fix: Fixes and tests for StopIteration on .convert()

Signed-off-by: Christoph Auer <cau@zurich.ibm.com>

docling/document_converter.py

@@ -159,7 +159,7 @@ class DocumentConverter:
         raises_on_error: bool = True,
         max_num_pages: int = sys.maxsize,
         max_file_size: int = sys.maxsize,
-    ) -> ConversionResult:
+    ) -> Optional[ConversionResult]:
 
         all_res = self.convert_all(
             source=[source],
@@ -167,7 +167,7 @@ class DocumentConverter:
             max_num_pages=max_num_pages,
             max_file_size=max_file_size,
         )
-        return next(all_res)
+        return next(all_res, None)
 
     @validate_call(config=ConfigDict(strict=True))
     def convert_all(
@@ -186,7 +186,10 @@ class DocumentConverter:
             limits=limits,
         )
         conv_res_iter = self._convert(conv_input, raises_on_error=raises_on_error)
+
+        had_result = False
         for conv_res in conv_res_iter:
+            had_result = True
             if raises_on_error and conv_res.status not in {
                 ConversionStatus.SUCCESS,
                 ConversionStatus.PARTIAL_SUCCESS,
@@ -197,6 +200,14 @@ class DocumentConverter:
             else:
                 yield conv_res
 
+        if not had_result:
+            if raises_on_error:
+                raise RuntimeError(
+                    f"Conversion failed because the provided file has no recognizable format or it wasn't in the list of allowed formats."
+                )
+            else:
+                return None
+
     def _convert(
         self, conv_input: _DocumentConversionInput, raises_on_error: bool
     ) -> Iterator[ConversionResult]:

tests/test_invalid_input.py

@@ -0,0 +1,51 @@
+from io import BytesIO
+from pathlib import Path
+
+import pytest
+
+from docling.backend.docling_parse_backend import DoclingParseDocumentBackend
+from docling.datamodel.base_models import ConversionStatus, DocumentStream, InputFormat
+from docling.datamodel.pipeline_options import PdfPipelineOptions
+from docling.document_converter import DocumentConverter, PdfFormatOption
+
+from .verify_utils import verify_conversion_result_v1, verify_conversion_result_v2
+
+GENERATE = False
+
+
+def get_pdf_path():
+
+    pdf_path = Path("./tests/data/2305.03393v1-pg9.pdf")
+    return pdf_path
+
+
+@pytest.fixture
+def converter():
+    converter = DocumentConverter()
+
+    return converter
+
+
+def test_convert_invalid_doc(converter: DocumentConverter):
+
+    # Test with unrecognizable file format (xyz)
+    result = converter.convert(
+        DocumentStream(name="input.xyz", stream=BytesIO(b"xyz")), raises_on_error=False
+    )
+    assert result is None  # No result comes back at all, since this file is skipped.
+
+    with pytest.raises(RuntimeError):
+        result = converter.convert(
+            DocumentStream(name="input.xyz", stream=BytesIO(b"xyz")),
+            raises_on_error=True,
+        )
+
+    # Test with too small filesize limit
+    result = converter.convert(get_pdf_path(), max_file_size=1, raises_on_error=False)
+    assert result is not None
+    assert result.status == ConversionStatus.FAILURE
+
+    with pytest.raises(RuntimeError):
+        result = converter.convert(
+            get_pdf_path(), max_file_size=1, raises_on_error=True
+        )