Use 'poppler' to check if pdfs are encrypted

scripts/pdf_unlock.py

@@ -34,7 +34,7 @@ then unlock the pdfs and overwrite the original versions.
 """
 
 __author__ = "D. Howe"
-__version__ = "0.1.0"
+__version__ = "0.2.0"
 __email__ = "d.howe@wrl.unsw.edu.au"
 
 import os
@@ -44,7 +44,58 @@ import argparse
 import subprocess
 from tqdm import tqdm
 
+
+def pdfinfo(pdf_name, *args):
+    """Get pdf info using the poppler tool 'pdfinfo'.
+
+    Args:
+        pdf_name: path to pdf file
+        args:     list of optional arguments, e.g. ['-isodates', '-v']
+
+    Returns:
+        A dictionary containing the lines of stdout from calling 'pdfinfo'
+    """
+    command_str = ['pdfinfo', *args, pdf_name]
+    result = subprocess.run(command_str, stdout=subprocess.PIPE)
+    stdout = result.stdout.decode('utf-8').split('\n')
+
+    pdf_info = {}
+    for line in stdout:
+        if line:
+            key, val = line.split(':', 1)
+            pdf_info[key] = val.strip()
+
+    return pdf_info
+
+
+def isencrypted(pdf_name):
+    """Check if a pdf is encrypted.
+
+    Args:
+        pdf_name: path to pdf file
+
+    Returns:
+        True if encrypted, otherwise False
+    """
+    pdf_info = pdfinfo(pdf_name)
+    status = pdf_info['Encrypted'].split(' ')[0]
+
+    if status == 'yes':
+        return True
+    else:
+        return False
+
+
 def pdf_unlock(pdf_file, overwrite=False):
+    """Rewrite pdf with Ghostscript, removing encryption.
+
+    Args:
+        pdf_file:  path to pdf file
+        overwrite: boolean flag
+
+    Returns:
+        A dictionary containing the lines of stdout from calling 'pdfinfo'
+    """
     # Create suffix for unlocked pdfs
     suffix = '-unlocked'
 
@@ -53,7 +104,12 @@ def pdf_unlock(pdf_file, overwrite=False):
     output_name = input_name + suffix
 
     # Skip file if it has already been processed
-    if input_name.endswith(suffix):
+    if input_name.endswith(suffix) or os.path.exists(output_name + '.pdf'):
+        return
+
+    # Skip file if it is not encrypted
+    if not isencrypted(pdf_file):
+        print('file skipped')
         return
 
     # Get name of Ghostscript executable