Add file id

Unstructured-IO · Oct 2, 2024 · 2d56f42 · 2d56f42
1 parent 135edf5
commit 2d56f42
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 0 deletions.
diff --git a/unstructured_ingest/v2/processes/connectors/fsspec/box.py b/unstructured_ingest/v2/processes/connectors/fsspec/box.py
@@ -90,6 +90,7 @@ def get_metadata(self, file_data: dict) -> FileDataSourceMetadata:
         record_locator = {
             "protocol": self.index_config.protocol,
             "remote_file_path": self.index_config.remote_url,
+            "file_id": file_data.get("id"),
         }
         return FileDataSourceMetadata(
             date_created=date_created,

diff --git a/unstructured_ingest/v2/processes/connectors/fsspec/dropbox.py b/unstructured_ingest/v2/processes/connectors/fsspec/dropbox.py
@@ -71,6 +71,7 @@ def get_metadata(self, file_data: dict) -> FileDataSourceMetadata:
         record_locator = {
             "protocol": self.index_config.protocol,
             "remote_file_path": self.index_config.remote_url,
+            "file_id": file_data.get("id"),
         }
         return FileDataSourceMetadata(
             date_created=date_created,

diff --git a/unstructured_ingest/v2/processes/connectors/fsspec/gcs.py b/unstructured_ingest/v2/processes/connectors/fsspec/gcs.py
@@ -123,6 +123,7 @@ def get_metadata(self, file_data: dict) -> FileDataSourceMetadata:
         record_locator = {
             "protocol": self.index_config.protocol,
             "remote_file_path": self.index_config.remote_url,
+            "file_id": file_data.get("id"),
         }
         return FileDataSourceMetadata(
             date_created=date_created,