Feature: page count (#7750)

---------

Co-authored-by: shamoon <4887959+shamoon@users.noreply.github.com>

src/documents/migrations/1053_document_page_count.py

@@ -0,0 +1,62 @@
+# Generated by Django 4.2.16 on 2024-09-21 15:44
+from pathlib import Path
+
+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 Path(settings.ORIGINALS_DIR / fname).resolve()
+
+
+def add_number_of_pages_to_page_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:
+            with pikepdf.Pdf.open(source_path(doc)) as pdf:
+                if pdf.pages is not None:
+                    doc.page_count = len(pdf.pages)
+                    doc.save()
+        except Exception as e:  # pragma: no cover
+            print(f"Error retrieving number of pages for {doc.filename}: {e}")
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("documents", "1052_document_transaction_id"),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name="document",
+            name="page_count",
+            field=models.PositiveIntegerField(
+                blank=False,
+                null=True,
+                unique=False,
+                db_index=False,
+            ),
+        ),
+        migrations.RunPython(
+            add_number_of_pages_to_page_count,
+            migrations.RunPython.noop,
+        ),
+    ]