merge: feat/exam-marker-cohort-seed (exam-marker foundation)

run/initialization/seed_cohort_9p_ph1.py

@@ -0,0 +1,218 @@
+"""
+seed_cohort_9p_ph1.py — Markable cohort for exam-marker testing.
+
+Creates N student accounts and enrols them ALL into a single class (default the
+Greenfield Year 9 Physics class `9P/Ph1`), so there is a real cohort to mark.
+
+Why: the canonical timetable seeds enrol "one student per year-group band"
+(seed_greenfield_timetable.py), so every class has <=1 student — too few for a
+results table / per-question stats. This seeder fills one class to a usable size.
+
+Mechanics (identical paths to the canonical seeds — nothing bespoke server-side):
+  - auth user:   POST {SUPABASE_URL}/auth/v1/admin/users
+  - profile:     upsert public.profiles  (school_id = institute)
+  - membership:  upsert public.institute_memberships  (role 'student')
+  - enrolment:   POST {API_BASE_URL}/database/timetable/classes/{class_id}/students
+                 (as a school_admin; class_students upsert → idempotent)
+
+Idempotent: re-running skips existing auth users and upserts everything else.
+
+Env required:  SUPABASE_URL, SERVICE_ROLE_KEY   (API_BASE_URL defaults to api-dev)
+Optional env:  COHORT_COUNT, COHORT_CLASS_CODE, COHORT_INSTITUTE_ID,
+               SEED_STUDENT_PASSWORD, SEED_SCHOOL_ADMIN_PASSWORD
+
+Run (dev):
+    SUPABASE_URL=... SERVICE_ROLE_KEY=... API_BASE_URL=http://192.168.0.64:18000 \
+    python3 -c "from run.initialization.seed_cohort_9p_ph1 import seed; seed()"
+"""
+import os
+import time
+import requests
+from typing import Dict, Any, List, Optional
+
+# Greenfield Academy (the school actually populated on dev .94)
+GREENFIELD_ID = os.getenv("COHORT_INSTITUTE_ID", "a1b2c3d4-e5f6-7890-abcd-ef1234567890")
+GREENFIELD_DOMAIN = "greenfieldacademy.test"
+GREENFIELD_ADMIN_EMAIL = f"admin@{GREENFIELD_DOMAIN}"
+
+CLASS_CODE = os.getenv("COHORT_CLASS_CODE", "9P/Ph1")
+COHORT_COUNT = int(os.getenv("COHORT_COUNT", "10"))
+
+# Realistic-ish names so the results table doesn't read "Pupil 01..10".
+COHORT_NAMES = [
+    ("Amelia", "Clarke"), ("Noah", "Bennett"), ("Olivia", "Foster"), ("Leo", "Hughes"),
+    ("Ava", "Patel"), ("Jacob", "Reid"), ("Mia", "Turner"), ("Harry", "Ellis"),
+    ("Isla", "Morgan"), ("Oscar", "Khan"), ("Freya", "Walsh"), ("Theo", "Ndlovu"),
+]
+
+DEFAULT_STUDENT_PASSWORD = "Student@Cc2025!"
+DEFAULT_SCHOOL_ADMIN_PASSWORD = "Admin@Cc2025!"
+
+
+def _ctx() -> Dict[str, str]:
+    return {
+        "supa_url": os.environ["SUPABASE_URL"].rstrip("/"),
+        "service_key": os.environ["SERVICE_ROLE_KEY"],
+        "api_base": os.environ.get("API_BASE_URL", "http://192.168.0.64:18000").rstrip("/"),
+        "student_pw": os.getenv("SEED_STUDENT_PASSWORD", DEFAULT_STUDENT_PASSWORD),
+        "admin_pw": os.getenv("SEED_SCHOOL_ADMIN_PASSWORD", DEFAULT_SCHOOL_ADMIN_PASSWORD),
+    }
+
+
+def _sb_headers(ctx: Dict[str, str]) -> Dict[str, str]:
+    return {
+        "apikey": ctx["service_key"],
+        "Authorization": f"Bearer {ctx['service_key']}",
+        "Content-Type": "application/json",
+    }
+
+
+def _sign_in(ctx: Dict[str, str], email: str, password: str) -> str:
+    r = requests.post(
+        f"{ctx['supa_url']}/auth/v1/token?grant_type=password",
+        headers={"apikey": ctx["service_key"], "Content-Type": "application/json"},
+        json={"email": email, "password": password},
+    )
+    r.raise_for_status()
+    return r.json()["access_token"]
+
+
+def _resolve_class_id(ctx: Dict[str, str]) -> Optional[str]:
+    r = requests.get(
+        f"{ctx['supa_url']}/rest/v1/classes",
+        headers=_sb_headers(ctx),
+        params={"class_code": f"eq.{CLASS_CODE}",
+                "institute_id": f"eq.{GREENFIELD_ID}",
+                "select": "id,name", "limit": "1"},
+    )
+    data = r.json() if r.ok else []
+    return data[0]["id"] if data else None
+
+
+def _existing_auth_users(ctx: Dict[str, str]) -> Dict[str, str]:
+    r = requests.get(
+        f"{ctx['supa_url']}/auth/v1/admin/users",
+        headers=_sb_headers(ctx), params={"per_page": 200},
+    )
+    r.raise_for_status()
+    return {u["email"]: u["id"] for u in r.json().get("users", [])}
+
+
+def _create_auth_user(ctx: Dict[str, str], spec: Dict) -> Optional[str]:
+    r = requests.post(
+        f"{ctx['supa_url']}/auth/v1/admin/users",
+        headers=_sb_headers(ctx),
+        json={
+            "email": spec["email"], "password": ctx["student_pw"], "email_confirm": True,
+            "user_metadata": {
+                "username": spec["username"], "full_name": spec["full_name"],
+                "display_name": spec["display_name"], "user_type": "student",
+            },
+        },
+    )
+    if r.status_code in (200, 201):
+        return r.json()["id"]
+    return None
+
+
+def _upsert(ctx: Dict[str, str], table: str, row: Dict, on_conflict: str) -> bool:
+    h = {**_sb_headers(ctx), "Prefer": "resolution=merge-duplicates,return=minimal"}
+    r = requests.post(f"{ctx['supa_url']}/rest/v1/{table}",
+                      headers=h, json=row, params={"on_conflict": on_conflict})
+    return r.ok
+
+
+def _cohort_specs() -> List[Dict]:
+    specs = []
+    for i in range(1, COHORT_COUNT + 1):
+        first, last = COHORT_NAMES[(i - 1) % len(COHORT_NAMES)]
+        prefix = f"cohort{i:02d}"
+        specs.append({
+            "email": f"{prefix}@{GREENFIELD_DOMAIN}",
+            "username": f"{prefix}.{GREENFIELD_DOMAIN.replace('.', '_')}",
+            "full_name": f"{first} {last}",
+            "display_name": first,
+        })
+    return specs
+
+
+def seed(count: Optional[int] = None) -> Dict[str, Any]:
+    global COHORT_COUNT
+    if count is not None:
+        COHORT_COUNT = count
+    ctx = _ctx()
+    results: Dict[str, Any] = {"class_code": CLASS_CODE, "requested": COHORT_COUNT,
+                              "created": 0, "reused": 0, "enrolled": 0, "errors": []}
+
+    print(f"COHORT SEED → {CLASS_CODE} @ {GREENFIELD_DOMAIN} (target {COHORT_COUNT} students)")
+
+    class_id = _resolve_class_id(ctx)
+    if not class_id:
+        results["errors"].append(f"class {CLASS_CODE} not found for institute {GREENFIELD_ID}")
+        print(f"  ✗ {results['errors'][-1]}")
+        return results
+    print(f"  class_id = {class_id}")
+
+    existing = _existing_auth_users(ctx)
+    specs = _cohort_specs()
+
+    # 1) accounts: auth user + profile + membership
+    uids: Dict[str, str] = {}
+    for spec in specs:
+        email = spec["email"]
+        uid = existing.get(email)
+        if uid:
+            results["reused"] += 1
+        else:
+            uid = _create_auth_user(ctx, spec)
+            if not uid:
+                results["errors"].append(f"create auth user {email}")
+                print(f"  ✗ create {email}")
+                continue
+            results["created"] += 1
+            time.sleep(0.15)
+        uids[email] = uid
+        ok_p = _upsert(ctx, "profiles", {
+            "id": uid, "email": email, "user_type": "student",
+            "username": spec["username"], "full_name": spec["full_name"],
+            "display_name": spec["display_name"], "school_id": GREENFIELD_ID,
+            "neo4j_sync_status": "pending",
+        }, on_conflict="id")
+        ok_m = _upsert(ctx, "institute_memberships", {
+            "profile_id": uid, "institute_id": GREENFIELD_ID, "role": "student", "metadata": {},
+        }, on_conflict="profile_id,institute_id")
+        if not (ok_p and ok_m):
+            results["errors"].append(f"profile/membership {email} (p={ok_p} m={ok_m})")
+
+    # 2) enrol all into the class (via API, as school admin)
+    admin_token = _sign_in(ctx, GREENFIELD_ADMIN_EMAIL, ctx["admin_pw"])
+    for spec in specs:
+        uid = uids.get(spec["email"])
+        if not uid:
+            continue
+        r = requests.post(
+            f"{ctx['api_base']}/database/timetable/classes/{class_id}/students",
+            headers={"Authorization": f"Bearer {admin_token}", "Content-Type": "application/json"},
+            json={"student_id": uid},
+        )
+        body = {}
+        try:
+            body = r.json()
+        except Exception:
+            pass
+        if r.ok and (body.get("status") == "ok" or body.get("row") or body.get("id")):
+            results["enrolled"] += 1
+            print(f"  ✓ {spec['email'].split('@')[0]} → {CLASS_CODE}")
+        else:
+            results["errors"].append(f"enrol {spec['email']}: {r.status_code} {str(body)[:120]}")
+            print(f"  ✗ enrol {spec['email']}: {r.status_code}")
+        time.sleep(0.1)
+
+    print(f"\nDONE: created {results['created']}, reused {results['reused']}, "
+          f"enrolled {results['enrolled']}/{COHORT_COUNT}, errors {len(results['errors'])}")
+    return results
+
+
+if __name__ == "__main__":
+    import json
+    print(json.dumps(seed(), indent=2, default=str))