app.py

"""
FastAPI Backend Application for Document Processing and Management

This application provides a comprehensive backend service for processing and managing various types of documents,
including images, SMS data, contacts, and YouTube links. It integrates with multiple external services and APIs
to provide document processing, storage, and management capabilities.

Main Components:
1. API Setup and Configuration:
   - FastAPI application with CORS middleware
   - Firebase, Google Sheets, Drive, and YouTube API integrations
   - Gemini AI model configuration for image processing

2. Core File Management Functions:
   - move_file_to_folder: Handles file organization in Google Drive
   - update_user_sheet: Manages data updates in Google Sheets with backup capabilities

3. Image Processing Functions:
   - request_gpt4o_completion: Processes images using GPT-4 Vision or Gemini
   - process_photo: Endpoint for handling image processing requests
   - update_user_sheets_and_folders: Organizes processed image data

4. Sheet Management Functions:
   - copy_sms_sheet: Creates new sheets for SMS data
   - edit_spreadsheet: Manages template-based sheet creation
   - check_user_doc: Handles user verification and transaction records

5. Data Synchronization Functions:
   - sync-data endpoint: Manages contacts and SMS data synchronization
   - update_fb_transaction: Updates Firebase transaction records

6. YouTube Integration:
   - share-target-handle endpoint: Processes YouTube links
   - Functions for video information extraction and summarization

7. Business Information Functions:
   - search_business_info: Business information lookup in reference spreadsheet

8. Utility Functions:
   - Helper functions for data conversion and formatting
   - Error handling and logging utilities

Dependencies:
- FastAPI for API framework
- Firebase Admin for user management
- Google API clients for Drive and Sheets integration
- Various AI models for image and text processing
"""

from fastapi import FastAPI, HTTPException, UploadFile, File, Request, BackgroundTasks, Form
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
from httpx import Timeout
import httpx
import logging
import base64
import firebase_admin
from firebase_admin import credentials
from firebase_admin import firestore
import json
from datetime import datetime
import pytz

from typing import List, Optional, Dict
from google.oauth2 import service_account
from googleapiclient.discovery import build

from io import BytesIO
import google.generativeai as genai
import os

from google.cloud import vision
from typing import Tuple
from base64 import b64encode

from openai import OpenAI
from fuzzywuzzy import fuzz

from youtube_transcript_api import YouTubeTranscriptApi
from langchain_community.document_loaders import YoutubeLoader

import re
import asyncio

# start application
app = FastAPI()

# allow all origins
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # Allows all origins
    allow_credentials=True,  # Allows credentials (such as cookies) to be sent with requests
    allow_methods=["*"],  # Allows all methods
    allow_headers=["*"],  # Allows all headers
)

# set up logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
logging.info("Application starting up")

request_timeout = Timeout(1000.0, connect=600.0)

# set up firebase


# Load Firebase configuration from a file
with open('/app/secrets/fb_secret.json', 'r') as f:
    firebase_configuration = json.load(f)

firebase_credentials = credentials.Certificate(firebase_configuration)
firebase_admin.initialize_app(firebase_credentials, {
    'projectId': firebase_configuration['project_id'],
})
firestore_db = firestore.client()
users_collection_ref = firestore_db.collection('users')
transactions_collection_ref = firestore_db.collection('transactions')

SCOPES = [
    'https://www.googleapis.com/auth/spreadsheets',
    'https://www.googleapis.com/auth/drive',
    'https://www.googleapis.com/auth/youtube.force-ssl'  # Add YouTube scope
]

with open('/app/secrets/openai_key.txt', 'r') as f:
    OPENAI_KEY = f.read()

APICredential = service_account.Credentials.from_service_account_file(
    '/app/secrets/googleapi_secret.json', scopes=SCOPES)

sheet_service = build('sheets', 'v4', credentials=APICredential)
drive_service = build('drive', 'v3', credentials=APICredential)

# Add YouTube API key setup
YOUTUBE_API_KEY = os.getenv('YOUTUBE_API_KEY')  # Add to your environment variables
youtube_service = build('youtube', 'v3', developerKey=YOUTUBE_API_KEY)  # Use API key instead of OAuth

with open('/app/secrets/gemini_key.txt', 'r') as f:
    api_key = f.read()

genai.configure(api_key=api_key)

# Create the model
generation_config = {
  "temperature": 1,
  "top_p": 0.95,
  "top_k": 64,
  "max_output_tokens": 8192,
  "response_mime_type": "application/json",
}

model = genai.GenerativeModel(
  model_name="gemini-1.5-flash",
  generation_config=generation_config,
  # safety_settings = Adjust safety settings
  # See https://ai.google.dev/gemini-api/docs/safety-settings
)



# first three are "등록일시", "썸네일", "바로가기", 
receipt_sheet_headers = ['발행일', '상호', '사업자번호', '합계금액', '내역', '카드번호', '카드종류', '승인번호', '기타']
# at index 1 '등록 일시', at last index '명함보기'
business_card_sheet_headers = ['이름', '회사명', '소속 및 직위', '주소', '전화' ,'휴대폰', '팩스','e-mail','홈페이지']
sms_sheet_headers = ['날짜', '계좌', '입금', '출금', '잔액', '문자']
contact_sheet_headers = ['이름',	'전화번호1',	'전화번호2',	'전화번호3',	'이메일주소1',	'이메일주소2',	'이메일주소3',	'그룹']

receipts_ss = 1395967745
business_cards_ss = 1733555840
contacts_ss = 1729750548
sms_ss = 1891574758
shared_links_ss = 289914391

kst = pytz.timezone('Asia/Seoul')


class SheetUpdateRequest(BaseModel):
    target_sheet_id: int
    data: list

def convert_dicts_to_list(dicts: List[Dict], headers: List[str], add_link: bool = False, 
                          image_id: str = '', link_text: str = '', 
                          link_position: int = -1) -> List[List[str]]:
    rows = []
    # timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
    timestamp = datetime.now(kst).strftime("%Y-%m-%d %H:%M:%S")
    
    for dict_item in dicts:
        # row = list(dict_item.values())
        # row = [str(value) for value in dict_item.values()]
        row = [str(dict_item.get(header, '')) for header in headers]
        
        # Always add timestamp to the first index
        row.insert(0, timestamp)
        
        if add_link:
            # Ensure the list length is at least 9
            while len(row) < 9:
                row.append('')
            if link_position == -1:  # Append link at the end
                row.append(f'=HYPERLINK("https://drive.google.com/file/d/{image_id}/view?usp=sharing", "{link_text}")')
            else:
                row.insert(link_position, f'=HYPERLINK("https://drive.google.com/file/d/{image_id}/view?usp=sharing", "{link_text}")')
                
        rows.append(row)
    # print(rows)
    return rows



async def download_file_from_drive(file_id: str) -> Tuple[bytes, str]:
    """
    Downloads a file from Google Drive using the file ID
    and returns the file content as bytes and the MIME type.
    """
    try:
        url = f"https://drive.google.com/uc?export=download&id={file_id}"
        
        async with httpx.AsyncClient(follow_redirects=True) as client:  # Enable redirect following
            response = await client.get(url)
            if response.status_code == 200:
                return response.content, response.headers.get('content-type', 'application/octet-stream')
            else:
                raise HTTPException(
                    status_code=500, 
                    detail=f"Failed to download file, status code: {response.status_code}"
                )

    except Exception as e:
        logging.error(f"Error downloading file from Drive: {e}")
        raise HTTPException(status_code=500, detail="Failed to download file from Drive")

async def request_gpt4o_completion(image_id: str, user_credit: int, use_gpt4: bool = False):
    """
    Processes images using either GPT-4 Vision or Gemini for text extraction and analysis.
    
    Args:
        image_id (str): The Google Drive ID of the image to process
        user_credit (int): Available user credits for processing
        use_gpt4 (bool): Whether to use GPT-4 Vision instead of Gemini
    
    Returns:
        str: JSON string containing extracted information
    """
    try:
        # Step 1: Download the image file from Google Drive
        file_content, mime_type = await download_file_from_drive(image_id)

        if use_gpt4:
            # Convert image to base64 for GPT-4 Vision
            base64_image = b64encode(file_content).decode('utf-8')
            
            client = OpenAI(api_key=OPENAI_KEY)
            
            response = client.chat.completions.create(
                model="gpt-4o",
                messages=[
                    {
                        "role": "system",
                        "content": """You are a receipt and business card data extractor. Extract information EXACTLY in the specified format.
For receipts, follow this EXACT format and order:
1. "발행일": Must be in format "M/D/YYYY HH:MM:SS"
2. "상호": Store/business name only
3. "사업자번호": Must be in format "###-##-#####" (if present)
4. "합계금액": Total amount with comma for thousands (e.g., "4,300")
5. "내역": List items with format "item price quantity" (e.g., "보통 7g 10통 430 4,300")
6. "카드번호": Must include asterisks for hidden numbers (e.g., "5286-6420-****-077")
7. "카드종류": Card type only
8. "승인번호": Approval number only
9. "기타": All remaining information as key-value pairs separated by commas
Example receipt:
{
  "receipts": [{
    "발행일": "3/6/2024 15:31:00",
    "상호": "시흥배곧동우체국",
    "사업자번호": "101-83-02925",
    "합계금액": "4,300",
    "내역": "보통 7g 10통 430 4,300",
    "카드번호": "5286-6420-****-077",
    "카드종류": "NH기업체크",
    "승인번호": "49980444",
    "기타": "전화: 031-8042-2047, 주소: 경기도 시흥시 정왕동 2543, 고객문의 전화: 1588-1300"
  }],
  "busi_cards": []
}
For business cards, follow this EXACT format and order:
1. "이름": Full name
2. "회사명": Company name
3. "소속 및 직위": Position and department
4. "주소": Address with postal code in format "주소 (우편번호) 코드"
5. "전화": Office phone with hyphens
6. "휴대폰": Mobile with hyphens (e.g., "010-0000-0000")
7. "팩스": Fax number with hyphens
8. "e-mail": Email address
9. "홈페이지": Website URL
Output must be valid JSON with "receipts" and "busi_cards" arrays."""
                    },
                    {
                        "role": "user",
                        "content": [
                            {"type": "text", "text": "Extract the information from this image following the exact format specified."},
                            {
                                "type": "image_url",
                                "image_url": {
                                    "url": f"data:{mime_type};base64,{base64_image}"
                                }
                            }
                        ]
                    }
                ],
                max_tokens=4096,
                temperature=1,
                response_format={ "type": "json_object" }
            )
            
            return response.choices[0].message.content

        else:
            # Existing Gemini implementation
            # Step 3: Get text data from the image using Google Cloud Vision
            client = vision.ImageAnnotatorClient()
            image = vision.Image(content=file_content)
            response = client.text_detection(image=image)
            texts = response.text_annotations

            if response.error.message:
                logging.error(f"Error in Google Cloud Vision API: {response.error.message}")
                raise Exception(
                    "{}\nFor more info on error messages, check: "
                    "https://cloud.google.com/apis/design/errors".format(response.error.message)
                )

            # Extract all detected text
            extracted_text = texts[0].description if texts else ""

            # Step 4: Prepare the prompt for Gemini
            prompt = f'''
            #Direction
            You are an employee of finance department of a company. For each task well done, you will be rewarded $1,000 bonus. 
            The boss has come to your office and gave you a task of transfering the content of receipts and business cards onto a google spreadsheet.
            He hands you two things. 
            One, is a paper containing guideline and example.
            Two, is an OCR text.
            #Guideline
            -strictly adhere to the given content of OCR text.
            -it could be of receipt(s) or business card(s) but not both
            -sort any other info that couldn't be classified as "기타"
            -format the extracted information as a JSON object with two main keys: 'receipts' and 'busi_cards'.
            -ensure the total number of receipts and business cards combined does not exceed 500.
            #Example
            ##Receipts
            Input
            은성마켓
            사업자번호: 727-27-01455
            대표자:허유성
            주
            소: 경기 시흥시 배곧3로 96
            전화 번호: 031-431-1544
            판매일: 24-02-23 20:10, 금요일 계산대:001
            NO. 상품명
            단가 수량 금액
            001 느타리버섯
            200092
            1,000
            1. 1,000 #
            002 양배추
            200071
            (#)면세물품:
            합
            계:
            신용카드지불:
            ★카드신용승인
            火
            1,800 1 1.800 #
            2,800
            2,800
            2,800
            가맹점명 은성마켓
            카드번호
            4673-09**-****-515*
            카드명 : KB국민체크
            전표매입: KB국민카드 무서명: 00111521584
            승인금액: 2,800
            (일시불)
            승인번호: 30014507, 전표No:201036
            제출 VAN: JTNET-리더기
            정상승인
            감사합니다.
            거래NO:0223007835 계산원: 관리자(001)
            2502230078357
            Output
            발행일: 2024/02/23 20:10
            상호: 은성마켓
            사업자번호: 727-27-01455
            합계금액: 2,800
            내역: 
                느타리버섯 1,000원 1개
                양배추 1,800원 1개
            카드번호: 4673-09**-****-515*
            카드종류: KB국민체크
            승인번호: 30014507
            기타: 대표자: 허유성, 주소: 경기 시흥시 배곧3로 96, 전화 번호: 031-431-1544, 면세물품: 2,800, 합계: 2,800, 신용카드지불:2,800, 전표No:201036, 제출 VAN: JTNET-리더기, 거래NO:0223007835 계산원: 관리자(001)
            ##Business Cards
            Input
            박장원
            아이프리마
            팀장/정책사업팀
            R&D사업실
            Over the MIRACLE
            경기도 화성시 정남면 만년로 98번길 55 10층 (우) 18523
            Mobile 010-9582-0925
            Tel (031) 000-0000
            Fax (031) 000-0000
            Email iprima@iprima.com
            www.iprima.com
            Output
            이름: 박장원
            회사명: 아이프리마
            소속 및 직위: 팀장/정책사업팀 R&D사업실
            주소: 경기도 화성시 정남면 만년로 98번길 55 (우편번호) 18523
            전화: (031) 000-0000
            휴대폰: (031) 000-0000
            팩스: 010-0000-0000
            e-mail: iprima@gmail.com
            홈페이지: www.iprima.com
            기타: Over the MIRACLE
            Image content: {extracted_text}
            '''
            logging.info("Prompt prepared for Gemini model: " + prompt)

            # Generate content using the Gemini model
            response = model.generate_content(
                contents=[prompt]
            )

            if response.text:
                logging.info("Parsing Gemini model response")
                json_response = json.loads(response.text)
                logging.info(f"JSON response parsed: {response.text}")
                return json.dumps(json_response)
            else:
                logging.error("Gemini model did not return a text response")
                raise Exception("Gemini model did not return a text response")

    except Exception as e:
        logging.error(f"Error in request_gpt4o_completion: {str(e)}")
        raise HTTPException(status_code=500, detail=f"Failure: {str(e)}")

    
async def move_file_to_folder(file_id, current_parents, new_parents):
    """
    Moves a file to specified folders in Google Drive while managing parent folders.
    
    Args:
        file_id (str): The ID of the file to move
        current_parents (list/str): Current parent folder(s) to remove from
        new_parents (list/str): New parent folder(s) to add to
    
    Returns:
        None
    """
    try:

        # Ensure current_parents is a list of parent IDs
        if not isinstance(current_parents, list):
            current_parents = [current_parents]
        if not isinstance(new_parents, list):
            new_parents = [new_parents]

        # Retrieve the existing parents to remove
        file = drive_service.files().get(fileId=file_id, fields='parents').execute()
        existing_parents = file.get('parents')
        logging.info(f"Existing parents of the file {file_id}: {existing_parents}")

        # Remove only the current parents
        parents_to_remove = [parent for parent in current_parents if parent in existing_parents]
        parents_to_add = [parent for parent in new_parents if parent not in existing_parents]

        previous_parents = ",".join(parents_to_remove)
        logging.info(f"Parents to remove: {parents_to_remove}")
        logging.info(f"Parents to add: {parents_to_add}")

        # Move the file to the new folders while keeping it in yungsoogi folder
        if parents_to_add:
            file = drive_service.files().update(
                fileId=file_id,
                addParents=",".join(parents_to_add),
                removeParents=previous_parents,
                fields='id, parents'
            ).execute()
            logging.info(f"File {file_id} moved to new folders {new_parents}.")
        else:
            logging.info(f"File {file_id} already in the target folders {new_parents}.")
    except Exception as e:
        logging.error(f"Failed to move file {file_id} to new folders: {e}")

async def update_user_sheet(spreadsheet_id:str, sheet_name:str, data: List[List[str]], is_reset: Optional[bool] = False):
    """
    Updates a user's Google Sheet with new data, optionally creating a backup.
    
    Args:
        spreadsheet_id (str): The ID of the target spreadsheet
        sheet_name (str): Name of the sheet to update
        data (List[List[str]]): The data to write to the sheet
        is_reset (bool, optional): Whether to create a backup before updating
    
    Returns:
        None
    """
    print(f"Spreadsheet ID: {spreadsheet_id}, Sheet Name: {sheet_name}, Data: {data}")
    
    # Get all the sheets in the spreadsheet
    sheet_metadata = sheet_service.spreadsheets().get(spreadsheetId=spreadsheet_id).execute()
    sheets = sheet_metadata.get('sheets', '')
    
    # Find the sheet by name
    sheet = next((sheet for sheet in sheets if sheet['properties']['title'] == sheet_name), None)
    
    if not sheet:
        raise ValueError(f"Sheet '{sheet_name}' not found in the spreadsheet.")
    
    sheet_id = sheet['properties']['sheetId']
    
    # Rest of the function remains the same
    no_of_rows = len(data)
    no_of_cols = len(data[0])

    if is_reset:
        new_sheet_name = f"{sheet_name} {datetime.now(kst).strftime('%Y-%m-%d %H:%M')}"
        duplicate_requests = [{
            "duplicateSheet": {
                "sourceSheetId": sheet_id,
                "newSheetName": new_sheet_name
            }
        }]
        duplicate_response = sheet_service.spreadsheets().batchUpdate(
            spreadsheetId=spreadsheet_id,
            body={
                'requests': duplicate_requests
            }
        ).execute()
        sheet_id = duplicate_response['replies'][0]['duplicateSheet']['properties']['sheetId']
        sheet_name = new_sheet_name

    # Insert new rows
    requests = [
        {
        "updateSheetProperties": {
            "properties": {
                "sheetId": sheet_id,
                "hidden": False
            },
            "fields": "hidden"
        }
    },
        {
        "insertDimension": {
            "range": {
                "sheetId": sheet_id,
                "dimension": "ROWS",
                "startIndex": 1,
                "endIndex": no_of_rows + 1
            },
            "inheritFromBefore": False,
        }
    }]
    
    response = sheet_service.spreadsheets().batchUpdate(
        spreadsheetId=spreadsheet_id,
        body={
            'requests': requests
        }
    ).execute()
    
    update_body = {
        'values': data
    }
    start_column = 'A'
    end_column = chr(ord(start_column) + no_of_cols - 1)
    range_to_update = f'{sheet_name}!{start_column}2:{end_column}{no_of_rows + 1}'
    
    result = sheet_service.spreadsheets().values().update(
        spreadsheetId=spreadsheet_id,
        range=range_to_update,
        valueInputOption='USER_ENTERED',
        body=update_body
    ).execute()

    # Format the cells
    for row_index, row in enumerate(data):
        for col_index, cell_value in enumerate(row):
            if cell_value.startswith('=HYPERLINK'):
                color = {"red": 0, "green": 0, "blue": 1}  # Blue for hyperlinks
            else:
                color = {"red": 0, "green": 0, "blue": 0}  # Black for other text
            
            requests = [
                {
                    "repeatCell": {
                        "range": {
                            "sheetId": sheet_id,
                            "startRowIndex": row_index + 1,
                            "endRowIndex": row_index + 2,
                            "startColumnIndex": col_index,
                            "endColumnIndex": col_index + 1
                        },
                        "cell": {
                            "userEnteredFormat": {
                                "textFormat": {
                                    "foregroundColor": color
                                }
                            }
                        },
                        "fields": "userEnteredFormat.textFormat.foregroundColor"
                    }
                }
            ]
            
            response = sheet_service.spreadsheets().batchUpdate(
                spreadsheetId=spreadsheet_id,
                body={
                    'requests': requests
                }
            ).execute()

async def copy_sms_sheet(phone_no: str, spreadsheet_id: str):
    """
    Creates a new sheet for SMS data by copying a template.
    
    Args:
        phone_no (str): Phone number to create sheet for
        spreadsheet_id (str): Target spreadsheet ID
    
    Returns:
        str: ID of the newly created sheet
    """
    requests = [
        {
            "duplicateSheet": {
                "sourceSheetId": sms_ss,
                "newSheetName": 'SMS ' + phone_no
            }
        }
    ]
    
    response = sheet_service.spreadsheets().batchUpdate(
        spreadsheetId=spreadsheet_id,
        body={
            'requests': requests
        }
    ).execute()
    print(response)
    new_sheet_id = response['replies'][0]['duplicateSheet']['properties']['sheetId']

    unhide_request = [{
        "updateSheetProperties": {
            "properties": {
                "sheetId": new_sheet_id,
                "hidden": False
            },
            "fields": "hidden"
        }
    }]

    sheet_service.spreadsheets().batchUpdate(
        spreadsheetId=spreadsheet_id,
        body={
            'requests': unhide_request
        }
    ).execute()

    return new_sheet_id
    
async def update_fb_transaction(transaction_ref: firestore.DocumentReference, no_sms: Optional[int] = 0, no_contacts: Optional[int] = 0, no_receipts: Optional[int] = 0, no_business_cards: Optional[int] = 0):
    try:
        # Update the transaction
        transaction_ref.update({
            "no_sms": firestore.Increment(no_sms),
            "no_contacts": firestore.Increment(no_contacts),
            "no_receipts": firestore.Increment(no_receipts),
            "no_business_cards": firestore.Increment(no_business_cards)
        })

        return {"status": "Success"}
    except Exception as e:
        logging.error(f"Error updating transaction: {e}")
        return {"status": f"Error: {str(e)}"}

async def get_youtube_transcript_url(video_id: str) -> str:
    try:
        headers = {
            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36',
            'Accept-Language': 'ko,en;q=0.9'
        }

        url = f"https://www.youtube.com/watch?v={video_id}"
        async with httpx.AsyncClient() as client:
            response = await client.get(url, headers=headers)
            page_content = response.text
            
            # Look for timedtext URLs
            timedtext_pattern = r'"baseUrl":"(https://www\.youtube\.com/api/timedtext[^"]+)"'
            matches = re.finditer(timedtext_pattern, page_content)
            
            urls = {}  # Dictionary to store language:url pairs
            for match in matches:
                url = match.group(1).replace('\\u0026', '&')
                # Extract language from URL
                lang_match = re.search(r'lang=(\w+)', url)
                if lang_match:
                    lang = lang_match.group(1)
                    urls[lang] = url
            
            if urls:
                # Try languages in order of preference
                if 'ko' in urls:
                    logging.info("Found Korean transcript")
                    return urls['ko']
                elif 'en' in urls:
                    logging.info("Found English transcript")
                    return urls['en']
                else:
                    # Get first available language
                    lang = next(iter(urls))
                    logging.info(f"Found {lang.upper()} transcript")
                    return urls[lang]
            else:
                logging.warning("No transcript URLs found in the page content")
                return "No transcript URLs found"
                
    except Exception as e:
        logging.error(f"Error in get_youtube_transcript_url: {str(e)}")
        return f"Error: {str(e)}"

async def process_youtube_video(video_id: str, video_title: str = "") -> dict:
    try:
        # Get video details from YouTube Data API
        video_response = youtube_service.videos().list(
            part='snippet',
            id=video_id
        ).execute()

        if not video_response.get('items'):
            return {
                "title": video_title,
                "summary": "비디오를 찾을 수 없습니다.",
                "key_points": [],
                "timestamps": {}
            }

        video_info = video_response['items'][0]['snippet']
        video_title = video_info.get('title', video_title)
        description = video_info.get('description', '')

        # Get transcript URL
        transcript_url = await get_youtube_transcript_url(video_id)
        if not transcript_url:
            return {
                "title": video_title,
                "summary": "자막을 찾을 수 없습니다.",
                "key_points": [],
                "timestamps": {}
            }

        # Get the actual captions
        async with httpx.AsyncClient() as client:
            response = await client.get(transcript_url)
            if response.status_code != 200:
                return {
                    "title": video_title,
                    "summary": "자막을 가져오는데 실패했습니다.",
                    "key_points": [],
                    "timestamps": {}
                }
            
            full_transcript = response.text

        # Generate summary using Gemini
        prompt = """
        You are a professional content summarizer. Summarize the following YouTube video transcript 
        in Korean. Focus on the key points and main ideas. Keep the summary concise but informative.
        
        Video Title: {title}
        Description: {description}
        
        Transcript:
        {transcript}
        
        Please format the response as a JSON object with the following structure:
        {{
            "title": "{title}",
            "summary": "Comprehensive summary in Korean",
            "key_points": ["Key point 1", "Key point 2", "..."],
            "timestamps": {{"topic": "timestamp"}} (if any significant timestamps are mentioned)
        }}
        """
        
        response = model.generate_content(
            prompt.format(
                title=video_title,
                description=description,
                transcript=full_transcript
            )
        )
        return json.loads(response.text)
        
    except Exception as e:
        logging.error(f"Error in summary generation: {str(e)}")
        return {
            "title": video_title,
            "summary": "동영상 처리 중 오류가 발생했습니다.",
            "key_points": [],
            "timestamps": {}
        }

def extract_youtube_id(url: str) -> str:
    """
    Extract YouTube video ID from various URL formats
    Returns None if URL is invalid
    """
    if "youtube.com/watch?v=" in url:
        return url.split("watch?v=")[1].split("&")[0]
    elif "youtu.be/" in url:
        return url.split("youtu.be/")[1].split("?")[0]
    return None

@app.post("/share-target-handle")
async def share_target_handle(request: Request):
    """
    Endpoint for processing shared YouTube links.
    Extracts video information and creates summaries in user's spreadsheet.
    
    Args:
        request (Request): FastAPI request object containing share data
    
    Returns:
        dict: Processing status and summary data
    """
    logging.info("Starting share-target-handle endpoint")
    try:
        # Parse request data
        request_data = await request.json()
        logging.info(f"Received request data: {request_data}")
        
        shared_data = request_data.get("shared_data", {})
        video_url = shared_data.get("text", "")
        video_title = shared_data.get("title", "")  # Get the video title
        user_id = request_data.get("user_id")
        logging.info(f"Extracted shared_data: {shared_data}, user_id: {user_id}")

        # Validate user exists
        user_ref = users_collection_ref.document(user_id)
        user = user_ref.get()
        if not user.exists:
            logging.error(f"User not found: {user_id}")
            return {"status": "error", "message": "User not found"}
            
        user_data = user.to_dict()
        logging.info(f"Found user data: {user_data}")

        # Check if spreadsheet exists
        if 'spreadsheet' not in user_data['gDrive_metadata']:
            return {"status": "error", "message": "User's spreadsheet not found"}

        # Check if shared links sheet exists
        sheet_metadata = sheet_service.spreadsheets().get(
            spreadsheetId=user_data['gDrive_metadata']['spreadsheet']
        ).execute()
        
        shared_links_sheet_exists = any(
            sheet['properties']['title'] == '공유링크' 
            for sheet in sheet_metadata.get('sheets', [])
        )

        # Create sheet if it doesn't exist
        if not shared_links_sheet_exists:
            try:
                logging.info(f"Copying template sheet to user's spreadsheet: {user_data['gDrive_metadata']['spreadsheet']}")
                
                # Copy template sheet using shared_links_ss
                copied_sheet = sheet_service.spreadsheets().sheets().copyTo(
                    spreadsheetId=template_sheet_id,
                    sheetId=shared_links_ss,
                    body={"destinationSpreadsheetId": user_data['gDrive_metadata']['spreadsheet']}
                ).execute()
                
                logging.info(f"Copied sheet to user's spreadsheet: {copied_sheet}")
                
                # Update sheet name to remove "Copy of" prefix
                rename_request = {
                    "updateSheetProperties": {
                        "properties": {
                            "sheetId": copied_sheet['sheetId'],
                            "title": "공유링크"
                        },
                        "fields": "title"
                    }
                }
                
                sheet_service.spreadsheets().batchUpdate(
                    spreadsheetId=user_data['gDrive_metadata']['spreadsheet'],
                    body={"requests": [rename_request]}
                ).execute()
                
                logging.info("Renamed sheet to '공유링크'")
                
            except Exception as e:
                logging.error(f"Failed to create shared links sheet: {e}")
                return {"status": "error", "message": f"Failed to create shared links sheet: {str(e)}"}

        # Process YouTube URL
        video_id = extract_youtube_id(video_url)
        if not video_id:
            return {"status": "error", "message": "Not a valid YouTube URL"}

        # Generate video summary
        summary_data = await process_youtube_video(video_id, video_title)

        # Prepare sheet data
        sheet_data = [[
            f'=HYPERLINK("{video_url}", "바로가기")',
            summary_data.get('title', video_title),  # Use video_title as fallback
            summary_data.get('summary', ''),
            datetime.now(kst).strftime("%Y-%m-%d %H:%M:%S")
        ]]

        # Update spreadsheet
        await update_user_sheet(
            spreadsheet_id=user_data['gDrive_metadata']['spreadsheet'],
            sheet_name="공유링크",
            data=sheet_data
        )

        return {
            "status": "success", 
            "message": "Video processed and saved successfully",
            "data": summary_data
        }

    except Exception as e:
        logging.error(f"Error in share-target-handle: {str(e)}", exc_info=True)
        return {"status": "error", "message": str(e)}

@app.post("/check-userdoc")
async def check_user_doc(user_id: str = Form(...)):
    """
    Verifies user existence and initializes transaction records if needed.
    
    Args:
        user_id (str): ID of the user to check
    
    Returns:
        dict: User existence status
    """
    user_ref = users_collection_ref.document(user_id)
    user = user_ref.get()
    if user.exists:
        return {"dbUserExists": True}
    transaction_ref = transactions_collection_ref.document(user_id)
    transaction = transaction_ref.get()
    if not transaction.exists:
        transaction_ref.set({
            "no_sms": 0,
            "no_contacts": 0,
            "no_receipts": 0,
            "no_business_cards": 0,
            "purchased_credit": {
                "image_detection": 100,
                "sync_data": 500,
            },
            "uid": user_id,
        })
    return {"dbUserExists": False}

template_sheet_id = '1i5mrmlTs5sPWtx2mBtc_f-zm1D3x21r1T_77guki8_8'

@app.post("/edit-spreadsheet")
async def edit_spreadsheet(target_sheet_id: str = Form(...)):
    """
    Endpoint for creating or updating a spreadsheet from a template.
    Replaces all sheets in the target spreadsheet with template sheets.
    
    Args:
        target_sheet_id (str): ID of the spreadsheet to edit
    
    Returns:
        dict: Status of the operation
    """
    try:
        # Retrieve the target spreadsheet's metadata
        target_spreadsheet = sheet_service.spreadsheets().get(spreadsheetId=target_sheet_id).execute()

        # Collect all sheet IDs in the target spreadsheet
        target_sheet_ids = [sheet['properties']['sheetId'] for sheet in target_spreadsheet.get('sheets', [])]

        # Prepare batch update request to delete existing sheets
        requests = []
        for sheet_id in target_sheet_ids:
            requests.append({
                "deleteSheet": {
                    "sheetId": sheet_id
                }
            })

        # Retrieve the template spreadsheet's metadata
        template_spreadsheet = sheet_service.spreadsheets().get(spreadsheetId=template_sheet_id).execute()

        # Iterate through each sheet in the template spreadsheet
        for sheet in template_spreadsheet.get('sheets', []):
            sheet_id = sheet['properties']['sheetId']

            # Use the copyTo method to copy the sheet to the target spreadsheet
            copy_request = sheet_service.spreadsheets().sheets().copyTo(
                spreadsheetId=template_sheet_id,
                sheetId=sheet_id,
                body={"destinationSpreadsheetId": target_sheet_id}
            )
            copy_response = copy_request.execute()

            # Get the copied sheet ID
            copied_sheet_id = copy_response['sheetId']

            # Rename the copied sheet to remove "Copy of"
            copied_sheet_title = sheet['properties']['title'].replace("Copy of ", "")
            rename_request = {
                "updateSheetProperties": {
                    "properties": {
                        "sheetId": copied_sheet_id,
                        "title": copied_sheet_title
                    },
                    "fields": "title"
                }
            }

            # Execute rename request
            sheet_service.spreadsheets().batchUpdate(
                spreadsheetId=target_sheet_id,
                body={"requests": [rename_request]}
            ).execute()

            logging.info(f"Sheet {sheet['properties']['title']} copied and renamed to {copied_sheet_title} in {target_sheet_id}.")

        if requests:
            # Execute batch update to delete sheets
            batch_update_request = {"requests": requests}
            sheet_service.spreadsheets().batchUpdate(
                spreadsheetId=target_sheet_id,
                body=batch_update_request
            ).execute()
            logging.info(f"Deleted existing sheets in target spreadsheet {target_sheet_id}.")
            
        logging.info(f"All sheets from template {template_sheet_id} copied to {target_sheet_id}.")
        return {"message": "Spreadsheet copied successfully"}

    except Exception as e:
        logging.error(f"Failed to copy sheets: {e}")
        return {"error": str(e)}
    
async def create_shortcut(file_id, parent_folder_id, name):
    try:
        file_metadata = {
            'name': name,
            'mimeType': 'application/vnd.google-apps.shortcut',
            'parents': [parent_folder_id],
            'shortcutDetails': {
                'targetId': file_id
            }
        }
        shortcut = drive_service.files().create(body=file_metadata, fields='id').execute()
        return shortcut.get('id')
    except Exception as e:
        logging.error(f"Failed to create shortcut: {e}")
        return None


async def search_business_info(company_name: str, address: str, spreadsheet_id: str = '1YhbIaFFcz6mrMpA9yCSri8zaPL49V6v7l1YYt0AiKss') -> dict:
    """
    Enhanced search for business information using fuzzy string matching and weighted scoring
    """
    try:
        logging.info(f"Searching business info for company: '{company_name}', address: '{address}'")
        
        # Clean and normalize input strings
        company_name = company_name.strip().lower()
        address = address.strip().lower()
        
        # Remove common company suffixes for better matching
        company_suffixes = ['(주)', '(유)', '주식회사', '유한회사']
        normalized_company = company_name
        for suffix in company_suffixes:
            normalized_company = normalized_company.replace(suffix.lower(), '').strip()
            
        # Get all rows from the spreadsheet
        range_name = "경기도!A2:P"  # Extending range to include all relevant columns
        result = sheet_service.spreadsheets().values().get(
            spreadsheetId=spreadsheet_id,
            range=range_name,
            valueRenderOption='UNFORMATTED_VALUE'
        ).execute()
        
        if not result.get('values'):
            logging.info("No data found in spreadsheet")
            return None
            
        # Process each row and calculate match scores
        matches = []
        for row in result.get('values', []):
            if len(row) < 15:  # Skip incomplete rows
                continue
                
            db_company = str(row[1]).strip().lower()  # Company name is in column B
            db_address = str(row[12]).strip().lower()  # Address is in column M
            
            # Normalize database company name
            normalized_db_company = db_company
            for suffix in company_suffixes:
                normalized_db_company = normalized_db_company.replace(suffix.lower(), '').strip()
            
            # Calculate match scores
            company_score = fuzz.token_sort_ratio(normalized_company, normalized_db_company)
            address_score = fuzz.token_sort_ratio(address, db_address)
            
            # Weight the scores (giving more weight to company name match)
            total_score = (company_score * 0.7) + (address_score * 0.3)
            
            # Only consider matches above certain thresholds
            if company_score > 60 or address_score > 80:
                matches.append({
                    'score': total_score,
                    'data': {
                        "사업자번호": str(row[2]) if len(row) > 2 else "",
                        "대표자명": str(row[11]) if len(row) > 11 else "",
                        "전화번호": str(row[13]) if len(row) > 13 else "",
                        "우편번호": str(row[14]) if len(row) > 14 else "",
                    },
                    'company_score': company_score,
                    'address_score': address_score
                })
                
                logging.info(f"Found potential match: {db_company}")
                logging.info(f"Scores - Company: {company_score}, Address: {address_score}, Total: {total_score}")
        
        if not matches:
            logging.info("No matches found meeting the threshold criteria")
            return None
            
        # Sort matches by total score and return the best match
        matches.sort(key=lambda x: x['score'], reverse=True)
        best_match = matches[0]
        
        logging.info(f"Best match found with score {best_match['score']}")
        logging.info(f"Company match score: {best_match['company_score']}")
        logging.info(f"Address match score: {best_match['address_score']}")
        
        return best_match['data']
        
    except Exception as e:
        logging.error(f"Error searching business info: {str(e)}", exc_info=True)
        return None

async def update_user_sheets_and_folders(user_data, transaction_ref, transaction_data, image_id):
    """
    Updates sheets with processed image data and organizes files in Google Drive.
    
    Args:
        user_data (dict): User's metadata and preferences
        transaction_ref: Firebase transaction reference
        transaction_data (dict): User's transaction data
        image_id (str): The ID of the processed image
    
    Returns:
        dict: Status of the update operation
    """
    try:
        parent_folders = [user_data['gDrive_metadata']['yungsoogi'], user_data['gDrive_metadata']['uploaded']]
        dataJson = await request_gpt4o_completion(image_id, transaction_data['purchased_credit']['image_detection'], use_gpt4=True)
        if dataJson is None:
            return {"error": "An error occurred while processing the image"}
        
        data = json.loads(dataJson)
        # Check if 'receipts' key exists and the length
        found_receipt_no = len(data['receipts']) if 'receipts' in data else 0
        found_business_cards_no = len(data['busi_cards']) if 'busi_cards' in data else 0
        new_folder = None
        secondary_folder = None
        # timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        timestamp = datetime.now(kst).strftime("%Y-%m-%d %H:%M:%S")
        # https://drive.google.com/file/d/1cFIA09PBqFjM8YvIz1D60p7HMGk1uBvo/view?usp=sharing
        download_image_url = f"https://drive.google.com/uc?id={image_id}&export=download"
        image_url = f"https://drive.google.com/file/d/{image_id}/view?usp=sharing"
        # 이미지를 보여주어야 할때 =image({image_url})를 사용하고 갈 수 있는 경로를 만들때는 =HYPERLINK({image_url}, '바로가기')를 사용하세요.

        # Determine the primary folder to move the image to and secondary folder for shortcut
        if found_receipt_no > 0:
            new_folder = user_data['gDrive_metadata']['receipts']
            receipts_data = convert_dicts_to_list(
                data['receipts'],
                receipt_sheet_headers,
                add_link=True, 
                image_id=image_id, 
                link_text="바로가기",
                link_position=1  # Link third
            )
            print(receipts_data)
            await update_user_sheet(user_data['gDrive_metadata']['spreadsheet'], "영수증", receipts_data)

        if found_business_cards_no > 0:
            # Enrich business card data with additional info
            for card in data['busi_cards']:
                if card.get('회사명') and card.get('주소'):
                    additional_info = await search_business_info(card['회사명'], card['주소'])
                    if additional_info:
                        # Update business number if empty
                        if not card.get('사업자번호') and additional_info.get('사업자번호'):
                            card['사업자번호'] = additional_info['사업자번호']
                        
                        # Add additional info to 기타
                        extra_info = []
                        if additional_info.get('대표자명'):
                            extra_info.append(f"대표자: {additional_info['대표자명']}")
                        if additional_info.get('우편번호'):
                            extra_info.append(f"우편번호: {additional_info['우편번호']}")
                        
                        if extra_info:
                            if '기타' in card:
                                card['기타'] += ", " + ", ".join(extra_info)
                            else:
                                card['기타'] = ", ".join(extra_info)

            if new_folder is None:
                new_folder = user_data['gDrive_metadata']['business_cards']
            else:
                secondary_folder = user_data['gDrive_metadata']['business_cards']
            business_cards_data = convert_dicts_to_list(
                data['busi_cards'],
                business_card_sheet_headers,
                add_link=True, 
                image_id=image_id, 
                link_text="명함보기",
                link_position=10
            )
            await update_user_sheet(user_data['gDrive_metadata']['spreadsheet'], "명함", business_cards_data)

        print(f"{found_receipt_no}, {found_business_cards_no}")
    
        status = await update_fb_transaction(
            transaction_ref,
            no_receipts=found_receipt_no,
            no_business_cards=found_business_cards_no
        )
    
        if status['status'] == "Success":
            if new_folder:
                await move_file_to_folder(image_id, parent_folders, [new_folder])
                if secondary_folder:
                    await create_shortcut(image_id, secondary_folder, f"Shortcut to {image_id}")
            return {"success": True}
        else:
            logging.error(f"Transaction update failed with status: {status['status']}")
            return {"error": "An error has occurred while updating the transaction"}
    except Exception as e:
        logging.error(f"An error occurred: {e}")
        # Move the image file to the error folder
        await move_file_to_folder(image_id, parent_folders, [user_data['gDrive_metadata']['error']])
        return {"error": str(e)}

@app.post("/process-image")
async def process_photo(background_tasks: BackgroundTasks, image_id: str = Form(...), user_id: str = Form(...)):
    """
    Endpoint for processing uploaded images. Handles receipt and business card detection.
    
    Args:
        background_tasks: FastAPI background tasks handler
        image_id (str): The ID of the uploaded image in Google Drive
        user_id (str): The ID of the user making the request
    
    Returns:
        dict: Processing status and any errors
    """
    transaction_ref = transactions_collection_ref.document(user_id)
    transaction = transaction_ref.get()
    
    if not transaction.exists:
        transaction_ref.set({
            "no_sms": 0,
            "no_contacts": 0,
            "no_receipts": 0,
            "no_business_cards": 0,
            "purchased_credit": {
                "image_detection": 100,
                "sync_data": 500,
            }
        })
        transaction = transaction_ref.get()
    
    transaction_data = transaction.to_dict()
    
    if transaction_data['no_receipts'] + transaction_data['no_business_cards'] >= transaction_data['purchased_credit']['image_detection']:
        return {"error": "You have reached the limit of the number of items you can process. Please upgrade your plan."}
    
    user_ref = users_collection_ref.document(user_id)
    user = user_ref.get()
    user_data = user.to_dict()
    
    background_tasks.add_task(update_user_sheets_and_folders, user_data, transaction_ref, transaction_data, image_id)
    return {"success": True}
    
@app.post("/sync-data")
async def process_photo(data_json: str = Form(...), user_id: str = Form(...)):
    """
    Endpoint for synchronizing contacts and SMS data.
    Handles creation of new SMS sheets and updates existing ones.
    
    Args:
        data_json (str): JSON string containing sync data
        user_id (str): ID of the user making the request
    
    Returns:
        dict: Sync operation status
    """
    # print(data_json)
    data: Dict[str, Any] = json.loads(data_json)
    transaction_ref = transactions_collection_ref.document(user_id)
    transaction = transaction_ref.get()
    
    if not transaction.exists:
        transaction_ref.set({
            "no_sms": 0,
            "no_contacts": 0,
            "no_receipts": 0,
            "no_business_cards": 0,
            "purchased_credit": {
                "image_detection": 100,
                "sync_data": 500,
            }
        })
        transaction = transaction_ref.get()
        
    transaction_data = transaction.to_dict()

    if transaction_data['no_contacts'] + transaction_data['no_sms'] >= transaction_data['purchased_credit']['sync_data']:
        return {"error": "You have reached the limit of the number of items you can process. Please upgrade your plan."}

    user_ref = users_collection_ref.document(user_id)
    user = user_ref.get()
    user_data = user.to_dict()

    try:
        if data['isContacts']:
            await update_user_sheet(user_data['gDrive_metadata']['spreadsheet'], "연락처", data['data'], True)
            transaction_data['no_contacts'] += len(data['data'])
            transaction_ref.update({'no_contacts': transaction_data['no_contacts']})
            
        else:
            if 'sms_ids' not in user_data:
                print('No sms_ids')
                user_data['sms_ids'] = {}

            existing_phones = set(user_data.get('sms_ids', {}).keys())
            new_phones = [phone['address'] for phone in data['data'] if phone['address'] not in existing_phones]
            
            print(f"Existing phones: {existing_phones}")
            
            for phone_address in new_phones:
                new_sheet_id = await copy_sms_sheet(phone_address, user_data['gDrive_metadata']['spreadsheet'])
                print(f"New sheet id: {new_sheet_id}")
                user_data['sms_ids'][phone_address] = {'sheet_id': new_sheet_id, 'last_updated': "1970-01-01T00:00:00.000"}
                print(user_data['sms_ids'])
                
            updated_sms_count = 0
            for phone in data['data']:
                last_updated = user_data['sms_ids'][phone['address']]['last_updated']
                print(f"Last updated: {last_updated}")
                
                # Convert last_updated to datetime for comparison
                last_updated_dt = datetime.fromisoformat(last_updated)
                update_message = [sms for sms in phone['sms'] if datetime.fromisoformat(sms[0]) > last_updated_dt]
                print(f"Update message: {update_message}")
                
                if not update_message:
                    continue
                
                sheet_id = user_data['sms_ids'][phone['address']]['sheet_id']
                print(f"Sheet ID: {sheet_id}")

                await update_user_sheet(user_data['gDrive_metadata']['spreadsheet'], sheet_id, 'SMS ' + phone['address'], update_message, False)
                
                user_data['sms_ids'][phone['address']]['last_updated'] = phone['sms'][0][0]
                updated_sms_count += len(update_message)

                
            user_ref.update({'sms_ids': user_data['sms_ids']})
            transaction_data['no_sms'] += updated_sms_count
            transaction_ref.update({'no_sms': transaction_data['no_sms']})

        
        return {"success": True}
    except Exception as e:
        logging.error(f"An error occurred: {e}")
        return {"error": str(e)}