bug: omit session handler from serialization to avoid mp issues (#2366)

### Description
The session handler variable can be anything, because it's specific to
the SDK being used for the connector. This can break the serialization
depending on what that is. To avoid this all together, the session
handler itself is not serialized. Instead, it needs to be recreated if
an object is serialized and then deserialized.

CHANGELOG.md

@@ -1,4 +1,4 @@
-## 0.11.9-dev3
+## 0.11.9-dev4
 
 ### Enhancements
 

unstructured/__version__.py

@@ -1 +1 @@
-__version__ = "0.11.9-dev3"  # pragma: no cover
+__version__ = "0.11.9-dev4"  # pragma: no cover

unstructured/ingest/interfaces.py

@@ -299,6 +299,8 @@ def add_props(self, as_dict: dict, props: t.List[str]):
 
     def to_dict(self, **kwargs) -> t.Dict[str, Json]:
         as_dict = _asdict(self, **kwargs)
+        if "_session_handle" in as_dict:
+            as_dict.pop("_session_handle", None)
         self.add_props(as_dict=as_dict, props=self.properties_to_serialize)
         if getattr(self, "_source_metadata") is not None:
             self.add_props(as_dict=as_dict, props=self.metadata_properties)

unstructured/ingest/pipeline/source.py

@@ -28,14 +28,8 @@ def get_single(self, doc: BaseSingleIngestDoc, ingest_doc_dict: dict) -> str:
             # Still need to fetch metadata if file exists locally
             doc.update_source_metadata()
         else:
-            # TODO: update all to use doc.to_json(redact_sensitive=True) once session handler
-            # can be serialized
-            try:
-                serialized_doc = doc.to_json(redact_sensitive=True)
-                logger.debug(f"Fetching {serialized_doc} - PID: {os.getpid()}")
-            except Exception as e:
-                logger.warning("failed to print full doc: ", e)
-                logger.debug(f"Fetching {doc.__class__.__name__} - PID: {os.getpid()}")
+            serialized_doc = doc.to_json(redact_sensitive=True)
+            logger.debug(f"Fetching {serialized_doc} - PID: {os.getpid()}")
             if self.retry_strategy:
                 self.retry_strategy(doc.get_file)
             else: