| 1 | //! Blame, archives, the Atom feed, the file finder, language statistics, |
| 2 | //! robots.txt and HTTPS clones, checked against the git CLI. |
| 3 | |
| 4 | mod common; |
| 5 | |
| 6 | use std::path::{Path, PathBuf}; |
| 7 | use std::process::Command; |
| 8 | use std::sync::Arc; |
| 9 | |
| 10 | use axum::body::{Body, to_bytes}; |
| 11 | use axum::http::{HeaderMap, Method, Request, StatusCode, header}; |
| 12 | use common::{Work, bare, git}; |
| 13 | use ferrenium::config::Config; |
| 14 | use ferrenium::index::languages; |
| 15 | use ferrenium::server; |
| 16 | use ferrenium::state::{AppState, Overrides}; |
| 17 | use tower::ServiceExt; |
| 18 | |
| 19 | struct Fixture { |
| 20 | bare: PathBuf, |
| 21 | head: String, |
| 22 | } |
| 23 | |
| 24 | /// Several authors editing one file, a rename, vendored/generated/doc |
| 25 | /// files, a `.gitattributes` and a tag. |
| 26 | fn fixture(root: &Path) -> Fixture { |
| 27 | let bare = bare(root, "ext.git", &[]); |
| 28 | let work = Work::clone_of(&bare); |
| 29 | let w = work.path(); |
| 30 | let write = |name: &str, data: &str| { |
| 31 | let path = w.join(name); |
| 32 | std::fs::create_dir_all(path.parent().unwrap()).unwrap(); |
| 33 | std::fs::write(path, data).unwrap(); |
| 34 | }; |
| 35 | // A day apart, so blame ages differ. |
| 36 | let day = std::cell::Cell::new(0); |
| 37 | let commit = |message: &str, author: &str| { |
| 38 | git(w, &["add", "-A"]); |
| 39 | day.set(day.get() + 1); |
| 40 | let date = format!("{} +0000", 1_700_000_000 + day.get() * 86_400); |
| 41 | let status = Command::new("git") |
| 42 | .args(["commit", "-q", "-m", message, "--author", author]) |
| 43 | .current_dir(w) |
| 44 | .env("GIT_CONFIG_NOSYSTEM", "1") |
| 45 | .env("GIT_CONFIG_GLOBAL", "/dev/null") |
| 46 | .env("GIT_COMMITTER_NAME", "Ada") |
| 47 | .env("GIT_COMMITTER_EMAIL", "ada@example.com") |
| 48 | .env("GIT_AUTHOR_DATE", &date) |
| 49 | .env("GIT_COMMITTER_DATE", &date) |
| 50 | .status() |
| 51 | .unwrap(); |
| 52 | assert!(status.success()); |
| 53 | }; |
| 54 | write("src/old.rs", "fn a() {}\nfn b() {}\nfn c() {}\n"); |
| 55 | write("README.md", "# Ext\n"); |
| 56 | commit("Start", "Ada <ada@example.com>"); |
| 57 | write( |
| 58 | "src/old.rs", |
| 59 | "fn a() {}\nfn b2() {}\nfn c() {}\nfn d() {}\n", |
| 60 | ); |
| 61 | commit("Edit b, add d", "Bob <bob@example.com>"); |
| 62 | git(w, &["mv", "src/old.rs", "src/main.rs"]); |
| 63 | commit("Rename", "Ada <ada@example.com>"); |
| 64 | write( |
| 65 | "src/main.rs", |
| 66 | "// top\nfn a() {}\nfn b2() {}\nfn c() {}\nfn d() {}\n", |
| 67 | ); |
| 68 | write("run.sh", "#!/bin/sh\necho run\n"); |
| 69 | write("vendor/lib.js", &"var x = 1;\n".repeat(500)); |
| 70 | write("docs/conf.py", &"x = 1\n".repeat(500)); |
| 71 | write("app.min.js", &"var y=2;".repeat(500)); |
| 72 | write("extra/big.c", &"int x;\n".repeat(500)); |
| 73 | write(".gitattributes", "extra/** linguist-vendored\n"); |
| 74 | commit("Add the rest", "Cy <cy@example.com>"); |
| 75 | git(w, &["tag", "v1.0"]); |
| 76 | work.push("main"); |
| 77 | git(w, &["push", "-q", "origin", "--tags"]); |
| 78 | let head = git(w, &["rev-parse", "HEAD"]); |
| 79 | Fixture { bare, head } |
| 80 | } |
| 81 | |
| 82 | struct Site { |
| 83 | _root: tempfile::TempDir, |
| 84 | state: Arc<AppState>, |
| 85 | app: axum::Router, |
| 86 | f: Fixture, |
| 87 | } |
| 88 | |
| 89 | /// `repos` is extra lines for the `[repos]` table. |
| 90 | fn site(repos: &str) -> Site { |
| 91 | let root = tempfile::tempdir().unwrap(); |
| 92 | let f = fixture(root.path()); |
| 93 | let toml = format!( |
| 94 | "[repos]\nscan_path = {:?}\n{repos}\n[watch]\nenabled = false\n\ |
| 95 | [server]\nbase_url = \"https://git.example.com\"\n", |
| 96 | root.path().to_str().unwrap() |
| 97 | ); |
| 98 | let mut config = Config::parse(&toml).expect("parse"); |
| 99 | config.finish().expect("valid"); |
| 100 | let state = AppState::new(config, Overrides::default()); |
| 101 | let config = state.config.load_full(); |
| 102 | state.index.full_scan(&config); |
| 103 | Site { |
| 104 | app: server::app(state.clone()), |
| 105 | state, |
| 106 | f, |
| 107 | _root: root, |
| 108 | } |
| 109 | } |
| 110 | |
| 111 | struct Res { |
| 112 | status: StatusCode, |
| 113 | headers: HeaderMap, |
| 114 | body: Vec<u8>, |
| 115 | } |
| 116 | |
| 117 | impl Res { |
| 118 | fn text(&self) -> String { |
| 119 | String::from_utf8_lossy(&self.body).into_owned() |
| 120 | } |
| 121 | } |
| 122 | |
| 123 | async fn request(app: &axum::Router, method: Method, uri: &str) -> Res { |
| 124 | let req = Request::builder() |
| 125 | .method(method) |
| 126 | .uri(uri) |
| 127 | .body(Body::empty()) |
| 128 | .unwrap(); |
| 129 | let res = app.clone().oneshot(req).await.unwrap(); |
| 130 | let status = res.status(); |
| 131 | let headers = res.headers().clone(); |
| 132 | let body = to_bytes(res.into_body(), usize::MAX) |
| 133 | .await |
| 134 | .unwrap() |
| 135 | .to_vec(); |
| 136 | Res { |
| 137 | status, |
| 138 | headers, |
| 139 | body, |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | async fn get(app: &axum::Router, uri: &str) -> Res { |
| 144 | request(app, Method::GET, uri).await |
| 145 | } |
| 146 | |
| 147 | async fn ok(app: &axum::Router, uri: &str) -> String { |
| 148 | let res = get(app, uri).await; |
| 149 | assert_eq!(res.status, StatusCode::OK, "{uri}: {}", res.text()); |
| 150 | res.text() |
| 151 | } |
| 152 | |
| 153 | /// Commit of every line, from the page's hunk heads. |
| 154 | fn blamed_lines(html: &str) -> Vec<String> { |
| 155 | let mut out = Vec::new(); |
| 156 | for part in html.split(r#"class="blame-meta "#).skip(1) { |
| 157 | let rows: usize = part |
| 158 | .split(r#"rowspan=""#) |
| 159 | .nth(1) |
| 160 | .unwrap() |
| 161 | .split('"') |
| 162 | .next() |
| 163 | .unwrap() |
| 164 | .parse() |
| 165 | .unwrap(); |
| 166 | let id = part.split("/commit/").nth(1).unwrap()[..40].to_string(); |
| 167 | out.extend(std::iter::repeat_n(id, rows)); |
| 168 | } |
| 169 | out |
| 170 | } |
| 171 | |
| 172 | #[tokio::test] |
| 173 | async fn blame_matches_git() { |
| 174 | let s = site(""); |
| 175 | for path in ["src/main.rs", "run.sh", "README.md"] { |
| 176 | let html = ok(&s.app, &format!("/ext/blame/main/{path}")).await; |
| 177 | let porcelain = git(&s.f.bare, &["blame", "--porcelain", "main", "--", path]); |
| 178 | let expected: Vec<String> = porcelain |
| 179 | .lines() |
| 180 | .filter(|l| { |
| 181 | l.len() > 41 |
| 182 | && l.as_bytes()[40] == b' ' |
| 183 | && l[..40].bytes().all(|b| b.is_ascii_hexdigit()) |
| 184 | }) |
| 185 | .map(|l| l[..40].to_string()) |
| 186 | .collect(); |
| 187 | assert_eq!(blamed_lines(&html), expected, "{path}"); |
| 188 | } |
| 189 | let html = ok(&s.app, "/ext/blame/main/src/main.rs").await; |
| 190 | // Hunk heads: author, age bar (newest is full strength), links. |
| 191 | assert!(html.contains("blame-meta age-8"), "{html}"); |
| 192 | assert!(html.contains("blame-meta age-0")); |
| 193 | assert!(html.contains(">Bob<")); |
| 194 | assert!(html.contains("5 lines · 3 commits"), "{html}"); |
| 195 | assert!(html.contains(r#"href="/ext/blob/main/src/main.rs">File<"#)); |
| 196 | // The blob page links here; directories redirect; binaries don't blame. |
| 197 | let blob = ok(&s.app, "/ext/blob/main/src/main.rs").await; |
| 198 | assert!(blob.contains(r#"href="/ext/blame/main/src/main.rs">Blame<"#)); |
| 199 | let res = get(&s.app, "/ext/blame/main/src").await; |
| 200 | assert_eq!(res.status, StatusCode::TEMPORARY_REDIRECT); |
| 201 | let res = get(&s.app, "/ext/blame/main/nope").await; |
| 202 | assert_eq!(res.status, StatusCode::NOT_FOUND); |
| 203 | } |
| 204 | |
| 205 | fn tar_listing(data: &[u8]) -> String { |
| 206 | let mut child = Command::new("tar") |
| 207 | .args(["tzf", "-"]) |
| 208 | .stdin(std::process::Stdio::piped()) |
| 209 | .stdout(std::process::Stdio::piped()) |
| 210 | .spawn() |
| 211 | .unwrap(); |
| 212 | use std::io::Write; |
| 213 | child.stdin.take().unwrap().write_all(data).unwrap(); |
| 214 | let out = child.wait_with_output().unwrap(); |
| 215 | // git also lists directories; only files matter. |
| 216 | let mut lines: Vec<&str> = std::str::from_utf8(&out.stdout) |
| 217 | .unwrap() |
| 218 | .lines() |
| 219 | .filter(|l| !l.ends_with('/')) |
| 220 | .collect(); |
| 221 | lines.sort(); |
| 222 | lines.join("\n") |
| 223 | } |
| 224 | |
| 225 | #[tokio::test] |
| 226 | async fn archives_match_git_archive() { |
| 227 | let s = site(""); |
| 228 | let res = get(&s.app, "/ext/archive/v1.0.tar.gz").await; |
| 229 | assert_eq!(res.status, StatusCode::OK); |
| 230 | assert_eq!(res.headers[header::CONTENT_TYPE], "application/gzip"); |
| 231 | assert!( |
| 232 | res.headers[header::CONTENT_DISPOSITION] |
| 233 | .to_str() |
| 234 | .unwrap() |
| 235 | .contains("filename=\"ext-v1.0.tar.gz\"") |
| 236 | ); |
| 237 | assert!(res.headers.get(header::CONTENT_ENCODING).is_none()); |
| 238 | let ours = tar_listing(&res.body); |
| 239 | let theirs = Command::new("git") |
| 240 | .args(["archive", "--format=tar.gz", "--prefix=ext-v1.0/", "v1.0"]) |
| 241 | .current_dir(&s.f.bare) |
| 242 | .output() |
| 243 | .unwrap() |
| 244 | .stdout; |
| 245 | assert_eq!(ours, tar_listing(&theirs)); |
| 246 | assert!(ours.contains("ext-v1.0/src/main.rs")); |
| 247 | |
| 248 | let res = get(&s.app, &format!("/ext/archive/{}.zip", s.f.head)).await; |
| 249 | assert_eq!(res.status, StatusCode::OK); |
| 250 | assert_eq!(res.headers[header::CONTENT_TYPE], "application/zip"); |
| 251 | assert_eq!(res.headers[header::CACHE_CONTROL], "public, max-age=86400"); |
| 252 | assert_eq!(&res.body[..2], b"PK"); |
| 253 | |
| 254 | for uri in [ |
| 255 | "/ext/archive/nope.zip", |
| 256 | "/ext/archive/main.rar", |
| 257 | "/ext/archive/main", |
| 258 | ] { |
| 259 | assert_eq!( |
| 260 | get(&s.app, uri).await.status, |
| 261 | StatusCode::NOT_FOUND, |
| 262 | "{uri}" |
| 263 | ); |
| 264 | } |
| 265 | let s = site("snapshots = [\"tar.gz\"]"); |
| 266 | assert_eq!( |
| 267 | get(&s.app, "/ext/archive/main.zip").await.status, |
| 268 | StatusCode::NOT_FOUND |
| 269 | ); |
| 270 | let refs = ok(&s.app, "/ext/refs").await; |
| 271 | assert!( |
| 272 | refs.contains(r#"href="/ext/archive/v1.0.tar.gz""#), |
| 273 | "{refs}" |
| 274 | ); |
| 275 | assert!(!refs.contains(".zip")); |
| 276 | } |
| 277 | |
| 278 | #[tokio::test] |
| 279 | async fn atom_feed() { |
| 280 | let s = site(""); |
| 281 | let res = get(&s.app, "/ext/atom").await; |
| 282 | assert_eq!(res.status, StatusCode::OK); |
| 283 | assert_eq!( |
| 284 | res.headers[header::CONTENT_TYPE], |
| 285 | "application/atom+xml; charset=utf-8" |
| 286 | ); |
| 287 | let xml = res.text(); |
| 288 | assert!(xml.starts_with("<?xml")); |
| 289 | assert_eq!(xml.matches("<entry>").count(), 4); |
| 290 | assert!(xml.contains("<id>https://git.example.com/ext/atom</id>")); |
| 291 | assert!(xml.contains(&format!( |
| 292 | r#"href="https://git.example.com/ext/commit/{}""#, |
| 293 | s.f.head |
| 294 | ))); |
| 295 | assert!(xml.contains("<name>Cy</name>")); |
| 296 | assert_eq!( |
| 297 | get(&s.app, "/ext/atom?h=nope").await.status, |
| 298 | StatusCode::NOT_FOUND |
| 299 | ); |
| 300 | // Advertised on the summary. |
| 301 | let html = ok(&s.app, "/ext").await; |
| 302 | assert!(html.contains( |
| 303 | r#"<link rel="alternate" type="application/atom+xml" title="Commits" href="/ext/atom">"# |
| 304 | )); |
| 305 | } |
| 306 | |
| 307 | #[tokio::test] |
| 308 | async fn find_files() { |
| 309 | let s = site(""); |
| 310 | let html = ok(&s.app, "/ext/find/main?q=MAIN").await; |
| 311 | assert!( |
| 312 | html.contains(r#"href="/ext/blob/main/src/main.rs""#), |
| 313 | "{html}" |
| 314 | ); |
| 315 | assert!(html.contains("1 of 8 files"), "{html}"); |
| 316 | let html = ok(&s.app, "/ext/find/main?q=zzz").await; |
| 317 | assert!(html.contains("[No matches]")); |
| 318 | let html = ok(&s.app, "/ext/find/main").await; |
| 319 | assert!(html.contains("8 files")); |
| 320 | assert!(html.contains(r#"aria-current="page">Find<"#)); |
| 321 | } |
| 322 | |
| 323 | #[tokio::test] |
| 324 | async fn language_statistics() { |
| 325 | let s = site(""); |
| 326 | let repo = s.state.index.load().repos["ext"].clone(); |
| 327 | let handle = ferrenium::git::handle(&repo.gix); |
| 328 | let tree = languages::root_tree(&handle, repo.meta.head.unwrap()).unwrap(); |
| 329 | let stats = languages::compute(&handle, tree).unwrap(); |
| 330 | let names: Vec<&str> = stats.languages.iter().map(|(n, _)| *n).collect(); |
| 331 | // Vendored, documentation, minified and attribute-vendored files, and |
| 332 | // prose (README), don't count. |
| 333 | assert_eq!(names, ["Rust", "Shell"], "{stats:?}"); |
| 334 | |
| 335 | // The summary shows the bar once the warmer has run. |
| 336 | let html = ok(&s.app, "/ext").await; |
| 337 | assert!(!html.contains("lang-bar")); |
| 338 | s.state |
| 339 | .cache |
| 340 | .languages(tree, || Ok::<_, std::convert::Infallible>(stats)) |
| 341 | .unwrap(); |
| 342 | let html = ok(&s.app, "/ext").await; |
| 343 | assert!(html.contains(r#"class="lang-bar""#), "{html}"); |
| 344 | assert_eq!( |
| 345 | html.matches(r#"<i class="o0"></i>"#).count() |
| 346 | + html.matches(r#"<i class="o1"></i>"#).count(), |
| 347 | 40 |
| 348 | ); |
| 349 | } |
| 350 | |
| 351 | #[tokio::test] |
| 352 | async fn robots_and_methods() { |
| 353 | let s = site(""); |
| 354 | let robots = ok(&s.app, "/robots.txt").await; |
| 355 | assert!(robots.contains("Disallow: /*/blame/")); |
| 356 | assert!(robots.contains("Disallow: /*/archive/")); |
| 357 | assert!(robots.contains("Disallow: /*?q=")); |
| 358 | assert_eq!( |
| 359 | request(&s.app, Method::POST, "/ext/tree/main").await.status, |
| 360 | StatusCode::METHOD_NOT_ALLOWED |
| 361 | ); |
| 362 | assert_eq!( |
| 363 | get(&s.app, "/ext/info/refs?service=git-receive-pack") |
| 364 | .await |
| 365 | .status, |
| 366 | StatusCode::FORBIDDEN |
| 367 | ); |
| 368 | assert_eq!( |
| 369 | get(&s.app, "/ext/info/refs").await.status, |
| 370 | StatusCode::FORBIDDEN |
| 371 | ); |
| 372 | assert_eq!( |
| 373 | request(&s.app, Method::POST, "/ext.git/git-receive-pack") |
| 374 | .await |
| 375 | .status, |
| 376 | StatusCode::FORBIDDEN |
| 377 | ); |
| 378 | assert_eq!( |
| 379 | get(&s.app, "/ext.git/tree/main").await.status, |
| 380 | StatusCode::NOT_FOUND |
| 381 | ); |
| 382 | let s = site("https_clone = false"); |
| 383 | assert_eq!( |
| 384 | get(&s.app, "/ext/info/refs?service=git-upload-pack") |
| 385 | .await |
| 386 | .status, |
| 387 | StatusCode::NOT_FOUND |
| 388 | ); |
| 389 | } |
| 390 | |
| 391 | #[tokio::test(flavor = "multi_thread")] |
| 392 | async fn clone_over_https() { |
| 393 | let s = site(""); |
| 394 | let listener = tokio::net::TcpListener::bind("127.0.0.1:0").await.unwrap(); |
| 395 | let addr = listener.local_addr().unwrap(); |
| 396 | let app = s.app.clone(); |
| 397 | tokio::spawn(async move { axum::serve(listener, app).await.unwrap() }); |
| 398 | |
| 399 | let dest = tempfile::tempdir().unwrap(); |
| 400 | let head = s.f.head.clone(); |
| 401 | let out = tokio::task::spawn_blocking(move || { |
| 402 | let mut results = Vec::new(); |
| 403 | for (url, dir, protocol) in [ |
| 404 | (format!("http://{addr}/ext.git"), "v2", "2"), |
| 405 | (format!("http://{addr}/ext"), "v0", "0"), |
| 406 | ] { |
| 407 | let target = dest.path().join(dir); |
| 408 | let status = Command::new("git") |
| 409 | .args([ |
| 410 | "-c", |
| 411 | &format!("protocol.version={protocol}"), |
| 412 | "clone", |
| 413 | "-q", |
| 414 | &url, |
| 415 | ]) |
| 416 | .arg(&target) |
| 417 | .env("GIT_CONFIG_NOSYSTEM", "1") |
| 418 | .env("GIT_CONFIG_GLOBAL", "/dev/null") |
| 419 | .env("GIT_TERMINAL_PROMPT", "0") |
| 420 | .status() |
| 421 | .unwrap(); |
| 422 | assert!(status.success(), "clone {url}"); |
| 423 | results.push(git(&target, &["rev-parse", "HEAD"])); |
| 424 | } |
| 425 | // Pushing is refused. |
| 426 | let work = dest.path().join("v2"); |
| 427 | std::fs::write(work.join("new.txt"), "x").unwrap(); |
| 428 | git(&work, &["add", "-A"]); |
| 429 | git(&work, &["commit", "-q", "-m", "x"]); |
| 430 | let push = Command::new("git") |
| 431 | .args(["push", "-q", "origin", "main"]) |
| 432 | .current_dir(&work) |
| 433 | .env("GIT_CONFIG_NOSYSTEM", "1") |
| 434 | .env("GIT_CONFIG_GLOBAL", "/dev/null") |
| 435 | .env("GIT_TERMINAL_PROMPT", "0") |
| 436 | .output() |
| 437 | .unwrap(); |
| 438 | assert!(!push.status.success()); |
| 439 | assert!(String::from_utf8_lossy(&push.stderr).contains("403")); |
| 440 | results |
| 441 | }) |
| 442 | .await |
| 443 | .unwrap(); |
| 444 | assert_eq!(out, vec![head.clone(), head]); |
| 445 | } |