Upload scripts/load_all_1800_students.py with huggingface_hub

scripts/load_all_1800_students.py

@@ -1,99 +1,99 @@
-"""
-Ingests the complete 1,800 Synthetic Students Dataset into SQLite Database.
-Creates student entities, backlog records, offers, and student login accounts.
-"""
-
-from __future__ import annotations
-
-import os
-import sys
-import time
-
-PROJECT_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
-if PROJECT_ROOT not in sys.path:
-    sys.path.insert(0, PROJECT_ROOT)
-
-if "pyarrow" not in sys.modules:
-    sys.modules["pyarrow"] = None
-
-from sqlalchemy import select
-from backend.app.api.auth.hasher import hash_password
-from backend.app.importers.student_importer import StudentImporter
-from backend.app.storage.database import SessionLocal
-from backend.app.storage.models.entities import StudentModel, UserModel
-from backend.app.storage.repositories.mappers import student_domain_to_entity
-
-
-def load_1800_students() -> None:
-    start_time = time.perf_counter()
-    csv_path = os.path.join(PROJECT_ROOT, "fixtures", "students", "synthetic_students_1800.csv")
-    if not os.path.exists(csv_path):
-        print(f"Error: CSV file not found at {csv_path}")
-        return
-
-    print(f"[*] Reading and parsing {csv_path}...")
-    with open(csv_path, "rb") as f:
-        content = f.read()
-
-    importer = StudentImporter()
-    report = importer.parse_file(content, "synthetic_students_1800.csv", dry_run=False)
-    print(f"[+] Successfully parsed {len(report.imported_students)} students from 9 branches.")
-
-    db = SessionLocal()
-    try:
-        # Precompute password hash once for all 1,800 student logins
-        default_student_pwd_hash = hash_password("123456")
-
-        # Existing roll numbers in DB
-        existing_rolls = set(db.execute(select(StudentModel.roll_no)).scalars().all())
-        existing_user_names = set(db.execute(select(UserModel.username)).scalars().all())
-
-        new_students_count = 0
-        new_users_count = 0
-
-        batch_size = 200
-        for i, s in enumerate(report.imported_students):
-            if s.roll_no not in existing_rolls:
-                entity = student_domain_to_entity(s)
-                db.add(entity)
-                existing_rolls.add(s.roll_no)
-                new_students_count += 1
-
-                if s.roll_no not in existing_user_names:
-                    u = UserModel(
-                        id=f"usr_{s.id}",
-                        username=s.roll_no,
-                        email=s.email,
-                        password_hash=default_student_pwd_hash,
-                        role="student",
-                        student_id=s.id,
-                        is_active=True,
-                    )
-                    db.add(u)
-                    existing_user_names.add(s.roll_no)
-                    new_users_count += 1
-
-            if (i + 1) % batch_size == 0:
-                db.commit()
-                print(f"   --> Committed {i + 1}/{len(report.imported_students)} students...")
-
-        db.commit()
-
-        total_in_db = db.execute(select(StudentModel)).scalars().all()
-        elapsed = time.perf_counter() - start_time
-        print("=" * 65)
-        print(f"[OK] Complete Ingestion Finished in {elapsed:.2f} seconds!")
-        print(f"    - Added new students: {new_students_count}")
-        print(f"    - Added student logins: {new_users_count}")
-        print(f"    - Total students in DB: {len(total_in_db)}")
-        print("=" * 65)
-    except Exception as exc:
-        db.rollback()
-        print(f"[!] Error during ingestion: {exc}")
-        raise
-    finally:
-        db.close()
-
-
-if __name__ == "__main__":
-    load_1800_students()
+"""
+Ingests the complete 1,800 Synthetic Students Dataset into SQLite Database.
+Creates student entities, backlog records, offers, and student login accounts.
+"""
+
+from __future__ import annotations
+
+import os
+import sys
+import time
+
+PROJECT_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
+if PROJECT_ROOT not in sys.path:
+    sys.path.insert(0, PROJECT_ROOT)
+
+if "pyarrow" not in sys.modules:
+    sys.modules["pyarrow"] = None
+
+from sqlalchemy import select
+from backend.app.api.auth.hasher import hash_password
+from backend.app.importers.student_importer import StudentImporter
+from backend.app.storage.database import SessionLocal
+from backend.app.storage.models.entities import StudentModel, UserModel
+from backend.app.storage.repositories.mappers import student_domain_to_entity
+
+
+def load_1800_students() -> None:
+    start_time = time.perf_counter()
+    csv_path = os.path.join(PROJECT_ROOT, "fixtures", "students", "synthetic_students_1800.csv")
+    if not os.path.exists(csv_path):
+        print(f"Error: CSV file not found at {csv_path}")
+        return
+
+    print(f"[*] Reading and parsing {csv_path}...")
+    with open(csv_path, "rb") as f:
+        content = f.read()
+
+    importer = StudentImporter()
+    report = importer.parse_file(content, "synthetic_students_1800.csv", dry_run=False)
+    print(f"[+] Successfully parsed {len(report.imported_students)} students from 9 branches.")
+
+    db = SessionLocal()
+    try:
+        # Precompute password hash once for all 1,800 student logins (admin123)
+        default_student_pwd_hash = hash_password("admin123")
+
+        # Existing roll numbers in DB
+        existing_rolls = set(db.execute(select(StudentModel.roll_no)).scalars().all())
+        existing_user_names = set(db.execute(select(UserModel.username)).scalars().all())
+
+        new_students_count = 0
+        new_users_count = 0
+
+        batch_size = 200
+        for i, s in enumerate(report.imported_students):
+            if s.roll_no not in existing_rolls:
+                entity = student_domain_to_entity(s)
+                db.add(entity)
+                existing_rolls.add(s.roll_no)
+                new_students_count += 1
+
+                if s.roll_no not in existing_user_names:
+                    u = UserModel(
+                        id=f"usr_{s.id}",
+                        username=s.roll_no,
+                        email=s.email,
+                        password_hash=default_student_pwd_hash,
+                        role="student",
+                        student_id=s.id,
+                        is_active=True,
+                    )
+                    db.add(u)
+                    existing_user_names.add(s.roll_no)
+                    new_users_count += 1
+
+            if (i + 1) % batch_size == 0:
+                db.commit()
+                print(f"   --> Committed {i + 1}/{len(report.imported_students)} students...")
+
+        db.commit()
+
+        total_in_db = db.execute(select(StudentModel)).scalars().all()
+        elapsed = time.perf_counter() - start_time
+        print("=" * 65)
+        print(f"[OK] Complete Ingestion Finished in {elapsed:.2f} seconds!")
+        print(f"    - Added new students: {new_students_count}")
+        print(f"    - Added student logins: {new_users_count}")
+        print(f"    - Total students in DB: {len(total_in_db)}")
+        print("=" * 65)
+    except Exception as exc:
+        db.rollback()
+        print(f"[!] Error during ingestion: {exc}")
+        raise
+    finally:
+        db.close()
+
+
+if __name__ == "__main__":
+    load_1800_students()