diff --git a/scraper/db.py b/scraper/db.py index 9b97517..5f2582b 100644 --- a/scraper/db.py +++ b/scraper/db.py @@ -168,6 +168,7 @@ def _upsert_postgres(jobs_list: list, db_url: str): """ records = [] + seen_hashes = set() now = datetime.datetime.now(datetime.timezone.utc) for j in jobs_list: @@ -175,6 +176,9 @@ def _upsert_postgres(jobs_list: list, db_url: str): if not job_url: continue job_hash = generate_job_hash(job_url) + if job_hash in seen_hashes: + continue + seen_hashes.add(job_hash) job_id = "job_" + str(uuid.uuid4()).replace("-", "")[:20] records.append((