Merge pull request #33675 from sokumon/pdf-issue

fix: scan pdfs before uploading

frappe/core/doctype/file/file.py

@@ -26,6 +26,7 @@ from frappe.utils import (
 )
 from frappe.utils.file_manager import is_safe_path
 from frappe.utils.image import optimize_image, strip_exif_data
+from frappe.utils.pdf import pdf_contains_js
 
 from .exceptions import (
 	AttachmentLimitReached,
@@ -137,8 +138,8 @@ class File(Document):
 		self.validate_file_path()
 		self.validate_file_url()
 		self.validate_file_on_disk()
-
 		self.file_size = frappe.form_dict.file_size or self.file_size
+		self.check_content()
 
 	def validate_attachment_references(self):
 		if not self.attached_to_doctype:
@@ -388,6 +389,10 @@ class File(Document):
 				exc=FileTypeNotAllowed,
 			)
 
+	def check_content(self):
+		if self.file_type == "PDF" and not pdf_contains_js(self._content):
+			frappe.throw(_("PDF cannot be uploaded, It contains unsafe content"))
+
 	def validate_duplicate_entry(self):
 		if not self.flags.ignore_duplicate_entry_error and not self.is_folder:
 			if not self.content_hash:
@@ -649,7 +654,7 @@ class File(Document):
 
 		if isinstance(self._content, str):
 			self._content = self._content.encode()
-
+		self.check_content()
 		with open(file_path, "wb+") as f:
 			f.write(self._content)
 			os.fsync(f.fileno())

frappe/utils/pdf.py

@@ -384,3 +384,32 @@ def get_wkhtmltopdf_version():
 			pass
 
 	return wkhtmltopdf_version or "0"
+
+
+def pdf_contains_js(file_content):
+	from io import BytesIO
+
+	reader = PdfReader(BytesIO(file_content))
+
+	def has_javascript(obj):
+		if isinstance(obj, dict):
+			for key, value in obj.items():
+				if key in ("/JS", "/JavaScript"):
+					return True
+				if has_javascript(value):
+					return True
+		elif isinstance(obj, list):
+			for item in obj:
+				if has_javascript(item):
+					return True
+		return False
+
+	root = reader.trailer.get("/Root", {})
+	if has_javascript(root):
+		return False
+
+	for page in reader.pages:
+		if has_javascript(page):
+			return False
+
+	return True