datatrash/src/multipart.rs

use crate::{config, mime_relations};
use actix_multipart::{Field, Multipart};
use actix_web::{error, http::header::DispositionParam, Error};
use futures_util::{StreamExt, TryStreamExt};
use mime::{Mime, APPLICATION_OCTET_STREAM, TEXT_PLAIN};
use std::{cmp::min, io::ErrorKind, path::Path};
use time::{Duration, OffsetDateTime};
use tokio::{
    fs::{self, File},
    io::AsyncWriteExt,
};

const MAX_UPLOAD_DURATION: Duration = Duration::days(31);
const DEFAULT_UPLOAD_DURATION: Duration = Duration::minutes(30);

pub(crate) struct UploadConfig {
    pub original_name: Option<String>,
    pub content_type: Mime,
    pub valid_till: OffsetDateTime,
    pub delete_on_download: bool,
}

pub(crate) async fn parse_multipart(
    payload: Multipart,
    file_path: &Path,
    config: &config::Config,
) -> Result<UploadConfig, error::Error> {
    match parse_multipart_inner(payload, file_path, config).await {
        Ok(data) => Ok(data),
        Err(err) => {
            match fs::remove_file(file_path).await {
                Err(err) if err.kind() != ErrorKind::NotFound => {
                    log::error!("could not remove file {:?}", err);
                }
                _ => {}
            }
            Err(err)
        }
    }
}

pub(crate) async fn parse_multipart_inner(
    mut payload: Multipart,
    file_path: &Path,
    config: &config::Config,
) -> Result<UploadConfig, error::Error> {
    let mut original_name: Option<String> = None;
    let mut content_type: Option<Mime> = None;
    let mut keep_for_seconds: Option<String> = None;
    let mut delete_on_download = false;
    let mut password = None;
    let mut size = 0;

    while let Ok(Some(mut field)) = payload.try_next().await {
        let name = get_field_name(&field)?.to_owned();
        match name.as_str() {
            "keep_for" => {
                keep_for_seconds = Some(parse_string(&name, &mut field).await?);
            }
            "file" => {
                let (mime, uploaded_name) = get_file_metadata(&field);
                if uploaded_name.is_none() || uploaded_name.as_deref() == Some("") {
                    continue;
                }
                original_name = uploaded_name;
                let first_bytes;
                (size, first_bytes) = create_file(file_path, field, config.max_file_size).await?;
                content_type = Some(
                    mime.filter(|mime| *mime != APPLICATION_OCTET_STREAM)
                        .map(mime_relations::get_alias)
                        .or_else(|| get_content_type(&first_bytes))
                        .unwrap_or(APPLICATION_OCTET_STREAM),
                );
            }
            "text" => {
                if original_name.is_some() {
                    continue;
                }
                let first_bytes;
                (size, first_bytes) = create_file(file_path, field, config.max_file_size).await?;
                content_type = Some(get_content_type(&first_bytes).unwrap_or(TEXT_PLAIN));
            }
            "delete_on_download" => {
                delete_on_download = parse_string(&name, &mut field).await? != "false";
            }
            "password" => {
                password = Some(parse_string(&name, &mut field).await?);
            }
            _ => {}
        };
    }

    let content_type =
        content_type.ok_or_else(|| error::ErrorBadRequest("no content type found"))?;
    let keep_for = keep_for_seconds
        .map(|k| k.parse())
        .transpose()
        .map_err(|e| error::ErrorBadRequest(format!("field keep_for is not a number: {e}")))?
        .map_or(DEFAULT_UPLOAD_DURATION, Duration::seconds);
    let valid_till = OffsetDateTime::now_utc() + keep_for;

    let upload_config = UploadConfig {
        original_name,
        content_type,
        valid_till,
        delete_on_download,
    };

    check_requirements(&upload_config, size, &password, &keep_for, config)?;

    Ok(upload_config)
}

fn check_requirements(
    upload_config: &UploadConfig,
    size: u64,
    password: &Option<String>,
    keep_for: &Duration,
    config: &config::Config,
) -> Result<(), error::Error> {
    if let Some(original_name) = upload_config.original_name.as_ref() {
        if original_name.len() > 255 {
            return Err(error::ErrorBadRequest("filename is too long"));
        }
    }

    if *keep_for > MAX_UPLOAD_DURATION {
        return Err(error::ErrorBadRequest(format!(
            "maximum allowed validity is {MAX_UPLOAD_DURATION}, but you specified {keep_for}"
        )));
    }

    if let Some(no_auth_limits) = &config.no_auth_limits {
        let requires_auth = *keep_for > no_auth_limits.max_time
            || *keep_for > no_auth_limits.large_file_max_time
                && size > no_auth_limits.large_file_size;
        // hIGh sECUriTy paSsWoRD CHEck
        if requires_auth && password.as_ref() != Some(&no_auth_limits.auth_password) {
            return Err(error::ErrorBadRequest(
                "upload requires authentication, but authentication was incorrect",
            ));
        }
    }

    Ok(())
}

fn get_field_name(field: &Field) -> Result<&str, error::Error> {
    Ok(field
        .content_disposition()
        .get_name()
        .ok_or(error::ParseError::Incomplete)?)
}

async fn parse_string(
    name: &str,
    field: &mut actix_multipart::Field,
) -> Result<String, error::Error> {
    let data = read_content(field).await?;
    String::from_utf8(data)
        .map_err(|_| error::ErrorBadRequest(format!("could not parse field {name} as utf-8")))
}

async fn read_content(field: &mut actix_multipart::Field) -> Result<Vec<u8>, error::Error> {
    let mut data = Vec::new();
    while let Some(chunk) = field.try_next().await.map_err(error::ErrorBadRequest)? {
        data.extend(chunk);
    }
    Ok(data)
}

async fn create_file(
    filename: &Path,
    field: Field,
    max_file_size: Option<u64>,
) -> Result<(u64, Vec<u8>), Error> {
    let mut file = File::create(&filename).await.map_err(|file_err| {
        log::error!("could not create file {:?}", file_err);
        error::ErrorInternalServerError("could not create file")
    })?;
    write_to_file(&mut file, field, max_file_size).await
}

async fn write_to_file(
    file: &mut File,
    mut field: Field,
    max_size: Option<u64>,
) -> Result<(u64, Vec<u8>), error::Error> {
    let mut first_bytes = Vec::with_capacity(2048);
    let mut written_bytes: u64 = 0;
    while let Some(chunk) = field.next().await {
        let chunk = chunk.map_err(error::ErrorBadRequest)?;
        written_bytes += chunk.len() as u64;
        validate_max_size(written_bytes, max_size)?;

        let remaining_first_bytes = min(2048 - first_bytes.len(), chunk.len());
        first_bytes.extend_from_slice(&chunk[..remaining_first_bytes]);

        file.write_all(&chunk).await.map_err(|write_err| {
            log::error!("could not write file {:?}", write_err);
            error::ErrorInternalServerError("could not write file")
        })?;
    }
    Ok((written_bytes, first_bytes))
}

fn validate_max_size(written_bytes: u64, max_size: Option<u64>) -> Result<(), Error> {
    if let Some(max_size) = max_size {
        if written_bytes > max_size {
            return Err(error::ErrorPayloadTooLarge(format!(
                "exceeded maximum file size of {max_size} bytes"
            )));
        }
    }
    Ok(())
}

fn get_file_metadata(field: &actix_multipart::Field) -> (Option<Mime>, Option<String>) {
    let mime = field.content_type().cloned();
    let filename = field
        .content_disposition()
        .parameters
        .iter()
        .find_map(|param| match param {
            DispositionParam::Filename(filename) => Some(filename.clone()),
            _ => None,
        });
    (mime, filename)
}

fn get_content_type(bytes: &[u8]) -> Option<Mime> {
    tree_magic_mini::from_u8(bytes).parse().ok()
}