File size: 5,518 Bytes
fcf024d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
### 1. **Imports and Dependencies:**
- **Gradio:** Gradio is used to create the interactive web interface where users can interact with the chatbot.
- **OpenAI:** The `openai` library is used to interact with OpenAI's language models (like GPT) to generate responses based on the provided input.
- **Base64 and PIL:** `base64` is used to convert images into a text-based format for processing, and `PIL` (Python Imaging Library) is used for handling and manipulating image files.
- **PyMuPDF (`fitz`):** This library is used for handling PDFs. It allows us to extract text from the uploaded PDF file.
### 2. **Functions:**
- **`extract_text_from_pdf(pdf_file):`**
- This function takes a PDF file and extracts all the text from it.
- It opens the PDF, processes each page, and collects the text.
- If there's an error (e.g., the PDF is corrupted), it catches the exception and returns an error message.
- **`generate_mcq_quiz(pdf_content, num_questions, openai_api_key, model_choice):`**
- This function generates a multiple-choice quiz from the extracted text of a PDF.
- It sends a request to OpenAI's model (via the `openai` library) to generate quiz questions based on the content of the PDF.
- The function constructs a prompt with the content, the number of questions, and asks OpenAI to generate questions in a multiple-choice format.
- It limits the text sent to OpenAI to 8,000 characters to avoid sending too much data.
- The response from OpenAI includes questions, possible answers, and explanations.
- **`generate_image_response(input_text, image, openai_api_key, model_choice):`**
- This function handles image input.
- When the user uploads an image, the image is converted to base64 (a text format) and sent to OpenAI with any accompanying text (from the user).
- OpenAI generates a response based on both the image and the text provided by the user. It could analyze the image and give a description or answer questions about it.
- **`chatbot(input_text, image, pdf_file, openai_api_key, model_choice, pdf_content, num_quiz_questions, pdf_quiz_mode, history):`**
- This is the main chatbot function. It checks which type of input (text, image, or PDF) is provided and performs the appropriate action.
- If the input type is **PDF**, it will either extract the content and generate a quiz or display a message prompting the user to upload a PDF.
- If the input type is **Image**, it will call the `generate_image_response` function to analyze the image and respond.
- It also handles the conversation history (so that the chat doesn't lose context).
- **`clear_history():`**
- This function clears the conversation history, resetting the chat.
- **`update_input_type(choice):`**
- Based on the user's choice (Image or PDF Quiz), this function hides or shows certain UI elements.
- For example, if the user chooses "PDF Quiz", it will show the PDF upload field and quiz settings but hide the image upload field.
### 3. **Gradio Interface:**
- **Custom CSS (`custom_css`):**
- This is custom styling that makes the chatbot look nice. It sets a background color, applies gradients to buttons, and customizes the chatbox and headers.
- **Interface Setup (`create_interface`):**
- The `gr.Blocks` function is used to create the layout of the interface.
- Inside the `gr.Blocks()`, different elements like textboxes, images, buttons, and chat history are defined:
- **Textbox for OpenAI API key:** The user needs to provide their API key to use OpenAI's models.
- **Radio buttons for Input Type:** The user selects between "Image" or "PDF(QUIZ)".
- **Textbox for user questions:** For image-based input, the user can type questions.
- **Image input field:** Users can upload an image for analysis.
- **PDF input field:** If the user selects "PDF(QUIZ)", they can upload a PDF to generate quiz questions.
- **Quiz settings:** Includes a slider to choose the number of questions and a checkbox to confirm that the user wants a quiz.
- **Buttons:**
- **Submit button:** Sends the user's input to the chatbot function (either an image, PDF, or question).
- **Clear button:** Clears the chat history and resets the input fields.
- **Chat history (`gr.Chatbot()`):**
- This displays the conversation history between the user and the chatbot, so the user can see the entire interaction.
### 4. **Logic for Handling User Inputs:**
- When the user selects **"Image"** as the input type, the interface will display fields for the image and a question text box. The bot will generate a response based on the image and any question the user asks.
- When the user selects **"PDF(QUIZ)"** as the input type, the interface will show a PDF upload field and allow the user to specify how many quiz questions they want to generate. The bot will extract text from the PDF and use OpenAI to create a quiz.
### 5. **Launching the Interface:**
- The `demo.launch()` function runs the web interface, making the chatbot available to users in their browser.
### 6. **Overall Flow:**
- The user chooses an input type (Image or PDF Quiz).
- Depending on the choice, the relevant fields (text, image, PDF) appear.
- The user submits their query or uploads their file, and the chatbot generates a response.
- The conversation history is maintained for context and can be cleared by the user. |