fix: make PyPDF backward compatible (#7996)

* fix: make PyPDF backward compatible

* Add release note

---------

Co-authored-by: Vladimir Blagojevic <dovlex@gmail.com>

haystack/components/converters/pypdf.py

@@ -106,8 +106,14 @@ class PyPDFToDocument:
         :returns:
             Deserialized component.
         """
-        converter_class = deserialize_type(data["init_parameters"]["converter"]["type"])
+
+        if converter := data["init_parameters"].get("converter"):
+            converter_class = deserialize_type(converter["type"])
             data["init_parameters"]["converter"] = converter_class.from_dict(data["init_parameters"]["converter"])
+        else:
+            # Ensures backwards compatibility with Pipelines dumped with < 2.3.0
+            data["init_parameters"]["converter"] = DefaultConverter()
+
         return default_from_dict(cls, data)
 
     @component.output_types(documents=List[Document])

releasenotes/notes/improve-pypdf-backwards-compatibility-bcc75871005e9aba.yaml

@@ -0,0 +1,4 @@
+---
+enhancements:
+ - |
+   Enhanced the PyPDF converter to ensure backwards compatibility with Pipelines dumped with versions older than 2.3.0. The update includes a conditional check to automatically default to the `DefaultConverter` if a specific converter is not provided, improving the component's robustness and ease of use.

test/components/converters/test_pypdf_to_document.py

@@ -40,6 +40,12 @@ class TestPyPDFToDocument:
         assert isinstance(instance, PyPDFToDocument)
         assert isinstance(instance.converter, DefaultConverter)
 
+    def test_from_dict_no_converter(self):
+        data = {"type": "haystack.components.converters.pypdf.PyPDFToDocument", "init_parameters": {}}
+        instance = PyPDFToDocument.from_dict(data)
+        assert isinstance(instance, PyPDFToDocument)
+        assert isinstance(instance.converter, DefaultConverter)
+
     @pytest.mark.integration
     def test_run(self, test_files_path, pypdf_converter):
         """