app.py

from langchain_community.llms import HuggingFacePipeline 
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
from components import caption_chain, tag_chain
from components import pexels, utils
import os, gc
import gradio as gr
import os
from moviepy.editor import VideoFileClip, concatenate_videoclips

# 모델과 토크나이저 로드
model = AutoModelForSeq2SeqLM.from_pretrained("declare-lab/flan-alpaca-large")
tokenizer = AutoTokenizer.from_pretrained("declare-lab/flan-alpaca-large")

# 파이프라인 설정
pipe = pipeline(
    'text2text-generation',
    model=model,
    tokenizer=tokenizer,
    max_length=120
)

# HuggingFacePipeline을 사용하여 LLM 초기화
local_llm = HuggingFacePipeline(pipeline=pipe)

# 체인 구성
llm_chain = caption_chain.chain(llm=local_llm)
sum_llm_chain = tag_chain.chain(llm=local_llm)

# Pexels API 키
pexels_api_key = os.getenv('pexels_api_key')


def pred(product_name, orientation):
    # 기존 코드 생략...
    folder_name, sentences = pexels.generate_videos(product_name, pexels_api_key, orientation, height, width, llm_chain, sum_llm_chain)
    video_files = [os.path.join(folder_name, f) for f in os.listdir(folder_name) if f.endswith('.mp4')]
    
    if not video_files:  # 비디오 파일이 없으면 메시지 반환
        return ["No videos were generated. Please check the input and try again.", ""]
    
    # 비디오 파일 결합
    video_path = combine_videos(video_files, folder_name)
    if not video_path:  # 비디오 결합에 실패하면 메시지 반환
        return ["Failed to combine videos.", ""]
    
    return ["\n".join(sentences), video_path]

def combine_videos(video_files, output_folder):
    if not video_files:
        print("No video files to combine.")
        return ""  # 빈 문자열 반환
    
    clips = [VideoFileClip(vf) for vf in video_files]
    final_clip = concatenate_videoclips(clips)
    output_path = os.path.join(output_folder, "final_video.mp4")
    final_clip.write_videofile(output_path)
    return output_path
    
# 예측 함수
def pred(product_name, orientation):
    # 비디오 방향과 해상도 설정
    if orientation == "Shorts/Reels/TikTok (1080 x 1920)":
        orientation = "portrait"  # 오타 수정
        height = 1920
        width = 1080
    elif orientation == "Youtube Videos (1920 x 1080)":
        orientation = "landscape"
        height = 1080
        width = 1920
    else:
        orientation = "square"
        height = 1080
        width = 1080

    # 비디오 생성 및 문장 추출
    folder_name, sentences = pexels.generate_videos(product_name, pexels_api_key, orientation, height, width, llm_chain, sum_llm_chain)
    gc.collect()

    # 비디오 파일이 실제로 생성되었는지 확인
    if not sentences:
        return ["No videos generated. Please try again.", ""]

    # 비디오 파일 결합

    video_path = combine_videos(video_files, folder_name)
    if not video_path or not os.path.exists(video_path):
        return ["Failed to combine videos.", ""]
    
    return ["\n".join(sentences), video_path]

# Gradio 인터페이스 설정 및 런칭
with gr.Blocks() as demo:
    gr.Markdown(
        """
        # Ads Generator
        Create video ads based on your product name using AI
        ### Note: the video generation takes about 2-4 minutes 
        """
    )
    dimension = gr.Dropdown(
        ["Shorts/Reels/TikTok (1080 x 1920)", "Facebook/Youtube Videos (1920 x 1080)", "Square (1080 x 1080)"], 
        label="Video Dimension", info="Choose dimension"
    )
    product_name = gr.Textbox(label="Product name")
    captions = gr.Label(label="Captions")
    video = gr.Video()
    btn = gr.Button("Submit")
    btn.click(pred, inputs=[product_name, dimension], outputs=[captions, video])

    # 여기에 추가적인 Gradio 인터페이스 구성 요소를 추가할 수 있습니다.

# Gradio 앱 실행
demo.launch()