revert version bump of unstructured

Author: aaronsteers

airbyte_cdk/sources/file_based/file_types/unstructured_parser.py

@@ -378,9 +378,15 @@ def _get_filetype(self, file: IOBase, remote_file: RemoteFile) -> Optional[FileT
         # detect_filetype is either using the file name or file content
         # if possible, try to leverage the file name to detect the file type
         # if the file name is not available, use the file content
-        file_type = detect_filetype(
-            file_path=remote_file.uri,
-        )
+        file_type: FileType | None = None
+        try:
+            file_type = detect_filetype(
+                file_path=remote_file.uri,
+            )
+        except Exception:
+            # Path doesn't exist locally. Try something else...
+            pass
+
         if file_type is not None and not file_type == FileType.UNK:
             return file_type
 

poetry.lock

@@ -73,7 +73,7 @@ nltk = { version = "3.9.1", optional = true }
 # This will ensure that even when you run poetry install or pip install, the compatible version of numpy will always be chosen.
 # airbyte-ci will try to install latest version when --use-local-cdk is used, resulting in the conflict.
 numpy = "<2"
-unstructured = { version = "^0.16.5", extras = ["docx", "pptx"], optional = true }
+unstructured = { version = "0.10.27", extras = ["docx", "pptx"], optional = true }
 "unstructured.pytesseract" = { version = ">=0.3.12", optional = true }
 pyjwt = "^2.8.0"
 cryptography = ">=42.0.5,<44.0.0"