Fix: solved embedded image serialization/deserialization with Pydantic. (#95)

Signed-off-by: Eric Musa <[email protected]>

Author: Eric-Musa

docling_core/types/doc/document.py

@@ -445,7 +445,7 @@ class ImageRef(BaseModel):
     mimetype: str
     dpi: int
     size: Size
-    uri: Union[AnyUrl, Path]
+    uri: Union[AnyUrl, Path] = Field(union_mode="left_to_right")
     _pil: Optional[PILImage.Image] = None
 
     @property
@@ -1864,7 +1864,7 @@ def load_from_json(cls, filename: Path) -> "DoclingDocument":
 
         """
         with open(filename, "r") as f:
-            return cls.model_validate(json.loads(f.read()))
+            return cls.model_validate_json(f.read())
 
     def save_as_yaml(
         self,

test/test_docling_doc.py

@@ -693,6 +693,18 @@ def _verify_saved_output(filename: str, paths: List[Path]):
         assert pred == gt, f"pred!=gt for {filename}"
 
 
+def _verify_loaded_output(filename: str, pred=None):
+    gt = DoclingDocument.load_from_json(Path(str(filename) + ".gt"))
+
+    pred = pred or DoclingDocument.load_from_json(Path(filename))
+    assert isinstance(pred, DoclingDocument)
+
+    assert (
+        pred.export_to_dict() == gt.export_to_dict()
+    ), f"pred.export_to_dict() != gt.export_to_dict() for {filename}"
+    assert pred == gt, f"pred!=gt for {filename}"
+
+
 def test_save_to_disk():
 
     doc: DoclingDocument = _construct_doc()
@@ -761,12 +773,18 @@ def test_save_to_disk():
     )
     _verify_saved_output(filename=filename, paths=paths)
 
+    doc_emb_loaded = DoclingDocument.load_from_json(filename)
+    _verify_loaded_output(filename=filename, pred=doc_emb_loaded)
+
     filename = Path("test/data/doc/constructed_doc.referenced.json")
     doc.save_as_json(
         filename=filename, artifacts_dir=image_dir, image_mode=ImageRefMode.REFERENCED
     )
     _verify_saved_output(filename=filename, paths=paths)
 
+    doc_ref_loaded = DoclingDocument.load_from_json(filename)
+    _verify_loaded_output(filename=filename, pred=doc_ref_loaded)
+
     ### YAML
 
     filename = Path("test/data/doc/constructed_doc.embedded.yaml")