Fix issue with subdirectories and zip compression, closes #609

src/python/txtai/archive/zip.py

@@ -18,12 +18,17 @@ def pack(self, path, output):
         with ZipFile(output, "w", ZIP_DEFLATED) as zfile:
             for root, _, files in sorted(os.walk(path)):
                 for f in files:
-                    zfile.write(os.path.join(root, f), arcname=f)
+                    # Generate archive name with relative path, if necessary
+                    name = os.path.join(os.path.relpath(root, path), f)
+
+                    # Write file to zip
+                    zfile.write(os.path.join(root, f), arcname=name)
 
     def unpack(self, path, output):
         with ZipFile(path, "r") as zfile:
             # Validate path if directory specified
             for fullpath in zfile.namelist():
+                fullpath = os.path.join(path, fullpath)
                 if os.path.dirname(fullpath) and not self.validate(path, fullpath):
                     raise IOError(f"Invalid zip entry: {fullpath}")
 

test/python/testarchive.py

@@ -20,6 +20,35 @@ class TestArchive(unittest.TestCase):
     Archive tests.
     """
 
+    def testDirectory(self):
+        """
+        Test directory included in compressed files
+        """
+
+        for extension in ["tar", "zip"]:
+            # Create archive instance
+            archive = ArchiveFactory.create()
+
+            # Create subdirectory in archive working path
+            path = os.path.join(archive.path(), "dir")
+            os.makedirs(path, exist_ok=True)
+
+            # Create file in archive working path
+            with open(os.path.join(path, "test"), "w", encoding="utf-8") as f:
+                f.write("test")
+
+            # Save archive
+            path = os.path.join(tempfile.gettempdir(), f"subdir.{extension}")
+            archive.save(path)
+
+            # Extract files from archive
+            archive = ArchiveFactory.create()
+            archive.load(path)
+
+            # Check if file properly extracted
+            path = os.path.join(archive.path(), "dir", "test")
+            self.assertTrue(os.path.exists(path))
+
     def testInvalidTar(self):
         """
         Test invalid tar file