Merge pull request #21 from openaleph/bug/password-7z

Handle password protected 7z

Author: simonwoerpel

ingestors/packages/__init__.py

@@ -5,9 +5,9 @@
 from pathlib import PurePath
 
 import py7zr
-from py7zr.exceptions import ArchiveError
+from py7zr.exceptions import ArchiveError, PasswordRequired
 
-from ingestors.exc import ProcessingException
+from ingestors.exc import ENCRYPTED_MSG, ProcessingException
 from ingestors.ingestor import Ingestor
 from ingestors.support.package import PackageSupport
 from ingestors.support.shell import ShellSupport
@@ -29,7 +29,10 @@ def unpack(self, file_path, entity, temp_dir):
 
         try:
             with py7zr.SevenZipFile(str(pure_file_path), mode="r") as extractor:
-                extractor.extractall(path=temp_dir)
+                try:
+                    extractor.extractall(path=temp_dir)
+                except PasswordRequired:
+                    raise ProcessingException(ENCRYPTED_MSG)
         except ArchiveError as e:
             raise ProcessingException(f"Error: {e}")
 

ingestors/support/package.py

@@ -1,11 +1,12 @@
-import shutil
 import logging
+import shutil
+
 from followthemoney import model
 
-from ingestors.support.temp import TempFileSupport
-from ingestors.support.encoding import EncodingSupport
 from ingestors.directory import DirectoryIngestor
 from ingestors.exc import ProcessingException
+from ingestors.support.encoding import EncodingSupport
+from ingestors.support.temp import TempFileSupport
 
 log = logging.getLogger(__name__)
 
@@ -36,7 +37,7 @@ def ingest(self, file_path, entity):
             self.manager.delegate(DirectoryIngestor, temp_dir, entity)
         except ProcessingException as e:
             raise ProcessingException(
-                "Could not unpack the contents of this file."
+                f"Could not unpack the contents of this file. {e}"
             ) from e
 
     def unpack(self, file_path, entity, temp_dir):

tests/fixtures/500_pages_password.7z

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 from pprint import pprint  # noqa
 
-from .support import TestCase
+from tests.support import TestCase
 
 
 class PackagesTest(TestCase):
@@ -22,3 +22,12 @@ def test_tar(self):
         self.manager.ingest(fixture_path, entity)
         self.assertEqual(entity.first("processingStatus"), self.manager.STATUS_SUCCESS)
         self.assertEqual(entity.schema.name, "Package")
+
+    def test_password_protected_7z(self):
+        fixture_path, entity = self.fixture("500_pages_password.7z")
+        self.manager.ingest(fixture_path, entity)
+        self.assertEqual(entity.first("processingStatus"), self.manager.STATUS_FAILURE)
+        self.assertEqual(
+            entity.get("processingError")[0],
+            "Could not unpack the contents of this file. The document might be protected with a password. Try removing the password protection and re-uploading the documents.",
+        )