diff --git a/src/khoj/processor/pdf/pdf_to_entries.py b/src/khoj/processor/pdf/pdf_to_entries.py index 78ba034e..81c2250f 100644 --- a/src/khoj/processor/pdf/pdf_to_entries.py +++ b/src/khoj/processor/pdf/pdf_to_entries.py @@ -70,9 +70,10 @@ class PdfToEntries(TextToEntries): f.write(bytes) try: loader = PyMuPDFLoader(f"{tmp_file}", extract_images=True) + pdf_entries_per_file = [page.page_content for page in loader.load()] except ImportError: loader = PyMuPDFLoader(f"{tmp_file}") - pdf_entries_per_file = [page.page_content for page in loader.load()] + pdf_entries_per_file = [page.page_content for page in loader.load()] entry_to_location_map += zip(pdf_entries_per_file, [pdf_file] * len(pdf_entries_per_file)) entries.extend(pdf_entries_per_file) except Exception as e: