wip

src/instructlab/sdg/utils/docprocessor.py

@@ -361,6 +361,7 @@ def _process_parsed_docling_json(self, json_fp: Path) -> Dataset:
             Dataset: Dataset object.
         """
         logger.info(f"Processing parsed docling json file: {json_fp}")
+        print(f"THIS IS KHALED: {json_fp=}")
         with open(json_fp, "r", encoding="utf-8") as f:
             data = json.load(f)
 
@@ -574,6 +575,7 @@ def chunk_pdfs(pdf_docs: List, filepaths: List, leaf_node_path: Path, model_name
 
     converter = DocumentConverter(pipeline_options=PipelineOptions())
     parsed_pdfs = [converter.convert_single(d) for d in pdf_docs]
+    print(f"THIS IS KHALED: {parsed_pdfs=}")
     parsed_dicts = [p.render_as_dict() for p in parsed_pdfs]
 
     docling_jsons_path = DOC_FILEPATH / "docling-jsons"