oolatoocs/src/bsky.rs

use crate::{config::BlueskyConfig, utils::convert_aspect_ratio, OolatoocsError};
use atrium_api::{
    app::bsky::feed::post::RecordData, com::atproto::repo::upload_blob::Output,
    types::string::Datetime, types::string::Language, types::string::RecordKey,
};
use bsky_sdk::{
    agent::config::{Config, FileStore},
    rich_text::RichText,
    BskyAgent,
};
use futures::{stream, StreamExt};
use image::ImageReader;
use log::{debug, error, warn};
use megalodon::entities::{
    attachment::{Attachment, AttachmentType},
    card::Card,
};
use regex::Regex;
use std::{error::Error, fs::exists, io::Cursor};
use webp::*;

/// Intermediary struct to deal with replies more easily
#[derive(Debug)]
pub struct BskyReply {
    pub record_uri: String,
    pub root_record_uri: String,
}

pub async fn get_session(config: &BlueskyConfig) -> Result<BskyAgent, Box<dyn Error>> {
    if exists(&config.config_path)? {
        let bluesky = BskyAgent::builder()
            .config(Config::load(&FileStore::new(&config.config_path)).await?)
            .build()
            .await?;

        if bluesky.api.com.atproto.server.get_session().await.is_ok() {
            bluesky
                .to_config()
                .await
                .save(&FileStore::new(&config.config_path))
                .await?;
            return Ok(bluesky);
        }
    }

    let bluesky = BskyAgent::builder().build().await?;
    bluesky.login(&config.handle, &config.password).await?;
    bluesky
        .to_config()
        .await
        .save(&FileStore::new(&config.config_path))
        .await?;

    Ok(bluesky)
}

pub async fn build_post_record(
    config: &BlueskyConfig,
    text: &str,
    language: &Option<String>,
    embed: Option<atrium_api::types::Union<atrium_api::app::bsky::feed::post::RecordEmbedRefs>>,
    reply_to: &Option<BskyReply>,
) -> Result<RecordData, Box<dyn Error>> {
    let mut rt = RichText::new_with_detect_facets(text).await?;

    let insert_chars = "…";

    let re = Regex::new(r#"(https?://)(www\.)?(\S{1,26})(\S*)"#).unwrap();

    while let Some(found) = re.captures(&rt.text.clone()) {
        if let Some(group) = found.get(4) {
            if !group.is_empty() {
                rt.insert(group.start(), insert_chars);
                rt.delete(
                    group.start() + insert_chars.len(),
                    group.start() + insert_chars.len() + group.len(),
                );
            }
        }
        if let Some(group) = found.get(1) {
            let www: usize = found.get(2).map_or(0, |x| x.len());
            rt.delete(group.start(), group.start() + www + group.len());
        }
    }

    let langs = language.clone().map(|s| vec![Language::new(s).unwrap()]);

    let reply = if let Some(x) = reply_to {
        let root_record = get_record(&config.handle, &rkey(&x.root_record_uri)).await?;
        let parent_record = get_record(&config.handle, &rkey(&x.record_uri)).await?;

        Some(
            atrium_api::app::bsky::feed::post::ReplyRefData {
                parent: atrium_api::com::atproto::repo::strong_ref::MainData {
                    cid: parent_record.data.cid.unwrap(),
                    uri: parent_record.data.uri.to_owned(),
                }
                .into(),
                root: atrium_api::com::atproto::repo::strong_ref::MainData {
                    cid: root_record.data.cid.unwrap(),
                    uri: root_record.data.uri.to_owned(),
                }
                .into(),
            }
            .into(),
        )
    } else {
        None
    };

    Ok(RecordData {
        created_at: Datetime::now(),
        embed,
        entities: None,
        facets: rt.facets,
        labels: None,
        langs,
        reply,
        tags: None,
        text: rt.text,
    })
}

async fn get_record(
    config: &str,
    rkey: &str,
) -> Result<
    atrium_api::types::Object<atrium_api::com::atproto::repo::get_record::OutputData>,
    Box<dyn Error>,
> {
    let bsky = BskyAgent::builder().build().await?;
    let record = bsky
        .api
        .com
        .atproto
        .repo
        .get_record(
            atrium_api::com::atproto::repo::get_record::ParametersData {
                cid: None,
                collection: atrium_api::types::string::Nsid::new("app.bsky.feed.post".to_string())?,
                repo: atrium_api::types::string::Handle::new(config.to_string())?.into(),
                rkey: RecordKey::new(rkey.to_string())?,
            }
            .into(),
        )
        .await?;

    Ok(record)
}

/// Generate an quote embed record
/// it is encapsulated in Option to prevent this function from failing
pub async fn generate_quote_records(
    config: &BlueskyConfig,
    quote_id: &str,
) -> Result<atrium_api::app::bsky::feed::post::RecordEmbedRefs, Box<dyn Error>> {
    // if we can’t match the quote_id, simply return None
    let quote_record = get_record(&config.handle, &rkey(quote_id)).await?;

    Ok(
        atrium_api::app::bsky::feed::post::RecordEmbedRefs::AppBskyEmbedRecordMain(Box::new(
            atrium_api::app::bsky::embed::record::MainData {
                record: atrium_api::com::atproto::repo::strong_ref::MainData {
                    cid: quote_record.data.cid.unwrap(),
                    uri: quote_record.data.uri.to_owned(),
                }
                .into(),
            }
            .into(),
        )),
    )
}

/// Generate an embed webcard record into Bsky
/// If the preview image does not exist or fails to upload, it is simply ignored
pub async fn generate_webcard_records(
    bsky: &BskyAgent,
    card: &Card,
) -> Result<atrium_api::app::bsky::feed::post::RecordEmbedRefs, Box<dyn Error + Send + Sync>> {
    let blob = match &card.image {
        Some(url) => upload_media(true, bsky, url).await?.blob.clone().into(),
        None => None,
    };

    let record_card = atrium_api::app::bsky::embed::external::ExternalData {
        description: card.description.clone(),
        thumb: blob,
        title: card.title.clone(),
        uri: card.url.clone(),
    };

    Ok(
        atrium_api::app::bsky::feed::post::RecordEmbedRefs::AppBskyEmbedExternalMain(Box::new(
            atrium_api::app::bsky::embed::external::MainData {
                external: record_card.into(),
            }
            .into(),
        )),
    )
}

/// Generate an array of Bsky media records
/// As Bsky does not support multiple video in a record or mix of video and images, video has the
/// highest priority
pub async fn generate_media_records(
    bsky: &BskyAgent,
    media_attach: &[Attachment],
) -> Result<atrium_api::app::bsky::feed::post::RecordEmbedRefs, Box<dyn Error + Send + Sync>> {
    let image_media_attach: Vec<_> = media_attach
        .iter()
        .filter(|x| x.r#type == AttachmentType::Image)
        .cloned()
        .collect();
    let video_media_attach: Vec<_> = media_attach
        .iter()
        .filter(|x| x.r#type == AttachmentType::Video || x.r#type == AttachmentType::Gifv)
        .cloned()
        .collect();

    // Bsky only tasks 1 video per post, so we’ll try to treat that first and exit
    if !video_media_attach.is_empty() {
        // treat only the very first video, ignore the rest
        let media = &video_media_attach[0];
        let blob = upload_media(false, bsky, &media.url).await?;

        return Ok(
            atrium_api::app::bsky::feed::post::RecordEmbedRefs::AppBskyEmbedVideoMain(Box::new(
                atrium_api::app::bsky::embed::video::MainData {
                    alt: media.description.clone(),
                    aspect_ratio: convert_aspect_ratio(
                        &media.meta.as_ref().and_then(|m| m.original.clone()),
                    ),
                    captions: None,
                    video: blob.data.blob,
                }
                .into(),
            )),
        );
    }

    // It wasn’t a video, then it’s an image or a gallery of 4 images
    let mut stream = stream::iter(image_media_attach)
        .map(|media| {
            let bsky = bsky.clone();
            tokio::task::spawn(async move {
                debug!("Treating media {}", &media.url);
                upload_media(true, &bsky, &media.url).await.map(|i| {
                    atrium_api::app::bsky::embed::images::ImageData {
                        alt: media
                            .description
                            .clone()
                            .map_or("".to_string(), |v| v.to_owned()),
                        aspect_ratio: convert_aspect_ratio(
                            &media.meta.as_ref().and_then(|m| m.original.clone()),
                        ),
                        image: i.data.blob,
                    }
                })
            })
        })
        .buffered(4);

    let mut images = Vec::new();

    while let Some(result) = stream.next().await {
        match result {
            Ok(Ok(v)) => images.push(v.into()),
            Ok(Err(e)) => warn!("Cannot treat a specific media: {}", e),
            Err(e) => error!("Something went wrong when joining main thread: {}", e),
        }
    }

    if !images.is_empty() {
        return Ok(
            atrium_api::app::bsky::feed::post::RecordEmbedRefs::AppBskyEmbedImagesMain(Box::new(
                atrium_api::app::bsky::embed::images::MainData { images }.into(),
            )),
        );
    }

    Err(OolatoocsError::new("Cannot embed media").into())
}

async fn upload_media(
    is_image: bool,
    bsky: &BskyAgent,
    u: &str,
) -> Result<Output, Box<dyn Error + Send + Sync>> {
    let dl = reqwest::get(u).await?;
    let content_length = dl.content_length().ok_or("Content length unavailable")?;
    let bytes = if content_length <= 1_000_000 || !is_image {
        dl.bytes().await?.as_ref().to_vec()
    } else {
        // this is an image and it’s over 1Mb long
        debug!("Img file too large: {}", content_length);
        // defaults to 95% quality for WebP compression
        let mut default_quality = 95f32;
        let img = ImageReader::new(Cursor::new(dl.bytes().await?))
            .with_guessed_format()?
            .decode()?;
        let encoder: Encoder = Encoder::from_image(&img)?;
        let mut webp: WebPMemory = encoder.encode(default_quality);

        while webp.len() > 1_000_000 {
            debug!("Img file too large at {}%, reducing…", default_quality);
            default_quality -= 5.0;
            webp = encoder.encode(default_quality);
        }

        webp.to_vec()
    };

    let record = bsky.api.com.atproto.repo.upload_blob(bytes).await?;

    Ok(record)
}

fn rkey(record_id: &str) -> String {
    record_id.split('/').nth(4).unwrap().to_string()
}

#[cfg(test)]
mod tests {
    use super::*;

    #[tokio::test]
    async fn test_build_post_record() {
        let text = "@factornews@piaille.fr Retrouvez-nous ici https://www.nintendojo.fr/articles/editos/le-mod-renovation-de-8bitdo-pour-manette-n64 et là https://www.nintendojo.fr/articles/analyses/vite-vu/vite-vu-morbid-the-lords-of-ire et un lien très court http://vsl.ie/TaMere et un autre https://p.nintendojo.fr/w/kV3CBbKKt1nPEChHhZiNve + http://www.xxx.com + https://www.youtube.com/watch?v=dQw4w9WgXcQ&pp=ygUJcmljayByb2xs";
        let expected_text = "@factornews@piaille.fr Retrouvez-nous ici nintendojo.fr/articles/edi… et là nintendojo.fr/articles/ana… et un lien très court vsl.ie/TaMere et un autre p.nintendojo.fr/w/kV3CBbKK… + xxx.com + youtube.com/watch?v=dQw4w9…";

        let bsky_conf = BlueskyConfig {
            handle: "tamerelol.bsky.social".to_string(),
            password: "dtc".to_string(),
            config_path: "nope".to_string(),
        };

        let created_record_data = build_post_record(&bsky_conf, text, &None, None, &None)
            .await
            .unwrap();

        assert_eq!(expected_text, &created_record_data.text);
    }
}