From a4780521940a77fca088e01a9aa574b16ce019de Mon Sep 17 00:00:00 2001
From: Florian Zirker <florian.zirker@kapsweyer.de>
Date: Fri, 23 Feb 2024 18:20:41 +0100
Subject: [PATCH] paperless: move script to this git

---
 .editorconfig                          |  6 ++
 .gitignore                             |  3 +-
 paperless/docker-compose.yaml          |  2 +-
 paperless/scripts/removePdfPassword.py | 79 ++++++++++++++++++++++++++
 4 files changed, 88 insertions(+), 2 deletions(-)
 create mode 100644 paperless/scripts/removePdfPassword.py

diff --git a/.editorconfig b/.editorconfig
index ec07101..b7a329e 100644
--- a/.editorconfig
+++ b/.editorconfig
@@ -16,3 +16,9 @@ tab_width = 2
 
 [*.md]
 trim_trailing_whitespace = false
+
+[*.py]
+end_of_line    = lf
+indent_style   = space
+indent_size    = 4
+tab_width      = 4
diff --git a/.gitignore b/.gitignore
index 8682eff..ba238b6 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 .env
-**/.vscode/*
\ No newline at end of file
+**/.vscode/*
+paperless/scripts/passwords.txt
diff --git a/paperless/docker-compose.yaml b/paperless/docker-compose.yaml
index 9e77420..53c1caf 100644
--- a/paperless/docker-compose.yaml
+++ b/paperless/docker-compose.yaml
@@ -40,7 +40,7 @@ services:
       retries: 5
     volumes:
       - ${VOLUMES_PATH}/paperless/data:/usr/src/paperless/data
-      - ${VOLUMES_PATH}/paperless/scripts:/usr/src/paperless/scripts
+      - ./scripts:/usr/src/paperless/scripts
       - ${MEDIA_PATH}:/usr/src/paperless/media
       - ${EXPORT_PATH}:/usr/src/paperless/export
       - ${CONSUME_PATH}:/usr/src/paperless/consume
diff --git a/paperless/scripts/removePdfPassword.py b/paperless/scripts/removePdfPassword.py
new file mode 100644
index 0000000..4931931
--- /dev/null
+++ b/paperless/scripts/removePdfPassword.py
@@ -0,0 +1,79 @@
+#!/usr/bin/env python
+
+"""Script for papeless-ngx to decrypt pdf files secured with a password
+Passwords to check ar in file /usr/src/paperless/scripts/passwords.txt
+Original from https://piep.tech/posts/automatic-password-removal-in-paperless-ngx/
+"""
+
+import sys
+import os
+import pikepdf
+
+
+PWD_PATH = "/usr/src/paperless/scripts/passwords.txt"
+#PWD_PATH = "input/passwords.txt"
+
+
+def read_passwords_from_file():
+    """Read Paswors from file"""
+    passwords = []
+
+    with open(PWD_PATH, "r", encoding="utf-8") as f:
+        passwords = f.readlines()
+
+    passwords = list(filter(len, map(str.strip, passwords)))
+    if not passwords:
+        print("Empty password file")
+
+    return passwords
+
+
+def unlock_pdf(file_path, password_list):
+    """Removes Password from PDF. A list of passwords is tried through."""
+    for pwd in password_list:
+        try:
+            with pikepdf.open(file_path, password=pwd, allow_overwriting_input=True) as pdf:
+                pdf.save(file_path)
+                return True
+        except pikepdf.PasswordError:
+            continue
+    return False
+
+
+def is_pdf_encrypted(file_path):
+    """Checks if a PDF file is encrypted"""
+    try:
+        with pikepdf.open(file_path) as pdf:
+            return pdf.is_encrypted
+    except pikepdf.PasswordError:
+        return True
+
+
+def is_pdf(file_path):
+    """Check if Filename ends with PDF. File is not opened"""
+    return os.path.splitext(file_path)[-1].lower() == ".pdf"
+
+
+if __name__ == "__main__":
+
+    doc_path = os.environ.get('DOCUMENT_WORKING_PATH')
+
+    if not os.path.exists(doc_path):
+        print(f"Document {doc_path} not found.")
+        sys.exit(1)
+
+    if not is_pdf(doc_path):
+        print(f"Document {doc_path} is not a PDF. So no decryption needed.")
+        sys.exit(0)
+
+    if not is_pdf_encrypted(doc_path):
+        print(f"Document {doc_path} was already decrypted and no password needed.")
+        sys.exit(0)
+
+    pwds = read_passwords_from_file()
+    if unlock_pdf(doc_path, pwds):
+        print(f"Document {doc_path} successfull decrypted.")
+        sys.exit(0)
+    else:
+        print(f"Document {doc_path} could not be decrypted.")
+        sys.exit(1)