path: root/src/bots/fdroid/mod.rs

//! F-Droid

mod index;

use atrium_api::app::bsky::richtext::facet::{ByteSliceData, LinkData, MainData, MainFeaturesItem};
use atrium_api::types::{Union, string::Language};
use reqwest::{
    StatusCode,
    header::{ETAG, IF_NONE_MATCH},
};
use serde::{Deserialize, Serialize};
use unicode_segmentation::UnicodeSegmentation;

use self::index::{AntiFeature, EN_LOCALE, Index, PHONE_DEVICE_TYPE, Package, Screenshot};
use crate::{
    MAX_IMAGE_SIZE, MAX_POST_GRAPHEMES, MAX_POST_IMAGES, http_client,
    report::{Post, PostImage, Report},
};

const REPOSITORY_URL: &str = "https://f-droid.org/repo";
const PACKAGE_URL_PREFIX: &str = "https://f-droid.org/packages/";
const ANTI_FEATURES_EXCLUDING_APPS: &[AntiFeature] =
    &[AntiFeature::KnownVuln, AntiFeature::NoSourceSince];

#[derive(Debug, Default, Deserialize, Serialize)]
pub struct BreakingState {
    pub last_repo_timestamp: Option<u64>,
    pub index_etag: Option<String>,
}

pub struct Breaking;

impl Report for Breaking {
    type State = BreakingState;

    const STATE_FILE: &'static str = "fdroid-breaking.json";

    async fn generate(state: Self::State) -> Result<(Vec<Post>, Self::State), String> {
        let mut request = http_client().get(format!("{REPOSITORY_URL}/index-v2.json"));
        if let Some(etag) = state.index_etag.as_deref() {
            request = request.header(IF_NONE_MATCH, etag);
        }

        let response = request
            .send()
            .await
            .map_err(|error| format!("failed to fetch F-Droid index: {error}"))?;
        if response.status() == StatusCode::NOT_MODIFIED {
            return Ok((Vec::new(), state));
        }

        let response = response
            .error_for_status()
            .map_err(|error| format!("failed to fetch F-Droid index: {error}"))?;
        let index_etag = response
            .headers()
            .get(ETAG)
            .and_then(|value| value.to_str().ok())
            .map(str::to_owned);
        let index = response
            .json::<Index>()
            .await
            .map_err(|error| format!("failed to parse F-Droid index: {error}"))?;

        let Some(timestamp) = state.last_repo_timestamp else {
            // We don't know the timestamp of the last index version.
            tracing::info!(
                timestamp = index.repo.timestamp,
                "recording initial repository timestamp; will announce new apps with the next index version"
            );
            let new_state = BreakingState {
                last_repo_timestamp: Some(index.repo.timestamp),
                index_etag,
            };
            return Ok((Vec::new(), new_state));
        };

        if index.repo.timestamp < timestamp {
            let error = format!(
                "F-Droid repository timestamp moved backwards from {timestamp} to {}",
                index.repo.timestamp
            );
            tracing::error!("{error}");
            return Err(error);
        }

        let mut apps: Vec<_> = index
            .packages
            .iter()
            .filter_map(|(id, package)| {
                (package.metadata.added > timestamp)
                    .then(|| App::from_package(id.as_str(), package))
            })
            .collect();

        tracing::info!(
            timestamp = index.repo.timestamp,
            new_apps = apps.len(),
            total_apps = index.packages.len(),
            "building posts"
        );

        apps.sort_unstable_by(|left, right| {
            left.name
                .to_lowercase()
                .cmp(&right.name.to_lowercase())
                .then_with(|| left.id.cmp(right.id))
        });

        Ok((
            download_images(build_announcement_posts(&apps)).await?,
            BreakingState {
                last_repo_timestamp: Some(index.repo.timestamp),
                index_etag,
            },
        ))
    }
}

struct App<'a> {
    id: &'a str,
    name: &'a str,
    summary: &'a str,
    screenshot: Option<&'a Screenshot>,
    anti_features: Vec<AntiFeature>,
}

impl<'a> App<'a> {
    fn from_package(id: &'a str, package: &'a Package) -> Self {
        Self {
            id,
            name: package
                .metadata
                .name
                .get(EN_LOCALE)
                .map_or(id, String::as_str),
            summary: package
                .metadata
                .summary
                .get(EN_LOCALE)
                .map_or("", String::as_str),
            screenshot: package
                .metadata
                .screenshots
                .get(PHONE_DEVICE_TYPE)
                .and_then(|screenshots| screenshots.get(EN_LOCALE))
                .and_then(|screenshots| {
                    screenshots
                        .iter()
                        // We could resize images but realistically they should fit.
                        .find(|screenshot| screenshot.size < MAX_IMAGE_SIZE)
                }),
            anti_features: package
                .versions
                .values()
                .flat_map(|version| version.anti_features.keys())
                // "Tethered network service" means an app depends on a network service,
                // which isn't easy to self-host. There's a lot of such network services out there
                // be it OpenStreetMap or a public transit API ... not much apps can do about that.
                // So we just disregard this anti-feature in this bot.
                .filter(|&anti_feature| anti_feature != &AntiFeature::TetheredNet)
                .cloned()
                .collect(),
        }
    }
}

fn build_announcement_posts(apps: &[App<'_>]) -> Vec<PendingPost> {
    let apps: Vec<_> = apps
        .iter()
        .filter(|app| {
            let Some(anti_feature) = ANTI_FEATURES_EXCLUDING_APPS
                .iter()
                .find(|anti_feature| app.anti_features.contains(anti_feature))
            else {
                return true;
            };

            tracing::info!(
                app_id = app.id,
                app_name = app.name,
                "omitting F-Droid app due to anti-feature {anti_feature:?}"
            );
            false
        })
        .collect();

    if apps.is_empty() {
        return Vec::new();
    }

    let mut builders = vec![PostBuilder::new(format!(
        "{} new F-Droid apps are available:",
        apps.len()
    ))];

    let (without_anti_features, with_anti_features): (Vec<_>, Vec<_>) = apps
        .into_iter()
        .partition(|app| app.anti_features.is_empty());
    let (with_screenshots, without_screenshots): (Vec<_>, Vec<_>) = without_anti_features
        .into_iter()
        .partition(|app| app.screenshot.is_some());

    for app in with_screenshots {
        let name = truncate(app.name, MAX_POST_GRAPHEMES);
        let screenshot = app.screenshot.unwrap();
        let image = PendingImage {
            url: format!("{REPOSITORY_URL}{}", screenshot.name),
            // Using the summary for the alt text isn't great
            // but the Bluesky lexicon provides no alternative.
            alt: format!("{}: {}", app.name, app.summary),
        };

        let post = builders.last().unwrap();
        let separator = if post.text.is_empty() {
            ""
        } else if post.images.is_empty() {
            "\n"
        } else {
            ", "
        };
        if !post.can_fit(separator, &name) || post.images.len() == MAX_POST_IMAGES {
            builders.push(PostBuilder::new(String::new()));
        }

        let post = builders.last_mut().unwrap();
        let separator = if post.text.is_empty() {
            ""
        } else if post.images.is_empty() {
            "\n"
        } else {
            ", "
        };
        post.push_app(separator, &name, &name, app.id, Some(image));
    }

    for app in without_screenshots {
        push_bullet_app(&mut builders, app);
    }

    if let Some((first, rest)) = with_anti_features.split_first() {
        const ANTI_FEATURES_HEADING: &str =
            "New apps with an anti-feature (other than tethered network):";
        let line_limit =
            MAX_POST_GRAPHEMES - grapheme_count(ANTI_FEATURES_HEADING) - grapheme_count("\n");
        let (name, line) = app_line(first, line_limit);

        // Keep anti-feature apps in their own section, starting at a post boundary.
        builders.push(PostBuilder::new(ANTI_FEATURES_HEADING.to_owned()));
        builders
            .last_mut()
            .unwrap()
            .push_app("\n", &line, &name, first.id, None);

        for app in rest {
            push_bullet_app(&mut builders, app);
        }
    }

    builders.into_iter().map(PostBuilder::finish).collect()
}

fn push_bullet_app(builders: &mut Vec<PostBuilder>, app: &App<'_>) {
    let (name, line) = app_line(app, MAX_POST_GRAPHEMES);
    let post = builders.last().unwrap();
    let separator = if post.text.is_empty() { "" } else { "\n" };
    if !post.can_fit(separator, &line) {
        builders.push(PostBuilder::new(String::new()));
    }

    let post = builders.last_mut().unwrap();
    let separator = if post.text.is_empty() { "" } else { "\n" };
    post.push_app(separator, &line, &name, app.id, None);
}

/// Build a bullet that fits within `limit`, preserving as much of the summary as possible.
fn app_line(app: &App<'_>, limit: usize) -> (String, String) {
    let prefix = "* ";
    let anti_features = anti_feature_annotation(&app.anti_features);
    let suffix = ": ";
    let fixed = grapheme_count(prefix)
        + grapheme_count(app.name)
        + grapheme_count(&anti_features)
        + grapheme_count(suffix);

    if fixed <= limit {
        let summary = truncate(app.summary, limit - fixed);
        (
            app.name.to_owned(),
            format!("{prefix}{}{anti_features}{suffix}{summary}", app.name),
        )
    } else {
        let reserved = grapheme_count(prefix) + grapheme_count(&anti_features);
        let name = truncate(app.name, limit.saturating_sub(reserved));
        let line = format!("{prefix}{name}{anti_features}");
        (name, line)
    }
}

fn anti_feature_annotation(anti_features: &[AntiFeature]) -> String {
    if anti_features.is_empty() {
        return String::new();
    }

    let mut icons: Vec<_> = anti_features
        .iter()
        .map(|anti_feature| match anti_feature {
            AntiFeature::Tracking => (0, "👁️"),
            AntiFeature::Ads => (1, "🪧"),
            AntiFeature::NonFreeNet => (2, "☁️"),
            AntiFeature::NonFreeDep => (3, "📦"),
            AntiFeature::NonFreeAssets => (4, "🖼️"),
            AntiFeature::NonFreeAdd => (5, "🧩"),
            AntiFeature::DisabledAlgorithm => (6, "🔓"), // shouldn't happen for new apps
            AntiFeature::Other(_) => (7, "❓"),
            AntiFeature::KnownVuln | AntiFeature::NoSourceSince | AntiFeature::TetheredNet => {
                unreachable!()
            }
        })
        .collect();
    icons.sort_unstable_by_key(|(order, _)| *order);
    icons.dedup_by_key(|(order, _)| *order);

    format!(
        " ({})",
        icons
            .into_iter()
            .map(|(_, icon)| icon)
            .collect::<Vec<_>>()
            .join("")
    )
}

fn truncate(text: &str, limit: usize) -> String {
    let length = grapheme_count(text);
    if length <= limit {
        return text.to_owned();
    }
    if limit == 0 {
        return String::new();
    }

    let mut shortened: String = text.graphemes(true).take(limit - 1).collect();
    shortened.push('…');
    shortened
}

fn grapheme_count(text: &str) -> usize {
    text.graphemes(true).count()
}

struct PendingPost {
    text: String,
    facets: Vec<atrium_api::app::bsky::richtext::facet::Main>,
    images: Vec<PendingImage>,
}

struct PendingImage {
    url: String,
    alt: String,
}

struct PostBuilder {
    text: String,
    graphemes: usize,
    facets: Vec<atrium_api::app::bsky::richtext::facet::Main>,
    images: Vec<PendingImage>,
}

impl PostBuilder {
    fn new(text: String) -> Self {
        let graphemes = grapheme_count(&text);
        Self {
            text,
            graphemes,
            facets: Vec::new(),
            images: Vec::new(),
        }
    }

    fn can_fit(&self, separator: &str, text: &str) -> bool {
        self.graphemes + grapheme_count(separator) + grapheme_count(text) <= MAX_POST_GRAPHEMES
    }

    fn push_app(
        &mut self,
        separator: &str,
        line: &str,
        name: &str,
        id: &str,
        image: Option<PendingImage>,
    ) {
        self.text.push_str(separator);
        let byte_start = self.text.len() + line.find(name).unwrap_or(0);
        self.text.push_str(line);
        let byte_end = byte_start + name.len();
        self.graphemes += grapheme_count(separator) + grapheme_count(line);

        self.facets.push(
            MainData {
                features: vec![Union::Refs(MainFeaturesItem::Link(Box::new(
                    LinkData {
                        uri: format!("{PACKAGE_URL_PREFIX}{id}/"),
                    }
                    .into(),
                )))],
                index: ByteSliceData {
                    byte_start,
                    byte_end,
                }
                .into(),
            }
            .into(),
        );
        if let Some(image) = image {
            self.images.push(image);
        }
    }

    fn finish(self) -> PendingPost {
        PendingPost {
            text: self.text,
            facets: self.facets,
            images: self.images,
        }
    }
}

async fn download_images(pending_posts: Vec<PendingPost>) -> Result<Vec<Post>, String> {
    let client = http_client();
    let english: Language = "en".parse().unwrap();
    let mut posts = Vec::with_capacity(pending_posts.len());

    for (post_index, post) in pending_posts.into_iter().enumerate() {
        let mut images = Vec::with_capacity(post.images.len());
        for image in post.images {
            let data = client
                .get(&image.url)
                .send()
                .await
                .and_then(reqwest::Response::error_for_status)
                .map_err(|error| {
                    format!(
                        "failed to download image {} for post {}: {error}",
                        image.url,
                        post_index + 1
                    )
                })?
                .bytes()
                .await
                .map_err(|error| {
                    format!(
                        "failed to download image {} for post {}: {error}",
                        image.url,
                        post_index + 1
                    )
                })?;
            images.push(PostImage {
                data: data.to_vec(),
                alt: image.alt,
            });
        }
        posts.push(Post {
            text: post.text,
            facets: post.facets,
            images,
            langs: vec![english.clone()],
            tags: vec!["F-Droid".to_owned()],
        });
    }

    Ok(posts)
}

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

    #[test]
    fn builds_announcement_posts() {
        assert!(build_announcement_posts(&[]).is_empty());

        let screenshot = Screenshot {
            name: "/screenshots/camera.png".to_owned(),
            size: 123,
        };
        let apps = [
            App {
                id: "org.example.camera",
                name: "Camera 📷",
                summary: "Take photos",
                screenshot: Some(&screenshot),
                anti_features: Vec::new(),
            },
            App {
                id: "org.example.notes",
                name: "Notes",
                summary: "Take notes",
                screenshot: None,
                anti_features: Vec::new(),
            },
            App {
                id: "org.example.tracker",
                name: "Tracker",
                summary: "Tracks users",
                screenshot: None,
                anti_features: vec![
                    AntiFeature::Other("UnknownFeature".to_owned()),
                    AntiFeature::Tracking,
                    AntiFeature::NonFreeNet,
                    AntiFeature::NonFreeDep,
                    AntiFeature::NonFreeAssets,
                    AntiFeature::NonFreeAdd,
                    AntiFeature::DisabledAlgorithm,
                    AntiFeature::Ads,
                ],
            },
            App {
                id: "org.example.vulnerable",
                name: "Vulnerable",
                summary: "Has a known vulnerability",
                screenshot: None,
                anti_features: vec![AntiFeature::KnownVuln],
            },
            App {
                id: "org.example.abandoned",
                name: "Abandoned",
                summary: "No longer has source available",
                screenshot: None,
                anti_features: vec![AntiFeature::NoSourceSince],
            },
        ];

        let posts = build_announcement_posts(&apps);

        assert_eq!(posts.len(), 2);
        assert_eq!(
            posts[0].text,
            "3 new F-Droid apps are available:\n\
             Camera 📷\n\
             * Notes: Take notes"
        );
        assert_eq!(
            posts[1].text,
            "New apps with an anti-feature (other than tethered network):\n\
             * Tracker (👁️🪧☁️📦🖼️🧩🔓❓): Tracks users"
        );
        assert_eq!(posts[0].images.len(), 1);
        assert_eq!(
            posts[0].images[0].url,
            "https://f-droid.org/repo/screenshots/camera.png"
        );
        assert_eq!(posts[0].images[0].alt, "Camera 📷: Take photos");
        assert!(posts[1].images.is_empty());

        let expected_links = [
            (&posts[0], "Camera 📷", "org.example.camera"),
            (&posts[0], "Notes", "org.example.notes"),
            (&posts[1], "Tracker", "org.example.tracker"),
        ];
        for (post, name, id) in expected_links {
            let facet = post
                .facets
                .iter()
                .find(|facet| &post.text[facet.index.byte_start..facet.index.byte_end] == name)
                .unwrap();
            match facet.features.as_slice() {
                [Union::Refs(MainFeaturesItem::Link(link))] => {
                    assert_eq!(link.uri, format!("{PACKAGE_URL_PREFIX}{id}/"));
                }
                features => panic!("expected one link feature, got {features:?}"),
            }
        }
    }
}