Logsday (rs)

This is my version of the famous Logsday website (which is currently not publically available, which is why this project aims to be publically available).

You will be able to upload a devlog exactly once a week.

Structure

• To run the server, cargo run
• Entry point for the server is src/main.rs
• Askama's templates (#[template(path = "index.html")]) live in ./templates directory
  • Note that Askama's templates are not actually html files. They're.. templates. With {{ ... }} being replaced with stuff before being sent to the client.
• Axum's Router uses nest_service("/static", ServeDir::new("public")) to map requests to /static to take from ./public directory. That way, any requests to CSS or JS will take from there.
• The database is ./sqlite.db. Needs to be created manually. Before running you might need to set an env variable DATABASE_URL="sqlite:sqlite.db" or DATABASE_URL="~/logsday/sqlite.db"

Paths

• / => templates/landing.html- home, undecided
• /signup => templates/signup.html
• /login => templates/login.html
• /logout - logout and redirect to /
• /mdguide => templates/mdguide.html
• /u => templates/viewuser.html - your profile; redirect to /login when not logged in
• /u/{username} => templates/viewuser.html - user's profile + list of their projects; when your page, give extra options
• /u/{username}/{project_slug} => templates/viewproject.html - project page + list of logs; when your project, give extra options
• /u/{username}/{project_slug}/{log_number} => templates/viewlog.html - log page; when your log, give extra options
• /new/project => templates/newproject.html - create a new project; redirect to /login when not logged in
• /new/log/{project_slug} => templates/newlog.html - create a new log for the specified project; redirect to /login when not logged in
• /new/media/{project_slug} - return a json list of existing files on GET; download the file and return a json of the uploaded file on POST; new log
• /new/media/{project_slug}/{log_number} - return a json list of existing files on GET; download the file and return a json of the uploaded file on POST; specified log
• /del/project/{project_slug} - delete the project; redirect to /login when not logged in
• /del/log/{project_slug}/{log_number} - delete the log for the specified project; redirect to /login when not logged in
• /del/media/{project_slug}/new/{file_name} - delete the given file from the newlog
• /del/media/{project_slug}/{log_number}/{file_name} - delete the given file from the specified log
• /uploads/{username}/{project_slug}/{log_number}/{filename} => uploads/{username}/{project_slug}/{log_number}/{filename} - uploaded files for every log go here
• /static/* => static/* - any static files that may be retrieved by the client (favicon, css/js, etc)
• /bits/navuser => templates/bits/login.html/nav_user.html - get the navbar login/signup or the logout/time until logsday; bits uris are htmx helpers

Tech Stack

• Database:
  • SQLite
• Web:
  • HTMX
  • marked.js live markdown preview
  • highlight.js highlighting code blocks
• Rust Crates:
  • axum for web server basics
  • aksama for template rendering
  • argon2 for password hashing
  • pulldown-cmark for markdown to html rendering
  • sqlx for interacting with sqlite db
  • infer for scanning magic bytes of files
  • image for converting image formats
  • scraper for looking through html files for linked/embedded files
  • tokio_cron_scheduler for cron jobs from Rust, for now
  • tower_sessions for easy sessions
  • axum_typed_multipart for convenience, for now

Notes

• A lot of research for this project was done with Gemini. I don't think it's a big deal, but thought I'd put it here. It's just very convenient. And I don't need the nuiance of deep-diving into the topics yet. If capstone taught me something - I shouldn't be as afraid to just do something with a moderate amount of planning. Overplanning can be overwhelming and unproductive (to me).
• POST responses should be either: Error message string (will be shown with htmx) or an HX_Redirect that redirects to the new page. Any exceptions will be noted here and in the code.
  • Exception 1: /new/media/... should return a json with {error} or {filename, filesize, filepath}
• Unix epoch starts on Thu, Jan 1, 1970. For an 8-day week, Unix epoch starts on Mon, Jan 1, 1970. In code, all weekdays are 0-indexed (Mon = 0, Tue = 1, etc).
• You will not be able to private a project/log. You can unlist it, but not private.

Comprehensive .md rendering rules

• CommonMark spec
  • server side uses pulldown_cmark with the following options enabled
• Additional options:
  • ENABLE_TABLES - idk the syntax for tables; may not be included later, if difficult to do
  • strikethrough
• Additional features:
  • ![](name.ext) represents a multimedia embed, depending on .ext. For supported extensions, refer to filestuff::media_type

SQLite Tables

CREATE TABLE users (
    uid INTEGER PRIMARY KEY AUTOINCREMENT,
    username TEXT NOT NULL UNIQUE,
    displayname TEXT NOT NULL,
    password TEXT NOT NULL,
    week_len INTEGER NOT NULL DEFAULT 8,
    logsday_weekday INTEGER NOT NULL DEFAULT 3, -- Logsday is between Wednesday and Thursday; Monday is 0; Sunday is 6/7
    schedule_last_changed INTEGER NOT NULL, -- when the user changed their Logsday selection last time; unix timestamp
    admin BOOLEAN NOT NULL DEFAULT FALSE,
    created_on INTEGER NOT NULL -- unix timestamp
);

CREATE TABLE projects (
    uid INTEGER PRIMARY KEY AUTOINCREMENT,
    user_uid INTEGER NOT NULL,
    title TEXT NOT NULL,
    slug TEXT NOT NULL,
    description TEXT,
    created_on INTEGER NOT NULL, -- unix timestamp

    UNIQUE(user_uid, slug)
    FOREIGN KEY (user_uid) REFERENCES users(uid) ON DELETE CASCADE
);

CREATE TABLE logs (
    uid INTEGER PRIMARY KEY AUTOINCREMENT,
    project_uid INTEGER NOT NULL,
    title TEXT NOT NULL,
    number INTEGER NOT NULL, -- this log's sequential number in the project
    created_on INTEGER NOT NULL, -- unix timestamp

    UNIQUE(project_uid, number)
    FOREIGN KEY (project_uid) REFERENCES projects(uid) ON DELETE CASCADE
);