app.py

import gradio as gr
import os
import base64
from datetime import datetime
import requests
import trello
from dotenv import load_dotenv
import urllib3
import wave
import audioop
import io
import speech_recognition as sr

# Disable SSL warning
urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)

# Load environment variables from .env file
load_dotenv()

SAMBANOVA_API_KEY = "34115dcb-baab-4390-ab5c-e501666f9f4e"
SAMBANOVA_URL = "https://api.sambanova.ai/v1/chat/completions"

# Initialize Trello client
trello_client = trello.TrelloClient(
    api_key=os.getenv('TRELLO_API_KEY'),
    token=os.getenv('TRELLO_TOKEN')
)

def get_trello_members():
    """Get all members from Trello workspace"""
    try:
        boards = trello_client.list_boards()
        if not boards:
            raise Exception("No Trello boards found")
        
        board = boards[0]
        members = board.get_members()
        return {(member.full_name or member.username): member.id for member in members}
    except Exception as e:
        print(f"Error fetching Trello members: {str(e)}")
        return {}

def process_audio_data(audio_path):
    """Process WAV audio file"""
    try:
        with wave.open(audio_path, 'rb') as wav_file:
            # Get audio parameters
            n_channels = wav_file.getnchannels()
            sampwidth = wav_file.getsampwidth()
            framerate = wav_file.getframerate()
            n_frames = wav_file.getnframes()
            
            # Read audio data
            audio_data = wav_file.readframes(n_frames)
            
            # Convert to mono if stereo
            if n_channels == 2:
                audio_data = audioop.tomono(audio_data, sampwidth, 1, 1)
            
            # Convert to 16-bit if needed
            if sampwidth != 2:
                audio_data = audioop.lin2lin(audio_data, sampwidth, 2)
            
            # Resample to 16kHz if needed
            if framerate != 16000:
                audio_data, _ = audioop.ratecv(audio_data, 2, 1, framerate, 16000, None)
                framerate = 16000
            
            return audio_data, framerate
            
    except Exception as e:
        print(f"Error processing audio: {str(e)}")
        raise

def transcribe_audio(audio_file):
    """Convert audio to text using Speech Recognition"""
    try:
        # Initialize recognizer
        recognizer = sr.Recognizer()
        
        # Handle different audio input formats
        if isinstance(audio_file, tuple):
            audio_path = audio_file[0]
        else:
            audio_path = audio_file
            
        print(f"Processing audio file: {audio_path}")
        
        try:
            # Load the audio file
            with sr.AudioFile(audio_path) as source:
                # Adjust for ambient noise
                recognizer.adjust_for_ambient_noise(source)
                # Read the audio data
                audio_data = recognizer.record(source)
                
                # Perform the transcription with increased timeout
                text = recognizer.recognize_google(
                    audio_data,
                    language='en-US',
                    show_all=False,
                    with_confidence=False
                )
                
                if not text:
                    raise Exception("No transcription results returned")
                    
                return text.strip()
                
        except sr.UnknownValueError:
            raise Exception("Speech could not be understood. Please try speaking more clearly.")
        except sr.RequestError as e:
            raise Exception(f"Could not request results from Google Speech Recognition service; {e}")
            
    except Exception as e:
        print(f"Transcription error details: {str(e)}")
        raise Exception(f"Transcription error: {str(e)}")

def improve_task_description(text):
    """Improve and summarize task description using SambaNova API"""
    try:
        prompt = f"""Please improve and structure this task description for better clarity and actionability:
        
Original task: {text}

Please provide:
1. A clear, concise task title
2. Key objectives
3. Suggested deadline (if not specified)
4. Any important details or requirements
"""

        headers = {
            'Authorization': f'Bearer {SAMBANOVA_API_KEY}',
            'Content-Type': 'application/json'
        }
        
        data = {
            'messages': [
                {'role': 'user', 'content': prompt}
            ],
            'model': 'Meta-Llama-3.1-8B-Instruct',
            'max_tokens': 2000,
            'temperature': 0.7
        }
        
        response = requests.post(
            SAMBANOVA_URL,
            headers=headers,
            json=data,
            verify=False,
            timeout=620
        )
        
        if response.status_code != 200:
            raise Exception(f"SambaNova API request failed: {response.text}")
            
        improved_text = response.json()['choices'][0]['message']['content']
        return improved_text
    except Exception as e:
        raise Exception(f"Error improving task description: {str(e)}")

def create_trello_card(task_description, selected_members):
    """Create a Trello card with the improved task description"""
    try:
        boards = trello_client.list_boards()
        if not boards:
            raise Exception("No Trello boards found")
        
        board = boards[0]
        print(f"Using board: {board.name}")  # Debug print
        
        lists = board.list_lists()
        if not lists:
            raise Exception("No lists found in the board")
        
        todo_list = lists[0]
        print(f"Using list: {todo_list.name}")  # Debug print
        
        # Create card with improved description
        card = todo_list.add_card(
            name=task_description.split('\n')[0],  # First line as title
            desc=task_description
        )
        
        # Debug prints
        print(f"Selected members: {selected_members}")
        print(f"Member types: {[type(m) for m in selected_members]}")
        
        # Assign members to card
        if selected_members:
            for member_id in selected_members:
                try:
                    card.add_member(member_id)
                except Exception as e:
                    print(f"Error adding member {member_id}: {str(e)}")
        
        return card.url
    except Exception as e:
        print(f"Trello card creation error details: {str(e)}")
        raise Exception(f"Error creating Trello card: {str(e)}")

def process_input(input_text, selected_members):
    """Process input text and create Trello card"""
    try:
        # Improve the task description
        improved_description = improve_task_description(input_text)
        
        # Create Trello card
        card_url = create_trello_card(improved_description, selected_members)
        
        # Get member names for display
        members_dict = get_trello_members()
        member_names = [name for name, mid in members_dict.items() 
                       if mid in selected_members]
        
        return f"""
Original Input:
--------------
{input_text}

Improved Task Description:
------------------------
{improved_description}

Task Created in Trello:
----------------------
Assigned to: {', '.join(member_names) if member_names else 'Not assigned'}
Card URL: {card_url}
"""
    except Exception as e:
        return f"Error processing input: {str(e)}"

def process_audio(audio_file, selected_members):
    """Process audio input and create Trello card"""
    try:
        if audio_file is None:
            return "Error: No audio file or text provided"
        print(f"Audio file type: {type(audio_file)}")  # Debug print
        print(f"Audio file content: {audio_file}")     # Debug print
        text = transcribe_audio(audio_file)
        
        return process_input(text, selected_members)
    except Exception as e:
        print(f"Audio processing error details: {str(e)}")  # Debug print
        return f"Error processing audio: {str(e)}"

def process_audio_with_members(audio, selected_members):
    """Process audio with selected members"""
    try:
        if audio is None:
            return "Error: Please provide an audio input (record or upload)"
            
        print(f"Received audio input: {type(audio)}")
        print(f"Audio content: {audio}")
        
        # Convert selected member names to member IDs
        members_dict = get_trello_members()
        selected_member_ids = []
        for name in (selected_members or []):
            if name in members_dict:
                selected_member_ids.append(members_dict[name])
            else:
                print(f"Warning: Member {name} not found in members dictionary")
        
        try:
            result = process_audio(audio, selected_member_ids)
            return result
        except Exception as e:
            error_msg = str(e)
            if "Speech could not be understood" in error_msg:
                return "Could not understand the speech. Please try again with clearer audio."
            elif "Could not request results" in error_msg:
                return "Network error. Please check your internet connection and try again."
            else:
                return f"Error processing audio: {error_msg}"
                
    except Exception as e:
        print(f"Error in process_audio_with_members: {str(e)}")
        return f"Error processing audio with members: {str(e)}"

def process_text_with_members(text, selected_members):
    """Process text with selected members"""
    try:
        # Convert selected member names to member IDs
        members_dict = get_trello_members()
        # Debug prints
        print(f"Members dict: {members_dict}")
        print(f"Selected members: {selected_members}")
        
        selected_member_ids = []
        for name in (selected_members or []):
            if name in members_dict:
                selected_member_ids.append(members_dict[name])
            else:
                print(f"Warning: Member {name} not found in members dictionary")
        
        return process_input(text, selected_member_ids)
    except Exception as e:
        print(f"Error in process_text_with_members: {str(e)}")
        return f"Error processing text with members: {str(e)}"

# Create Gradio interface
with gr.Blocks(title="TaskWhisper - Smart Task Manager") as interface:
    gr.Markdown("# 🎙️ TaskWhisper - Smart Task Manager")
    gr.Markdown("Record audio or type your task. The AI will help improve and structure your task description.")
    
    # Get Trello members for the dropdown
    members = get_trello_members()
    
    with gr.Tab("Audio Input"):
        audio_input = gr.Audio(
            label="Record or Upload Audio",
            sources=["microphone", "upload"],
            type="filepath",
            format="wav",
            interactive=True
        )
        gr.Markdown("""
        *Instructions:*
        - Use microphone to record directly
        - Or upload an audio file (WAV format)
        - Speak clearly for better results
        - Keep background noise minimal
        """)
        member_dropdown_audio = gr.Dropdown(
            choices=list(members.keys()),
            multiselect=True,
            label="Assign to Members",
            info="Select one or more members to assign the task",
            value=[]
        )
        audio_button = gr.Button("Process Audio")
    
    with gr.Tab("Text Input"):
        text_input = gr.Textbox(
            lines=3,
            placeholder="Type your task here (e.g., 'Need to prepare quarterly report with sales data by next Friday')",
            label="Text Input"
        )
        member_dropdown_text = gr.Dropdown(
            choices=list(members.keys()),
            multiselect=True,
            label="Assign to Members",
            info="Select one or more members to assign the task",
            value=[]  # Initialize with empty selection
        )
        text_button = gr.Button("Process Text")
    
    output = gr.Textbox(
        label="Task Details",
        lines=15
    )
    
    # Set up event handlers
    audio_button.click(
        fn=process_audio_with_members,
        inputs=[audio_input, member_dropdown_audio],
        outputs=output
    )
    
    text_button.click(
        fn=process_text_with_members,
        inputs=[text_input, member_dropdown_text],
        outputs=output
    )

if __name__ == "__main__":
    interface.launch(share=True)