Download app.py from AsgharHussain/Researcher_Assistant_Bot: direct link, hf CLI and curl.
- Browser
- Download file 4.87 kB
-
https://huggingface.co/spaces/AsgharHussain/Researcher_Assistant_Bot/resolve/main/app.py
- Command line
-
hf download hf://spaces/AsgharHussain/Researcher_Assistant_Bot/app.py
-
curl -L -o app.py https://huggingface.co/spaces/AsgharHussain/Researcher_Assistant_Bot/resolve/main/app.py
4.87 kB
| # app.py | |
| import gradio as gr | |
| from huggingface_hub import InferenceClient | |
| import PyPDF2 | |
| import io | |
| # How many characters to include per uploaded PDF (avoid huge inputs / token blowup) | |
| MAX_CHARS_PER_PDF = 20000 | |
| def extract_text_from_pdf_fileobj(file_obj) -> str: | |
| """ | |
| Extract text from a file-like object containing a PDF (Gradio gives a tempfile path | |
| accessible via file.name, but file might also be an in-memory BytesIO). | |
| Returns the first MAX_CHARS_PER_PDF characters. | |
| """ | |
| try: | |
| # file_obj may be a _TemporaryFileWrapper with .name path or a dict/file-like; handle both | |
| if hasattr(file_obj, "name"): | |
| reader = PyPDF2.PdfReader(file_obj.name) | |
| else: | |
| # try to read bytes | |
| file_obj.seek(0) | |
| reader = PyPDF2.PdfReader(io.BytesIO(file_obj.read())) | |
| except Exception as e: | |
| # best-effort: return empty string if cannot parse | |
| return f"[Could not read PDF: {e}]" | |
| text_parts = [] | |
| try: | |
| for page in reader.pages: | |
| try: | |
| text_parts.append(page.extract_text() or "") | |
| except Exception: | |
| # ignore pages that fail | |
| continue | |
| except Exception: | |
| # some PDFs may throw during iteration — fallback | |
| pass | |
| combined = "\n".join(text_parts) | |
| if len(combined) > MAX_CHARS_PER_PDF: | |
| combined = combined[:MAX_CHARS_PER_PDF] + "\n\n[TRUNCATED]" | |
| return combined | |
| def build_context_from_uploaded_files(uploaded_files): | |
| """ | |
| Given the list returned by gr.File (list of file-like objects), return a single string | |
| that summarizes / contains the extracted text to be sent as extra context. | |
| """ | |
| if not uploaded_files: | |
| return "" | |
| ctx_parts = [] | |
| for f in uploaded_files: | |
| try: | |
| # `f` is a TemporaryFile object from gradio with .name attribute | |
| extracted = extract_text_from_pdf_fileobj(f) | |
| header = f"--- Begin extracted text from uploaded file: {getattr(f, 'name', 'uploaded_pdf')} ---\n" | |
| footer = f"\n--- End of {getattr(f, 'name', 'uploaded_pdf')} ---\n\n" | |
| ctx_parts.append(header + extracted + footer) | |
| except Exception as e: | |
| ctx_parts.append(f"[Error extracting {getattr(f, 'name', 'uploaded_pdf')}: {e}]") | |
| return "\n".join(ctx_parts) | |
| def respond( | |
| message, | |
| history: list[dict[str, str]], | |
| uploaded_files, # NEW: list of uploaded PDF files | |
| system_message, | |
| max_tokens, | |
| temperature, | |
| top_p, | |
| hf_token: gr.OAuthToken, | |
| ): | |
| """ | |
| For more information on `huggingface_hub` Inference API support, please check the docs: | |
| https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference | |
| """ | |
| # Build an extra 'context' system message from uploaded PDFs (if any) | |
| uploaded_context = build_context_from_uploaded_files(uploaded_files) | |
| client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b") | |
| # system message first | |
| messages = [{"role": "system", "content": system_message}] | |
| # if we have uploaded PDF content, add it as another system message so the model sees it | |
| if uploaded_context: | |
| messages.append({"role": "system", "content": "Context from uploaded PDFs:\n\n" + uploaded_context}) | |
| # replay conversation history (assumed to be a list of role/content dicts) | |
| messages.extend(history) | |
| # finally the user message | |
| messages.append({"role": "user", "content": message}) | |
| response = "" | |
| for message in client.chat_completion( | |
| messages, | |
| max_tokens=max_tokens, | |
| stream=True, | |
| temperature=temperature, | |
| top_p=top_p, | |
| ): | |
| choices = message.choices | |
| token = "" | |
| if len(choices) and choices[0].delta.content: | |
| token = choices[0].delta.content | |
| response += token | |
| yield response | |
| """ | |
| For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface | |
| """ | |
| chatbot = gr.ChatInterface( | |
| respond, | |
| type="messages", | |
| additional_inputs=[ | |
| # NEW: add a File uploader as the first additional input | |
| gr.File(label="Upload PDFs (optional)", file_count="multiple", file_types=[".pdf"]), | |
| gr.Textbox(value="You are a friendly Chatbot.", label="System message"), | |
| gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"), | |
| gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"), | |
| gr.Slider( | |
| minimum=0.1, | |
| maximum=1.0, | |
| value=0.95, | |
| step=0.05, | |
| label="Top-p (nucleus sampling)", | |
| ), | |
| ], | |
| ) | |
| with gr.Blocks() as demo: | |
| with gr.Sidebar(): | |
| gr.LoginButton() | |
| chatbot.render() | |
| if __name__ == "__main__": | |
| demo.launch() | |