databricks-tellr-app 0.4.3.dev40__tar.gz → 0.4.3.dev42__tar.gz

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: databricks-tellr-app
-Version: 0.4.3.dev40
+Version: 0.4.3.dev42
 Summary: Tellr application package for Databricks Apps
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/databricks_tellr_app.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: databricks-tellr-app
-Version: 0.4.3.dev40
+Version: 0.4.3.dev42
 Summary: Tellr application package for Databricks Apps
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "databricks-tellr-app"
-version = "0.4.3.dev40"
+version = "0.4.3.dev42"
 description = "Tellr application package for Databricks Apps"
 readme = "README.md"
 requires-python = ">=3.10"

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/src/api/main.py

@@ -31,10 +31,7 @@ from src.api.routes.settings import (
     identities_router,
     slide_styles_router,
 )
-from src.api.services.export_job_queue import (
-    start_export_cleanup_worker,
-    start_export_worker,
-)
+from src.api.services.export_job_queue import start_export_worker
 from src.api.services.job_queue import recover_stuck_requests, start_worker
 from src.core.database import (
     is_lakebase_environment,
@@ -52,7 +49,6 @@ IS_TESTING = ENVIRONMENT == "test"
 # Worker task references for cleanup
 _worker_task = None
 _export_worker_task = None
-_export_cleanup_task: asyncio.Task | None = None
 _cleanup_task = None
 _timeout_task = None
 _frontend_assets_stack: ExitStack | None = None
@@ -61,8 +57,7 @@ _frontend_assets_stack: ExitStack | None = None
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     """Lifespan context manager for startup/shutdown events."""
-    global _worker_task, _export_worker_task, _export_cleanup_task
-    global _cleanup_task, _timeout_task, _frontend_assets_stack
+    global _worker_task, _export_worker_task, _cleanup_task, _timeout_task, _frontend_assets_stack
 
     # Startup
     logger.info(f"Starting AI Slide Generator API (environment: {ENVIRONMENT})")
@@ -129,10 +124,6 @@ async def lifespan(app: FastAPI):
         _export_worker_task = await start_export_worker()
         logger.info("Export job queue worker started")
 
-        # Start the export job TTL cleanup worker
-        _export_cleanup_task = await start_export_cleanup_worker()
-        logger.info("Export job cleanup worker started")
-
         # Start the MCP job timeout sweeper
         from src.api.services.job_queue import mark_timed_out_jobs_loop
         _timeout_task = asyncio.create_task(mark_timed_out_jobs_loop())
@@ -178,14 +169,6 @@ async def lifespan(app: FastAPI):
             pass
         logger.info("Export job queue worker stopped")
 
-    if _export_cleanup_task:
-        _export_cleanup_task.cancel()
-        try:
-            await _export_cleanup_task
-        except asyncio.CancelledError:
-            pass
-        logger.info("Export job cleanup worker stopped")
-
     if _cleanup_task:
         _cleanup_task.cancel()
         try:

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/src/api/routes/export.py

@@ -807,11 +807,13 @@ async def poll_pptx_export(job_id: str):
 
 
 @router.get("/pptx/download/{job_id}")
-async def download_pptx_export(job_id: str):
+async def download_pptx_export(job_id: str, background_tasks: BackgroundTasks):
     """Download completed PPTX export.
 
     Args:
         job_id: Job ID from start_pptx_export_async
+        background_tasks: FastAPI background tasks for cleanup
+
     Returns:
         FileResponse with PPTX file
 
@@ -822,6 +824,7 @@ async def download_pptx_export(job_id: str):
     _require_export_job_access(job_id)
     from src.api.services.export_job_queue import (
         get_export_job_status,
+        cleanup_export_job,
     )
 
     job = get_export_job_status(job_id)
@@ -849,6 +852,9 @@ async def download_pptx_export(job_id: str):
         extra={"job_id": job_id, "pptx_filename": filename},
     )
 
+    # Schedule cleanup after download
+    background_tasks.add_task(cleanup_export_job, job_id)
+
     return FileResponse(
         path=output_path,
         filename=filename,
@@ -1289,4 +1295,4 @@ async def export_pptx_huashu_from_html(request: ExportPPTXEditableRequest):
         content=pptx_bytes,
         media_type="application/vnd.openxmlformats-officedocument.presentationml.presentation",
         headers=headers,
-    )
+    )

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/src/api/services/export_job_queue.py

@@ -29,9 +29,6 @@ logger = logging.getLogger(__name__)
 # Only needs to work within a single process.
 export_queue: asyncio.Queue = asyncio.Queue()
 
-EXPORT_CLEANUP_INTERVAL_SECONDS = 300
-EXPORT_JOB_MAX_AGE_MINUTES = 30
-
 
 def generate_job_id() -> str:
     """Generate a unique job ID."""
@@ -513,34 +510,6 @@ async def start_export_worker() -> asyncio.Task:
     return asyncio.create_task(export_worker())
 
 
-async def export_cleanup_loop() -> None:
-    """Periodically remove terminal export jobs that have exceeded their TTL."""
-    logger.info("Export job cleanup worker started")
-    while True:
-        try:
-            await asyncio.sleep(EXPORT_CLEANUP_INTERVAL_SECONDS)
-            cleaned = await asyncio.to_thread(
-                cleanup_stale_jobs,
-                EXPORT_JOB_MAX_AGE_MINUTES,
-            )
-            if cleaned:
-                logger.info("Cleaned up %d stale export job(s)", cleaned)
-        except asyncio.CancelledError:
-            logger.info("Export job cleanup worker shutting down")
-            raise
-        except Exception as e:
-            logger.warning(
-                "Export job cleanup iteration failed",
-                exc_info=True,
-                extra={"error": str(e)},
-            )
-
-
-async def start_export_cleanup_worker() -> asyncio.Task:
-    """Start the background export job cleanup task."""
-    return asyncio.create_task(export_cleanup_loop())
-
-
 def cleanup_export_job(job_id: str) -> None:
     """Clean up job data and temporary files.
 

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/src/services/converter_jail/jail.py

@@ -115,6 +115,11 @@ def netns_available() -> bool:
         except Exception:
             result = False
     _netns_cache = result
+    # NOTE (F-CR-30): in Databricks Apps `unshare --net` is typically
+    # unavailable. We deliberately do NOT fail closed: that would disable all
+    # PPTX/GSlides production conversion. Residual risk is accepted because the
+    # child env is whitelisted via build_scrubbed_env (no credentials) and the
+    # host-side AST import allowlist runs before every jail exec.
     if not result:
         logger.warning(
             "Network namespace unavailable — converter jail runs without netns. "

{databricks_tellr_app-0.4.3.dev40 → databricks_tellr_app-0.4.3.dev42}/src/services/html_to_google_slides.py

@@ -1355,6 +1355,15 @@ class HtmlToGoogleSlidesConverter:
         import shutil as _shutil
         import tempfile as _tempfile
         from src.services.converter_jail import protocol, run_gslides_jail
+        from src.services.converter_jail.ast_guard import DisallowedImport, check_imports
+
+        # Same host-side AST allowlist as _build_gslides_job_dir (F-CR-30): a
+        # rejected snippet is never written or exec'd; caller falls back.
+        try:
+            check_imports(code)
+        except (DisallowedImport, SyntaxError) as exc:
+            logger.warning("GSlides retry snippet rejected pre-jail: %s", exc)
+            return None
 
         job_dir = _tempfile.mkdtemp(prefix="gslides_retry_")
         try: