Coverage for backend / app / job_email_scraping / migrate_email_ids.py: 0%

1"""Migration script to update external_email_id from IMAP sequence numbers to UIDs. 

2 

3Fetches all emails from the past 1000 days via IMAP, matches them to database 

4records by subject + date_received, and updates the database. 

5Multiple IMAP matches for a single DB record are highlighted as ambiguous. 

6""" 

7 

8import sys 

9 

10from app.config import settings 

11from app.database import get_db 

12from app.emails.email_service import EmailService 

13from app.job_email_scraping.models import JobEmail 

14 

15 

16def run_migration(dry_run: bool = True): 

17 """Run the migration. 

18 :param dry_run: If True, only print what would change without updating the DB.""" 

19 

20 db = next(get_db()) 

21 email_service = EmailService(settings.scraper_email_username, settings.scraper_email_password) 

22 

23 vps_emails = email_service.get_emails(timedelta_days=10000) 

24 emails_db = db.query(JobEmail).order_by(JobEmail.external_email_id).all() 

25 emails_db = [email for email in emails_db if not email.external_email_id.startswith("jam")] 

26 emails_db = sorted(emails_db, key=lambda email: int(email.external_email_id)) 

27 

28 for email_db in emails_db[::-1]: 

29 matching = [ 

30 vps_email 

31 for vps_email in vps_emails 

32 if vps_email["date"] == email_db.date_received and vps_email["subject"] == email_db.subject 

33 ] 

34 if len(matching) != 1: 

35 raise AssertionError(f"Should be 1, Is {len(matching)}") 

36 print( 

37 f"Current external_email_id is {email_db.external_email_id}, new UID is {matching[0]["id"]}. Difference is {int(email_db.external_email_id) - int(matching[0]["id"])}" 

38 ) 

39 if dry_run: 

40 pass 

41 else: 

42 email_db.external_email_id = matching[0]["id"] 

43 db.commit() 

44 

45 

46if __name__ == "__main__": 

47 dry = "--apply" not in sys.argv 

48 if dry: 

49 print("=== DRY RUN MODE (pass --apply to update the database) ===\n") 

50 else: 

51 print("=== APPLYING CHANGES ===\n") 

52 

53 run_migration(dry_run=dry)