margin-left: auto;
border: none;
+7 -56
View File
@@ -9,7 +9,7 @@
'use client';
import { useEffect, useRef, useState, type CSSProperties } from 'react';
import { useEffect, useRef, useState } from 'react';
import { useRouter, usePathname, useSearchParams } from 'next/navigation';
import { useComparison } from '@/hooks/useComparison';
@@ -85,9 +85,7 @@ export function ComparisonView({
replaceSchools(urlSchools);
}
}
// Re-seed when a client-side navigation lands on a different ?urns= set
// (initialUrns/initialData are new props on the same component instance).
}, [isInitialized, initialUrns.join(',')]); // eslint-disable-line react-hooks/exhaustive-deps
}, [isInitialized]); // eslint-disable-line react-hooks/exhaustive-deps
const urnKey = selectedSchools.map((s) => s.urn).join(',');
@@ -130,18 +128,8 @@ export function ComparisonView({
const covered = urnKey.split(',').every((urn) => have[urn] != null);
if (covered) return;
// Guard against out-of-order responses: while the basket hydrates from
// localStorage it can transiently hold a DIFFERENT school set than the
// URL, firing a fetch for schools the user is no longer comparing. That
// stale response must not replace data for the current set — it blanked
// every section until a hard refresh. Cleanup marks the run cancelled
// when urnKey moves on, so only the current selection's response is
// applied (replacing the map keeps it bounded and guarantees a re-added
// school is refetched fresh rather than served a lingering old entry).
let cancelled = false;
fetchComparison(urnKey, { cache: 'no-store' })
.then((data) => {
if (cancelled) return;
setComparisonData(data.comparison);
setNationalAverages(data.national_averages);
setBenchmarks(data.benchmarks);
@@ -152,9 +140,6 @@ export function ComparisonView({
// destroy a working comparison the user is looking at.
console.error('Failed to fetch comparison:', err);
});
return () => {
cancelled = true;
};
}, [urnKey, isInitialized]);
const primarySchools = selectedSchools.filter((school) => {
@@ -188,11 +173,7 @@ export function ComparisonView({
if (!metricFitsPhase) {
setSelectedMetric(newPhase === 'secondary' ? 'attainment_8_score' : 'rwm_expected_pct');
}
// selectedSchools is a dep because the basket hydrates after mount: the
// first run sees an empty basket and bails, so it must re-fire when the
// schools arrive. primarySchools/secondarySchools/metrics/selectedMetric
// are intentionally omitted (derived or would cause loops).
}, [comparisonData, selectedSchools]); // eslint-disable-line react-hooks/exhaustive-deps
}, [comparisonData]); // eslint-disable-line react-hooks/exhaustive-deps
const handlePhaseChange = (phase: 'primary' | 'secondary') => {
phaseLockedByUser.current = true;
@@ -357,28 +338,8 @@ export function ComparisonView({
/>
) : (
<>
{/* Sticky school bar — column identity while scrolling. On desktop
it shares the sections' grid template (via --school-count) so
each chip sits exactly over the column it labels. */}
<div
className={styles.schoolBar}
style={{ '--school-count': activeSchools.length } as CSSProperties}
aria-label="Schools in this comparison"
>
{/* Fills the 200px label rail on desktop (hidden on mobile).
All-through schools must not be miscounted as "primary
schools"/"secondary schools" — mixed baskets get "· primary
view" phrasing instead. */}
<div className={styles.barCaption}>
<span className={styles.barCaptionEyebrow}>Comparing</span>
<span className={styles.barCaptionCount}>
{activeSchools.every((sch) =>
sch.phase?.toLowerCase().includes(comparePhase),
)
? `${activeSchools.length} ${comparePhase} school${activeSchools.length === 1 ? '' : 's'}`
: `${activeSchools.length} schools · ${comparePhase} view`}
</span>
</div>
{/* Sticky school bar — column identity while scrolling */}
<div className={styles.schoolBar} aria-label="Schools in this comparison">
{activeSchools.map((school, index) => (
<div
key={school.urn}
@@ -396,13 +357,7 @@ export function ComparisonView({
<span className={styles.chipNameShort}>{shortName(school.school_name)}</span>
</a>
<span className={styles.chipMeta}>
{[
/all.?through/i.test(school.phase ?? '') ? 'All-through' : null,
school.local_authority,
school.school_type,
]
.filter(Boolean)
.join(' · ')}
{[school.local_authority, school.school_type].filter(Boolean).join(' · ')}
</span>
</span>
<button
@@ -434,11 +389,7 @@ export function ComparisonView({
benchmarks={benchmarks}
isSecondary={!isPrimary}
/>
<CompareAdmissions
schools={activeSchools}
data={activeComparisonData}
isSecondary={!isPrimary}
/>
<CompareAdmissions schools={activeSchools} data={activeComparisonData} />
<CompareCommunity
schools={activeSchools}
data={activeComparisonData}
+2 -2
View File
@@ -271,10 +271,10 @@ export function HomeView({ initialSchools, filters, totalSchools, howItWorks, ed
freshness, standing in for the hidden eyebrow too) on phones,
where every line above the fold costs. */}
<span className={styles.heroDescriptionFull}>
<strong>24,000+ primary and secondary schools</strong> with Key Stage 2 SATs, GCSE results, Ofsted grades, progress scores and admissions data side by side, in one place.
<strong>27,000+ primary and secondary schools</strong> with Key Stage 2 SATs, GCSE results, Ofsted grades, progress scores and admissions data side by side, in one place.
</span>
<span className={styles.heroDescriptionCompact}>
<strong>24,000+ English schools</strong> SATs, GCSEs, Ofsted &amp; admissions, side by side. Updated for 2026/27.
<strong>27,000+ English schools</strong> SATs, GCSEs, Ofsted &amp; admissions, side by side. Updated for 2026/27.
</span>
</p>
</div>
+1 -7
View File
@@ -123,13 +123,7 @@
.modal {
width: 100%;
max-width: 100%;
/* Bottom sheet sized against the overlay (which tracks the visual
viewport), NOT vh: when the keyboard is open the overlay is short, so
max-height:100% keeps the whole sheet — input and results — above the
keyboard. min-height gives a comfortable default without a tiny stub,
but is capped at 100% so it never exceeds the visible area. */
min-height: min(55vh, 100%);
max-height: 100%;
max-height: 95vh;
border-bottom-left-radius: 0;
border-bottom-right-radius: 0;
animation: slideUp 0.3s ease;
+2 -30
View File
@@ -5,7 +5,7 @@
'use client';
import { useEffect, useCallback, useRef } from 'react';
import { useEffect, useCallback } from 'react';
import { createPortal } from 'react-dom';
import styles from './Modal.module.css';
@@ -18,8 +18,6 @@ interface ModalProps {
}
export function Modal({ isOpen, onClose, children, title, size = 'medium' }: ModalProps) {
const overlayRef = useRef<HTMLDivElement>(null);
const handleEscape = useCallback((e: KeyboardEvent) => {
if (e.key === 'Escape') {
onClose();
@@ -41,32 +39,6 @@ export function Modal({ isOpen, onClose, children, title, size = 'medium' }: Mod
};
}, [isOpen, handleEscape]);
// Pin the overlay to the VISUAL viewport, not the layout viewport. On mobile
// the on-screen keyboard shrinks the visual viewport but not the layout one,
// so a `position: fixed; inset: 0` overlay keeps full height — leaving the
// bottom-anchored sheet (and the dim backdrop's lower half) hidden behind
// the keyboard. Tracking visualViewport.height/offsetTop keeps the whole
// overlay — backdrop and sheet — inside the visible area, above the keyboard.
useEffect(() => {
if (!isOpen) return;
const vv = typeof window !== 'undefined' ? window.visualViewport : null;
const el = overlayRef.current;
if (!vv || !el) return;
const sync = () => {
el.style.top = `${vv.offsetTop}px`;
el.style.height = `${vv.height}px`;
el.style.bottom = 'auto';
};
sync();
vv.addEventListener('resize', sync);
vv.addEventListener('scroll', sync);
return () => {
vv.removeEventListener('resize', sync);
vv.removeEventListener('scroll', sync);
};
}, [isOpen]);
if (!isOpen || typeof window === 'undefined') return null;
const handleOverlayClick = (e: React.MouseEvent) => {
@@ -76,7 +48,7 @@ export function Modal({ isOpen, onClose, children, title, size = 'medium' }: Mod
};
return createPortal(
<div ref={overlayRef} className={styles.overlay} onClick={handleOverlayClick}>
<div className={styles.overlay} onClick={handleOverlayClick}>
<div className={`${styles.modal} ${styles[size]}`}>
<div className={styles.header}>
{title && <h2 className={styles.title}>{title}</h2>}
@@ -2,6 +2,14 @@
padding: 1.5rem;
}
.title {
font-size: 1.5rem;
font-weight: 700;
color: var(--text-primary, #1a1612);
margin-bottom: 1.5rem;
font-family: var(--font-playfair), 'Playfair Display', serif;
}
.warning {
background: var(--accent-gold-bg);
border: 1px solid var(--accent-gold, #c9a227);
@@ -111,16 +119,12 @@
font-size: 1rem;
font-weight: 600;
color: var(--text-primary, #1a1612);
margin-bottom: 0.25rem;
margin-bottom: 0.5rem;
white-space: nowrap;
overflow: hidden;
text-overflow: ellipsis;
}
.resultButton {
flex: 0 0 auto;
}
.schoolMeta {
display: flex;
gap: 1rem;
@@ -155,30 +159,21 @@
padding: 1rem;
}
/* One scroll container on mobile: the modal content itself scrolls, so the
results list must not add its own inner scroll (double scrollbars, and
the input would be trapped above a short 400px window when the keyboard
shrinks the sheet). */
.results {
max-height: none;
overflow: visible;
.title {
font-size: 1.25rem;
}
/* Compact stacked card: name + meta, then a full-width action so the tap
target is obvious and the card doesn't waste vertical space. */
.resultItem {
flex-direction: column;
align-items: stretch;
gap: 0.625rem;
padding: 0.875rem;
}
.resultButton {
.addButton {
width: 100%;
}
.schoolMeta {
flex-wrap: wrap;
gap: 0.25rem 1rem;
flex-direction: column;
gap: 0.25rem;
}
}
+5 -3
View File
@@ -81,8 +81,10 @@ export function SchoolSearchModal({ isOpen, onClose }: SchoolSearchModalProps) {
};
return (
<Modal isOpen={isOpen} onClose={handleClose} title="Add School to Comparison">
<Modal isOpen={isOpen} onClose={handleClose}>
<div className={styles.modalContent}>
<h2 className={styles.title}>Add School to Comparison</h2>
{!canAddMore && (
<div className={styles.warning}>
Maximum 5 schools can be compared. Remove a school to add another.
@@ -127,9 +129,9 @@ export function SchoolSearchModal({ isOpen, onClose }: SchoolSearchModalProps) {
<button
onClick={() => handleAddSchool(school)}
disabled={alreadySelected || !canAddMore}
className={`${styles.resultButton} ${
className={
alreadySelected ? "btn btn-active" : "btn btn-secondary"
}`}
}
>
{alreadySelected ? "✓ Comparing" : "+ Compare"}
</button>
@@ -148,35 +148,9 @@ export function CompareAcademics({
}
return null;
});
// DfE stopped publishing Progress 8 from 2024/25: those GCSE year groups
// sat no KS2 tests (COVID), so there is no baseline to measure progress
// from. A bare "No data" reads as a gap on our side — say why. Judged
// PER SCHOOL on its own latest data year: a school whose data simply
// stops earlier (an unrelated gap) must not borrow the COVID explanation
// from a neighbour that does have 2024/25 data.
const p8NotPublished = urns.map((urn) => {
const rows = data[String(urn)]?.yearly_data ?? [];
const y = rows.length ? Math.trunc(rows[rows.length - 1].year) : 0;
return y >= 202425;
});
const grade5 = latestValues(data, urns, 'english_maths_strong_pass_pct');
const ebacc = latestValues(data, urns, 'ebacc_entry_pct');
const att8Anchor = nationalAverages?.secondary?.attainment_8_score;
const grade5Anchor = nationalAverages?.secondary?.english_maths_strong_pass_pct;
const ebaccAnchor = nationalAverages?.secondary?.ebacc_entry_pct;
// Every headline number gets its England anchor + verdict chip, so the
// "anchored against the England average" promise holds for the grade-5
// and EBacc rows too, not just Attainment 8.
const anchorChip = (value: number | null, anchor: number | null | undefined, tol: number) => {
if (value == null || anchor == null) return null;
const v = verdict(value, anchor, tol);
return (
<Chip tone={v === 'above' ? 'good' : v === 'below' ? 'warn' : 'neutral'}>
{v === 'above' ? 'Above' : v === 'below' ? 'Below' : 'Close to'} England average
</Chip>
);
};
return (
<Section
@@ -189,8 +163,7 @@ export function CompareAcademics({
<Cell key={school.urn} school={school} index={i}>
{att8[i] != null ? (
<>
<span className={s.big}>{(att8[i] as number).toFixed(1)}</span>{' '}
{anchorChip(att8[i], att8Anchor, 2)}
<span className={s.big}>{(att8[i] as number).toFixed(1)}</span>
{att8Anchor != null && (
<span className={s.small}>England average {att8Anchor.toFixed(1)}</span>
)}
@@ -216,11 +189,6 @@ export function CompareAcademics({
>
{banding[i]}
</Chip>
) : p8NotPublished[i] ? (
<span className={s.small}>
Not published this GCSE year group sat no KS2 tests (COVID), so DfE has no
baseline to measure progress from
</span>
) : (
<span className={s.small}>No data</span>
)}
@@ -232,38 +200,14 @@ export function CompareAcademics({
</RowLabel>
{schools.map((school, i) => (
<Cell key={school.urn} school={school} index={i}>
{grade5[i] != null ? (
<>
<span className={s.big} style={{ fontSize: '1.1rem' }}>
{Math.round(grade5[i] as number)}%
</span>{' '}
{anchorChip(grade5[i], grade5Anchor, 3)}
{grade5Anchor != null && (
<span className={s.small}>England average {Math.round(grade5Anchor)}%</span>
)}
</>
) : (
<span className={s.small}>No data</span>
)}
{grade5[i] != null ? `${Math.round(grade5[i] as number)}%` : <span className={s.small}>No data</span>}
</Cell>
))}
<RowLabel tip="% entering the English Baccalaureate subject combination.">EBacc entry</RowLabel>
{schools.map((school, i) => (
<Cell key={school.urn} school={school} index={i}>
{ebacc[i] != null ? (
<>
<span className={s.big} style={{ fontSize: '1.1rem' }}>
{Math.round(ebacc[i] as number)}%
</span>{' '}
{anchorChip(ebacc[i], ebaccAnchor, 3)}
{ebaccAnchor != null && (
<span className={s.small}>England average {Math.round(ebaccAnchor)}%</span>
)}
</>
) : (
<span className={s.small}>No data</span>
)}
{ebacc[i] != null ? `${Math.round(ebacc[i] as number)}%` : <span className={s.small}>No data</span>}
</Cell>
))}
</SectionGrid>
@@ -274,24 +218,6 @@ export function CompareAcademics({
const national = nationalAverages?.primary;
const disadvantaged = latestValues(data, urns, 'rwm_expected_disadvantaged_pct');
const disadvantagedAnchor = benchmarks?.primary?.disadvantaged_rwm_expected_pct ?? null;
// Cohort size behind the disadvantaged figure (spec §8.5): these are small
// groups where single pupils move the percentage — show roughly how many
// pupils the figure rests on. Taken from the SAME yearly row that supplies
// the displayed percentage: resolving eligible_pupils and the
// disadvantaged share independently could mix years and misstate the
// cohort behind the figure.
const cohorts = urns.map((urn) => {
const rows = data[String(urn)]?.yearly_data ?? [];
for (let i = rows.length - 1; i >= 0; i--) {
const row = rows[i];
if (row.rwm_expected_disadvantaged_pct != null) {
if (row.eligible_pupils == null || row.disadvantaged_pct == null) return null;
const cohort = Math.round((row.eligible_pupils * row.disadvantaged_pct) / 100);
return cohort > 0 ? cohort : null;
}
}
return null;
});
return (
<Section
@@ -344,9 +270,6 @@ export function CompareAcademics({
<span className={s.big} style={{ fontSize: '1.1rem' }}>
{Math.round(value)}%
</span>{' '}
{cohorts[i] != null && (
<span className={s.small}>of ~{cohorts[i]} disadvantaged pupils</span>
)}{' '}
{disadvantagedAnchor != null && (
<Chip tone={verdict(value, disadvantagedAnchor, 5) === 'below' ? 'warn' : 'good'}>
{verdict(value, disadvantagedAnchor, 5) === 'above' &&
@@ -7,7 +7,7 @@
'use client';
import { admissionsForPhase, summariseAdmissions } from '@/lib/compareLogic';
import { summariseAdmissions } from '@/lib/compareLogic';
import type { ComparisonData, School } from '@/lib/types';
import { CHART_COLORS } from '@/lib/utils';
import { Cell, Chip, Measure, Section, SectionGrid, sectionStyles as s } from './sectionShared';
@@ -15,17 +15,11 @@ import { Cell, Chip, Measure, Section, SectionGrid, sectionStyles as s } from '.
export function CompareAdmissions({
schools,
data,
isSecondary = false,
}: {
schools: School[];
data: Record<string, ComparisonData>;
isSecondary?: boolean;
}) {
// Admissions rounds are phase-specific: an all-through school's Year 7
// round must never stand in for Reception on the primary tab (and vice
// versa) — beside pure primaries it reads as Reception odds.
const rows = schools.map((school) => admissionsForPhase(data[String(school.urn)], isSecondary));
const roundLabel = isSecondary ? 'Year 7' : 'Reception';
const rows = schools.map((school) => data[String(school.urn)]?.admissions ?? null);
const anyData = rows.some(Boolean);
const entryYear = rows.find(Boolean)?.year;
const entryLabel = entryYear
@@ -34,10 +28,7 @@ export function CompareAdmissions({
if (!anyData) {
return (
<Section
title="Getting a place"
how={`No ${roundLabel} admissions data is available for these schools yet.`}
>
<Section title="Getting a place" how="No admissions data is available for these schools yet.">
<></>
</Section>
);
@@ -72,9 +63,7 @@ export function CompareAdmissions({
<strong>{a.places_offered.toLocaleString('en-GB')}</strong> places
</>
) : (
<span className={s.small}>
We don&apos;t hold {roundLabel} admissions data for this school
</span>
<span className={s.small}>No data</span>
)}
</Cell>
);
@@ -115,28 +104,15 @@ export function CompareAdmissions({
{schools.map((school, i) => {
const a = rows[i];
const summary = summariseAdmissions(a);
const info = data[String(school.urn)]?.school_info;
const selective = (info?.admissions_policy ?? '').toLowerCase() === 'selective';
const faith =
!!info?.religious_denomination &&
!/^(none|does not apply|not applicable)$/i.test(info.religious_denomination);
let text: string | null = null;
if (summary.firstPrefPct != null) {
if (selective) {
// Selective schools: the entrance test decides, whatever the
// offer percentage looks like — never the distance template.
text =
'Entry is by entrance test — the school is selective; distance and preference rank dont decide places.';
} else if (summary.firstPrefPct >= 100) {
if (summary.firstPrefPct >= 100) {
text = `Every family who put ${school.school_name} first got a place.`;
} else if (summary.firstPrefPct >= 90) {
text = `Nearly every family who put ${school.school_name} first got a place.`;
} else if (a?.oversubscribed) {
text = isSecondary
? faith
? 'More first-choice applications than places — check the schools admission criteria (faith-based criteria may apply).'
: 'More first-choice applications than places — check the schools admission criteria (catchment or distance often decides, but criteria vary).'
: 'More first-choice applications than places — check the schools admission criteria (for most non-faith primaries, distance decides).';
text =
'More first-choice applications than places — check the schools admission criteria (for most non-faith primaries, distance decides).';
} else {
text = `${summary.firstPrefPct}% of first-choice families received an offer.`;
}
@@ -8,7 +8,6 @@
'use client';
import {
admissionsForPhase,
latestValues,
ofstedDisplay,
summariseAdmissions,
@@ -146,11 +145,7 @@ export function CompareAtAGlance({
<Measure label="Getting a place">
{schools.map((school, i) => {
// Phase-matched round only — an all-through school's Year 7 round
// must not masquerade as Reception odds on the primary tab.
const summary = summariseAdmissions(
admissionsForPhase(data[String(school.urn)], isSecondary),
);
const summary = summariseAdmissions(data[String(school.urn)]?.admissions);
return (
<Cell key={school.urn} school={school} index={i}>
{summary.chip ? (
@@ -170,14 +165,8 @@ export function CompareAtAGlance({
{schools.map((school, i) => {
const census = data[String(school.urn)]?.census;
const pupils = census?.total_pupils ?? school.total_pupils ?? null;
// An all-through school's roll covers every age group, so judging
// it against the single-phase median ("Much larger than average")
// is meaningless — label the roll honestly instead.
const isAllThrough = /all.?through/i.test(school.phase ?? '');
let sizeNote: string | null = null;
if (isAllThrough) {
sizeNote = 'Whole-school roll (all-through, all ages)';
} else if (pupils != null && medianPupils != null) {
if (pupils != null && medianPupils != null) {
if (pupils >= medianPupils * 1.5) sizeNote = 'Much larger than average';
else if (pupils >= medianPupils * 1.1) sizeNote = 'Larger than average';
else if (pupils <= medianPupils * 0.66) sizeNote = 'Much smaller than average';
@@ -33,10 +33,7 @@ export function CompareCommunity({
const bench = isSecondary ? benchmarks?.secondary : benchmarks?.primary;
const fsmChip = (value: number | null) => {
// FSM is anchored only against a real FSM benchmark (census-sourced,
// pupil-weighted). disadvantaged_pct is a different measure (FSM6+CLA)
// — never fall back across definitions; no anchor means no chip.
const anchor = bench?.fsm_pct ?? null;
const anchor = bench?.fsm_pct ?? bench?.disadvantaged_pct ?? null;
if (value == null || anchor == null) return null;
const v = verdict(value, anchor, 3);
return (
@@ -48,25 +45,10 @@ export function CompareCommunity({
);
};
const anyAllThrough = schools.some((school) => /all.?through/i.test(school.phase ?? ''));
return (
<Section
title="Who goes there"
how={
<>
The school&apos;s community, from the latest school census. State-school averages are
computed from our dataset and shown for context there&apos;s no &ldquo;right&rdquo;
number here.
{anyAllThrough && (
<>
{' '}
For all-through schools these figures cover the whole school, all ages not just
the {isSecondary ? 'secondary' : 'primary'} phase.
</>
)}
</>
}
how="The school's community, from the latest school census. State-school averages are computed from our dataset and shown for context — there's no “right” number here."
>
<SectionGrid schools={schools}>
<Measure label="Pupils on roll">
@@ -108,21 +108,14 @@ function JudgementDetailCell({
);
}
const legacyAreas: Array<[string, number | null | undefined]> = [
const legacyAreas: Array<[string, number | null]> = [
['Quality of education', ofsted.quality_of_education],
['Behaviour & attitudes', ofsted.behaviour_attitudes],
['Personal development', ofsted.personal_development],
['Leadership & management', ofsted.leadership_management],
['Early years provision', ofsted.early_years_provision],
['Sixth form provision', ofsted.sixth_form_provision],
];
// Only real Ofsted grades (14) are judgements. The MI file uses sentinel
// codes for "not applicable / no judgement" (9, and 0/8 variants) — those
// must never render as a rating chip.
const published = legacyAreas.filter(
(entry): entry is [string, number] =>
entry[1] != null && entry[1] >= 1 && entry[1] <= 4,
);
const published = legacyAreas.filter(([, grade]) => grade != null);
if (published.length === 0) {
return (
@@ -187,18 +180,10 @@ export function CompareOfsted({
<Measure label="Inspected">
{schools.map((school, i) => {
const ofsted = data[String(school.urn)]?.ofsted;
// A report card is dated by its OWN inspection date. The legacy
// inspection_date belongs to an older inspection and must never
// be shown against a report card (report cards exist only from
// Nov 2025).
const dateIso =
displays[i].kind === 'report_card'
? ofsted?.rc_inspection_date ?? null
: ofsted?.inspection_date ?? null;
const age = yearsSince(dateIso);
const age = yearsSince(ofsted?.inspection_date ?? null);
return (
<Cell key={school.urn} school={school} index={i}>
{formatInspectionDate(dateIso)}{' '}
{formatInspectionDate(ofsted?.inspection_date ?? null)}{' '}
{age != null && age > 4 && <Chip tone="neutral">4+ years ago</Chip>}
</Cell>
);
@@ -239,17 +224,8 @@ export function CompareOfsted({
`https://reports.ofsted.gov.uk/provider/21/${school.urn}`;
return (
<Cell key={school.urn} school={school} index={i}>
{/* Short visible label to save space on mobile (the coloured
school name already leads the row); the full name stays in
aria-label so screen readers can tell the links apart. */}
<a
className={s.link}
href={url}
target="_blank"
rel="noopener noreferrer"
aria-label={`${school.school_name}'s Ofsted page`}
>
Ofsted page
<a className={s.link} href={url} target="_blank" rel="noopener noreferrer">
{school.school_name}&apos;s Ofsted page
</a>
</Cell>
);
@@ -60,10 +60,18 @@
margin: 0 0 1rem;
}
/* ComparisonChart owns its own canvas height now (a definite px value per
breakpoint), with the mobile chip legend above and the gap note below it
flowing at natural size. This box therefore only needs to not constrain
that height — no fixed height, or the note would again eat the plot. */
/* ComparisonChart runs Chart.js with maintainAspectRatio:false, so it fills
its container's height — which must be *definite*. A min-height alone does
not resolve the chart wrapper's height:100%, leaving Chart.js to fall back
to its ~150px default (a squashed sliver). Give it a real height. */
.chartBox {
min-height: 0;
height: 420px;
}
@media (max-width: 640px) {
/* Taller on mobile: the mobile-only school chips sit above the canvas and
wrap to two rows for 3+ schools, so the plot keeps a usable height. */
.chartBox {
height: 360px;
}
}
@@ -119,7 +119,6 @@ export function TrendsExplorer({
metric={metric}
metricLabel={metricLabel}
nationalByYear={nationalByYear}
isSecondary={!isPrimaryPhase}
/>
</div>
</div>
@@ -72,18 +72,15 @@
align-items: center;
gap: 0.4rem;
flex-basis: 100%;
/* Slightly larger than the values below it so the school each row belongs
to is easy to read on mobile (hidden on desktop, where the column header
names the school). */
font-size: 0.9rem;
font-size: 0.8rem;
font-weight: 600;
color: var(--sc, var(--text-secondary));
margin-bottom: 0.2rem;
margin-bottom: 0.15rem;
}
.cellDot {
width: 10px;
height: 10px;
width: 9px;
height: 9px;
border-radius: 50%;
background: var(--dot, var(--text-muted));
flex: none;
+6 -11
View File
@@ -53,10 +53,6 @@ export interface CompareChart {
/** True when England published a 2021/22 figure but no school has one —
* the UI shows: "DfE didn't publish school-level figures for 2021/22". */
showUnpublished202122Note: boolean;
/** Years where the England overlay has a value but no school does — the
* chart shows a dashed-line-only stretch that needs explaining (KS2 and
* KS4 have different honest explanations, so the component owns the copy). */
englandOnlyYears: number[];
}
export function buildCompareChart(
@@ -96,12 +92,11 @@ export function buildCompareChart(
}
}
const englandOnlyYears = years.filter(
(year, i) =>
englandDataset?.data[i] != null && schoolDatasets.every((ds) => ds.data[i] == null),
);
const idx202122 = years.indexOf(202122);
const showUnpublished202122Note =
idx202122 >= 0 &&
englandDataset?.data[idx202122] != null &&
schoolDatasets.every((ds) => ds.data[idx202122] == null);
const showUnpublished202122Note = englandOnlyYears.includes(202122);
return { years, schoolDatasets, englandDataset, showUnpublished202122Note, englandOnlyYears };
return { years, schoolDatasets, englandDataset, showUnpublished202122Note };
}
-41
View File
@@ -143,41 +143,6 @@ export interface AdmissionsSummary {
interest: string | null;
}
/**
* Pick the admissions round for the ACTIVE phase tab. An all-through school
* can carry only a Year 7 (Secondary) round — rendering that beside pure
* primaries' Reception rounds made 433-forms-for-173-places read as
* Reception odds. Rows matching the target phase win (latest year first);
* rows tagged with the OTHER phase are never substituted. Untagged rows
* (legacy data, no school_phase) are used only when no row carries a phase.
*/
export function admissionsForPhase(
data:
| { admissions?: SchoolAdmissions | null; admissions_history?: SchoolAdmissions[] }
| null
| undefined,
isSecondary: boolean,
): SchoolAdmissions | null {
if (!data) return null;
const rows: SchoolAdmissions[] = [
...(data.admissions_history ?? []),
...(data.admissions ? [data.admissions] : []),
];
if (rows.length === 0) return null;
const target = isSecondary ? 'secondary' : 'primary';
const byYearDesc = (a: SchoolAdmissions, b: SchoolAdmissions) => (b.year ?? 0) - (a.year ?? 0);
const matching = rows
.filter((r) => r.school_phase?.toLowerCase() === target)
.sort(byYearDesc);
if (matching.length > 0) return matching[0];
const tagged = rows.some((r) => r.school_phase != null);
if (!tagged) return [...rows].sort(byYearDesc)[0];
return null;
}
export function summariseAdmissions(
a: SchoolAdmissions | null | undefined,
): AdmissionsSummary {
@@ -192,12 +157,6 @@ export function summariseAdmissions(
if (pct != null) {
if (pct >= 100) {
chip = { tone: 'good', text: 'All first choices offered' };
} else if (pct < 50) {
// Banded, not one blanket chip: "Over 1 in 4" on a school where more
// than half missed out understated the worst cases by half.
chip = { tone: 'warn', text: 'More than half of first choices missed out' };
} else if (pct < 67) {
chip = { tone: 'warn', text: 'About 1 in 3 first choices missed out' };
} else if (pct < 75) {
chip = { tone: 'warn', text: 'Over 1 in 4 first choices missed out' };
} else {
-4
View File
@@ -79,16 +79,12 @@ export interface School {
export interface OfstedInspection {
framework: 'OEIF' | 'ReportCard' | null;
inspection_date: string | null;
/** Start date of the report-card inspection itself (Nov 2025+); null otherwise. */
rc_inspection_date?: string | null;
inspection_type: string | null;
// OEIF fields (old framework, pre-Nov 2025)
overall_effectiveness: 1 | 2 | 3 | 4 | null;
quality_of_education: number | null;
behaviour_attitudes: number | null;
personal_development: number | null;
/** Sixth-form judgement where applicable; sentinel 9 = not applicable. */
sixth_form_provision?: number | null;
leadership_management: number | null;
early_years_provision: number | null;
previous_overall: number | null;
+1 -1
View File
@@ -180,7 +180,7 @@ with DAG(
dbt_build_ees = BashOperator(
task_id="dbt_build",
bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ees_ks2+ stg_legacy_ks2+ stg_ees_ks4+ stg_legacy_ks4+ stg_ees_census+ stg_ees_admissions+ stg_ees_ks2_national+ stg_ees_ks4_national+",
bash_command=f"cd {PIPELINE_DIR}/transform && {DBT_BIN} build --profiles-dir . --target production --select stg_ees_ks2+ stg_legacy_ks2+ stg_ees_ks4+ stg_legacy_ks4+ stg_ees_census+ stg_ees_admissions+ stg_ees_ks2_national+",
)
sync_typesense_ees = BashOperator(
@@ -564,74 +564,6 @@ class EESKs2NationalStream(Stream):
yield record
# ── KS4 National Headlines (national level only — one row per year) ──────────
# Dataset: "National characteristics summary data" (Key stage 4 performance).
# Official England state-funded headline measures, 2018/19 → latest.
# Suppressed values ('z', 'x') → NULL downstream. Progress 8 is legitimately
# absent in years with no KS2 baseline (e.g. 2024/25) — that is DfE policy,
# not missing data.
_KS4_NATIONAL_CSV_URL = (
"https://explore-education-statistics.service.gov.uk/data-catalogue/"
"data-set/1b649e16-01e8-435b-a814-56be2faf9054/csv"
)
_KS4_NATIONAL_COL_MAP = {
"attainment8_average": "attainment_8_score",
"progress8_average": "progress_8_score",
"engmath_94_percent": "english_maths_standard_pass_pct",
"engmath_95_percent": "english_maths_strong_pass_pct",
"ebacc_entering_percent": "ebacc_entry_pct",
"ebacc_94_percent": "ebacc_standard_pass_pct",
"ebacc_95_percent": "ebacc_strong_pass_pct",
"ebacc_aps_average": "ebacc_avg_score",
}
class EESKs4NationalStream(Stream):
"""National KS4 headline averages — one row per academic year.
Filters to geographic_level == 'National', establishment_type_group ==
'All state-funded', breakdown_topic == 'Total', breakdown == 'Total'
so only the England-wide all-pupils row per year is emitted.
"""
name = "ees_ks4_national"
primary_keys = ["time_period"]
replication_key = None
schema = th.PropertiesList(
th.Property("time_period", th.StringType, required=True),
*[th.Property(out, th.StringType) for out in _KS4_NATIONAL_COL_MAP.values()],
).to_dict()
def get_records(self, context):
import pandas as pd
self.logger.info("Downloading KS4 national headlines: %s", _KS4_NATIONAL_CSV_URL)
resp = requests.get(_KS4_NATIONAL_CSV_URL, timeout=60)
resp.raise_for_status()
df = pd.read_csv(io.BytesIO(resp.content), dtype=str, keep_default_na=False)
df.columns = [c.strip().lower() for c in df.columns]
for col, want in [
("geographic_level", "national"),
("establishment_type_group", "all state-funded"),
("breakdown_topic", "total"),
("breakdown", "total"),
]:
if col in df.columns:
df = df[df[col].str.strip().str.lower() == want]
self.logger.info("Emitting %d national KS4 rows", len(df))
for _, row in df.iterrows():
record = {"time_period": row.get("time_period", "").strip()}
for csv_col, field in _KS4_NATIONAL_COL_MAP.items():
record[field] = row.get(csv_col, "").strip()
yield record
# ── Legacy KS2 (pre-COVID wide format from DfE performance tables) ────────────
# The DfE "Compare School Performance" site published school-level KS2 CSVs
# in a wide format (one row per school, ~300 columns). EES only has school-level
@@ -971,7 +903,6 @@ class TapUKEES(Tap):
LegacyKS2Stream(self),
LegacyKS4Stream(self),
EESKs2NationalStream(self),
EESKs4NationalStream(self),
]
@@ -89,10 +89,6 @@ COLUMN_PRIORITY = {
"rc_leadership_governance": ["Leadership and governance"],
"rc_early_years": ["Early years (where applicable)"],
"rc_sixth_form": ["Post-16 provision (where applicable)"],
# Date of the latest FULL inspection — in the renewed framework this is
# the report-card inspection's own start date (col "Inspection start
# date"), distinct from the legacy OEIF graded/ungraded dates above.
"rc_inspection_date": ["Inspection start date"],
"report_url": [
"Web Link (opens in new window)",
"Web link to Ofsted provider page",
@@ -102,44 +98,16 @@ COLUMN_PRIORITY = {
def discover_csv_url() -> str | None:
"""Scrape GOV.UK page to find the latest MI CSV download link.
The page lists a decade of monthly files, oldest first — take the
newest 'latest inspections as at <date>' link by parsing its date,
never matches[0] (that is a 2017 file).
"""
"""Scrape GOV.UK page to find the latest MI CSV download link."""
resp = requests.get(GOV_UK_PAGE, timeout=30)
resp.raise_for_status()
csv_links = re.findall(
# Look for CSV attachment links
matches = re.findall(
r'href="(https://assets\.publishing\.service\.gov\.uk/[^"]+\.csv)"',
resp.text,
)
months = {
'january': 1, 'february': 2, 'march': 3, 'april': 4, 'may': 5, 'june': 6,
'july': 7, 'august': 8, 'september': 9, 'october': 10, 'november': 11, 'december': 12,
'jan': 1, 'feb': 2, 'mar': 3, 'apr': 4, 'jun': 6,
'jul': 7, 'aug': 8, 'sep': 9, 'oct': 10, 'nov': 11, 'dec': 12,
}
parsed_links = []
for link in csv_links:
normalized = link.lower().replace('-', '_')
if 'latest_inspections_as_at' not in normalized:
continue
match = re.search(r'as_at_(\d{1,2})_([a-z]+)_(\d{4})', normalized)
if match:
day, month_str, year = match.groups()
month = months.get(month_str)
if month:
try:
parsed_links.append((datetime(int(year), month, int(day)), link))
except ValueError:
continue
parsed_links.sort(reverse=True)
if parsed_links:
return parsed_links[0][1]
if csv_links:
return csv_links[-1]
if matches:
return matches[0]
# Fall back to ODS
matches = re.findall(
r'href="(https://assets\.publishing\.service\.gov\.uk/[^"]+\.ods)"',
@@ -223,7 +191,6 @@ class OfstedInspectionsStream(Stream):
th.Property("rc_leadership_governance", th.StringType),
th.Property("rc_early_years", th.StringType),
th.Property("rc_sixth_form", th.StringType),
th.Property("rc_inspection_date", th.StringType),
th.Property("report_url", th.StringType),
).to_dict()
@@ -34,7 +34,6 @@ select
rc_leadership_governance,
rc_early_years,
rc_sixth_form,
rc_inspection_date,
report_url
from ranked
where rn = 1
@@ -133,16 +133,6 @@ models:
- name: year
tests: [not_null]
- name: fact_census_benchmarks
description: >
State-school context benchmarks from the pupil census — one row per
phase (primary/secondary), latest census year. fsm_pct/eal_pct are
pupil-weighted means; consumers label them "state-school average
(computed from our dataset)", never "England average".
columns:
- name: phase
tests: [not_null, unique]
- name: fact_admissions
description: School admissions — one row per URN per year
columns:
@@ -1,39 +0,0 @@
{{ config(materialized='table') }}
-- Mart: state-school context benchmarks from the pupil census — one row per
-- phase, latest census year. Computed at import time (never per request).
-- fsm_pct / eal_pct are pupil-weighted means, i.e. "what % of pupils", not
-- "the median school" — this matches how DfE quotes national FSM/EAL rates.
-- Consumers must label these "state-school average (computed from our
-- dataset)" (spec §8.6), never "England average".
with latest as (
select max(year) as year from {{ ref('fact_pupil_characteristics') }}
),
classified as (
select
case
when p.phase_type_grouping ilike '%primary%' then 'primary'
when p.phase_type_grouping ilike '%secondary%' then 'secondary'
end as phase,
p.total_pupils,
p.fsm_pct,
p.eal_pct,
l.year
from {{ ref('fact_pupil_characteristics') }} p
join latest l on p.year = l.year
where p.total_pupils is not null and p.total_pupils > 0
)
select
phase,
max(year) as year,
round((sum(fsm_pct * total_pupils) filter (where fsm_pct is not null)
/ nullif(sum(total_pupils) filter (where fsm_pct is not null), 0))::numeric, 1) as fsm_pct,
round((sum(eal_pct * total_pupils) filter (where eal_pct is not null)
/ nullif(sum(total_pupils) filter (where eal_pct is not null), 0))::numeric, 1) as eal_pct,
round(percentile_cont(0.5) within group (order by total_pupils))::integer as median_pupils
from classified
where phase is not null
group by phase
@@ -1,22 +1,25 @@
{{ config(materialized='table') }}
-- Mart: OFFICIAL DfE KS4 national headline averages — one row per academic
-- year (England, state-funded, all pupils), from the EES national dataset.
-- Replaces the previous unweighted school-level means, which were 715
-- points off every headline measure and produced an arithmetically
-- impossible national Progress 8. gcse_grade_91_pct has no official
-- national series and is NULL (schema kept for the API model).
-- Mart: Computed national KS4 averages — one row per academic year.
-- Unlike fact_ks2_national_averages (official DfE figures), DfE publishes no
-- KS4 national-headline dataset we ingest yet, so these are means computed
-- across the state schools in our dataset. Computed once at build time so the
-- API never has to aggregate the full performance table per request.
-- Semantics match the API's previous per-request computation: rows where
-- attainment_8_score is non-null; per-column means ignore NULLs.
select
year,
attainment_8_score,
progress_8_score,
english_maths_standard_pass_pct,
english_maths_strong_pass_pct,
ebacc_entry_pct,
ebacc_standard_pass_pct,
ebacc_strong_pass_pct,
ebacc_avg_score,
cast(null as double precision) as gcse_grade_91_pct
from {{ ref('stg_ees_ks4_national') }}
round(avg(attainment_8_score)::numeric, 2) as attainment_8_score,
round(avg(progress_8_score)::numeric, 2) as progress_8_score,
round(avg(english_maths_standard_pass_pct)::numeric, 2) as english_maths_standard_pass_pct,
round(avg(english_maths_strong_pass_pct)::numeric, 2) as english_maths_strong_pass_pct,
round(avg(ebacc_entry_pct)::numeric, 2) as ebacc_entry_pct,
round(avg(ebacc_standard_pass_pct)::numeric, 2) as ebacc_standard_pass_pct,
round(avg(ebacc_strong_pass_pct)::numeric, 2) as ebacc_strong_pass_pct,
round(avg(ebacc_avg_score)::numeric, 2) as ebacc_avg_score,
round(avg(gcse_grade_91_pct)::numeric, 2) as gcse_grade_91_pct
from {{ ref('fact_ks4_performance') }}
where attainment_8_score is not null
group by year
order by year
@@ -23,6 +23,5 @@ select
rc_leadership_governance,
rc_early_years,
rc_sixth_form,
rc_inspection_date,
report_url
from {{ ref('stg_ofsted_inspections') }}
@@ -51,9 +51,6 @@ sources:
- name: ees_ks2_national
description: KS2 national headline averages from DfE EES data catalogue — one row per academic year
- name: ees_ks4_national
description: Official KS4 national headline averages from DfE EES data catalogue — one row per academic year
# Phonics: no school-level data on EES (only national/LA level)
- name: fbit_finance
@@ -1,20 +0,0 @@
{{ config(materialized='table') }}
-- Staging model: official DfE KS4 national headline averages — one row per
-- academic year (England, all state-funded, all pupils). Source: EES data
-- catalogue "National characteristics summary data". Suppressed values
-- ('z', 'x') are coerced to NULL by safe_numeric — Progress 8 is 'z' in
-- years with no KS2 baseline (e.g. 2024/25): legitimately unpublished.
select
cast(trim(time_period) as integer) as year,
{{ safe_numeric('attainment_8_score') }} as attainment_8_score,
{{ safe_numeric('progress_8_score') }} as progress_8_score,
{{ safe_numeric('english_maths_standard_pass_pct') }} as english_maths_standard_pass_pct,
{{ safe_numeric('english_maths_strong_pass_pct') }} as english_maths_strong_pass_pct,
{{ safe_numeric('ebacc_entry_pct') }} as ebacc_entry_pct,
{{ safe_numeric('ebacc_standard_pass_pct') }} as ebacc_standard_pass_pct,
{{ safe_numeric('ebacc_strong_pass_pct') }} as ebacc_strong_pass_pct,
{{ safe_numeric('ebacc_avg_score') }} as ebacc_avg_score
from {{ source('raw', 'ees_ks4_national') }}
where time_period ~ '^[0-9]+$'
@@ -49,12 +49,6 @@ renamed as (
{{ parse_report_card_grade('rc_early_years') }}::integer as rc_early_years,
{{ parse_report_card_grade('rc_sixth_form') }}::integer as rc_sixth_form,
-- Start date of the latest FULL inspection (the report-card
-- inspection in the renewed framework). Guarded in the final select:
-- only kept when the row actually carries report-card grades, because
-- in legacy-format files this column is the legacy inspection date.
to_date(nullif(trim(rc_inspection_date), 'NULL'), 'DD/MM/YYYY') as rc_inspection_date_raw,
nullif(trim(report_url), 'NULL') as report_url
from source
where urn is not null
@@ -64,17 +58,5 @@ renamed as (
)
)
select
*,
case
when rc_safeguarding_met is not null
or rc_inclusion is not null
or rc_curriculum_teaching is not null
or rc_achievement is not null
or rc_attendance_behaviour is not null
or rc_personal_development is not null
or rc_leadership_governance is not null
then rc_inspection_date_raw
end as rc_inspection_date
from renamed
select * from renamed
where inspection_date is not null