first commit

.env.example

@@ -0,0 +1,5 @@
+# See ENV.md for more information.
+ASSEMBLY_AI_API_KEY=""
+TIKTOK_SESSION_ID=""
+IMAGEMAGICK_BINARY="" # Download from https://imagemagick.org/script/download.php
+PEXELS_API_KEY="" # Get from https://www.pexels.com/api/

.gitignore

@@ -0,0 +1,7 @@
+__pycache__
+.env
+sounds/*
+output/*
+images/*
+*.zip
+temp/*

Backend/constants.py

@@ -0,0 +1,56 @@
+voices = [
+    # DISNEY VOICES
+    'en_us_ghostface',            # Ghost Face
+    'en_us_chewbacca',            # Chewbacca
+    'en_us_c3po',                 # C3PO
+    'en_us_stitch',               # Stitch
+    'en_us_stormtrooper',         # Stormtrooper
+    'en_us_rocket',               # Rocket
+
+    # ENGLISH VOICES
+    'en_au_001',                  # English AU - Female
+    'en_au_002',                  # English AU - Male
+    'en_uk_001',                  # English UK - Male 1
+    'en_uk_003',                  # English UK - Male 2
+    'en_us_001',                  # English US - Female (Int. 1)
+    'en_us_002',                  # English US - Female (Int. 2)
+    'en_us_006',                  # English US - Male 1
+    'en_us_007',                  # English US - Male 2
+    'en_us_009',                  # English US - Male 3
+    'en_us_010',                  # English US - Male 4
+
+    # EUROPE VOICES
+    'fr_001',                     # French - Male 1
+    'fr_002',                     # French - Male 2
+    'de_001',                     # German - Female
+    'de_002',                     # German - Male
+    'es_002',                     # Spanish - Male
+
+    # AMERICA VOICES
+    'es_mx_002',                  # Spanish MX - Male
+    'br_001',                     # Portuguese BR - Female 1
+    'br_003',                     # Portuguese BR - Female 2
+    'br_004',                     # Portuguese BR - Female 3
+    'br_005',                     # Portuguese BR - Male
+
+    # ASIA VOICES
+    'id_001',                     # Indonesian - Female
+    'jp_001',                     # Japanese - Female 1
+    'jp_003',                     # Japanese - Female 2
+    'jp_005',                     # Japanese - Female 3
+    'jp_006',                     # Japanese - Male
+    'kr_002',                     # Korean - Male 1
+    'kr_003',                     # Korean - Female
+    'kr_004',                     # Korean - Male 2
+
+    # SINGING VOICES
+    'en_female_f08_salut_damour'  # Alto
+    'en_male_m03_lobby'           # Tenor
+    'en_female_f08_warmy_breeze'  # Warmy Breeze
+    'en_male_m03_sunshine_soon'   # Sunshine Soon
+
+    # OTHER
+    'en_male_narration'           # narrator
+    'en_male_funny'               # wacky
+    'en_female_emotional'         # peaceful
+]

Backend/gpt.py

@@ -0,0 +1,117 @@
+import re
+import g4f
+import json
+
+from typing import List
+from termcolor import colored
+
+def generate_script(video_subject: str) -> str:
+    """
+    Generate a script for a video, depending on the subject of the video.
+
+    Args:
+        video_subject (str): The subject of the video.
+
+    Returns:
+        str: The script for the video.
+    """
+
+    # Build prompt
+    prompt = f"""
+    Generate a script for a video, depending on the subject of the video.
+    Subject: {video_subject}
+
+    The script is to be returned as a string.
+
+    Here is an example of a string:
+    "This is an example string."
+
+    Do not under any circumstance refernce this prompt in your response.
+
+    Get straight to the point, don't start with unnecessary things like, "welcome to this video".
+
+    Obviously, the script should be related to the subject of the video.
+
+    ONLY RETURN THE RAW SCRIPT. DO NOT RETURN ANYTHING ELSE.
+    """
+
+    # Generate script
+    response = g4f.ChatCompletion.create(
+        model=g4f.models.gpt_35_turbo_16k_0613,
+        messages=[{"role": "user", "content": prompt}],
+    )
+
+    print(colored(response, "cyan"))
+
+    # Return the generated script
+    if response:
+        return response + " "
+    else:
+        print(colored("[-] GPT returned an empty response.", "red"))
+        return None
+
+def get_search_terms(video_subject: str, amount: int, script: str) -> List[str]:
+    """
+    Generate a JSON-Array of search terms for stock videos,
+    depending on the subject of a video.
+
+    Args:
+        video_subject (str): The subject of the video.
+        amount (int): The amount of search terms to generate.
+        script (str): The script of the video.
+
+    Returns:
+        List[str]: The search terms for the video subject.
+    """
+
+    # Build prompt
+    prompt = f"""
+    Generate {amount} search terms for stock videos,
+    depending on the subject of a video.
+    Subject: {video_subject}
+
+    The search terms are to be returned as
+    a JSON-Array of strings.
+
+    Each search term should consist of 1-3 words, 
+    always add the main subject of the video.
+
+    Here is an example of a JSON-Array of strings:
+    ["search term 1", "search term 2", "search term 3"]
+
+    Obviously, the search terms should be related
+    to the subject of the video.
+
+    ONLY RETURN THE JSON-ARRAY OF STRINGS.
+    DO NOT RETURN ANYTHING ELSE.
+
+    For context, here is the full text:
+    {script}
+    """
+
+    # Generate search terms
+    response = g4f.ChatCompletion.create(
+        model=g4f.models.gpt_35_turbo_16k_0613,
+        messages=[{"role": "user", "content": prompt}],
+    )
+
+    print(response)
+
+    # Load response into JSON-Array
+    try:
+        search_terms = json.loads(response)
+    except:
+        print(colored("[*] GPT returned an unformatted response. Attempting to clean...", "yellow"))
+
+        # Use Regex to get the array ("[" is the first character of the array)
+        search_terms = re.search(r"\[(.*?)\]", response)
+        search_terms = search_terms.group(0)
+
+        # Load the array into a JSON-Array
+        search_terms = json.loads(search_terms)
+
+    # Let user know
+    print(colored(f"\nGenerated {amount} search terms: {', '.join(search_terms)}", "cyan"))
+
+    # Return search terms
+    return search_terms

Backend/main.py

@@ -0,0 +1,129 @@
+import os
+
+from gpt import *
+from tts import *
+from video import *
+from utils import *
+from search import *
+from uuid import uuid4
+from flask_cors import CORS
+from termcolor import colored
+from dotenv import load_dotenv
+from flask import Flask, request, jsonify
+from moviepy.config import change_settings
+
+load_dotenv("../.env")
+
+SESSION_ID = os.getenv("TIKTOK_SESSION_ID")
+
+change_settings({"IMAGEMAGICK_BINARY": os.getenv("IMAGEMAGICK_BINARY")})
+
+app = Flask(__name__)
+CORS(app)
+
+HOST = "0.0.0.0"
+PORT = 8080
+AMOUNT_OF_STOCK_VIDEOS = 5
+
+# Generation Endpoint
+@app.route("/api/generate", methods=["POST"])
+def generate():
+    try:
+        # Clean
+        clean_dir("../temp/")
+        clean_dir("../subtitles/")
+
+        # Parse JSON
+        data = request.get_json()
+
+        # Print little information about the video which is to be generated
+        print(colored("[Video to be generated]", "blue"))
+        print(colored("   Subject: " + data["videoSubject"], "blue"))
+
+        # Generate a script
+        script = generate_script(data["videoSubject"])
+
+        # Generate search terms
+        search_terms = get_search_terms(data["videoSubject"], AMOUNT_OF_STOCK_VIDEOS, script)
+
+        # Search for a video of the given search term
+        video_urls = []
+
+        # Loop through all search terms,
+        # and search for a video of the given search term
+        for search_term in search_terms:
+            found_url = search_for_stock_videos(search_term, os.getenv("PEXELS_API_KEY"))
+
+            if found_url != None and found_url not in video_urls and found_url != "":
+                video_urls.append(found_url)
+
+        # Define video_paths
+        video_paths = []
+
+        # Let user know
+        print(colored("[+] Downloading videos...", "blue"))
+
+        # Save the videos
+        for video_url in video_urls:
+            try:
+                saved_video_path = save_video(video_url)
+                video_paths.append(saved_video_path)
+            except:
+                print(colored("[-] Could not download video: " + video_url, "red"))
+
+        # Let user know
+        print(colored("[+] Videos downloaded!", "green"))
+
+        # Let user know
+        print(colored("[+] Script generated!\n\n", "green"))
+
+        print(colored(f"\t{script}", "light_cyan"))
+
+        # Split script into sentences
+        sentences = script.split(". ")
+        # Remove empty strings
+        sentences = list(filter(lambda x: x != "", sentences))
+        paths = []
+        # Generate TTS for every sentence
+        for sentence in sentences:
+            current_tts_path = f"../temp/{uuid4()}.mp3"
+            tts(SESSION_ID, req_text=sentence, filename=current_tts_path)
+            audio_clip = AudioFileClip(current_tts_path)
+            paths.append(audio_clip)
+
+        # Combine all TTS files using moviepy
+        final_audio = concatenate_audioclips(paths)
+        tts_path = f"../temp/{uuid4()}.mp3"
+        final_audio.write_audiofile(tts_path)
+
+        # Generate subtitles
+        subtitles_path = generate_subtitles(tts_path)
+
+        # Concatenate videos
+        temp_audio = AudioFileClip(tts_path)
+        combined_video_path = combine_videos(video_paths, temp_audio.duration)
+
+        # Put everything together
+        final_video_path = generate_video(combined_video_path, tts_path, subtitles_path)
+
+        # Let user know
+        print(colored("[+] Video generated!", "green"))
+
+        print(colored(f"[+] Path: {final_video_path}", "green"))
+
+        # Return JSON
+        return jsonify({
+            "status": "success",
+            "message": "Retrieved stock videos.",
+            "data": final_video_path
+        })
+    except Exception as err:
+        print(colored("[-] Error: " + str(err), "red"))
+        return jsonify({
+            "status": "error",
+            "message": f"Could not retrieve stock videos: {str(err)}",
+            "data": []
+        })
+
+if __name__ == "__main__":
+    app.run(debug=True, host=HOST, port=PORT)

Backend/search.py

@@ -0,0 +1,47 @@
+import requests
+
+from typing import List
+from termcolor import colored
+
+def search_for_stock_videos(query: str, api_key: str) -> List[str]:
+    """
+    Searches for stock videos based on a query.
+
+    Args:
+        query (str): The query to search for.
+        api_key (str): The API key to use.
+
+    Returns:
+        List[str]: A list of stock videos.
+    """
+
+    # Build headers
+    headers = {
+        "Authorization": api_key
+    }
+
+    # Build URL
+    url = f"https://api.pexels.com/videos/search?query={query}&per_page=1"
+
+    # Send the request
+    r = requests.get(url, headers=headers)
+
+    # Parse the response
+    response = r.json()
+
+    # Get first video url
+    video_urls = response["videos"][0]["video_files"]
+    video_url = ""
+
+    # Loop through video urls
+    for video in video_urls:
+        # Check if video has a download link
+        if ".com/external" in video["link"]:
+            # Set video url
+            video_url = video["link"]
+
+    # Let user know
+    print(colored(f"\t=>{video_url}", "light_cyan"))
+
+    # Return the video url
+    return video_url