mirror of
https://github.com/paperless-ngx/paperless-ngx.git
synced 2026-11-21 04:52:40 +00:00
Chore: resolve the zone OCR source file in one place
This commit is contained in:
1 parent
36de14d2c5
commit
5182b719c5
4 files changed
+184
-121
No files matched your search
+27
-23
@@ -109,6 +109,30 @@ class FieldUpdate:
|
||||
custom_field: CustomField | None = None
|
||||
|
||||
|
||||
@dataclass(frozen=True, kw_only=True, slots=True)
|
||||
class ZoneSourceFile:
|
||||
path: Path
|
||||
is_image: bool
|
||||
|
||||
|
||||
def resolve_zone_source(document: Document) -> ZoneSourceFile | None:
|
||||
"""
|
||||
Pick the stored file zone OCR reads: the archive PDF when it exists on disk,
|
||||
else the source file when it exists, else None.
|
||||
"""
|
||||
if document.has_archive_version:
|
||||
archive_path = document.archive_path
|
||||
if archive_path is not None and archive_path.is_file():
|
||||
return ZoneSourceFile(path=archive_path, is_image=False)
|
||||
source_path = document.source_path
|
||||
if source_path.is_file():
|
||||
return ZoneSourceFile(
|
||||
path=source_path,
|
||||
is_image=document.mime_type.startswith("image/"),
|
||||
)
|
||||
return None
|
||||
|
||||
|
||||
class ZoneOcrReason(StrEnum):
|
||||
"""Why a zone produced no value for its target."""
|
||||
|
||||
@@ -253,7 +277,6 @@ def _visible_templates(
|
||||
|
||||
def run_zone_extraction(
|
||||
document: Document,
|
||||
original_file: Path | None = None,
|
||||
templates: list[OcrTemplate] | None = None,
|
||||
) -> list[ZoneOcrResult]:
|
||||
"""
|
||||
@@ -269,14 +292,13 @@ def run_zone_extraction(
|
||||
return []
|
||||
|
||||
label = f"document {document.pk}"
|
||||
doc_path = _resolve_doc_path(document, original_file)
|
||||
is_archive = doc_path is not None and doc_path == document.archive_path
|
||||
source = resolve_zone_source(document)
|
||||
plans = plan_zone_extraction(
|
||||
doc_path,
|
||||
source.path if source else None,
|
||||
document.page_count,
|
||||
templates,
|
||||
label=label,
|
||||
is_image=not is_archive and document.mime_type.startswith("image/"),
|
||||
is_image=source.is_image if source else False,
|
||||
)
|
||||
results = apply_zone_plans(get_root_document(document), plans, label=label)
|
||||
if zone_results_written(results):
|
||||
@@ -417,24 +439,6 @@ def zone_results_written(results: list[ZoneOcrResult]) -> bool:
|
||||
return any(result.reason is None for result in results)
|
||||
|
||||
|
||||
def _resolve_doc_path(
|
||||
document: Document,
|
||||
original_file: Path | None,
|
||||
) -> Path | None:
|
||||
"""Prefer the archive, then the source, then the given file."""
|
||||
candidates: list[Path | None] = []
|
||||
if document.has_archive_version:
|
||||
candidates.append(document.archive_path)
|
||||
candidates.append(document.source_path)
|
||||
if original_file is not None:
|
||||
candidates.append(original_file)
|
||||
|
||||
for path in candidates:
|
||||
if path is not None and Path(path).is_file():
|
||||
return Path(path)
|
||||
return None
|
||||
|
||||
|
||||
def _resolve_page_idx(page_value: int | None, page_count: int | None) -> int:
|
||||
"""Resolve a 1-based page or -1 (last page), defaulting to the first page."""
|
||||
if page_value == -1:
|
||||
|
||||
Reference in new issue
Block a user