fix: real nexus webhook format, atomic dedup, tested live

guarddog_nexus/harvester.py

@@ -4,7 +4,7 @@ import datetime
 import os
 import tempfile
 
-from sqlalchemy import select
+from sqlalchemy.exc import IntegrityError
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from guarddog_nexus.config import config
@@ -41,17 +41,6 @@ async def harvest(
 
     package_name, package_version = info
 
-    existing = await session.scalar(
-        select(Scan.id).where(
-            Scan.package_name == package_name,
-            Scan.package_version == package_version,
-            Scan.repository == repository,
-        )
-    )
-    if existing:
-        log.info("Already scanned %s==%s, skipping", package_name, package_version)
-        return None
-
     scan = Scan(
         package_name=package_name,
         package_version=package_version,
@@ -61,7 +50,13 @@ async def harvest(
         status=ScanStatus.PENDING.value,
     )
     session.add(scan)
-    await session.commit()
+    try:
+        await session.commit()
+    except IntegrityError:
+        await session.rollback()
+        log.info("Already scanned %s==%s (unique), skipping", package_name, package_version)
+        return None
+
     await session.refresh(scan)
 
     os.makedirs(config.temp_dir, exist_ok=True)