Sync DeepSeek OCR HF job code

ds_batch_ocr/stages.py

@@ -138,13 +138,11 @@ def _dataset_features() -> Features:
             "source_image_path": HfImage(),
             "document_with_boxes_image_path": HfImage(),
             "document_markdown_text": Value("string"),
-            "figures": Sequence(
-                {
-                    "figure_id": Value("string"),
-                    "image_path": Value("string"),
-                    "description": Value("string"),
-                }
-            ),
+            "figures": {
+                "figure_id": Sequence(Value("string")),
+                "image_path": Sequence(Value("string")),
+                "description": Sequence(Value("string")),
+            },
             "document_markdown_path": Value("string"),
             "document_final_markdown_path": Value("string"),
             "document_final_markdown_text": Value("string"),
@@ -157,30 +155,54 @@ def _dataset_path(base_dir: Path) -> Path:
     return base_dir / DATASET_FILENAME
 
 
-def _build_dataset_records_iter(documents: Iterable[Dict[str, Any]]) -> Iterable[Dict[str, Any]]:
-    for doc in documents:
-        document_with_boxes_path = doc.get("document_with_boxes_path") 
-        document_with_boxes_relpath = str(document_with_boxes_path) 
+def _figures_to_columnar(figures: Optional[Iterable[Dict[str, Any]]]) -> Dict[str, List[str]]:
+    ids: List[str] = []
+    paths: List[str] = []
+    descriptions: List[str] = []
 
-        figure_entries: List[Dict[str, Any]] = []
-        for figure in doc.get("figures") or []:
+    if figures:
+        for figure in figures:
             if not isinstance(figure, dict):
                 continue
 
-            raw_image_path = figure.get("image_path")
-            image_relpath = str(raw_image_path)
+            ids.append(str(figure.get("figure_id") or figure.get("id") or ""))
+            paths.append(str(figure.get("image_path") or ""))
+            descriptions.append(str(figure.get("description") or ""))
 
-            figure_entries.append(
-                {
-                    "figure_id": str(
-                        figure.get("figure_id")
-                        or figure.get("id")
-                    ),
-                    "image_path": image_relpath,
-                    "description": str(figure.get("description")),
-                    
-                }
-            )
+    return {
+        "figure_id": ids,
+        "image_path": paths,
+        "description": descriptions,
+    }
+
+
+def _figures_from_columnar(figures: Optional[Dict[str, Any]]) -> List[Dict[str, str]]:
+    if not isinstance(figures, dict):
+        return []
+
+    ids = list(figures.get("figure_id") or [])
+    paths = list(figures.get("image_path") or [])
+    descriptions = list(figures.get("description") or [])
+
+    length = max(len(ids), len(paths), len(descriptions))
+    result: List[Dict[str, str]] = []
+    for idx in range(length):
+        result.append(
+            {
+                "figure_id": str(ids[idx]) if idx < len(ids) else "",
+                "image_path": str(paths[idx]) if idx < len(paths) else "",
+                "description": str(descriptions[idx]) if idx < len(descriptions) else "",
+            }
+        )
+    return result
+
+
+def _build_dataset_records_iter(documents: Iterable[Dict[str, Any]]) -> Iterable[Dict[str, Any]]:
+    for doc in documents:
+        document_with_boxes_path = doc.get("document_with_boxes_path") 
+        document_with_boxes_relpath = str(document_with_boxes_path) 
+
+        figure_entries = _figures_to_columnar(doc.get("figures"))
         yield {
             "sample_id": str(doc.get("sample_id")),
             "dataset_index": int(doc.get("dataset_index") or 0),
@@ -230,6 +252,9 @@ def _push_dataset_records(
                 else:
                     data = dict(record)
 
+                if "figures" in data and not isinstance(data["figures"], dict):
+                    data["figures"] = _figures_to_columnar(data.get("figures"))
+
                 yield data
 
         write_jsonl_iter(dataset_path, _record_iterator())
@@ -255,8 +280,7 @@ def _load_dataset_records(path: Path) -> List[Dict[str, Any]]:
             if not line:
                 continue
             record = json.loads(line)
-            if record.get("figures") is None:
-                record["figures"] = []
+            record["figures"] = _figures_from_columnar(record.get("figures"))
             records.append(record)
     return records