paperless-ngx/src/documents/migrations/1053_document_pages_count.py

75 lines
2.0 KiB
Python

# Generated by Django 4.2.16 on 2024-09-21 15:44
import os
import pikepdf
from django.conf import settings
from django.db import migrations
from django.db import models
from django.utils.termcolors import colorize as colourise
def source_path(self):
if self.filename:
fname = str(self.filename)
return os.path.join(settings.ORIGINALS_DIR, fname)
def add_number_of_pages_to_pages_count(apps, schema_editor):
Document = apps.get_model("documents", "Document")
if not Document.objects.all().exists():
return
for doc in Document.objects.filter(mime_type="application/pdf"):
print(
" {} {} {}".format(
colourise("*", fg="green"),
colourise("Calculating number of pages for", fg="white"),
colourise(doc.filename, fg="cyan"),
),
)
try:
pdf = pikepdf.open(source_path(doc))
if pdf.pages is not None:
doc.pages_count = len(pdf.pages)
doc.save()
except Exception as e: # pragma: no cover
print(f"Error retrieving number of pages for {doc.filename}: {e}")
def remove_number_of_pages_to_pages_count(apps, schema_editor):
Document = apps.get_model("documents", "Document")
if not Document.objects.all().exists():
return
for document in Document.objects.filter(mime_type="application/pdf"):
document.pages_count = 0
document.save()
class Migration(migrations.Migration):
dependencies = [
("documents", "1052_document_transaction_id"),
]
operations = [
migrations.AddField(
model_name="document",
name="pages_count",
field=models.PositiveIntegerField(
blank=False,
null=True,
unique=False,
db_index=False,
),
),
migrations.RunPython(
add_number_of_pages_to_pages_count,
remove_number_of_pages_to_pages_count,
),
]