datatrash/src/main.rs

mod deleter;
mod file_kind;
mod multipart;

use actix_files::{Files, NamedFile};
use actix_multipart::Multipart;
use actix_web::{
    error,
    http::header::{Charset, ContentDisposition, DispositionParam, DispositionType, ExtendedValue},
    middleware, web, App, Error, HttpRequest, HttpResponse, HttpServer,
};
use async_std::{
    channel::{self, Sender},
    fs,
    path::{Path, PathBuf},
    task,
};
use env_logger::Env;
use file_kind::FileKind;
use futures::TryStreamExt;
use mime::Mime;
use multipart::UploadConfig;
use rand::prelude::SliceRandom;
use sqlx::{
    postgres::{PgPool, PgPoolOptions, PgRow},
    Row,
};
use std::env;

const INDEX_HTML: &str = include_str!("../template/index.html");
const UPLOAD_HTML: &str = include_str!("../template/upload.html");
const VIEW_HTML: &str = include_str!("../template/view.html");

const ID_CHARS: &[char] = &[
    'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'j', 'k', 'm', 'n', 'p', 'q', 'r', 's', 't', 'u', 'v',
    'w', 'x', 'y', 'z', '1', '2', '3', '4', '5', '6', '7', '8', '9',
];

async fn index(req: web::HttpRequest) -> Result<HttpResponse, Error> {
    let upload_url = format!("{}/upload", get_host_url(&req));
    let index_html = INDEX_HTML.replace("{upload_url}", upload_url.as_str());
    Ok(HttpResponse::Ok()
        .content_type("text/html")
        .body(index_html))
}

// multipart data
// required: either 'file' or 'text'
// optional: 'keep_for' default to 30 minutes
async fn upload(
    req: web::HttpRequest,
    payload: Multipart,
    db: web::Data<PgPool>,
    expiry_watch_sender: web::Data<Sender<()>>,
    config: web::Data<Config>,
) -> Result<HttpResponse, Error> {
    let file_id = gen_file_id();
    let mut filename = config.files_dir.clone();
    filename.push(&file_id);

    let parsed_multipart =
        multipart::parse_multipart(payload, &file_id, &filename, config.max_file_size).await;
    let UploadConfig {
        original_name,
        valid_till,
        kind,
        delete_on_download,
    } = match parsed_multipart {
        Ok(data) => data,
        Err(err) => {
            if filename.exists().await {
                fs::remove_file(filename).await.map_err(|_| {
                    error::ErrorInternalServerError(
                        "could not parse multipart; could not remove file",
                    )
                })?;
            }
            return Err(err);
        }
    };

    let db_insert = sqlx::query(
        "INSERT INTO Files (file_id, file_name, valid_till, kind, delete_on_download) \
        VALUES ($1, $2, $3, $4, $5)",
    )
    .bind(&file_id)
    .bind(&original_name)
    .bind(valid_till.naive_local())
    .bind(kind.to_string())
    .bind(delete_on_download)
    .execute(db.as_ref())
    .await;
    if db_insert.is_err() {
        fs::remove_file(filename).await.map_err(|_| {
            error::ErrorInternalServerError(
                "could not insert file into database; could not remove file",
            )
        })?;
        return Err(error::ErrorInternalServerError(
            "could not insert file into database",
        ));
    }

    log::info!(
        "{} create new file {} (valid_till: {}, kind: {}, delete_on_download: {})",
        req.connection_info().realip_remote_addr().unwrap_or("-"),
        file_id,
        valid_till,
        kind,
        delete_on_download
    );

    expiry_watch_sender.send(()).await.unwrap();

    let redirect = if kind == FileKind::BINARY {
        let encoded_name = urlencoding::encode(&original_name);
        format!("/upload/{}/{}", file_id, encoded_name)
    } else {
        format!("/upload/{}", file_id)
    };

    let url = get_file_url(&req, &file_id, Some(&original_name));
    Ok(HttpResponse::SeeOther()
        .header("location", redirect)
        .body(format!("{}\n", url)))
}

fn gen_file_id() -> String {
    let mut rng = rand::thread_rng();
    let mut id = String::with_capacity(5);
    for _ in 0..5 {
        id.push(*ID_CHARS.choose(&mut rng).expect("ID_CHARS is not empty"));
    }
    id
}

fn get_host_url(req: &web::HttpRequest) -> String {
    let conn = req.connection_info();
    format!("{}://{}", conn.scheme(), conn.host())
}

fn get_file_url(req: &web::HttpRequest, id: &str, name: Option<&str>) -> String {
    if let Some(name) = name {
        let encoded_name = urlencoding::encode(name);
        format!("{}/{}/{}", get_host_url(req), id, encoded_name)
    } else {
        format!("{}/{}", get_host_url(req), id)
    }
}

async fn uploaded(req: web::HttpRequest) -> Result<HttpResponse, Error> {
    let id = req.match_info().query("id");
    let name = req.match_info().get("name");
    let url = get_file_url(&req, id, name);
    let upload_html = UPLOAD_HTML.replace("{url}", url.as_str());
    Ok(HttpResponse::Ok()
        .content_type("text/html")
        .body(upload_html))
}

async fn download(
    req: HttpRequest,
    db: web::Data<PgPool>,
    config: web::Data<Config>,
) -> Result<HttpResponse, Error> {
    let id = req.match_info().query("id");
    let mut rows =
        sqlx::query("SELECT file_id, file_name, delete_on_download from files WHERE file_id = $1")
            .bind(id)
            .fetch(db.as_ref());
    let row: PgRow = rows
        .try_next()
        .await
        .map_err(|_| error::ErrorInternalServerError("could not run select statement"))?
        .ok_or_else(|| error::ErrorNotFound("file does not exist or has expired"))?;

    let file_id: String = row.get("file_id");
    let file_name: String = row.get("file_name");
    let delete_on_download: bool = row.get("delete_on_download");
    let mut path = config.files_dir.clone();
    path.push(&file_id);

    let download = req.query_string().contains("dl");
    let (content_type, mut content_disposition) = get_content_types(&path, &file_name);
    let response = if content_type.type_() == mime::TEXT && !download {
        let content = fs::read_to_string(path).await.map_err(|_| {
            error::ErrorInternalServerError("this file should be here but could not be found")
        })?;
        let encoded = htmlescape::encode_minimal(&content);
        let view_html = VIEW_HTML.replace("{text}", &encoded);
        let response = HttpResponse::Ok().content_type("text/html").body(view_html);
        Ok(response)
    } else {
        if download {
            content_disposition.disposition = DispositionType::Attachment;
        }
        let file = NamedFile::open(path)
            .map_err(|_| {
                error::ErrorInternalServerError("this file should be here but could not be found")
            })?
            .set_content_type(content_type)
            .set_content_disposition(content_disposition);
        file.into_response(&req)
    };
    if delete_on_download {
        deleter::delete_by_id(&db, &file_id, &config.files_dir)
            .await
            .map_err(|_| error::ErrorInternalServerError("could not delete file"))?;
    }
    response
}

fn get_content_types(path: &Path, filename: &str) -> (Mime, ContentDisposition) {
    let std_path = std::path::Path::new(path.as_os_str());
    let ct = tree_magic_mini::from_filepath(std_path)
        .unwrap_or("application/octet-stream")
        .parse::<Mime>()
        .expect("tree_magic_mini should not produce invalid mime");

    let disposition = match ct.type_() {
        mime::IMAGE | mime::TEXT | mime::AUDIO | mime::VIDEO => DispositionType::Inline,
        _ => DispositionType::Attachment,
    };

    let cd = ContentDisposition {
        disposition,
        parameters: get_disposition_params(filename),
    };

    (ct, cd)
}

fn get_disposition_params(filename: &str) -> Vec<DispositionParam> {
    let mut parameters = vec![DispositionParam::Filename(filename.to_owned())];
    if !filename.is_ascii() {
        parameters.push(DispositionParam::FilenameExt(ExtendedValue {
            charset: Charset::Ext(String::from("UTF-8")),
            language_tag: None,
            value: filename.to_owned().into_bytes(),
        }))
    }
    parameters
}

async fn not_found() -> Result<HttpResponse, Error> {
    Ok(HttpResponse::NotFound()
        .content_type("text/plain")
        .body("not found"))
}

fn get_db_url() -> String {
    if let Ok(database_url) = env::var("DATABASE_URL") {
        return database_url;
    }

    let auth = if let Ok(user) = env::var("DATABASE_USER") {
        if let Ok(pass) = env::var("DATABASE_PASS") {
            format!("{}:{}@", user, pass)
        } else {
            format!("{}@", user)
        }
    } else {
        String::new()
    };

    format!(
        "postgresql://{auth}{host}/{name}",
        auth = auth,
        host = env::var("DATABASE_HOST").unwrap_or_else(|_| "localhost".to_string()),
        name = env::var("DATABASE_NAME").unwrap_or_else(|_| "datatrash".to_string())
    )
}

async fn setup_db() -> PgPool {
    let conn_url = &get_db_url();
    log::info!("Using Connection string {}", conn_url);

    let pool = PgPoolOptions::new()
        .max_connections(5)
        .connect_timeout(std::time::Duration::from_secs(5))
        .connect(conn_url)
        .await
        .expect("could not create db pool");

    for query in include_str!("../init-db.sql").split_inclusive(";") {
        sqlx::query(query)
            .execute(&pool)
            .await
            .expect("could not initialize database schema");
    }

    pool
}

#[derive(Clone)]
struct Config {
    files_dir: PathBuf,
    max_file_size: Option<u64>,
}

#[actix_web::main]
async fn main() -> std::io::Result<()> {
    env_logger::Builder::from_env(Env::default().default_filter_or("info,sqlx=warn")).init();

    let pool: PgPool = setup_db().await;
    let max_file_size = env::var("UPLOAD_MAX_BYTES")
        .ok()
        .and_then(|variable| variable.parse().ok())
        .unwrap_or(8 * 1024 * 1024);
    let max_file_size = if max_file_size == 0 {
        None
    } else {
        Some(max_file_size)
    };
    let config = Config {
        files_dir: PathBuf::from(env::var("FILES_DIR").unwrap_or_else(|_| "./files".to_owned())),
        max_file_size,
    };
    fs::create_dir_all(&config.files_dir)
        .await
        .expect("could not create directory for storing files");
    let (sender, receiver) = channel::bounded(8);

    log::info!("omnomnom");

    task::spawn(deleter::delete_old_files(
        receiver,
        pool.clone(),
        config.files_dir.clone(),
    ));

    let db = web::Data::new(pool);
    let expiry_watch_sender = web::Data::new(sender);
    let bind_address = env::var("BIND_ADDRESS").unwrap_or_else(|_| "0.0.0.0:8000".to_owned());

    HttpServer::new({
        move || {
            App::new()
                .wrap(middleware::Logger::new(r#"%{r}a "%r" =%s %bbytes %Tsec"#))
                .app_data(db.clone())
                .app_data(expiry_watch_sender.clone())
                .data(config.clone())
                .service(web::resource("/").route(web::get().to(index)))
                .service(web::resource("/upload").route(web::post().to(upload)))
                .service(
                    web::resource(["/upload/{id}", "/upload/{id}/{name}"])
                        .route(web::get().to(uploaded)),
                )
                .service(Files::new("/static", "static").disable_content_disposition())
                .service(
                    web::resource(["/{id:[a-z0-9]{5}}", "/{id:[a-z0-9]{5}}/{name}"])
                        .route(web::get().to(download)),
                )
                .default_service(web::route().to(not_found))
        }
    })
    .bind(bind_address)?
    .run()
    .await
}