Merge pull request #725 from alephdata/release/4.1.2

Release/4.1.2

Author: tillprochaska

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 4.1.0
+current_version = 4.1.2
 tag_name = {new_version}
 commit = True
 tag = True

ingestors/__init__.py

@@ -2,7 +2,7 @@
 
 import logging
 
-__version__ = "4.1.0"
+__version__ = "4.1.2"
 
 logging.getLogger("chardet").setLevel(logging.INFO)
 logging.getLogger("PIL").setLevel(logging.INFO)

ingestors/packages/__init__.py

@@ -4,6 +4,9 @@
 import tarfile
 from pathlib import PurePath
 
+import py7zr
+from py7zr.exceptions import ArchiveError
+
 from ingestors.ingestor import Ingestor
 from ingestors.support.package import PackageSupport
 from ingestors.support.shell import ShellSupport
@@ -24,9 +27,11 @@ def unpack(self, file_path, entity, temp_dir):
                 *pure_file_path.parts[1:-1], reconstructed_filename
             )
 
-        self.exec_command(
-            "7z", "x", str(pure_file_path), "-y", "-r", "-bb0", "-bd", f"-oc:{temp_dir}"
-        )
+        try:
+            with py7zr.SevenZipFile(str(pure_file_path), mode="r") as z:
+                z.extractall(path=temp_dir)
+        except ArchiveError as e:
+            raise ProcessingException(f"Error: {e}")
 
 
 class SingleFilePackageIngestor(PackageSupport, Ingestor):

requirements.txt

@@ -14,6 +14,7 @@ numpy<2.0.0 # pinned because otherwise spacy requires an incompatible numpy
 fingerprints==1.1.1
 fasttext==0.9.3
 pika==1.3.2
+py7zr==1.0.0
 
 # Development
 pytest==8.3.5

setup.py

@@ -5,7 +5,7 @@
 
 setup(
     name="ingest",
-    version="4.1.0",
+    version="4.1.2",
     author="Organized Crime and Corruption Reporting Project",
     packages=find_packages(exclude=["tests"]),
     package_dir={"ingestors": "ingestors"},

tests/fixtures/bad7zip.7z

@@ -0,0 +1 @@
+This is a secret!