| 1 | //! Periodic background jobs: repository scanning, disk usage caching, etc. |
| 2 | //! |
| 3 | //! Jobs run on configurable intervals and log errors without affecting the |
| 4 | //! server. The runner is spawned in `anvil-cli/src/main.rs::serve()` alongside |
| 5 | //! the HTTP server and CI worker. |
| 6 | |
| 7 | use std::{ |
| 8 | path::Path, |
| 9 | time::Duration, |
| 10 | }; |
| 11 | |
| 12 | use async_trait::async_trait; |
| 13 | use tokio::time::interval; |
| 14 | |
| 15 | use crate::{ |
| 16 | App, |
| 17 | Result, |
| 18 | language, |
| 19 | models::Repository, |
| 20 | preview_images, |
| 21 | repos, |
| 22 | usage, |
| 23 | }; |
| 24 | |
| 25 | /// A periodic task to be run on an interval. |
| 26 | #[async_trait] |
| 27 | pub trait PeriodicJob: Send + Sync { |
| 28 | /// Run the job once. Errors are logged but do not stop the runner. |
| 29 | async fn run(&self, app: &App) -> Result<()>; |
| 30 | |
| 31 | /// Human-readable name for logging. |
| 32 | fn name(&self) -> &str; |
| 33 | } |
| 34 | |
| 35 | /// Spawns a background task that runs jobs on their configured intervals. |
| 36 | /// The task runs until the app shuts down. |
| 37 | pub async fn spawn_runner(app: App, jobs: Vec<(Duration, Box<dyn PeriodicJob>)>) { |
| 38 | tokio::spawn(async move { |
| 39 | let mut tasks = Vec::new(); |
| 40 | for (interval_duration, job) in jobs { |
| 41 | let mut ticker = interval(interval_duration); |
| 42 | let app = app.clone(); |
| 43 | tasks.push(tokio::spawn(async move { |
| 44 | loop { |
| 45 | ticker.tick().await; |
| 46 | if let Err(e) = job.run(&app).await { |
| 47 | tracing::error!( |
| 48 | job = job.name(), |
| 49 | error = %e, |
| 50 | "periodic job failed" |
| 51 | ); |
| 52 | } |
| 53 | } |
| 54 | })); |
| 55 | } |
| 56 | // All tasks are spawned and run forever; if any exits, others continue. |
| 57 | // We could join them here with `futures::future::join_all(tasks).await`, |
| 58 | // but detaching is fine — tasks clean up automatically on shutdown. |
| 59 | }); |
| 60 | } |
| 61 | |
| 62 | /// Job that detects programming languages in repositories. |
| 63 | pub struct LanguageDetectionJob; |
| 64 | |
| 65 | #[async_trait::async_trait] |
| 66 | impl PeriodicJob for LanguageDetectionJob { |
| 67 | async fn run(&self, app: &App) -> Result<()> { |
| 68 | let repos_dir = app.config.repositories_dir(); |
| 69 | let mut db = app.db.clone(); |
| 70 | let repos_list = Repository::all().exec(&mut db).await?; |
| 71 | let repo_count = repos_list.len(); |
| 72 | |
| 73 | for repo in repos_list { |
| 74 | let repo_path = repos_dir |
| 75 | .join(repo.owner_id.to_string()) |
| 76 | .join(&repo.name) |
| 77 | .join(".git"); |
| 78 | if repo_path.exists() |
| 79 | && let Ok(detection) = language::detect(&repo_path).await |
| 80 | { |
| 81 | let _ = |
| 82 | repos::update_languages(&app.db, repo.id, &detection.primary, &detection.json) |
| 83 | .await; |
| 84 | } |
| 85 | } |
| 86 | tracing::info!("language detection job completed for {} repos", repo_count); |
| 87 | Ok(()) |
| 88 | } |
| 89 | |
| 90 | fn name(&self) -> &str { |
| 91 | "language_detection" |
| 92 | } |
| 93 | } |
| 94 | |
| 95 | /// Job that extracts preview images from repository READMEs. |
| 96 | pub struct PreviewImageJob; |
| 97 | |
| 98 | #[async_trait::async_trait] |
| 99 | impl PeriodicJob for PreviewImageJob { |
| 100 | async fn run(&self, app: &App) -> Result<()> { |
| 101 | let repos_dir = app.config.repositories_dir(); |
| 102 | let mut db = app.db.clone(); |
| 103 | let repos_list = Repository::all().exec(&mut db).await?; |
| 104 | |
| 105 | let mut updated = 0; |
| 106 | for repo in repos_list { |
| 107 | let repo_git_path = repos_dir |
| 108 | .join(repo.owner_id.to_string()) |
| 109 | .join(&repo.name) |
| 110 | .join(".git"); |
| 111 | if repo_git_path.exists() { |
| 112 | // Try to extract preview image from README in the repository |
| 113 | if let Some(image_url) = extract_readme_image(&repo_git_path).await { |
| 114 | // TODO: download and store the image, or just reference it |
| 115 | // For now, we'll just store the URL as a hash placeholder |
| 116 | if !image_url.is_empty() { |
| 117 | let _ = repos::update_preview_image(&app.db, repo.id, &image_url).await; |
| 118 | updated += 1; |
| 119 | } |
| 120 | } |
| 121 | } |
| 122 | } |
| 123 | tracing::info!("preview image job completed: {} repos updated", updated); |
| 124 | Ok(()) |
| 125 | } |
| 126 | |
| 127 | fn name(&self) -> &str { |
| 128 | "preview_images" |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | /// Job that caches disk usage for the admin dashboard. |
| 133 | pub struct DiskUsageCacheJob; |
| 134 | |
| 135 | #[async_trait::async_trait] |
| 136 | impl PeriodicJob for DiskUsageCacheJob { |
| 137 | async fn run(&self, app: &App) -> Result<()> { |
| 138 | // Compute disk usage and cache it |
| 139 | if let Ok(usage_data) = usage::compute(app).await { |
| 140 | let json = serde_json::to_string(&usage_data)?; |
| 141 | // Store in admin_cache table |
| 142 | crate::admin_cache::upsert(&app.db, "disk_usage", &json).await?; |
| 143 | tracing::info!("disk usage cache job completed"); |
| 144 | } |
| 145 | Ok(()) |
| 146 | } |
| 147 | |
| 148 | fn name(&self) -> &str { |
| 149 | "disk_usage_cache" |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | /// Try to extract the first image URL from a repository's README file. |
| 154 | /// Scans the repository for a README file, reads it, and returns the first image URL found. |
| 155 | async fn extract_readme_image(repo_path: &Path) -> Option<String> { |
| 156 | // Find README file in the repository (case-insensitive) |
| 157 | let readme_names = [ |
| 158 | "README.md", |
| 159 | "readme.md", |
| 160 | "README.MD", |
| 161 | "README.rst", |
| 162 | "README.txt", |
| 163 | ]; |
| 164 | |
| 165 | for readme_name in &readme_names { |
| 166 | let readme_path = repo_path.parent()?.join(readme_name); |
| 167 | if readme_path.exists() |
| 168 | && let Ok(content) = std::fs::read_to_string(&readme_path) |
| 169 | && let Some(url) = preview_images::extract_preview_url(&content) |
| 170 | { |
| 171 | return Some(url); |
| 172 | } |
| 173 | } |
| 174 | |
| 175 | None |
| 176 | } |