Files
paperless-ngx/src/documents/versioning.py
T
stumpylog 210d97a522 Fix: index document versions as their root document
The search index and the LLM index hold root documents only, a root being
indexed with its newest version's content. Every write path therefore had to
remember to hand them the root. Several did not: adding or deleting a note on
a version, restoring a trashed document together with its versions, and
reprocessing a version all wrote the version into the index under its own id,
where it could be returned as a separate search result. The paths that walk the
whole library did the same: the document_index reindex command put every
version in the search index, a full LLM index rebuild embedded every version,
and an incremental LLM update scoped to a version id indexed it under its own
id. Asking for documents like a version looked the version's own id up in the
index and silently found nothing.

WriteBatch.add_or_update, WriteBatch.add_or_update_ids and
llm_index_add_or_update_document now resolve a version to its root themselves.
The reindex command and update_llm_index only walk root documents, and an
incremental LLM update is scoped to the roots of the given ids through
versioning.root_document_ids, which add_or_update_ids shares. more_like_id
returns the root's id for a version. The reprocess task no longer picks the root
for the indexes and only still clears the caches of both documents.
2026-10-08 15:22:02 -07:00

295 lines
10 KiB
Python

from __future__ import annotations
from dataclasses import dataclass
from enum import StrEnum
from typing import TYPE_CHECKING
from typing import Any
from django.db.models import F
from django.db.models import OuterRef
from django.db.models import Prefetch
from django.db.models import QuerySet
from django.db.models import Subquery
from django.db.models import Window
from django.db.models.functions import Coalesce
from django.db.models.functions import RowNumber
from documents.models import Document
if TYPE_CHECKING:
from collections.abc import Iterable
from rest_framework.request import Request
def root_document_ids(ids: Iterable[int]) -> QuerySet[int]:
"""
The ids of the root documents of the given documents: a root stands for
itself and a version for its root. Only the indexes' bookkeeping needs
this, since they hold root documents only.
"""
return (
Document.objects.filter(pk__in=ids)
.annotate(root_id=Coalesce("root_document_id", "id"))
.order_by()
.values_list("root_id", flat=True)
.distinct()
)
def versions_newest_first(documents: QuerySet[Document]) -> QuerySet[Document]:
"""
Sorts versions so the newest one comes first using version_index and not on id,
because an existing document can be merged in as a version
"""
return documents.order_by(F("version_index").desc(nulls_last=True), "-id")
def annotate_effective_content(documents: QuerySet[Document]) -> QuerySet[Document]:
"""
Annotates documents with the content of their newest version unless the
queryset already carries the annotation, falling back to their own, so
get_effective_content() can answer from the row rather than querying for
the versions of each document.
"""
if "effective_content" in documents.query.annotations:
return documents
return documents.annotate(
effective_content=Coalesce(
Subquery(
versions_newest_first(
Document.objects.filter(root_document=OuterRef("pk")),
).values("content")[:1],
),
F("content"),
),
)
LATEST_VERSION_CONTENT_PREFETCH_ATTR = "_latest_version_content_prefetch"
def latest_version_content_prefetch() -> Prefetch:
"""
A Prefetch for Document.versions scoped to just the newest version's
content, for get_effective_content()'s fallback when no SQL annotation
is present.
Deliberately not merged into a metadata-only "versions" prefetch (the one
used for the serialized versions list): that one fetches every historical
version of every document, and pulling full OCR content for versions
nobody will read wastes DB transfer/memory at scale. This one is windowed
down to a single row per root, then bounded by Prefetch's own IN-list to
whatever page/result set it's attached to -- one cheap bulk query total,
not one per document and not one per version.
"""
return Prefetch(
"versions",
queryset=(
Document.objects.filter(
root_document_id__isnull=False,
deleted_at__isnull=True,
)
.annotate(
rn=Window(
RowNumber(),
partition_by=F("root_document_id"),
order_by=[
F("version_index").desc(nulls_last=True),
F("id").desc(),
],
),
)
.filter(rn=1)
.only("id", "root_document_id", "content")
),
to_attr=LATEST_VERSION_CONTENT_PREFETCH_ATTR,
)
def has_prefetched_effective_content(document: Document) -> bool:
"""
True if document.get_effective_content() can answer without an extra
per-instance query -- an SQL ``effective_content`` annotation, the lean
latest_version_content_prefetch(), or the metadata-only "versions"
prefetch is already present on the instance.
Callers that haven't set any of those up (e.g. views that build their
own querysets independently of DocumentViewSet.get_queryset(), like
TrashView or GlobalSearchView) intentionally don't pay for version-aware
content resolution -- see DocumentSerializer.to_representation(), which
uses this to decide whether to call get_effective_content() at all.
"""
if hasattr(document, "effective_content"):
return True
if getattr(document, LATEST_VERSION_CONTENT_PREFETCH_ATTR, None) is not None:
return True
prefetched_cache = getattr(document, "_prefetched_objects_cache", None)
return isinstance(prefetched_cache, dict) and "versions" in prefetched_cache
def sort_versions_newest_first(documents: list[Document]) -> list[Document]:
"""
Same sorting as versions_newest_first()
"""
return sorted(
documents,
key=lambda doc: (doc.version_index or 0, doc.id),
reverse=True,
)
class VersionResolutionError(StrEnum):
INVALID = "invalid"
NOT_FOUND = "not_found"
@dataclass(frozen=True, slots=True)
class VersionResolution:
document: Document | None
error: VersionResolutionError | None = None
def _document_manager(*, include_deleted: bool) -> Any:
return Document.global_objects if include_deleted else Document.objects
def get_request_version_param(request: Request) -> str | None:
if hasattr(request, "query_params"):
return request.query_params.get("version")
return None
def get_root_document(doc: Document, *, include_deleted: bool = False) -> Document:
# Use root_document_id to avoid a query when this is already a root.
# If root_document isn't available, fall back to the document itself.
if doc.root_document_id is None:
return doc
if doc.root_document is not None:
return doc.root_document
manager = _document_manager(include_deleted=include_deleted)
root_doc = manager.only("id").filter(id=doc.root_document_id).first()
return root_doc or doc
def get_latest_version_for_root(
root_doc: Document,
*,
include_deleted: bool = False,
) -> Document:
manager = _document_manager(include_deleted=include_deleted)
latest = versions_newest_first(manager.filter(root_document=root_doc)).first()
return latest or root_doc
def latest_version(document: Document) -> Document:
"""
The newest version of a root document, or the document itself if it has
no versions or is a version. Reads a prefetched "versions" lookup when
there is one and queries otherwise.
"""
if document.root_document_id is not None or document.pk is None:
return document
prefetched_cache = getattr(document, "_prefetched_objects_cache", None)
prefetched_versions = (
prefetched_cache.get("versions") if isinstance(prefetched_cache, dict) else None
)
if prefetched_versions is None:
return get_latest_version_for_root(document)
if not prefetched_versions:
return document
return sort_versions_newest_first(list(prefetched_versions))[0]
def resolve_requested_version_for_root(
root_doc: Document,
request: Request,
*,
include_deleted: bool = False,
) -> VersionResolution:
version_param = get_request_version_param(request)
if not version_param:
return VersionResolution(
document=get_latest_version_for_root(
root_doc,
include_deleted=include_deleted,
),
)
try:
version_id = int(version_param)
except (TypeError, ValueError):
return VersionResolution(document=None, error=VersionResolutionError.INVALID)
manager = _document_manager(include_deleted=include_deleted)
candidate = manager.only("id", "root_document_id").filter(id=version_id).first()
if candidate is None:
return VersionResolution(document=None, error=VersionResolutionError.NOT_FOUND)
if candidate.id != root_doc.id and candidate.root_document_id != root_doc.id:
return VersionResolution(document=None, error=VersionResolutionError.NOT_FOUND)
return VersionResolution(document=candidate)
def resolve_effective_document(
request_doc: Document,
request: Request,
*,
include_deleted: bool = False,
) -> VersionResolution:
root_doc = get_root_document(request_doc, include_deleted=include_deleted)
if get_request_version_param(request) is not None:
return resolve_requested_version_for_root(
root_doc,
request,
include_deleted=include_deleted,
)
if request_doc.root_document_id is None:
return VersionResolution(
document=get_latest_version_for_root(
root_doc,
include_deleted=include_deleted,
),
)
return VersionResolution(document=request_doc)
_EFFECTIVE_DOCUMENT_CACHE_ATTR = "_effective_document_resolution_cache"
def resolve_effective_document_by_pk(
pk: int,
request: Request,
*,
include_deleted: bool = False,
) -> VersionResolution:
# Django's `condition()` decorator (used for ETag/Last-Modified) invokes the
# etag_func and last_modified_func separately, and the view itself may resolve
# again -- all against the same request. Cache per-request so a single thumb/
# metadata/preview request doesn't redo this resolution multiple times.
cache = getattr(request, _EFFECTIVE_DOCUMENT_CACHE_ATTR, None)
if cache is None:
cache = {}
setattr(request, _EFFECTIVE_DOCUMENT_CACHE_ATTR, cache)
key = (pk, include_deleted)
if key in cache:
return cache[key]
manager = _document_manager(include_deleted=include_deleted)
request_doc = manager.only("id", "root_document_id").filter(pk=pk).first()
if request_doc is None:
resolution = VersionResolution(
document=None,
error=VersionResolutionError.NOT_FOUND,
)
else:
resolution = resolve_effective_document(
request_doc,
request,
include_deleted=include_deleted,
)
cache[key] = resolution
return resolution