Update app.py

app.py

@@ -1,27 +1,185 @@
-from flask import Flask, request, jsonify
-from transformers import pipeline
+import requests
+import json
+import os
+import logging
+from datetime import datetime
+from dotenv import load_dotenv
+from simple_salesforce import Salesforce
+from flask import Flask, jsonify, request, render_template, redirect, url_for
 
-# Initialize the Hugging Face model pipeline
-model_name = "distilbert-base-uncased"  # Change to the model you need
-nlp = pipeline("text-classification", model=model_name)
+# Configure logging
+logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
+logger = logging.getLogger(__name__)
 
-# Create Flask app
+# Load environment variables
+load_dotenv()
+
+# Hugging Face API configuration
+HUGGING_FACE_API_URL = os.getenv("HUGGING_FACE_API_URL", "https://api-inference.huggingface.co/models/distilgpt2")
+HUGGING_FACE_API_TOKEN = os.getenv("HUGGING_FACE_API_TOKEN")
+
+# Salesforce configuration
+SALESFORCE_USERNAME = os.getenv("SALESFORCE_USERNAME")
+SALESFORCE_PASSWORD = os.getenv("SALESFORCE_PASSWORD")
+SALESFORCE_SECURITY_TOKEN = os.getenv("SALESFORCE_SECURITY_TOKEN")
+SALESFORCE_DOMAIN = os.getenv("SALESFORCE_DOMAIN", "login")
+
+# Validate environment variables
+if not HUGGING_FACE_API_TOKEN:
+    logger.error("HUGGING_FACE_API_TOKEN is not set")
+    raise ValueError("HUGGING_FACE_API_TOKEN environment variable is not set")
+if not HUGGING_FACE_API_URL.startswith("https://api-inference.huggingface.co/models/"):
+    logger.error("Invalid HUGGING_FACE_API_URL: %s", HUGGING_FACE_API_URL)
+    raise ValueError("HUGGING_FACE_API_URL must point to a valid Hugging Face model")
+if not all([SALESFORCE_USERNAME, SALESFORCE_PASSWORD, SALESFORCE_SECURITY_TOKEN]):
+    logger.error("Salesforce credentials are incomplete")
+    raise ValueError("Salesforce credentials must be set")
+
+# Initialize Flask app
 app = Flask(__name__)
 
-@app.route("/predict", methods=["POST"])
-def predict():
-    # Get the data from the request
-    data = request.get_json()
-    text = data.get("text", "")
+def generate_coaching_output(data):
+    """
+    Generate daily checklist and tips using Hugging Face LLM.
+    """
+    logger.info("Generating coaching output for supervisor %s", data['supervisor_id'])
+    milestones_json = json.dumps(data['milestones'], indent=2)
+    prompt = f"""
+You are an AI Coach for construction site supervisors. Based on the following data, generate a daily checklist, three focus tips, and a motivational quote. Ensure outputs are concise, actionable, and tailored to the supervisor's role, project status, and reflection log.
+
+Supervisor Role: {data['role']}
+Project Milestones: {milestones_json}
+Reflection Log: {data['reflection_log']}
+Weather: {data['weather']}
+
+Format the response as JSON:
+{{
+    "checklist": ["item1", "item2", ...],
+    "tips": ["tip1", "tip2", "tip3"],
+    "quote": "motivational quote"
+}}
+"""
+
+    headers = {
+        "Authorization": f"Bearer {HUGGING_FACE_API_TOKEN}",
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "inputs": prompt,
+        "parameters": {
+            "max_length": 200,
+            "temperature": 0.7,
+            "top_p": 0.9
+        }
+    }
+
+    try:
+        response = requests.post(HUGGING_FACE_API_URL, headers=headers, json=payload, timeout=5)
+        response.raise_for_status()
+        result = response.json()
+        generated_text = result[0]["generated_text"] if isinstance(result, list) else result["generated_text"]
+
+        start_idx = generated_text.find('{')
+        end_idx = generated_text.rfind('}') + 1
+        if start_idx == -1 or end_idx == 0:
+            logger.error("No valid JSON found in LLM output")
+            raise ValueError("No valid JSON found in LLM output")
+        
+        json_str = generated_text[start_idx:end_idx]
+        output = json.loads(json_str)
+        logger.info("Successfully generated coaching output")
+        return output
+
+    except requests.exceptions.HTTPError as e:
+        logger.error("Hugging Face API HTTP error: %s", e)
+        return None
+    except (json.JSONDecodeError, ValueError) as e:
+        logger.error("Error parsing LLM output: %s", e)
+        return None
+    except Exception as e:
+        logger.error("Unexpected error in Hugging Face API call: %s", e)
+        return None
+
+def save_to_salesforce(output, supervisor_id, project_id):
+    """
+    Save coaching output to Salesforce Supervisor_AI_Coaching__c object.
+    """
+    if not output:
+        logger.error("No coaching output to save")
+        return False
+
+    try:
+        sf = Salesforce(
+            username=SALESFORCE_USERNAME,
+            password=SALESFORCE_PASSWORD,
+            security_token=SALESFORCE_SECURITY_TOKEN,
+            domain=SALESFORCE_DOMAIN
+        )
+        logger.info("Connected to Salesforce")
+
+        coaching_record = {
+            "Supervisor_ID__c": supervisor_id,
+            "Project_ID__c": project_id,
+            "Daily_Checklist__c": "\n".join(output["checklist"]),
+            "Suggested_Tips__c": "\n".join(output["tips"]),
+            "Quote__c": output["quote"],
+            "Generated_Date__c": datetime.now().strftime("%Y-%m-%d")
+        }
+
+        sf.Supervisor_AI_Coaching__c.upsert(
+            f"Supervisor_ID__c/{supervisor_id}_{datetime.now().strftime('%Y-%m-%d')}",
+            coaching_record
+        )
+        logger.info("Successfully saved coaching record to Salesforce for supervisor %s", supervisor_id)
+        return True
+
+    except Exception as e:
+        logger.error("Salesforce error: %s", e)
+        return False
+
+@app.route('/', methods=['GET'])
+def redirect_to_ui():
+    """
+    Redirect root URL to the UI.
+    """
+    return redirect(url_for('ui'))
+
+@app.route('/ui', methods=['GET'])
+def ui():
+    """
+    Serve the HTML user interface.
+    """
+    return render_template('index.html')
+
+@app.route('/generate', methods=['POST'])
+def generate_endpoint():
+    """
+    Endpoint to generate coaching output based on supervisor data.
+    """
+    try:
+        data = request.get_json()
+        if not data or not all(key in data for key in ['supervisor_id', 'role', 'project_id', 'milestones', 'reflection_log', 'weather']):
+            return jsonify({"status": "error", "message": "Invalid or missing supervisor data"}), 400
 
-    if not text:
-        return jsonify({"error": "Text input is required"}), 400
-    
-    # Use Hugging Face model for inference
-    result = nlp(text)
+        coaching_output = generate_coaching_output(data)
+        if coaching_output:
+            success = save_to_salesforce(coaching_output, data["supervisor_id"], data["project_id"])
+            if success:
+                return jsonify({"status": "success", "output": coaching_output}), 200
+            else:
+                return jsonify({"status": "error", "message": "Failed to save to Salesforce"}), 500
+        else:
+            return jsonify({"status": "error", "message": "Failed to generate coaching output"}), 500
+    except Exception as e:
+        logger.error("Error in generate endpoint: %s", e)
+        return jsonify({"status": "error", "message": str(e)}), 500
 
-    return jsonify({"prediction": result})
+@app.route('/health', methods=['GET'])
+def health_check():
+    """
+    Health check endpoint.
+    """
+    return jsonify({"status": "healthy", "message": "Application is running"}), 200
 
 if __name__ == "__main__":
-    # Run the app
-    app.run(host="0.0.0.0", port=5000)
+    app.run(host="0.0.0.0", port=int(os.getenv("PORT", 7860)))