diff --git a/example_streamlit_app.py b/example_streamlit_app.py index ffbd8b6..65eb8a6 100644 --- a/example_streamlit_app.py +++ b/example_streamlit_app.py @@ -7,7 +7,7 @@ import base64 async def highlight_pdf(user_input, pdf_file, make_comments): highlighter = Highlighter(comment=make_comments) pdf_buffer = io.BytesIO(pdf_file.read()) - highlighted_pdf_buffer = await highlighter.highlight(user_input, pdf_buffer=pdf_buffer) + highlighted_pdf_buffer = await highlighter.highlight(user_input, pdf_filename=pdf_buffer) return highlighted_pdf_buffer def main(): diff --git a/highlight_pdf.py b/highlight_pdf.py index 6195b35..2a2b3cd 100644 --- a/highlight_pdf.py +++ b/highlight_pdf.py @@ -334,7 +334,11 @@ class Highlighter: ): self.llm = LLM(**self.llm_params) - pdf = pymupdf.open(filename) + #! Fix this + if not isinstance(filename, io.BytesIO): + pdf = pymupdf.open(filename) + else: + pdf = pymupdf.open(stream=filename, filetype="pdf") output_pdf = pymupdf.open() vectorizer = TfidfVectorizer() diff --git a/highlighted_example_pdf_document.pdf b/highlighted_example_pdf_document.pdf deleted file mode 100644 index 8540f1f..0000000 Binary files a/highlighted_example_pdf_document.pdf and /dev/null differ