feat: clear print work tempfiles on expiration

Relates #47

Author: danillatency

src/bot/shared_messages.py

@@ -62,17 +62,22 @@ async def go_to_default_state(callback_or_message: CallbackQuery | Message, stat
 
 async def usual_error_answer(event: ErrorEvent):
     answer = None
-    if "message to delete" in str(event.exception):
-        answer = (
-            "A necessary message was lost 🤔\n\n"
-            + "Ignore if everything looks good\n"
-            + html.bold("Otherwise, try to send the file or /scan again\n")
-        )
-    elif "Server disconnected" in str(event.exception):
-        answer = "Telegram has issues with handling our service now 📉\n\n" + html.bold(
+    try:
+        answer = f"{event.exception.response.json()['detail']} 😕\n\n" + html.bold(
             "Try to send the file or /scan again"
         )
-    elif "Request timeout" in str(event.exception):
-        answer = "Telegram or our API is busy now 📉\n\n" + html.bold("Try to send the file or /scan again")
+    except (AttributeError, ValueError):
+        if "message to delete" in str(event.exception):
+            answer = (
+                "A necessary message was lost 🤔\n\n"
+                + "Ignore if everything looks good\n"
+                + html.bold("Otherwise, try to send the file or /scan again\n")
+            )
+        elif "Server disconnected" in str(event.exception):
+            answer = "Telegram has issues with handling our service now 📉\n\n" + html.bold(
+                "Try to send the file or /scan again"
+            )
+        elif "Request timeout" in str(event.exception):
+            answer = "Telegram or our API is busy now 📉\n\n" + html.bold("Try to send the file or /scan again")
     logger.error(f"{event.exception}")
     return answer

src/modules/printing/repository.py

@@ -1,13 +1,19 @@
 __all__ = ["printing_repository"]
 
+import asyncio
+import os
+import pathlib
 import re
 import time
+from asyncio import Task
+from tempfile import _TemporaryFileWrapper
 
 import bs4
 import cups
 import httpx
 from cachetools import TTLCache
 
+from src.api.dependencies import USER_AUTH
 from src.api.logging_ import logger
 from src.config import settings
 from src.config_schema import Printer
@@ -40,6 +46,30 @@ def callback(prompt):
         # Cache printer toner status for 5 minutes
         self._printer_toner_status_cache = TTLCache(maxsize=100, ttl=5 * 60)
 
+        self.tempfiles: dict[tuple[str, str], tuple[_TemporaryFileWrapper[bytes], Task[None]]] = {}
+        self.tempfile_expiration_time = 6 * 60 * 60
+
+    def store_tempfile(self, innohassle_user_id, f):
+        self.tempfiles[(innohassle_user_id, pathlib.Path(f.name).name)] = (
+            f,
+            asyncio.create_task(self.wait_for_tempfile_expiration(innohassle_user_id, f)),
+        )
+
+    async def wait_for_tempfile_expiration(self, innohassle_user_id, f):
+        await asyncio.sleep(self.tempfile_expiration_time)
+        await self.remove_tempfile(innohassle_user_id, pathlib.Path(f.name).name, True)
+
+    async def remove_tempfile(self, innohassle_user_id, filename, expired=False):
+        if (innohassle_user_id, filename) in self.tempfiles:
+            os.unlink(self.get_tempfile_path(innohassle_user_id, filename))
+            if not expired:
+                self.tempfiles[(innohassle_user_id, filename)][1].cancel()
+            del self.tempfiles[(innohassle_user_id, filename)]
+            return True
+
+    def get_tempfile_path(self, innohassle_user_id, filename):
+        return self.tempfiles[(innohassle_user_id, filename)][0].name
+
     def get_printer(self, cups_name: str) -> Printer | None:
         for elem in settings.api.printers_list:
             if elem.cups_name == cups_name:
@@ -172,9 +202,15 @@ def _parse_paper_percentage(self, html: str) -> int | None:
                             return int((level / maxcapacity) * 100)
         return None
 
-    def print_file(self, printer: Printer, file_name: str, title: str, options: PrintingOptions) -> int:
+    def print_file(
+        self, innohassle_user_id: USER_AUTH, filename: str, printer: Printer, options: PrintingOptions
+    ) -> int:
         options_dict = options.model_dump(by_alias=True, exclude_none=True)
-        return self.server.printFile(printer.cups_name, file_name, title, options=options_dict)
+        job_id = self.server.printFile(
+            printer.cups_name, self.get_tempfile_path(innohassle_user_id, filename), "job", options=options_dict
+        )
+        self.remove_tempfile(innohassle_user_id, filename)
+        return job_id
 
     def get_job_status(self, job_id: int) -> JobAttributes:
         attributes = self.server.getJobAttributes(

src/modules/printing/routes.py

@@ -1,5 +1,4 @@
 import asyncio
-import os
 import tempfile
 from pathlib import Path
 from typing import Any
@@ -18,7 +17,6 @@
 from src.modules.printing.repository import printing_repository
 
 router = APIRouter(prefix="/print", tags=["Print"])
-tempfiles: dict[tuple[str, str], Any] = {}
 
 
 @router.get("/job_status")
@@ -33,11 +31,13 @@ async def job_status(job_id: int, _innohassle_user_id: USER_AUTH) -> JobAttribut
 
 @router.get("/get_file", responses={404: {"description": "No such file"}})
 def get_file(filename: str, innohassle_user_id: USER_AUTH) -> FileResponse:
-    if (innohassle_user_id, filename) in tempfiles:
-        full_path = tempfiles[(innohassle_user_id, filename)].name
-        return FileResponse(full_path, headers={"Content-Disposition": f"attachment; filename={filename}"})
+    if (innohassle_user_id, filename) in printing_repository.tempfiles:
+        return FileResponse(
+            printing_repository.get_tempfile_path(innohassle_user_id, filename),
+            headers={"Content-Disposition": f"attachment; filename={filename}"},
+        )
     else:
-        raise HTTPException(404, "No such file")
+        raise HTTPException(404, "No such file. It was removed from our servers due to expiration")
 
 
 @router.get("/get_printers")
@@ -69,7 +69,7 @@ async def get_printer_status(printer_cups_name: str, _innohassle_user_id: USER_A
 @router.post("/prepare", responses={400: {"description": "Unsupported format"}})
 async def prepare_printing(file: UploadFile, innohassle_user_id: USER_AUTH) -> PreparePrintingResponse:
     """
-    Convert a file to pdf and return the path to the converted file
+    Convert a file to PDF and return the path to the converted file
     """
 
     if not file.size:
@@ -80,7 +80,7 @@ async def prepare_printing(file: UploadFile, innohassle_user_id: USER_AUTH) -> P
     if ext == ".pdf":
         f = tempfile.NamedTemporaryFile(dir=settings.api.temp_dir, suffix=".pdf")
         f.write(await file.read())
-        tempfiles[(innohassle_user_id, Path(f.name).name)] = f
+        printing_repository.store_tempfile(innohassle_user_id, f)
         return PreparePrintingResponse(filename=Path(f.name).name, pages=len(PyPDF2.PdfReader(f).pages))
     elif ext in [".doc", ".docx", ".png", ".txt", ".jpg", ".md", ".bmp", ".xlsx", ".xls", ".odt", ".ods"]:
         with (
@@ -92,7 +92,7 @@ async def prepare_printing(file: UploadFile, innohassle_user_id: USER_AUTH) -> P
             # Run conversion in a background thread
             await asyncio.to_thread(converting_repository.any2pdf, in_f.name, out_f.name)
             in_f.close()
-            tempfiles[(innohassle_user_id, Path(out_f.name).name)] = out_f
+            printing_repository.store_tempfile(innohassle_user_id, out_f)
             return PreparePrintingResponse(filename=Path(out_f.name).name, pages=len(PyPDF2.PdfReader(out_f).pages))
     else:
         raise HTTPException(400, f"no support of the {ext} format")
@@ -110,18 +110,15 @@ async def actual_print(
     """
     logger.info(f"Printing options: {printing_options}")
 
-    if (innohassle_user_id, filename) in tempfiles:
+    if (innohassle_user_id, filename) in printing_repository.tempfiles:
         printer = printing_repository.get_printer(printer_cups_name)
         if not printer:
             raise HTTPException(400, "No such printer")
-        full_path = tempfiles[(innohassle_user_id, filename)].name
-        job_id = printing_repository.print_file(printer, full_path, "job", printing_options)
+        job_id = printing_repository.print_file(innohassle_user_id, filename, printer, printing_options)
         logger.info(f"Job {job_id} has started")
-        os.unlink(full_path)
-        del tempfiles[(innohassle_user_id, filename)]
         return job_id
     else:
-        raise HTTPException(404, "No such file")
+        raise HTTPException(404, "No such file. It was removed from our servers due to expiration")
 
 
 @router.post("/cancel", responses={404: {"description": "No such file"}, 400: {"description": "No such printer"}})
@@ -132,12 +129,8 @@ async def cancel_printing(job_id: int, _innohassle_user_id: USER_AUTH) -> None:
 
 @router.post("/cancel_preparation", responses={404: {"description": "No such file"}})
 async def cancel_preparation(filename: str, innohassle_user_id: USER_AUTH) -> None:
-    if (innohassle_user_id, filename) in tempfiles:
-        full_path = tempfiles[(innohassle_user_id, filename)].name
-        os.unlink(full_path)
-        del tempfiles[(innohassle_user_id, filename)]
-    else:
-        raise HTTPException(404, "No such file")
+    if not printing_repository.remove_tempfile(innohassle_user_id, filename):
+        raise HTTPException(404, "No such file. It was removed from our servers due to expiration")
 
 
 @router.post("/debug/getPrinterAttributes")