path: root/src/download/mod.rs

// yt - A fully featured command line YouTube client
//
// Copyright (C) 2024 Benedikt Peetz <benedikt.peetz@b-peetz.de>
// SPDX-License-Identifier: GPL-3.0-or-later
//
// This file is part of Yt.
//
// You should have received a copy of the License along with this program.
// If not, see <https://www.gnu.org/licenses/gpl-3.0.txt>.

use std::{collections::HashMap, str::FromStr, sync::Arc, time::Duration};

use crate::{
    app::App,
    download::download_options::download_opts,
    storage::video_database::{
        downloader::{get_next_uncached_video, set_video_cache_path},
        extractor_hash::ExtractorHash,
        getters::get_video_yt_dlp_opts,
        Video, YtDlpOptions,
    },
};

use anyhow::{bail, Context, Result};
use bytes::Bytes;
use futures::{future::BoxFuture, FutureExt};
use log::{debug, error, info, warn};
use tokio::{fs, task::JoinHandle, time};

mod download_options;

#[derive(Debug)]
pub struct CurrentDownload {
    task_handle: JoinHandle<Result<()>>,
    extractor_hash: ExtractorHash,
}

impl CurrentDownload {
    fn new_from_video(app: Arc<App>, video: Video) -> Self {
        let extractor_hash = video.extractor_hash.clone();

        let task_handle = tokio::spawn(async move {
            Downloader::actually_cache_video(&app, &video)
                .await
                .with_context(|| format!("Failed to cache video: '{}'", video.title))?;
            Ok(())
        });

        Self {
            task_handle,
            extractor_hash,
        }
    }
}

pub struct Downloader {
    current_download: Option<CurrentDownload>,
    video_size_cache: HashMap<ExtractorHash, u64>,
}

impl Downloader {
    pub fn new() -> Self {
        Self {
            current_download: None,
            video_size_cache: HashMap::new(),
        }
    }

    /// The entry point to the Downloader.
    /// This Downloader will periodically check if the database has changed, and then also
    /// change which videos it downloads.
    /// This will run, until the database doesn't contain any watchable videos
    pub async fn consume(&mut self, app: Arc<App>, max_cache_size: u64) -> Result<()> {
        while let Some(next_video) = get_next_uncached_video(&app).await? {
            let cache_allocation = Self::get_current_cache_allocation(&app).await?;
            let video_size = self.get_approx_video_size(&app, &next_video).await?;
            if cache_allocation + video_size >= max_cache_size {
                warn!(
                    "Can't download video: '{}' ({}) as it's too large for the cache ({} of {} allocated).",
                    next_video.title, Bytes::new(video_size), Bytes::new(cache_allocation), Bytes::new(max_cache_size)
                );

                // Wait and hope, that a large video is deleted from the cache.
                time::sleep(Duration::from_secs(10)).await;
                continue;
            }

            if let Some(_) = &self.current_download {
                let current_download = self.current_download.take().expect("Is Some");

                if current_download.task_handle.is_finished() {
                    current_download.task_handle.await??;
                    continue;
                }

                if next_video.extractor_hash != current_download.extractor_hash {
                    info!(
                    "Noticed, that the next video is not the video being downloaded, replacing it ('{}' vs. '{}')!",
                        next_video.extractor_hash.into_short_hash(&app).await?, current_download.extractor_hash.into_short_hash(&app).await?
                    );

                    // Replace the currently downloading video
                    current_download.task_handle.abort();

                    let new_current_download =
                        CurrentDownload::new_from_video(Arc::clone(&app), next_video);

                    self.current_download = Some(new_current_download);
                } else {
                    debug!(
                        "Currently downloading '{}'",
                        current_download
                            .extractor_hash
                            .into_short_hash(&app)
                            .await?
                    );
                    // Reset the taken value
                    self.current_download = Some(current_download);
                }
            } else {
                info!(
                    "No video is being downloaded right now, setting it to '{}'",
                    next_video.title
                );
                let new_current_download =
                    CurrentDownload::new_from_video(Arc::clone(&app), next_video);
                self.current_download = Some(new_current_download);
            }

            time::sleep(Duration::new(1, 0)).await;
        }

        info!("Finished downloading!");
        Ok(())
    }

    pub async fn get_current_cache_allocation(app: &App) -> Result<u64> {
        fn dir_size(mut dir: fs::ReadDir) -> BoxFuture<'static, Result<u64>> {
            async move {
                let mut acc = 0;
                while let Some(entry) = dir.next_entry().await? {
                    let size = match entry.metadata().await? {
                        data if data.is_dir() => {
                            let path = entry.path();
                            let read_dir = fs::read_dir(path).await?;

                            dir_size(read_dir).await?
                        }
                        data => data.len(),
                    };
                    acc += size;
                }
                Ok(acc)
            }
            .boxed()
        }

        let val = dir_size(fs::read_dir(&app.config.paths.download_dir).await?).await;
        if let Ok(val) = val {
            info!("Cache dir has a size of '{}'", Bytes::new(val));
        }
        val
    }

    async fn get_approx_video_size(&mut self, app: &App, video: &Video) -> Result<u64> {
        if let Some(value) = self.video_size_cache.get(&video.extractor_hash) {
            Ok(*value)
        } else {
            // the subtitle file size should be negligible
            let add_opts = YtDlpOptions {
                subtitle_langs: "".to_owned(),
            };
            let opts = &download_opts(&app, add_opts);

            let result = yt_dlp::extract_info(&opts, &video.url, false, true)
                .await
                .with_context(|| {
                    format!("Failed to extract video information: '{}'", video.title)
                })?;

            let size = if let Some(val) = result.filesize {
                val
            } else if let Some(val) = result.filesize_approx {
                val
            } else {
                bail!("Failed to find a filesize for video: '{}'", video.title);
            };

            assert_eq!(
                self.video_size_cache
                    .insert(video.extractor_hash.clone(), size),
                None
            );

            Ok(size)
        }
    }

    async fn actually_cache_video(app: &App, video: &Video) -> Result<()> {
        debug!("Download started: {}", &video.title);

        let addional_opts = get_video_yt_dlp_opts(&app, &video.extractor_hash).await?;

        let result = yt_dlp::download(&[video.url.clone()], &download_opts(&app, addional_opts))
            .await
            .with_context(|| format!("Failed to download video: '{}'", video.title))?;

        assert_eq!(result.len(), 1);
        let result = &result[0];

        set_video_cache_path(app, &video.extractor_hash, Some(&result)).await?;

        info!(
            "Video '{}' was downlaoded to path: {}",
            video.title,
            result.display()
        );

        Ok(())
    }
}