Culturepedia

It is a platform that allows you to conveniently explore cultural content containing performances of various genres, provides personalized recommendations by analyzing user tastes, and helps you make better choices through in-depth reviews.

• Planning and Documentation: 2024.09.23 ~ 2024.09.27
• Development: 2024.09.27 ~ 2024.10.23

• Backend: Python, Django REST Framework (DRF), LLM
• Frontend: HTML, CSS, JS, Bootstrap
• Database: SQLite, PostgreSQL
• Version Control: Github, AWS
• IDE: VSCode
• APIs Used: KOPIS API (Korea Performance Information System), OpenAI API
• Other Tools and Libraries:
  • APScheduler==3.10.4
  • asgiref==3.8.1
  • certifi==2024.8.30
  • charset-normalizer==3.3.2
  • Django==4.2
  • django-apscheduler==0.7.0
  • django-seed==0.3.1
  • djangorestframework==3.15.2
  • djangorestframework-simplejwt==5.3.1
  • ElementTreeFactory==1.0
  • Faker==30.1.0
  • idna==3.10
  • pillow==10.4.0
  • psycopg2==2.9.9
  • PyJWT==2.9.0
  • python-dateutil==2.9.0.post0
  • pytz==2024.2
  • requests==2.32.3
  • six==1.16.0
  • sqlparse==0.5.1
  • toposort==1.10
  • typing_extensions==4.12.2
  • tzdata==2024.2
  • tzlocal==5.2
  • urllib3==2.2.3
  • xmltodict==0.13.0

• Sign up

  • Users can create an account by providing mandatory information (Email, Password, Username) and optional details (Gender, Birthday). After signing up, users can access features like favorites, writing reviews, and receiving performance recommendations.

• Log in

  • Users can log in using their email and password. Authentication is handled via JWT (JSON Web Token).

• Log Out

  • Users can log out, and their JWT token will be invalidated.

• Profile Edit

  • Users can update their username, password, gender, and birthday.

• Account Deletion

  • Users can delete their account by entering their password. This removes their data from the system.

• Profile View: Users can view their profile details (email, username, gender, birthday) and access their performance viewing history.

  • Viewing History
    • Users can see the performances they have reviewed.
  • Favorites
    • Users can view a list of performances they have marked as favorites.

  

• Browse Performances
  • Users can browse performance rankings by category, sort them by sales or newest, and filter performances by region. Clicking on a performance poster or title takes users to the performance’s detail page.
• Search Performances
  • Users can search performances by keywords (title, actor, venue, production company).
• Performance Details
  • Users can view detailed information about a performance, including the title, start and end dates, venue, cast, crew, runtime, age restrictions, producers, pricing, poster, synopsis, genre, and status.
• Write Reviews
  • Users can write reviews for performances they have attended, including a star rating out of 5.
• Edit Reviews
  • Users can edit reviews they have written.
• Delete Reviews
  • Users can delete reviews they have written.
• Add to Favorites
  • Users can add performances to their favorites list.
• Remove from Favorites
  • Users can remove performances from their favorites list.
• Performance Recommendations
  • Based on the user’s reviews (ratings of 3 stars or higher), favorites, and search tags, the system recommends personalized performances. If there’s insufficient user data, users can manually choose categories, characteristics, moods, and regions for recommendations.

• Sign Up (accounts)
  • Allows new users to register.
• Log In (signin)
  • Users can log in using their email and password.
• Log Out (signout)
  • Users can log out and terminate their session.
• Profile Edit (modify)
  • Users can update their profile information.
• Account Deletion (delete)
  • Users can delete their account.
• Profile View (profile)
  • Displays the user's profile.
• Performance List View (performances)
  • All users can view performance data.
• Performance Search (search)
  • Performances can be searched by title, actor, venue, or production company.
• Performance Details View (detail)
  • Displays detailed information about a performance.
• Performance Reviews (create/edit/delete)
  • Users can write, edit, and delete reviews for performances.
• Add/Remove Favorites
  • Users can add performances to or remove them from their favorites list.

• UI/UX Design
  • Build the user interface (UI) using Bootstrap.
• Sign Up/Log In/Log Out Pages
  • Pages for managing user accounts.
• Performance Review and Favorites Pages
  • Provide UI for users to write reviews and manage their favorite performances.
• Profile Page
  • Implement a page where users can view and edit their profile information.

• Review Table
  • Stores review data for each performance.
• Favorites Table
  • Stores a list of performances favorited by users.
• User Table
  • Stores user information and profiles.

• KOPIS API
  • Provides performance lists and detailed performance information.
• OpenAI API
  • Utilizes LLM to offer additional AI-powered features.

https://www.figma.com/design/bc7ezCAoLV0OBzk35nNMSQ/Culturepedia-WireFrame?m=auto&t=M5PEKIqkya6poh6A-1

https://documenter.getpostman.com/view/38012126/2sAY4rE4t4

CULTUREPEDIA
│
├── accounts/              # Handles user authentication and permissions
├── culturepedia/          # Project configuration files (e.g., settings.py)
├── performances/          # Contains features for performance browsing, searching, and recommendations
│   └── bots.py            # Handles interactions with the OpenAI API
│   └── tasks.py           # Scheduler for saving data from the KOPIS API to the local database
├── static/                # Static files (CSS, images, JavaScript, HTML) served to the client for faster loading
│   ├── css/               # Stylesheets that define the look and feel of the website
│   ├── img/               # Performance-related images used across the site
│   ├── js/                # Client-side scripts that handle dynamic interactions on the web pages
├── facility.py            # Stores detailed data about performance venues from the KOPIS API
├── kopis_api_detail.py    # Stores detailed data about specific performances from the KOPIS API
├── kopis_api.py           # Stores performance list data from the KOPIS API
├── db.sqlite3             # SQLite database file
├── .gitignore             # Specifies which files to ignore in version control
├── manage.py              # Django project management script
├── requirements.txt       # List of required Python packages

API DB vs Server DB : Storing large amounts of performance data locally can put a strain on the server, so it is recommended to query the API DB as it updates in real time.

• Due to the complicated code and traffic problems, it has been changed to store it in a local DB to inquire and retrieve it.

def resizing_images(url_list, quality=85):      # quaility: 70 ~ 90, default: 85
    
    img_folder = os.path.join(settings.STATICFILES_DIRS[0], "img")
    if not os.path.exists(img_folder):
        os.makedirs(img_folder)
    
    for url in url_list:
        try:
            # Download an image from the URL
            response = requests.get(url)
            response.raise_for_status()  # HTTP 요청 에러 체크

            # Image File Name Extract
            filename = os.path.basename(url)

            # Create a folder by performance.
            folder_name = f"{img_folder}/{filename.split('_')[1]}"
            if not os.path.exists(folder_name):
                os.mkdir(folder_name)

            # Open the image in memory.
            im = Image.open(BytesIO(response.content))
            
            # Save Image
            im.save(os.path.join(folder_name, filename), quality=quality)
            
        except Exception as e:
            pass

Issue	Cause	Solution
An AI was created to extract hashtags using given field values, but the hashtag_list contains unnecessary content in addition to the hashtags.	The gpt-4o-mini model outputs unnecessary messages, causing incorrect behavior.	1. Use the gpt-4o model instead of gpt-4o-mini. Result: When entering the same query, no unnecessary messages are output. Drawback: Relatively more expensive. 2. Continue using gpt-4o-mini but modify the query (e.g., "Please provide hashtags." ⇒ "Just provide the hashtags."). Result: Only hashtags are output without any unnecessary messages.

1. Use jobstores to store scheduler logs in the DB and prevent execution if a scheduler with the same ID already exists.
   ⇒ Duplicate execution occurs, with two IDs being created (failed).

scheduler.add_jobstore(jobstores.DjangoJobStore(), "default")  # Save to Django DB

2. Set a scheduler execution status flag as a global variable to check execution status and prevent duplicate execution.
   ⇒ Duplicate execution still occurs (failed).

scheduler_running = False  # Adding a Global Variable Flag
start_scheduler():
	scheduler_running = True

3. When starting the server using subprocess, two processes (1. performance lookup, reviews, recommendations, etc. / 2. API DB storage, etc.) are running simultaneously, which causes the start_scheduler inside ready to be executed twice.

Add code to the ready method to prevent duplicate execution.

• Before

def ready(self):
	start_scheduler()

• After

def ready(self):
	if os.environ.get('RUN_MAIN', None) is not None:    # Run scheduler if RUN_MAIN is 'true'
		print(' RUN_MAIN :', os.environ.get('RUN_MAIN', None))
		from .tasks import start_scheduler
		start_scheduler()