app_mg_script_gen.py

import streamlit as st
import os
import hashlib

# configuration must be at the top.
st.set_page_config(
    page_title="Motion Graphic Script Generator",
    page_icon="app_images/construct.webp",
    layout="centered",
    initial_sidebar_state="expanded"
)

### hash code function for the encryption
def hash_code(input_code):
    """Hashes the access code using SHA-256."""
    return hashlib.sha256(input_code.encode()).hexdigest()

### retrieve hash code 
ACCESS_CODE_HASH = os.getenv("ACCESS_CODE_HASH")

if not ACCESS_CODE_HASH:
    st.error("⚠️ Hashed access code not found. Please set ACCESS_CODE_HASH.")
    st.stop()

### Authentication Logic
if "authenticated" not in st.session_state:
    st.session_state.authenticated = False

if not st.session_state.authenticated:
    st.title("🔒 Access Restricted")
    access_code_input = st.text_input("Enter Access Code:", type="password")

    if st.button("Submit"):
        if hash_code(access_code_input) == ACCESS_CODE_HASH:
            st.session_state.authenticated = True
            st.rerun() 
        else:
            st.error("Incorrect access code. Please try again.")

    st.stop()  # Prevent unauthorized access



APP_URL = "https://motion-graphic-script-gen.streamlit.app/"
APP_IMAGE = "construct.webp"
PUBLISHED = True

APP_TITLE = "Motion Graphic Script Generator"
APP_INTRO = """Use this application to generate motion graphic scripts."""

SYSTEM_PROMPT = """Develop and refine academic video scripts (750 - 800 words) for motion graphic videos that blend information and visuals to improve comprehension and retention.

Visuals: Incorporate advanced visual elements like 2.5D animations, high-quality renderings, and immersive simulations to transform static information into dynamic content. 

Output format:
Format scripts as a table, with columns for approximate time, text, and visual cues. Because we should have a change in visuals every 15 seconds, I create the script in 15 - 20-second increments, with each increment being a row in the table. You work on the assumption that each minute of video will have 120 words of text.

Scripts will include the following sections:
 1) Begin with an engaging hook.
 2) Add a paragraph that establishes the relevance of the learning material to real-world practices.
 3) Present the key theory or ideas of the content.
 4) Provide an explanation of how that links to solving the problems presented at the start.
 5) A conclusion that pulls the video together, tying key points explored in the script.
"""

RAG_IMPLEMENTATION = True  # Enable RAG integration
SOURCE_DOCUMENT = "rag_docs/ABETSIS_C1_M0_V1.pdf"  # Path to your PDF document

# Required Libraries
# Required Libraries
import os
import fitz  # PyMuPDF for PDF processing
import openai

# PDF Text Extraction Function
def extract_text_from_pdf(pdf_path):
    """
    Extract text from a PDF file using PyMuPDF (fitz).
    """
    text = ""
    with fitz.open(pdf_path) as pdf:
        for page in pdf:
            text += page.get_text("text")  # Extract plain text from each page
    return text

# Prompt Builder Function
def build_user_prompt(user_input):
    """
    Dynamically build the user prompt with user-provided inputs and document content.
    """
    try:
        # Retrieve user inputs
        learning_objectives = user_input.get("learning_objectives", "").strip()
        learning_content = user_input.get("learning_content", "").strip()
        academic_stage = user_input.get("academic_stage_radio", "").strip()

        # Debugging: Print retrieved values
        print("Learning Objectives:", learning_objectives)
        print("Learning Content:", learning_content)
        print("Academic Stage:", academic_stage)

        # Validate required inputs
        if not learning_objectives:
            raise ValueError("The 'Learning Objectives' field is required.")
        if not learning_content:
            raise ValueError("The 'Learning Content' field is required.")
        if not academic_stage:
            raise ValueError("An 'Academic Stage' must be selected.")

        # Load document content for RAG
        document_text = ""
        if RAG_IMPLEMENTATION and os.path.exists(SOURCE_DOCUMENT):
            document_text = extract_text_from_pdf(SOURCE_DOCUMENT)
            document_text = document_text[:2000]  # Truncate text to fit within token limits

        # Construct the user prompt
        user_prompt = f"""
        {SYSTEM_PROMPT}

        Example Template/Training Data:
        {document_text}

        The motion graphic script should be aligned with the provided objectives: {learning_objectives}.
        Base the motion graphic script on the following learning content: {learning_content}.
        Please align the motion graphic script to the following academic stage level: {academic_stage}.
        """
        return user_prompt

    except Exception as e:
        raise ValueError(f"Error building prompt: {str(e)}")

# Configuration for the App
PHASES = {
    "generate_discussion": {
        "name": "Motion Graphic Script Generator",
        "fields": {
            "learning_objectives": {
                "type": "text_area",
                "label": "Enter the relevant module-level learning objective(s):",
                "height": 500
            },
            "learning_content": {
                "type": "text_area",
                "label": "Enter relevant learning content that will serve as the basis for the motion graphic script.",
                "height": 500
            },
            "academic_stage_radio": {
                "type": "radio",
                "label": "Select the academic stage of the students:",
                "options": [
                    "Lower Primary",
                    "Middle Primary",
                    "Upper Primary",
                    "Lower Secondary",
                    "Upper Secondary",
                    "Undergraduate",
                    "Postgraduate"
                ]
            }
        },
        "phase_instructions": """
        Provide the relevant details (learning objectives, content, and academic stage) to generate a discussion prompt.
        """,
        "user_prompt": [
            {
                "condition": {},
                "prompt": "The motion graphic should be aligned with the provided objectives: {learning_objectives}."
            },
            {
                "condition": {},
                "prompt": "Base the motion graphic script on the following content. {learning_content}"
            },
            {
                "condition": {},
                "prompt": "Please align the motion graphic script to the following academic stage level: {academic_stage_radio}."
            }
        ],
        "ai_response": True,
        "allow_revisions": True,
        "show_prompt": True,
        "read_only_prompt": False
    }
}

PREFERRED_LLM = "gpt-4o"
LLM_CONFIG_OVERRIDE = {"gpt-4o": {
    "family": "openai",
    "model": "gpt-4o",
    "temperature": 0.5,
    "top_p": 0.9,
    "frequency_penalty": 0.5,
    "presence_penalty": 0.3
}}

SIDEBAR_HIDDEN = True

### Logout Button in Sidebar
st.sidebar.button("Logout", on_click=lambda: st.session_state.update({"authenticated": False}))

# Entry Point
from core_logic.main import main
if __name__ == "__main__":
    main(config=globals())