Update app.py

app.py

@@ -1,69 +1,91 @@
-from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
+from smolagents import CodeAgent, HfApiModel, load_tool, tool
 import datetime
-import requests
 import pytz
 import yaml
+from transformers import pipeline
 from tools.final_answer import FinalAnswerTool
 from Gradio_UI import GradioUI
 
-# Herramienta para clasificar alertas P1, P2, P3 basado en criterios de la tabla
+# Initialize ASR model from Hugging Face
+asr_pipeline = pipeline("automatic-speech-recognition", model="facebook/wav2vec2-large-960h")
+
+# Tool: Convert time between time zones
 @tool
-def classify_alert(service_criticity: str, disruption: str, affectation_time: str, magnitude: str, business_workaround: str) -> str:
-    """Clasifica una alerta en P1, P2 o P3 según los criterios especificados.
+def convert_time(time_str: str, from_tz: str, to_tz: str) -> str:
+    """Convert time from one timezone to another.
     Args:
-        service_criticity: 'High' o 'Low'
-        disruption: 'Full', 'Degraded', 'None'
-        affectation_time: 'Up 15 mins' o 'Less 15 mins'
-        magnitude: 'High', 'Low', 'None'
-        business_workaround: 'Yes' o 'No'
+        time_str: Time in 'YYYY-MM-DD HH:MM:SS' format.
+        from_tz: Original timezone.
+        to_tz: Target timezone.
     """
-    if service_criticity == "High" and disruption in ["Full", "Degraded"] and magnitude == "High" and business_workaround == "No":
-        return "P1 - Crítico"
-    elif service_criticity == "High" and (disruption in ["Full", "Degraded"] or affectation_time == "Up 15 mins"):
-        return "P2 - Importante"
-    else:
-        return "P3 - Menor"
+    try:
+        from_zone = pytz.timezone(from_tz)
+        to_zone = pytz.timezone(to_tz)
+        local_time = datetime.datetime.strptime(time_str, "%Y-%m-%d %H:%M:%S")
+        local_time = from_zone.localize(local_time)
+        converted_time = local_time.astimezone(to_zone).strftime("%Y-%m-%d %H:%M:%S")
+        return f"Time in {to_tz}: {converted_time}"
+    except Exception as e:
+        return f"Error converting time: {str(e)}"
+
 
-# Herramienta para generar resúmenes de reportes de infraestructura
+# Tool: Assign alerts to departments based on keywords
 @tool
-def summarize_report(report_text: str) -> str:
-    """Genera un resumen de un reporte sobre la infraestructura de la plataforma.
+def assign_alert(alert_message: str) -> str:
+    """Assign an alert to a department based on keywords.
     Args:
-        report_text: Texto del reporte.
+        alert_message: The alert description.
     """
-    api_url = "https://api-inference.huggingface.co/models/facebook/bart-large-cnn"
-    headers = {"Authorization": "Bearer YOUR_HF_API_KEY"}
-    payload = {"inputs": report_text}
-    
-    response = requests.post(api_url, headers=headers, json=payload)
-    if response.status_code == 200:
-        return response.json()[0]['summary_text']
-    else:
-        return "Error al generar el resumen."
+    alert_keywords = {
+        "network": "Network Operations",
+        "server": "Infrastructure",
+        "database": "Database Team",
+        "security": "Security Operations",
+        "application": "Development Team"
+    }
+    for keyword, department in alert_keywords.items():
+        if keyword.lower() in alert_message.lower():
+            return f"Alert assigned to: {department}"
+    return "Alert could not be assigned automatically. Please check manually."
+
+
+# Tool: Transcribe audio and generate a timeline using Hugging Face ASR
+@tool
+def transcribe_audio(audio_path: str) -> str:
+    """Transcribe audio and log events with timestamps.
+    Args:
+        audio_path: Path to the audio file.
+    """
+    try:
+        result = asr_pipeline(audio_path)
+        transcript = result["text"]
+        timestamped_events = [f"{datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')} - {line}" for line in transcript.split('.')]
+        return "\n".join(timestamped_events)
+    except Exception as e:
+        return f"Error transcribing audio: {str(e)}"
 
-# Definir el modelo
-model = HfApiModel(
-    model_name="meta-llama/Llama-2-13b-chat-hf", 
-    api_key="YOUR_HF_API_KEY"
-)
 
 final_answer = FinalAnswerTool()
 
-image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)
+model = HfApiModel(
+    max_tokens=2096,
+    temperature=0.5,
+    model_id='Qwen/Qwen2.5-Coder-32B-Instruct',
+    custom_role_conversions=None,
+)
 
-# Cargar plantillas de prompts
 with open("prompts.yaml", 'r') as stream:
     prompt_templates = yaml.safe_load(stream)
-    
+
 agent = CodeAgent(
     model=model,
-    tools=[final_answer, classify_alert, summarize_report, image_generation_tool],
+    tools=[final_answer, convert_time, assign_alert, transcribe_audio],
     max_steps=6,
     verbosity_level=1,
     grammar=None,
     planning_interval=None,
-    name=None,
-    description=None,
+    name="Monitoring Assistant",
+    description="An agent that helps with alert assignment, time conversion, monitoring tasks, and real-time transcription.",
     prompt_templates=prompt_templates
 )