v3.py

import pyaudio
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg
import tkinter as tk
import speech_recognition as sr
import threading
import ctypes
from matplotlib.animation import FuncAnimation
import pyttsx3
from deepmultilingualpunctuation import PunctuationModel
import pyautogui
import time
import openai
import os
import tkinterwin
import friday
from gnewsclient import gnewsclient
import datetime
import wikipedia
import webbrowser
import requests

pyautogui.FAILSAFE=False

#vars for weather
api_key = "5995f5e32100e6a622ffb2f0d088cb02"
base_url = "http://api.openweathermap.org/data/2.5/weather?"

#news client
client = gnewsclient.NewsClient(language='english',location='india',max_results=3)

# Define the number of blocks horizontally and vertically
num_blocks_horizontal = 10
num_blocks_vertical = 7

# Define the size of each block in pixels
block_width = pyautogui.size().width // num_blocks_horizontal
block_height = pyautogui.size().height // num_blocks_vertical

# setting initial state
state="start"

# smart run
def perform(text):
    try:
        completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo",
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what is to be performed, Give me python script for achieving it using pyautogui, os and other required modules. The code should come as a single output, i.e don't output the code in various parts. Make the code robust enough and make sure that you maximise windows and perform such similar tasks so it is easier to track coordinates and work. Make sure the coordinates on screen are extremely accurate and that you wait enough for one part of the job to be done before starting other one. make the codes windows os friendly and do not give comments in the snippet. D not use locateonscreen as images are not available"},
            {"role": "user", "content" : text}]
            )
        print(completion['choices'][0]['message']['content'])
        output=completion['choices'][0]['message']['content']
        output=output.replace("```python","```")
        output=(output.split("```"))[1].split("```")[0]
        if output.startswith("python"):
            output=output.lstrip("python")
        with open("Codes\\workfile.py", "w") as f:
            f.write(output)
        os.system("python Codes\\workfile.py") 
        speak("Task Performed Successfully")
    except:
        speak("Sorry, I could not perform the task. Please try again.")

def webperform(text):
    try:
        completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo",
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what is to be performed, Give me selenium python script for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. Set chromedriver path as 'C:/Everything/chromedriver.exe'. Do not create code that might raise NoSuchElementException.wrap code in a try-except block to catch the NoSuchElementException exception and handle it gracefully, for example, by retrying the operation after waiting for some time or logging the error. Make sure you wait for the js to execute before continuing."},
            {"role": "user", "content" : text}]
            )
        print(completion['choices'][0]['message']['content'])
        output=completion['choices'][0]['message']['content']
        output=output.replace("```python","```")
        output=(output.split("```"))[1].split("```")[0]
        if output.startswith("python"):
            output=output.lstrip("python")
        output=output.replace("driver.quit()","while len(driver.window_handles) > 0: pass\ndriver.quit()")
        with open("Codes\\workfile.py", "w") as f:
            f.write(output)
        os.system("python Codes\\workfile.py")
        speak("Task Performed Successfully")
    except:
        speak("Sorry, I could not perform the task. Please try again.")

def gptreply(text):
    completion = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages = [{"role": "system", "content" : "You are Friday AI, created by Parth Gupta. Answer as concisely as possible."},
        {"role": "user", "content" : text}]
        )
    print(completion['choices'][0]['message']['content'])
    speak(completion['choices'][0]['message']['content']) 

# setting openaiapi
openai.api_key = os.getenv("OPENAI_KEY")

# Initializing the Punctuator Engine
# model = PunctuationModel()
model ="test"

#define engine for speech
engine = pyttsx3.init('sapi5')
voices = engine.getProperty('voices')
engine.setProperty('voice', voices[1].id)


# function for code execution in a new thread
def code_exec(prog_lang,*args):
    if prog_lang=='python':
        os.system("code Codes\\codefile.py")
        os.system("python Codes\\codefile.py")
    if prog_lang=='html':
        os.system("code Codes\\codefile.html")
        os.system("start Codes\\codefile.html")
    if prog_lang=='java':
        os.system("code Codes\\codefile.java")
        os.system("javac Codes\\codefile.java")
        os.system("java -classpath Codes\\codefile.class")
    if prog_lang=='c++':
        os.system("code Codes\\codefile.cpp")
        os.system("g++ Codes\\codefile.cpp -o codefile.exe")
        os.system("codefile.exe")
    if prog_lang=='cs':
        os.system("code Codes\\codefile.cs")
        os.system("csc Codes\\codefile.cs")
        os.system("Codes\\codefile.exe")
    if prog_lang=="c":
        os.system("code Codes\\codefile.c")
        os.system("gcc Codes\\codefile.c -o codefile.exe")
        os.system("codefile.exe")


#function for speech
def speak(audio):
    try:
        engine.say(audio)
        engine.runAndWait()
    except:
        pass

# Parameters
CHUNKSIZE = 1024 # number of audio samples per frame
RATE = 44100 # sampling rate in Hz
UPDATE_INTERVAL = 20 # update interval for the plot in ms

# Initialize PyAudio
p = pyaudio.PyAudio()

# Open audio stream
stream = p.open(format=pyaudio.paInt16,
                channels=1,
                rate=RATE,
                input=True,
                frames_per_buffer=CHUNKSIZE)

# Initialize plot
fig, ax = plt.subplots(facecolor='black', figsize=(3,1), dpi=100)
plt.axis('off')
line, = ax.plot(np.random.rand(CHUNKSIZE), color='red', linewidth=1)
ax.set_ylim(-1, 1)

# Function to update plot
def update_plot(frame):
    # Read audio from stream
    data = stream.read(CHUNKSIZE, exception_on_overflow=False)
    # Convert byte data to numpy array
    samples = np.frombuffer(data, dtype=np.int16)
    # Normalize samples
    samples = samples / 2**15
    # Update plot
    line.set_ydata(samples)
    return line,

# Create animation
ani = FuncAnimation(fig, update_plot, blit=True, interval=UPDATE_INTERVAL)

# grid for voice mouse
def grid(text):
    global state
    try:
        block_num=int(text)

        # Calculate the row and column of the block based on the block number
        block_col = (block_num - 1) % num_blocks_horizontal
        block_row = (block_num - 1) // num_blocks_horizontal

        # Calculate the x and y coordinates of the center of the block
        center_x = block_col * block_width + block_width // 2
        center_y = block_row * block_height + block_height // 2

        # Move the mouse to the center of the block
        pyautogui.moveTo(center_x, center_y)
        state="voicemouse"
        tkinterwin.destroy_win()
    except Exception as e:
        print(e)


# Function to check text for keywords
def check_text(text):
    global state
    # activation modes
    if 'activate type' in text:
        state="type"
        speak("mode set to typing")
    if 'activate voice mouse' in text:
        state="voicemouse"
        speak("mode set to voice mouse")
    if 'activate code' in text:
        state="code"
        speak("mode set to coding")
    if 'friday terminate' in text:
            speak("terminating. adios")
            os.abort()

    # friday gpt
    if "friday" in text and "chrome" in text:
        speak("working on it")
        text=text.replace("friday","")
        threading.Thread(target=webperform, args=([text])).start()
    elif "friday" in text and "open" in text:
        speak("working on it")
        text=text.replace("friday","")
        threading.Thread(target=perform, args=([text])).start()
    elif "friday" in text:
        text=text.replace("friday","")
        threading.Thread(target=gptreply, args=([text])).start()
    else:
        #friday features
        if "youtube" in text:
            threading.Thread(target=friday.youtube, args=([text])).start()
            speak("results opened in webbrowser")
        if "google" in text:
            threading.Thread(target=friday.google, args=([text])).start()
            speak("results opened in webbrowser")
        if "screenshot" in text:
            threading.Thread(target=friday.screenshot, args=([])).start()
            speak("Screenshot Saved and Opened for Preview")
        if "weather" in text:
            query=text
            query=query.split("in")[1].strip()
            complete_url = base_url + "appid=" + api_key + "&q=" + query
            response = requests.get(complete_url)
            x=response.json()
            if x["cod"] != "404":
                y = x["main"]
                current_temperature = y["temp"]
                current_pressure = y["pressure"]
                current_humidity = y["humidity"]
                z = x["weather"]
                weather_description = z[0]["description"]
                weatext=(" Temperature = " +
                    str(round(current_temperature-273.15)) +
                    " \N{DEGREE SIGN}C\nPressure = " +
                    str(current_pressure) +
                    " hPa\nHumidity = " +
                    str(current_humidity) +
                    " Percent\nDescription = " +
                    str(weather_description).title())
                speak("The Temperature in "+query+" is "+str(round(current_temperature-273.15))+"degree celsius and the weather can be described as "+str(weather_description))
            else:
                speak("City Not Found")
        if "time" in text:
            strTime = datetime.datetime.now().strftime("%H hours and %M minutes")
            speak(strTime)
        if "wikipedia" in text or "what are" in text or "what is" in text:
            query=text
            query = query.replace("wikipedia", "")
            query = query.replace("what are", "")
            query = query.replace("what is","")
            try:
                results = wikipedia.summary(query, sentences=1)
                speak(results)
            except:
                query = query.replace(" ","+")
                webbrowser.open("www.google.com/search?q="+query)
                speak("opened in webbrowser")
        if "calculate" in text or "evaluate" in text:
            query=text
            query = query.replace(" ","")
            query = query.replace("calculate","")
            query = query.replace("evaluate","")
            if 'into' in query:
                query = query.replace("into","*")
            if 'by' in query:
                query = query.replace("by","/")
            try:
                result=eval(query)
            except:
                result="invalid"
            speak(result)
        if "remember" in text or "remind me" in text:
            query=text
            if 'what' not in query:
                query=query.replace("remember ","")
                query=query.replace("remind me","")
                fil = open("remind.txt","a")
                fil.write(query+"\n")
                fil.close()
                speak("sure i will remember that for you")
            else:
                with open('remind.txt','r') as fil:
                    remtemp = fil.read()
                fil.close()
                os.remove("remind.txt")
                speak("You told me to remember "+remtemp)
        if 'open' in text or 'run' in text or 'launch' in text:
            text=text.replace("open ","")
            text=text.replace("run ","")
            text=text.replace("launch ","")
            text=text.lower()
            if text=="":
                pass
            else:
                speak("opening "+text)
                pyautogui.press('win')
                pyautogui.typewrite(text)
                time.sleep(1)
                pyautogui.press('enter')
        if "shutdown" in text:
            os.system("shutdown -s")
        else:
            pass

#function to code
def codify(text):
    global state
    if "end coding" in text:
        state="start"
        speak("you have stopped coding")
    else:
        if 'python' in text:
            speak("working on it")
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what will a code do. Give me the code for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. If creating functions, always include code for main as well"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            output=completion['choices'][0]['message']['content']
            output=output.replace("```python","```")
            output=(output.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.py", "w") as f:
                f.write(output)
            threading.Thread(target=code_exec, args=(['python'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        elif 'html' in text:
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on how a webpage should look like and what will its function be. Give me the code for it but don't explain how the code works. The code should contain css and javscript code so the page is responsive. Ise the <script> and <style> tags instead of creating separate files"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            html=completion['choices'][0]['message']['content']
            html=html.replace("```html","```")
            html=(html.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.html", "w") as f:
                f.write(html)
            threading.Thread(target=code_exec, args=(['html'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        elif 'java' in text:
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what will a code do. Give me the code for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. If creating functions, always include code for main as well"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            output=completion['choices'][0]['message']['content']
            output=output.replace("```cpp","```")
            output=(output.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.java", "w") as f:
                f.write(output)
            threading.Thread(target=code_exec, args=(['java'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        elif 'c plus plus' in text:
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what will a code do. Give me the code for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. If creating functions, always include code for main as well"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            output=completion['choices'][0]['message']['content']
            output=output.replace("```cpp","```")
            output=(output.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.cpp", "w") as f:
                f.write(output)
            threading.Thread(target=code_exec, args=(['c++'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        elif 'c sharp' in text:
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what will a code do. Give me the code for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. If creating functions, always include code for main as well"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            output=completion['choices'][0]['message']['content']
            output=output.replace("```cs","```")
            output=(output.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.cs", "w") as f:
                f.write(output)
            threading.Thread(target=code_exec, args=(['cs'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        elif ' c ' in text:
            state="start"
            completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo", 
            messages = [{"role": "system", "content" : "Answer as concisely as possible. I will be giving you a prompt on what will a code do. Give me the code for it but don't explain how the code works. The code should come as a single output, i.e don't output the code in various parts. If creating functions, always include code for main as well"},
            {"role": "user", "content" : text}]
            )
            print(completion['choices'][0]['message']['content'])
            output=completion['choices'][0]['message']['content']
            output=output.replace("```c","```")
            output=(output.split("```"))[1].split("```")[0]
            with open("Codes\\codefile.c", "w") as f:
                f.write(output)
            threading.Thread(target=code_exec, args=(['c'])).start()
            speak("the code is opened in vscode and running. coding mode ended")
        else:
            speak("i'm sorry, i couldn't understand what you meant. please specify the language you want the code in.")

#function for voice mouse
def voice_mouse(text):
    global state,gridwin
    if "mode reset" in text:
        state="start"
        speak("you have stopped voice mouse")
    if "left click" in text:
        count=text.count("left click")
        for i in range(count):
            pyautogui.click()
    if "right click" in text:
        count=text.count("right click")
        for i in range(count):
            pyautogui.click(button='right')
    if "scroll up" in text:
        count=text.count("scroll up")
        pyautogui.scroll(count*100)
    if "scroll down" in text:
        count=text.count("scroll down")
        pyautogui.scroll(count*-100)
    if "up" in text:
        count=text.count("up")
        pyautogui.moveRel(0, count*-100, duration=0.2)
    if "down" in text:
        count=text.count("down")
        pyautogui.moveRel(0, count*100, duration=0.2)
    if "left" in text:
        count=text.count("left")
        pyautogui.moveRel(count*-100, 0, duration=0.2)
    if "right" in text:
        count=text.count("right")
        pyautogui.moveRel(count*100, 0, duration=0.2)
    if "mouse map" in text:
        state="grid"
        gridwin=threading.Thread(target=tkinterwin.display_grid)
        gridwin.start()
        speak("choose a number on the grid")

# function to type
def type_text(text):
    global state
    if "mode reset" in text:
        text=text.replace("mode reset","")
        state="start"
        speak("you have stopped typing")
    if text=="":
        pass
    else:
        punctuated_text = model.restore_punctuation(text)
        pyautogui.typewrite(punctuated_text,0.1)

# Define a function to recognize speech
def recognize_speech():
    global state
    r = sr.Recognizer()
    while True:
        with sr.Microphone() as source:
            print("Speak now...")
            audio = r.listen(source)
            print("Processing...")
            text=""
        try:
            text = r.recognize_google(audio)
            text=text.lower()
        except sr.UnknownValueError:
            print("Sorry, could not understand audio")
        except sr.RequestError as e:
            print("Could not request results from Google Speech Recognition service; {0}".format(e))
        finally:
            if state=="type":
                threading.Thread(target=type_text, args=(text,)).start()
                print("You said: " + text)
            elif state=="voicemouse":
                threading.Thread(target=voice_mouse, args=(text,)).start()
                print("You said: " + text)
            elif state=="code":
                threading.Thread(target=codify, args=(text,)).start()
                print("You said: " + text)
            elif state=="grid":
                threading.Thread(target=grid, args=(text,)).start()
                print("You said: " + text)
            else:
                threading.Thread(target=check_text, args=(text,)).start()
                print("You said: " + text)

# Start a new thread for speech recognition
speech_thread = threading.Thread(target=recognize_speech)
speech_thread.start()

# Create tkinter window
root = tk.Tk()
root.overrideredirect(True)
root.geometry("300x100+{}+{}".format(ctypes.windll.user32.GetSystemMetrics(0) - 320, 20))
root.resizable(False, False)
root.attributes("-alpha", 0.6)
root.attributes("-topmost", True)

# Create canvas for plot
canvas = tk.Canvas(root, width=300, height=100, highlightthickness=0)
canvas.pack()

# Embed plot in canvas
plot_widget = FigureCanvasTkAgg(fig, master=canvas)
plot_widget.draw()
plot_widget.get_tk_widget().place(relx=0.5, rely=0.5, anchor="center")

# Start tkinter event loop
root.mainloop()

# Stop and close audio stream
stream.stop_stream()
stream.close()
p.terminate()