Fix test_parser_with_table

libs/community/langchain_community/document_loaders/parsers/images.py

@@ -48,7 +48,10 @@ class BaseImageBlobParser(BaseBlobParser):
 
             with blob.as_bytes_io() as buf:
                 if blob.mimetype == "application/x-npy":
+                    try:
                         img = Img.fromarray(numpy.load(buf))
+                    except EOFError:
+                        return  # Ignore too small images
                 else:
                     img = Img.open(buf)
                 content = self._analyze_image(img)

libs/community/langchain_community/document_loaders/parsers/pdf.py

@@ -1476,7 +1476,7 @@ class PDFPlumberParser(BaseBlobParser):
             raise ValueError("mode must be single or page")
         if extract_tables and extract_tables not in ["csv", "markdown", "html"]:
             raise ValueError("mode must be csv, markdown or html")
-        if not extract_images and not images_parser:
+        if extract_images and not images_parser:
             images_parser = RapidOCRBlobParser()
         self.password = password
         self.extract_images = extract_images
@@ -1554,6 +1554,10 @@ class PDFPlumberParser(BaseBlobParser):
                     elif isinstance(content, list):  # Table
                         page_text.append(_JOIN_TABLES + self._convert_table(content))
                     else:  # Image
+                        if self.images_parser:
+                            try:
+                                from PIL import Image as Img
+                                Img.fromarray(content) # Check if image is valid
                                 image_bytes = io.BytesIO()
                                 numpy.save(image_bytes, content)
                                 blob = Blob.from_data(
@@ -1567,6 +1571,10 @@ class PDFPlumberParser(BaseBlobParser):
                                         blob, text_from_image, self.images_inner_format
                                     )
                                 )
+                            except TypeError:
+                                pass
+                            except EOFError:
+                                pass
 
                 all_text = _merge_text_and_extras(extras, "".join(page_text).strip())
 

libs/community/tests/integration_tests/document_loaders/parsers/test_pdf_parsers.py

@@ -247,9 +247,6 @@ def test_parser_with_table(
     mode: str,
     extract_tables: str,
 ) -> None:
-
-    from langchain_community.document_loaders.parsers.images import BaseImageBlobParser
-
     parser_class = getattr(pdf_parsers, parser_factory)
 
     parser = parser_class(