Create app.py

app.py

@@ -0,0 +1,279 @@
+import os
+import logging
+import json
+from typing import Dict
+import gradio as gr
+import requests
+from langchain import PromptTemplate, LLMChain
+from langchain_groq import ChatGroq
+from google import genai  # For Gemini API usage
+
+# ============================ #
+#        Logging Setup         #
+# ============================ #
+logging.basicConfig(level=logging.WARNING)
+logger = logging.getLogger(__name__)
+
+# ============================ #
+#      API Key Handling        #
+# ============================ #
+def clean_api_key(key: str) -> str:
+    return ''.join(c for c in key if ord(c) < 128).strip()
+
+for key in ["GEMINI_API_KEY", "GROQ_API_KEY", "DID_API_KEY"]:
+    if not os.environ.get(key):
+        raise ValueError(f"Environment variable {key} is not set. Please set it in the Hugging Face Space secrets.")
+
+gemini_api_key = clean_api_key(os.environ["GEMINI_API_KEY"])
+groq_api_key = clean_api_key(os.environ["GROQ_API_KEY"])
+did_api_key = clean_api_key(os.environ["DID_API_KEY"])
+
+# Initialize Gemini client for calculations
+client = genai.Client(api_key=gemini_api_key)
+model_name = "gemini-2.0-flash-exp"  # Adjust based on your model
+
+# ============================ #
+#    Initialize LangChain      #
+# ============================ #
+chat_model = ChatGroq(model="Gemma2-9b-It", groq_api_key=groq_api_key)
+
+# ============================== #
+#  Chain Definitions and Helpers #
+# ============================== #
+
+# ============================== #
+#  Chain 1: Report Generation    #
+# ============================== #
+report_prompt_template = PromptTemplate(
+    input_variables=["qa_summary"],
+    template=(
+        "You are a wellness assistant. The user provided the following answers:\n\n"
+        "{qa_summary}\n\n"
+        "Based on these answers, provide a brief, actionable wellness report. "
+        "Include simple suggestions to improve their sleep, exercise, stress management, and diet. "
+        "Be concise and helpful.\n\n"
+        "Report:"
+    )
+)
+report_chain = LLMChain(llm=chat_model, prompt=report_prompt_template)
+
+def generate_short_report_for_session(responses: Dict[str, str]) -> str:
+    qa_summary = "\n".join([f"{q}: {a}" for q, a in responses.items()])
+    raw_report = report_chain.run(qa_summary=qa_summary)
+    clean_report = raw_report.replace("*", "").replace("**", "")
+    return clean_report
+
+# ============================== #
+#  Chain 2: Problem Severity Analysis #
+# ============================== #
+problem_prompt_template = PromptTemplate(
+    input_variables=["responses", "internal_report"],
+    template=(
+        "You are a wellness analyst. You have the following user responses to health-related questions:\n"
+        "{responses}\n\n"
+        "You also have an internal analysis report:\n"
+        "{internal_report}\n\n"
+        "From these inputs, determine a 'problem severity percentage' for the user in the following areas: "
+        "sleep, exercise, stress, and diet. "
+        "Return your answer in JSON format with keys: sleep_problem, exercise_problem, stress_problem, diet_problem.\n"
+        "Ensure severity percentages are numbers from 0 to 100.\n\n"
+        "JSON Output:"
+    )
+)
+problem_chain = LLMChain(llm=chat_model, prompt=problem_prompt_template)
+
+def analyze_problems_with_chain(responses: Dict[str, str], internal_report: str) -> Dict[str, float]:
+    responses_str = "\n".join([f"{q}: {a}" for q, a in responses.items()])
+    raw_text = problem_chain.run(responses=responses_str, internal_report=internal_report)
+    try:
+        start_idx = raw_text.find('{')
+        end_idx = raw_text.rfind('}') + 1
+        json_str = raw_text[start_idx:end_idx]
+        problems = json.loads(json_str)
+        for key in ["sleep_problem", "exercise_problem", "stress_problem", "diet_problem"]:
+            if key not in problems:
+                problems[key] = 0.0
+        problems = {k: float(v) for k, v in problems.items()}
+        return problems
+    except Exception as e:
+        logger.error(f"Error parsing problem percentages from LLM: {e}")
+        return {
+            "sleep_problem": 0.0,
+            "exercise_problem": 0.0,
+            "stress_problem": 0.0,
+            "diet_problem": 0.0
+        }
+
+# ============================== #
+#  Chain 3: Package Recommendation  #
+# ============================== #
+recommend_prompt_template = PromptTemplate(
+    input_variables=["problems"],
+    template=(
+        "Given the following problem severity percentages:\n"
+        "{problems}\n\n"
+        "Using these rules:\n"
+        "- If sleep_problem > 70: Recommend Sleep Improvement Package\n"
+        "- If stress_problem > 70: Recommend Stress Reduction Package\n"
+        "- If exercise_problem > 70: Recommend Exercise Enhancement Package\n"
+        "- If all problems are between 30 and 70: Recommend Balanced Wellness Package\n"
+        "- If no severe problems: Recommend General Wellness Package\n\n"
+        "What are the recommended wellness packages?"
+    )
+)
+recommend_chain = LLMChain(llm=chat_model, prompt=recommend_prompt_template)
+
+def generate_recommendations(problems: Dict[str, float]) -> str:
+    recommendations = recommend_chain.run(problems=json.dumps(problems))
+    return recommendations.strip()
+
+# ============================== #
+#  Chain 4: Final Summary Generation  #
+# ============================== #
+final_prompt_template = PromptTemplate(
+    input_variables=["report", "problems", "recommendation"],
+    template=(
+        "Based on the following information:\n"
+        "Report:\n{report}\n\n"
+        "Problem Severity Percentages:\n{problems}\n\n"
+        "Recommended Packages:\n{recommendation}\n\n"
+        "Generate a short summary suitable for video narration that synthesizes this information."
+    )
+)
+final_chain = LLMChain(llm=chat_model, prompt=final_prompt_template)
+
+def generate_final_summary(report: str, problems: Dict[str, float], recommendation: str) -> str:
+    summary = final_chain.run(
+        report=report,
+        problems=json.dumps(problems),
+        recommendation=recommendation
+    )
+    return summary.strip()
+
+# ============================== #
+#  Chain 5: Shorten Final Summary  #
+# ============================== #
+shorten_prompt_template = PromptTemplate(
+    input_variables=["final_summary"],
+    template=(
+        "Shorten the following summary to make it concise and engaging for video narration. "
+        "Ensure all key points remain intact:\n\n"
+        "{final_summary}\n\n"
+        "Shortened Summary:"
+    )
+)
+shorten_chain = LLMChain(llm=chat_model, prompt=shorten_prompt_template)
+
+def shorten_summary(final_summary: str) -> str:
+    shortened = shorten_chain.run(final_summary=final_summary)
+    return shortened.strip()
+
+# ============================== #
+#       Questions and Gradio     #
+# ============================== #
+questions = [
+    "How many hours of sleep do you get each night?",
+    "How often do you exercise in a week?",
+    "What is your current stress level on a scale from 1 to 10?",
+    "What are your primary wellness goals?",
+    "Do you follow any specific diet or have any dietary restrictions?",
+    "How would you describe your current eating habits?",
+    "How much time do you spend on relaxation or mindfulness activities daily?",
+    "Do you experience any recurring health issues or pain?",
+    "How do you manage stress on a daily basis?",
+    "What does your typical daily routine look like?"
+]
+
+def process_answers(
+    sleep: str,
+    exercise: str,
+    stress: str,
+    goals: str,
+    diet: str,
+    eating: str,
+    relaxation: str,
+    health_issues: str,
+    manage_stress: str,
+    routine: str
+):
+    # Map user inputs to questions
+    responses = {
+        questions[0]: sleep,
+        questions[1]: exercise,
+        questions[2]: stress,
+        questions[3]: goals,
+        questions[4]: diet,
+        questions[5]: eating,
+        questions[6]: relaxation,
+        questions[7]: health_issues,
+        questions[8]: manage_stress,
+        questions[9]: routine
+    }
+
+    # Execute chains sequentially using the collected responses
+    report = generate_short_report_for_session(responses)
+    problems = analyze_problems_with_chain(responses, report)
+    recommendation = generate_recommendations(problems)
+    final_summary = generate_final_summary(report, problems, recommendation)  # for user 
+    shortened_summary = shorten_summary(final_summary)  # for video 
+
+    # Prepare individual outputs for each section
+    wellness_report = f"**Wellness Report**\n------------------\n{report.strip()}"
+    
+    identified_problems = (
+        "**Identified Problems**\n"
+        "-----------------------\n"
+        f"Sleep Problem: {problems.get('sleep_problem', 'N/A')}%\n"
+        f"Exercise Problem: {problems.get('exercise_problem', 'N/A')}%\n"
+        f"Stress Problem: {problems.get('stress_problem', 'N/A')}%\n"
+        f"Diet Problem: {problems.get('diet_problem', 'N/A')}%"
+    )
+    
+    recommendations = (
+        "**Recommendations**\n"
+        "--------------------\n"
+        f"{recommendation.strip()}"
+    )
+    
+    summary_shown = (
+        "**Summary (SHOWN TO USER)**\n"
+        "-----------------\n"
+        f"{final_summary.strip()}"
+    )
+    
+    final_summary_video = (
+        "**Final Summary (FOR VIDEO CREATION)**\n"
+        "-----------------\n"
+        f"{shortened_summary.strip()}"
+    )
+
+    # Return each section as a separate output
+    return wellness_report, identified_problems, recommendations, summary_shown, final_summary_video
+
+# Create Gradio interface with separate textboxes for each question
+iface = gr.Interface(
+    fn=process_answers,
+    inputs=[
+        gr.Textbox(label=questions[0]),
+        gr.Textbox(label=questions[1]),
+        gr.Textbox(label=questions[2]),
+        gr.Textbox(label=questions[3]),
+        gr.Textbox(label=questions[4]),
+        gr.Textbox(label=questions[5]),
+        gr.Textbox(label=questions[6]),
+        gr.Textbox(label=questions[7]),
+        gr.Textbox(label=questions[8]),
+        gr.Textbox(label=questions[9])
+    ],
+    outputs=[
+        gr.Markdown(label="Wellness Report"),
+        gr.Markdown(label="Identified Problems"),
+        gr.Markdown(label="Recommendations"),
+        gr.Markdown(label="Summary (SHOWN TO USER)"),
+        gr.Markdown(label="Final Summary (FOR VIDEO CREATION)")
+    ],
+    title="Wellness Report Generator",
+    description="Answer the questions to generate a wellness report, problem analysis, recommendations, and a final summary."
+)
+
+iface.launch()