143 lines
4.5 KiB
Python
143 lines
4.5 KiB
Python
"""Asynchronous conversion. This is the path for large documents."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import logging
|
|
|
|
from fastapi import APIRouter, Response, status
|
|
from fastapi.responses import FileResponse
|
|
|
|
from ..deps import get_container
|
|
from ..errors import ConversionError, JobNotFoundError
|
|
from ..models import ConvertRequest, JobAccepted, JobState
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
router = APIRouter(prefix="/jobs", tags=["joby"])
|
|
|
|
|
|
def _result_url(job_id: str) -> str:
|
|
root = get_container().settings.root_path.rstrip("/")
|
|
return f"{root}/jobs/{job_id}/result"
|
|
|
|
|
|
CREATE_DESCRIPTION = """
|
|
Zaradi konverzi do fronty a hned vrati 202 s identifikatorem jobu. Telo je
|
|
stejne jako u POST /convert.
|
|
|
|
Postup sledujte pres GET /jobs/{job_id}, hotove PDF stahnete z
|
|
GET /jobs/{job_id}/result. Volitelny `callback_url` dostane po dokonceni POST
|
|
se stejnym telem, jake vraci GET /jobs/{job_id}.
|
|
|
|
**Uchovavani souboru.** Hotove PDF zustava na disku sluzby, aby slo stahnout.
|
|
Meziprodukty renderu se po dokonceni smazou, zustava jen vysledek. Dostupny je
|
|
`JOB_RESULT_TTL_SECONDS` (vychozi 1 hodina) od dokonceni jobu, presny cas je
|
|
v poli `expires_at`. Pak se soubor i zaznam jobu smazou a dalsi dotaz vraci 404.
|
|
Stahovat lze opakovane. Neuspesny job se maze hned. Uloziste je docasny adresar
|
|
procesu, restart sluzby vysledky ztrati.
|
|
"""
|
|
|
|
|
|
@router.post(
|
|
"",
|
|
response_model=JobAccepted,
|
|
status_code=status.HTTP_202_ACCEPTED,
|
|
summary="Zaradi konverzi do fronty",
|
|
description=CREATE_DESCRIPTION,
|
|
)
|
|
async def create_job(request: ConvertRequest) -> JobAccepted:
|
|
job = get_container().jobs.submit(request)
|
|
return JobAccepted(
|
|
job_id=job.id,
|
|
status=job.state.status,
|
|
created_at=job.state.created_at,
|
|
result_url=_result_url(job.id),
|
|
)
|
|
|
|
|
|
STATE_DESCRIPTION = """
|
|
Stav jobu, postup renderu a vysledek kontrol.
|
|
|
|
Stavy: `queued`, `running`, `done`, `failed`, `cancelled`, `expired`.
|
|
|
|
`progress.pass_number` rozlisuje prvni a druhy pruchod. Druhy nastava jen
|
|
u dokumentu s generovanym obsahem, takze ukazatel postupu probehne dvakrat.
|
|
|
|
`expires_at` je cas, do ktereho je vysledek ke stazeni. `missing_assets` je
|
|
seznam assetu, ktere se nepodarilo nacist, `warnings` obsahuje veci, na ktere
|
|
sluzba upozornuje, aniz by kvuli nim konverze selhala.
|
|
"""
|
|
|
|
|
|
@router.get(
|
|
"/{job_id}",
|
|
response_model=JobState,
|
|
summary="Stav jobu",
|
|
description=STATE_DESCRIPTION,
|
|
)
|
|
async def job_state(job_id: str) -> JobState:
|
|
job = get_container().jobs.get(job_id)
|
|
state = job.state.model_copy()
|
|
if state.status == "done":
|
|
state.result_url = _result_url(job_id)
|
|
return state
|
|
|
|
|
|
@router.get(
|
|
"/{job_id}/result",
|
|
summary="Stahne hotove PDF",
|
|
description=(
|
|
"Stahne vysledek jobu. Soubor se streamuje, nenacita se cely do pameti. "
|
|
"Stahovat lze opakovane, dokud vysledek nevyprsi. Doba dostupnosti je "
|
|
"JOB_RESULT_TTL_SECONDS od dokonceni jobu, presny cas je v poli expires_at "
|
|
"u GET /jobs/{job_id}."
|
|
),
|
|
response_class=Response,
|
|
responses={
|
|
200: {"content": {"application/pdf": {}}, "description": "Hotove PDF."},
|
|
404: {"description": "Job neexistuje nebo uz expiroval."},
|
|
409: {"description": "Job jeste nedobehl nebo skoncil chybou."},
|
|
},
|
|
)
|
|
async def job_result(job_id: str):
|
|
container = get_container()
|
|
job = container.jobs.get(job_id)
|
|
|
|
if job.state.status != "done":
|
|
error = ConversionError(
|
|
f"Vysledek neni k dispozici, job je ve stavu {job.state.status}.",
|
|
{"status": job.state.status},
|
|
)
|
|
error.error_code = "result_not_ready"
|
|
error.status_code = 409
|
|
raise error
|
|
|
|
path = container.storage.result_path(job_id)
|
|
if not path.exists():
|
|
raise JobNotFoundError("Soubor s vysledkem uz neexistuje, job pravdepodobne expiroval.")
|
|
|
|
filename = job.request.filename or "dokument.pdf"
|
|
return FileResponse(
|
|
path,
|
|
media_type="application/pdf",
|
|
filename=filename,
|
|
headers={
|
|
"X-Page-Count": str(job.state.page_count or 0),
|
|
"X-Engine-Used": job.state.engine_used or "",
|
|
},
|
|
)
|
|
|
|
|
|
@router.delete(
|
|
"/{job_id}",
|
|
response_model=JobState,
|
|
summary="Zrusi job nebo smaze jeho vysledek",
|
|
description=(
|
|
"Zrusi bezici job nebo smaze hotovy vysledek hned, bez cekani na expiraci. "
|
|
"Soubor se z uloziste odstrani okamzite, zaznam jobu zustava jeste po dobu "
|
|
"TTL, aby bylo videt, co se s nim stalo."
|
|
),
|
|
)
|
|
async def delete_job(job_id: str) -> JobState:
|
|
return get_container().jobs.cancel(job_id).state
|