Duh, actually use the settings

This commit is contained in:
shamoon
2025-04-23 09:54:48 -07:00
parent 952a10018d
commit 7b6f81e53e
2 changed files with 50 additions and 34 deletions

View File

@@ -22,6 +22,7 @@ from documents.plugins.helpers import ProgressStatusOptions
from documents.utils import copy_basic_file_stats
from documents.utils import copy_file_with_basic_stats
from documents.utils import maybe_override_pixel_limit
from paperless.config import BarcodeConfig
if TYPE_CHECKING:
from collections.abc import Callable
@@ -46,7 +47,7 @@ class Barcode:
Returns True if the barcode value equals the configured separation value,
False otherwise
"""
return self.value == settings.CONSUMER_BARCODE_STRING
return self.value == self.settings.CONSUMER_BARCODE_STRING
@property
def is_asn(self) -> bool:
@@ -54,7 +55,7 @@ class Barcode:
Returns True if the barcode value matches the configured ASN prefix,
False otherwise
"""
return self.value.startswith(settings.CONSUMER_ASN_BARCODE_PREFIX)
return self.value.startswith(self.settings.barcode_asn_prefix)
class BarcodePlugin(ConsumeTaskPlugin):
@@ -67,17 +68,23 @@ class BarcodePlugin(ConsumeTaskPlugin):
- ASN from barcode detection is enabled or
- Barcode support is enabled and the mime type is supported
"""
if settings.CONSUMER_BARCODE_TIFF_SUPPORT:
if self.settings.barcode_enable_tiff_support:
supported_mimes: set[str] = {"application/pdf", "image/tiff"}
else:
supported_mimes = {"application/pdf"}
return (
settings.CONSUMER_ENABLE_ASN_BARCODE
or settings.CONSUMER_ENABLE_BARCODES
or settings.CONSUMER_ENABLE_TAG_BARCODE
self.settings.barcode_enable_asn
or self.settings.barcodes_enabled
or self.settings.barcode_enable_tag
) and self.input_doc.mime_type in supported_mimes
def get_settings(self) -> BarcodeConfig:
"""
Returns the settings for this plugin (Django settings or app config)
"""
return BarcodeConfig()
def setup(self) -> None:
self.temp_dir = tempfile.TemporaryDirectory(
dir=self.base_tmp_dir,
@@ -86,6 +93,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
self.pdf_file: Path = self.input_doc.original_file
self._tiff_conversion_done = False
self.barcodes: list[Barcode] = []
self.settings = self.get_settings()
def run(self) -> None:
# Some operations may use PIL, override pixel setting if needed
@@ -99,7 +107,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
# try reading tags from barcodes
if (
settings.CONSUMER_ENABLE_TAG_BARCODE
self.settings.barcode_enable_tag
and (tags := self.tags) is not None
and len(tags) > 0
):
@@ -110,7 +118,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
logger.info(f"Found tags in barcode: {tags}")
# Lastly attempt to split documents
if settings.CONSUMER_ENABLE_BARCODES and (
if self.settings.barcodes_enabled and (
separator_pages := self.get_separation_pages()
):
# We have pages to split against
@@ -155,10 +163,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
# Update/overwrite an ASN if possible
# After splitting, as otherwise each split document gets the same ASN
if (
settings.CONSUMER_ENABLE_ASN_BARCODE
and (located_asn := self.asn) is not None
):
if self.settings.barcode_enable_asn and (located_asn := self.asn) is not None:
logger.info(f"Found ASN in barcode: {located_asn}")
self.metadata.asn = located_asn
@@ -245,8 +250,8 @@ class BarcodePlugin(ConsumeTaskPlugin):
# Get limit from configuration
barcode_max_pages: int = (
num_of_pages
if settings.CONSUMER_BARCODE_MAX_PAGES == 0
else settings.CONSUMER_BARCODE_MAX_PAGES
if self.settings.barcode_max_pages == 0
else self.settings.barcode_max_pages
)
if barcode_max_pages < num_of_pages: # pragma: no cover
@@ -261,7 +266,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
# Convert page to image
page = convert_from_path(
self.pdf_file,
dpi=settings.CONSUMER_BARCODE_DPI,
dpi=self.settings.barcode_dpi,
output_folder=self.temp_dir.name,
first_page=current_page_number + 1,
last_page=current_page_number + 1,
@@ -272,7 +277,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
logger.debug(f"Image is at {page_filepath}")
# Upscale image if configured
factor = settings.CONSUMER_BARCODE_UPSCALE
factor = self.settings.barcode_upscale
if factor > 1.0:
logger.debug(
f"Upscaling image by {factor} for better barcode detection",
@@ -308,7 +313,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
def asn(self) -> int | None:
"""
Search the parsed barcodes for any ASNs.
The first barcode that starts with CONSUMER_ASN_BARCODE_PREFIX
The first barcode that starts with barcode_asn_prefix
is considered the ASN to be used.
Returns the detected ASN (or None)
"""
@@ -317,7 +322,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
# Ensure the barcodes have been read
self.detect()
# get the first barcode that starts with CONSUMER_ASN_BARCODE_PREFIX
# get the first barcode that starts with barcode_asn_prefix
asn_text: str | None = next(
(x.value for x in self.barcodes if x.is_asn),
None,
@@ -326,7 +331,7 @@ class BarcodePlugin(ConsumeTaskPlugin):
if asn_text:
logger.debug(f"Found ASN Barcode: {asn_text}")
# remove the prefix and remove whitespace
asn_text = asn_text[len(settings.CONSUMER_ASN_BARCODE_PREFIX) :].strip()
asn_text = asn_text[len(self.settings.barcode_asn_prefix) :].strip()
# remove non-numeric parts of the remaining string
asn_text = re.sub(r"\D", "", asn_text)
@@ -356,9 +361,9 @@ class BarcodePlugin(ConsumeTaskPlugin):
for raw in tag_texts.split(","):
try:
tag_str: str | None = None
for regex in settings.CONSUMER_TAG_BARCODE_MAPPING:
for regex in self.settings.barcode_tag_mapping:
if re.match(regex, raw, flags=re.IGNORECASE):
sub = settings.CONSUMER_TAG_BARCODE_MAPPING[regex]
sub = self.settings.barcode_tag_mapping[regex]
tag_str = (
re.sub(regex, sub, raw, flags=re.IGNORECASE)
if sub
@@ -394,13 +399,13 @@ class BarcodePlugin(ConsumeTaskPlugin):
"""
# filter all barcodes for the separator string
# get the page numbers of the separating barcodes
retain = settings.CONSUMER_BARCODE_RETAIN_SPLIT_PAGES
retain = self.settings.barcode_retain_split_pages
separator_pages = {
bc.page: retain
for bc in self.barcodes
if bc.is_separator and (not retain or (retain and bc.page > 0))
} # as below, dont include the first page if retain is enabled
if not settings.CONSUMER_ENABLE_ASN_BARCODE:
if not self.settings.barcode_enable_asn:
return separator_pages
# add the page numbers of the ASN barcodes

View File

@@ -55,17 +55,6 @@ class OcrConfig(OutputTypeConfig):
max_image_pixel: float | None = dataclasses.field(init=False)
color_conversion_strategy: str = dataclasses.field(init=False)
user_args: dict[str, str] | None = dataclasses.field(init=False)
barcodes_enabled: bool = dataclasses.field(init=False)
barcode_enable_tiff_support: bool = dataclasses.field(init=False)
barcode_string: str = dataclasses.field(init=False)
barcode_retain_split_pages: bool = dataclasses.field(init=False)
barcode_enable_asn: bool = dataclasses.field(init=False)
barcode_asn_prefix: str = dataclasses.field(init=False)
barcode_upscale: float = dataclasses.field(init=False)
barcode_dpi: int = dataclasses.field(init=False)
barcode_max_pages: int = dataclasses.field(init=False)
barcode_enable_tag: bool = dataclasses.field(init=False)
barcode_tag_mapping: dict[str, str] = dataclasses.field(init=False)
def __post_init__(self) -> None:
super().__post_init__()
@@ -109,6 +98,28 @@ class OcrConfig(OutputTypeConfig):
user_args = {}
self.user_args = user_args
@dataclasses.dataclass
class BarcodeConfig(BaseConfig):
"""
Barcodes settings
"""
barcodes_enabled: bool = dataclasses.field(init=False)
barcode_enable_tiff_support: bool = dataclasses.field(init=False)
barcode_string: str = dataclasses.field(init=False)
barcode_retain_split_pages: bool = dataclasses.field(init=False)
barcode_enable_asn: bool = dataclasses.field(init=False)
barcode_asn_prefix: str = dataclasses.field(init=False)
barcode_upscale: float = dataclasses.field(init=False)
barcode_dpi: int = dataclasses.field(init=False)
barcode_max_pages: int = dataclasses.field(init=False)
barcode_enable_tag: bool = dataclasses.field(init=False)
barcode_tag_mapping: dict[str, str] = dataclasses.field(init=False)
def __post_init__(self) -> None:
app_config = self._get_config_instance()
self.barcodes_enabled = (
app_config.barcodes_enabled or settings.CONSUMER_ENABLE_BARCODES
)