move id_to_hash to interfaces.py

Unstructured-IO · Apr 4, 2024 · bec0b90 · bec0b90
1 parent 793ef37
commit bec0b90
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/unstructured/ingest/interfaces.py b/unstructured/ingest/interfaces.py
@@ -608,6 +608,10 @@ def process_file(
         logger.info(f"Processing {self.filename}")
 
         elements = self.partition_file(partition_config=partition_config, **partition_kwargs)
+
+        for idx, element in enumerate(elements):
+            element.id_to_hash(idx)
+
         element_dicts = elements_to_dicts(elements)
 
         self.isd_elems_no_filename: t.List[t.Dict[str, t.Any]] = []

diff --git a/unstructured/ingest/pipeline/partition.py b/unstructured/ingest/pipeline/partition.py
@@ -50,9 +50,6 @@ def run(self, ingest_doc_dict) -> Optional[str]:
                 **partition_kwargs,
             )
 
-            for idx, element in enumerate(elements):
-                element.id_to_hash(idx)
-
             with open(json_path, "w", encoding="utf8") as output_f:
                 logger.info(f"writing partitioned content to {json_path}")
                 json.dump(elements, output_f, ensure_ascii=False, indent=2, sort_keys=True)