add xlsx upload and document limits (98b49387) · Commits · certic / Jama

configuration.md

+15 −0

Original line number	Diff line number	Diff line
		@@ -54,6 +54,17 @@ CORS. Après mise à jour, chaque installation doit vérifier ces variables :
		\| `JAMA_PARTIAL_UPLOADS_DIR` \| `${JAMA_VAR_DIR}/partial_uploads` \| Répertoire des morceaux d'upload incomplets. \|
		\| `JAMA_UPLOAD_MAX_CHUNK_SIZE` \| `134217728` \| Taille maximale acceptée pour un morceau d'upload, en octets. La valeur par défaut correspond à 128 Mio. \|
		\| `JAMA_UPLOAD_MAX_TOTAL_CHUNKS` \| `10000` \| Nombre maximal de morceaux autorisés pour un upload. \|
		\| `JAMA_XLSX_UPLOADS_DIR` \| `${JAMA_VAR_DIR}/xlsx_uploads` \| Répertoire privé conservant les imports XLSX jusqu'à leur traitement en arrière-plan. \|
		\| `JAMA_XLSX_UPLOAD_MAX_SIZE` \| `26214400` \| Taille compressée maximale d'un import XLSX, en octets. \|
		\| `JAMA_XLSX_MAX_UNCOMPRESSED_SIZE` \| `104857600` \| Taille décompressée cumulée maximale des entrées d'une archive XLSX. \|
		\| `JAMA_XLSX_MAX_ZIP_ENTRIES` \| `1000` \| Nombre maximal d'entrées dans l'archive XLSX. \|
		\| `JAMA_XLSX_MAX_COMPRESSION_RATIO` \| `200` \| Ratio de compression maximal accepté pour une entrée XLSX. \|
		\| `JAMA_XLSX_MAX_ROWS` \| `50000` \| Nombre maximal de lignes dans la feuille importée. \|
		\| `JAMA_XLSX_MAX_COLUMNS` \| `256` \| Nombre maximal de colonnes dans la feuille importée. \|
		\| `JAMA_XLSX_MAX_CELLS` \| `500000` \| Nombre maximal de cellules parcourues pendant un import. \|
		\| `JAMA_XLSX_MAX_CELL_LENGTH` \| `32767` \| Longueur maximale d'une valeur textuelle de cellule. \|
		\| `JAMA_XLSX_MAX_AFFECTED_OBJECTS` \| `100000` \| Nombre maximal de ressources et collections mises à jour après expansion des cascades. \|
		\| `JAMA_XLSX_MAX_METADATA_WRITES` \| `500000` \| Nombre maximal estimé de valeurs de métadonnées écrites après expansion des cascades. \|
		\| `JAMA_TMP_THUMBNAILS_DIR` \| répertoire temporaire système \| Répertoire de cache des miniatures temporaires générées par les vues `resources`. \|
		\| `JAMA_TMP_THUMBNAILS_MAX_AGE_SECONDS` \| `604800` \| Âge maximal des miniatures temporaires avant suppression par la tâche de nettoyage. La valeur par défaut correspond à 7 jours. Mettre `0` pour désactiver le nettoyage. \|
		\| `JAMA_THUMBNAIL_MAX_SIZE` \| `2000` \| Taille maximale acceptée pour les miniatures simples générées par `resources/simple_thumb`. \|
		@@ -62,6 +73,10 @@ CORS. Après mise à jour, chaque installation doit vérifier ces variables :
		\| `JAMA_MEDIA_SUBPROCESS_TIMEOUT_SECONDS` \| `1800` \| Timeout appliqué aux outils médias lancés en sous-processus (`ffmpeg`, `ffprobe`, `convert`, `vips`, `pdftoppm`, `pdftotext`, etc.). La valeur par défaut correspond à 30 minutes. Mettre `0` pour désactiver le timeout. \|
		\| `JAMA_STATIC_ROOT` \| `${JAMA_VAR_DIR}/static` \| Destination de `collectstatic` pour les fichiers statiques en production. \|

		L'endpoint d'import de métadonnées XLSX exige l'en-tête `X-Project`. Le fichier
		est validé puis conservé dans `JAMA_XLSX_UPLOADS_DIR` jusqu'à son traitement par
		la tâche d'arrière-plan, qui le supprime après réussite ou échec.

		## Base de données et cache

		\| Variable \| Défaut \| Description \|

src/jama/admin_services/tasks.py

+14 −2

Original line number	Diff line number	Diff line
		@@ -90,6 +90,11 @@ def _task_result_project_candidates(task_result: DBTaskResult) -> list[tuple[str
		if collection_id is not None:
		candidates.append((TASK_RESULT_COLLECTION_LOOKUP, collection_id))
		elif task_name == "update_data_from_xlsx_rows":
		xlsx_project_id = _object_id(_argument(args, kwargs, "project_id", 1))
		if xlsx_project_id is not None:
		candidates.append((TASK_RESULT_PROJECT_LOOKUP, xlsx_project_id))
		xlsx_user_task_id = _object_id(_argument(args, kwargs, "user_task_id", 3))
		if xlsx_user_task_id is None and xlsx_project_id is None:
		xlsx_user_task_id = _object_id(_argument(args, kwargs, "user_task_id", 2))
		if xlsx_user_task_id is not None:
		candidates.append((TASK_RESULT_USER_TASK_LOOKUP, xlsx_user_task_id))
		@@ -177,7 +182,14 @@ def task_result_project(task_result: DBTaskResult) -> Project \| None:
		_argument(args, kwargs, "collection_id", 1)
		)
		if task_name == "update_data_from_xlsx_rows":
		return _project_from_user_task(_argument(args, kwargs, "user_task_id", 2))
		project_id = _object_id(_argument(args, kwargs, "project_id", 1))
		user_task_id = _argument(args, kwargs, "user_task_id", 3)
		if user_task_id is None and project_id is None:
		user_task_id = _argument(args, kwargs, "user_task_id", 2)
		return (
		_project_from_user_task(user_task_id)
		or Project.objects.filter(pk=project_id).first()
		)
		return None

src/jama/settings.py

+61 −0

Original line number	Diff line number	Diff line
		@@ -156,6 +156,49 @@ def _default_env_file_content(secret: str, var_dir: Path) -> str:
		"10000",
		"Maximum number of chunks accepted for one upload.",
		),
		(
		"JAMA_XLSX_UPLOADS_DIR",
		str((var_dir / "xlsx_uploads").resolve()),
		"Directory used to hold XLSX imports while background tasks process them.",
		),
		(
		"JAMA_XLSX_UPLOAD_MAX_SIZE",
		str(25 * 1024**2),
		"Maximum compressed XLSX upload size, in bytes.",
		),
		(
		"JAMA_XLSX_MAX_UNCOMPRESSED_SIZE",
		str(100 * 1024**2),
		"Maximum total uncompressed size of an XLSX archive, in bytes.",
		),
		(
		"JAMA_XLSX_MAX_ZIP_ENTRIES",
		"1000",
		"Maximum number of files in an XLSX archive.",
		),
		(
		"JAMA_XLSX_MAX_COMPRESSION_RATIO",
		"200",
		"Maximum compression ratio accepted for an XLSX archive entry.",
		),
		("JAMA_XLSX_MAX_ROWS", "50000", "Maximum worksheet rows per import."),
		("JAMA_XLSX_MAX_COLUMNS", "256", "Maximum worksheet columns per import."),
		("JAMA_XLSX_MAX_CELLS", "500000", "Maximum worksheet cells per import."),
		(
		"JAMA_XLSX_MAX_CELL_LENGTH",
		"32767",
		"Maximum string length accepted in one XLSX cell.",
		),
		(
		"JAMA_XLSX_MAX_AFFECTED_OBJECTS",
		"100000",
		"Maximum resource and collection updates after expanding cascades.",
		),
		(
		"JAMA_XLSX_MAX_METADATA_WRITES",
		"500000",
		"Maximum estimated metadata values written after expanding cascades.",
		),
		(
		"JAMA_TMP_THUMBNAILS_DIR",
		tmp_dir,
		@@ -342,6 +385,20 @@ JAMA_RPC_MAX_BODY_SIZE = int(os.getenv("JAMA_RPC_MAX_BODY_SIZE", 1024 * 1024))
		JAMA_RPC_MAX_BATCH_SIZE = int(os.getenv("JAMA_RPC_MAX_BATCH_SIZE", 100))
		JAMA_UPLOAD_MAX_CHUNK_SIZE = int(os.getenv("JAMA_UPLOAD_MAX_CHUNK_SIZE", 128 * 1024**2))
		JAMA_UPLOAD_MAX_TOTAL_CHUNKS = int(os.getenv("JAMA_UPLOAD_MAX_TOTAL_CHUNKS", 10000))
		JAMA_XLSX_UPLOAD_MAX_SIZE = int(os.getenv("JAMA_XLSX_UPLOAD_MAX_SIZE", 25 * 1024**2))
		JAMA_XLSX_MAX_UNCOMPRESSED_SIZE = int(
		os.getenv("JAMA_XLSX_MAX_UNCOMPRESSED_SIZE", 100 * 1024**2)
		)
		JAMA_XLSX_MAX_ZIP_ENTRIES = int(os.getenv("JAMA_XLSX_MAX_ZIP_ENTRIES", 1000))
		JAMA_XLSX_MAX_COMPRESSION_RATIO = int(os.getenv("JAMA_XLSX_MAX_COMPRESSION_RATIO", 200))
		JAMA_XLSX_MAX_ROWS = int(os.getenv("JAMA_XLSX_MAX_ROWS", 50000))
		JAMA_XLSX_MAX_COLUMNS = int(os.getenv("JAMA_XLSX_MAX_COLUMNS", 256))
		JAMA_XLSX_MAX_CELLS = int(os.getenv("JAMA_XLSX_MAX_CELLS", 500000))
		JAMA_XLSX_MAX_CELL_LENGTH = int(os.getenv("JAMA_XLSX_MAX_CELL_LENGTH", 32767))
		JAMA_XLSX_MAX_AFFECTED_OBJECTS = int(
		os.getenv("JAMA_XLSX_MAX_AFFECTED_OBJECTS", 100000)
		)
		JAMA_XLSX_MAX_METADATA_WRITES = int(os.getenv("JAMA_XLSX_MAX_METADATA_WRITES", 500000))
		JAMA_THUMBNAIL_MAX_SIZE = int(os.getenv("JAMA_THUMBNAIL_MAX_SIZE", 2000))
		JAMA_THUMBNAIL_MAX_CROP_PIXELS = int(
		os.getenv("JAMA_THUMBNAIL_MAX_CROP_PIXELS", 100000)
		@@ -370,6 +427,10 @@ PARTIAL_UPLOADS_DIR = os.getenv("JAMA_PARTIAL_UPLOADS_DIR") or str(
		Path(VAR_DIR, "partial_uploads").resolve()
		)
		os.makedirs(PARTIAL_UPLOADS_DIR, exist_ok=True)
		XLSX_UPLOADS_DIR = os.getenv("JAMA_XLSX_UPLOADS_DIR") or str(
		Path(VAR_DIR, "xlsx_uploads").resolve()
		)
		os.makedirs(XLSX_UPLOADS_DIR, exist_ok=True)

		MEDIA_FILES_DIR = os.getenv("JAMA_FILES_DIR") or str(
		Path(VAR_DIR, "media_source_files").resolve()

src/jama/tests.py

+29 −0

Original line number	Diff line number	Diff line
		@@ -44,6 +44,7 @@ from resources.models import (
		Resource,
		Role,
		ObjectPermission,
		UserTask,
		)


		@@ -83,6 +84,17 @@ class SettingsEnvTemplateTestCase(SimpleTestCase):
		"JAMA_PARTIAL_UPLOADS_DIR",
		"JAMA_UPLOAD_MAX_CHUNK_SIZE",
		"JAMA_UPLOAD_MAX_TOTAL_CHUNKS",
		"JAMA_XLSX_UPLOADS_DIR",
		"JAMA_XLSX_UPLOAD_MAX_SIZE",
		"JAMA_XLSX_MAX_UNCOMPRESSED_SIZE",
		"JAMA_XLSX_MAX_ZIP_ENTRIES",
		"JAMA_XLSX_MAX_COMPRESSION_RATIO",
		"JAMA_XLSX_MAX_ROWS",
		"JAMA_XLSX_MAX_COLUMNS",
		"JAMA_XLSX_MAX_CELLS",
		"JAMA_XLSX_MAX_CELL_LENGTH",
		"JAMA_XLSX_MAX_AFFECTED_OBJECTS",
		"JAMA_XLSX_MAX_METADATA_WRITES",
		"JAMA_TMP_THUMBNAILS_DIR",
		"JAMA_TMP_THUMBNAILS_MAX_AGE_SECONDS",
		"JAMA_THUMBNAIL_MAX_SIZE",
		@@ -276,6 +288,23 @@ class AdminTaskServicesTestCase(TestCase):
		self.assertEqual(project_labels, ["batch project"] * 3)
		self.assertLessEqual(len(queries), 3)

		def test_task_results_infer_xlsx_project_from_new_task_signature(self):
		project = Project.objects.create(label="XLSX task project")
		user = User.objects.create(username="xlsx-task-user")
		user_task = UserTask.objects.create(
		owner=user, project=project, description="XLSX import"
		)
		task_result = self.create_task_result(
		status="FAILED",
		task_path="resources.tasks.update_data_from_xlsx_rows",
		args_kwargs={
		"args": [user.pk, project.pk, "/private/import.xlsx", user_task.pk],
		"kwargs": {},
		},
		)

		self.assertEqual(task_services.task_result_project(task_result), project)

		def test_task_results_queryset_defers_detail_fields(self):
		task_result = self.create_task_result(
		status="FAILED",

src/resources/tasks.py

+172 −42

Original line number	Diff line number	Diff line
		from __future__ import annotations

		import logging
		from collections import Counter
		from typing import List
		from django.conf import settings
		from django.urls import reverse
		@@ -13,6 +16,12 @@ from pathlib import Path
		from django.core.mail import mail_admins
		from django.tasks import task
		from crontask import cron
		from resources.xlsx import (
		XLSXLimitExceeded,
		XLSXValidationError,
		iter_xlsx_rows,
		resolve_xlsx_upload_path,
		)

		logger = logging.getLogger(__name__)

		@@ -241,61 +250,182 @@ def recursive_set_metas_to_collection(
		return True


		def _xlsx_object_id(value, field_name: str) -> int \| None:
		if value in (None, ""):
		return None
		if isinstance(value, bool):
		raise XLSXValidationError(f"invalid {field_name}")
		if isinstance(value, str):
		if not value.strip().isdigit():
		raise XLSXValidationError(f"invalid {field_name}")
		object_id = int(value.strip())
		else:
		try:
		object_id = int(value)
		except (TypeError, ValueError) as error:
		raise XLSXValidationError(f"invalid {field_name}") from error
		if object_id != value:
		raise XLSXValidationError(f"invalid {field_name}")
		if object_id < 1:
		raise XLSXValidationError(f"invalid {field_name}")
		return object_id


		def _xlsx_row_references(
		xlsx_path: Path,
		) -> list[tuple[int \| None, int \| None, bool, int]]:
		references = []
		for row in iter_xlsx_rows(xlsx_path):
		resource_id = _xlsx_object_id(row.get("resource_pk"), "resource_pk")
		collection_id = _xlsx_object_id(row.get("collection_pk"), "collection_pk")
		cascade = str(row.get("cascade") or "").strip().lower() == "y"
		metadata_value_count = sum(
		str(value).count(models.XLSX_MULTIPLE_VALUES_SEPARATOR) + 1
		for column_name, value in row.items()
		if ":" in column_name and value is not None
		)
		references.append((resource_id, collection_id, cascade, metadata_value_count))
		return references


		def _validate_xlsx_import_scope(project: models.Project, xlsx_path: Path) -> None:
		references = _xlsx_row_references(xlsx_path)
		resource_ids = {
		resource_id for resource_id, _, _, _ in references if resource_id is not None
		}
		collection_ids = {
		collection_id
		for _, collection_id, _, _ in references
		if collection_id is not None
		}
		resources_by_id = (
		models.Resource.objects.filter(pk__in=resource_ids, deleted_at__isnull=True)
		.only("id", "ptr_project_id")
		.in_bulk()
		)
		collections_by_id = (
		models.Collection.objects.filter(pk__in=collection_ids, deleted_at__isnull=True)
		.only("id", "project_id")
		.in_bulk()
		)
		if set(resources_by_id) != resource_ids or any(
		resource.ptr_project_id != project.pk for resource in resources_by_id.values()
		):
		raise XLSXValidationError("resource does not belong to the import project")
		if set(collections_by_id) != collection_ids or any(
		collection.project_id != project.pk for collection in collections_by_id.values()
		):
		raise XLSXValidationError("collection does not belong to the import project")

		affected_objects = sum(
		resource_id is not None for resource_id, _, _, _ in references
		) + sum(collection_id is not None for _, collection_id, _, _ in references)
		metadata_writes = sum(
		metadata_value_count * ((resource_id is not None) + (collection_id is not None))
		for resource_id, collection_id, _, metadata_value_count in references
		)
		cascade_counts = Counter(
		(collection_id, metadata_value_count)
		for _, collection_id, cascade, metadata_value_count in references
		if cascade and collection_id is not None
		)
		maximum_affected_objects = settings.JAMA_XLSX_MAX_AFFECTED_OBJECTS
		maximum_metadata_writes = settings.JAMA_XLSX_MAX_METADATA_WRITES
		cascade_sizes = {}
		for (
		collection_id,
		metadata_value_count,
		), occurrence_count in cascade_counts.items():
		collection = collections_by_id[collection_id]
		if collection_id not in cascade_sizes:
		cascade_sizes[collection_id] = (
		collection.descendants_count() + collection.descendant_resources_count()
		)
		cascade_size = cascade_sizes[collection_id]
		affected_objects += cascade_size * occurrence_count
		metadata_writes += cascade_size * metadata_value_count * occurrence_count
		if maximum_affected_objects > 0 and affected_objects > maximum_affected_objects:
		raise XLSXLimitExceeded(
		"XLSX cascade exceeds the configured affected-object limit"
		)
		if maximum_metadata_writes > 0 and metadata_writes > maximum_metadata_writes:
		raise XLSXLimitExceeded(
		"XLSX cascade exceeds the configured metadata-write limit"
		)

		if maximum_affected_objects > 0 and affected_objects > maximum_affected_objects:
		raise XLSXLimitExceeded(
		"XLSX import exceeds the configured affected-object limit"
		)
		if maximum_metadata_writes > 0 and metadata_writes > maximum_metadata_writes:
		raise XLSXLimitExceeded(
		"XLSX import exceeds the configured metadata-write limit"
		)


		@task
		def update_data_from_xlsx_rows(
		user_id: int, xlsx_rows: List[dict], user_task_id: int = None
		):
		user_id: int,
		project_id: int,
		xlsx_path: str,
		user_task_id: int = None,
		) -> bool:
		from django.contrib.auth.models import User
		from resources.models import UserTask
		from resources.acl import UserAccess
		from rpc.methods import (
		ServiceException,
		update_collection_from_xlsx_row,
		update_resource_from_xlsx_row,
		)

		project = None

		# Try to find the project
		for r in xlsx_rows:
		if r.get("resource_pk"):
		first_resource = models.Resource.objects.filter(
		pk=r.get("resource_pk")
		user_task = UserTask.objects.filter(
		pk=user_task_id,
		owner_id=user_id,
		project_id=project_id,
		).first()
		if first_resource:
		project = first_resource.ptr_project
		break
		if r.get("collection_pk"):
		first_collection = models.Collection.objects.filter(
		pk=r.get("collection_pk")
		).first()
		if first_collection:
		project = first_collection.project
		break
		user_task = None
		user = User.objects.filter(pk=user_id).first()
		if user:
		if user_task_id is not None:
		user_task = UserTask.objects.filter(pk=user_task_id).first()
		resolved_xlsx_path = None
		try:
		resolved_xlsx_path = resolve_xlsx_upload_path(xlsx_path)
		user = User.objects.filter(pk=user_id, is_active=True).first()
		project = models.Project.objects.filter(pk=project_id).first()
		if not user or not project:
		raise XLSXValidationError("invalid XLSX import owner or project")
		if user_task_id is not None and not user_task:
		raise XLSXValidationError("invalid XLSX import task")
		UserAccess(user, project).check_project_access()
		if user_task:
		user_task.project = project
		user_task.started_at = timezone.now()
		user_task.save()
		from rpc.methods import (
		update_resource_from_xlsx_row,
		update_collection_from_xlsx_row,
		ServiceException,
		)
		user_task.save(update_fields=["started_at"])

		for xlsx_row in xlsx_rows:
		_validate_xlsx_import_scope(project, resolved_xlsx_path)
		for xlsx_row in iter_xlsx_rows(resolved_xlsx_path):
		try:
		if xlsx_row.get("resource_pk"):
		update_resource_from_xlsx_row(user, xlsx_row)
		if xlsx_row.get("collection_pk"):
		update_collection_from_xlsx_row(user, xlsx_row)
		except ServiceException as e:
		logger.warning(e)
		except ServiceException as error:
		logger.warning(error)
		except (ServiceException, XLSXValidationError) as error:
		logger.warning("XLSX import rejected: %s", error)
		if user_task:
		user_task.finished_at = timezone.now()
		user_task.save()
		user_task.failed_at = timezone.now()
		user_task.save(update_fields=["failed_at"])
		return False
		except Exception:
		if user_task:
		user_task.failed_at = timezone.now()
		user_task.save(update_fields=["failed_at"])
		raise
		else:
		logger.warning(
		f"No such user({user_id}) for update_data_from_xlsx_rows task, canceling."
		)
		if user_task:
		user_task.finished_at = timezone.now()
		user_task.save(update_fields=["finished_at"])
		return True
		finally:
		if resolved_xlsx_path:
		resolved_xlsx_path.unlink(missing_ok=True)


		@cron("0 * * * *")

Original line number	Diff line number	Diff line
		@@ -54,6 +54,17 @@ CORS. Après mise à jour, chaque installation doit vérifier ces variables :
		\| `JAMA_PARTIAL_UPLOADS_DIR` \| `${JAMA_VAR_DIR}/partial_uploads` \| Répertoire des morceaux d'upload incomplets. \|
		\| `JAMA_UPLOAD_MAX_CHUNK_SIZE` \| `134217728` \| Taille maximale acceptée pour un morceau d'upload, en octets. La valeur par défaut correspond à 128 Mio. \|
		\| `JAMA_UPLOAD_MAX_TOTAL_CHUNKS` \| `10000` \| Nombre maximal de morceaux autorisés pour un upload. \|
		\| `JAMA_XLSX_UPLOADS_DIR` \| `${JAMA_VAR_DIR}/xlsx_uploads` \| Répertoire privé conservant les imports XLSX jusqu'à leur traitement en arrière-plan. \|
		\| `JAMA_XLSX_UPLOAD_MAX_SIZE` \| `26214400` \| Taille compressée maximale d'un import XLSX, en octets. \|
		\| `JAMA_XLSX_MAX_UNCOMPRESSED_SIZE` \| `104857600` \| Taille décompressée cumulée maximale des entrées d'une archive XLSX. \|
		\| `JAMA_XLSX_MAX_ZIP_ENTRIES` \| `1000` \| Nombre maximal d'entrées dans l'archive XLSX. \|
		\| `JAMA_XLSX_MAX_COMPRESSION_RATIO` \| `200` \| Ratio de compression maximal accepté pour une entrée XLSX. \|
		\| `JAMA_XLSX_MAX_ROWS` \| `50000` \| Nombre maximal de lignes dans la feuille importée. \|
		\| `JAMA_XLSX_MAX_COLUMNS` \| `256` \| Nombre maximal de colonnes dans la feuille importée. \|
		\| `JAMA_XLSX_MAX_CELLS` \| `500000` \| Nombre maximal de cellules parcourues pendant un import. \|
		\| `JAMA_XLSX_MAX_CELL_LENGTH` \| `32767` \| Longueur maximale d'une valeur textuelle de cellule. \|
		\| `JAMA_XLSX_MAX_AFFECTED_OBJECTS` \| `100000` \| Nombre maximal de ressources et collections mises à jour après expansion des cascades. \|
		\| `JAMA_XLSX_MAX_METADATA_WRITES` \| `500000` \| Nombre maximal estimé de valeurs de métadonnées écrites après expansion des cascades. \|
		\| `JAMA_TMP_THUMBNAILS_DIR` \| répertoire temporaire système \| Répertoire de cache des miniatures temporaires générées par les vues `resources`. \|
		\| `JAMA_TMP_THUMBNAILS_MAX_AGE_SECONDS` \| `604800` \| Âge maximal des miniatures temporaires avant suppression par la tâche de nettoyage. La valeur par défaut correspond à 7 jours. Mettre `0` pour désactiver le nettoyage. \|
		\| `JAMA_THUMBNAIL_MAX_SIZE` \| `2000` \| Taille maximale acceptée pour les miniatures simples générées par `resources/simple_thumb`. \|
		@@ -62,6 +73,10 @@ CORS. Après mise à jour, chaque installation doit vérifier ces variables :
		\| `JAMA_MEDIA_SUBPROCESS_TIMEOUT_SECONDS` \| `1800` \| Timeout appliqué aux outils médias lancés en sous-processus (`ffmpeg`, `ffprobe`, `convert`, `vips`, `pdftoppm`, `pdftotext`, etc.). La valeur par défaut correspond à 30 minutes. Mettre `0` pour désactiver le timeout. \|
		\| `JAMA_STATIC_ROOT` \| `${JAMA_VAR_DIR}/static` \| Destination de `collectstatic` pour les fichiers statiques en production. \|

		L'endpoint d'import de métadonnées XLSX exige l'en-tête `X-Project`. Le fichier
		est validé puis conservé dans `JAMA_XLSX_UPLOADS_DIR` jusqu'à son traitement par
		la tâche d'arrière-plan, qui le supprime après réussite ou échec.

		## Base de données et cache

		\| Variable \| Défaut \| Description \|