| 1 | //! Repository browsing (M2) and creation. |
| 2 | |
| 3 | use std::collections::HashMap; |
| 4 | use std::path::Path; |
| 5 | |
| 6 | use axum::extract::{Path as UrlPath, Query, State}; |
| 7 | use axum::response::{IntoResponse, Redirect, Response}; |
| 8 | use axum::Form; |
| 9 | use df_db::ids::new_id; |
| 10 | use df_store::StoreError; |
| 11 | use serde::Deserialize; |
| 12 | use uuid::Uuid; |
| 13 | |
| 14 | use crate::error::{AppError, AppResult}; |
| 15 | use crate::repo_ctx::RepoContext; |
| 16 | use crate::state::{AppState, CsrfToken, CurrentUser, Nonce}; |
| 17 | use crate::views::repo as v; |
| 18 | use crate::views::{self, Chrome}; |
| 19 | |
| 20 | /// Map a store error to an HTTP outcome. |
| 21 | /// |
| 22 | /// `NoSuchRepo` becomes a 404 rather than a 500: it means the database and the |
| 23 | /// filesystem disagree, which is a real state after a restore (spec §10) and |
| 24 | /// should look like a missing repository, not a crash. |
| 25 | fn store_err(e: StoreError) -> AppError { |
| 26 | match e { |
| 27 | StoreError::NoSuchRepo | StoreError::NoSuchRevision | StoreError::NoSuchPath => { |
| 28 | AppError::NotFound |
| 29 | } |
| 30 | StoreError::IsDirectory => AppError::BadRequest("that path is a directory".into()), |
| 31 | StoreError::Path(p) => AppError::BadRequest(p.to_string()), |
| 32 | other => AppError::Internal(anyhow::anyhow!(other)), |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | /// `GET /{owner}/{repo}` — code view at the default bookmark. |
| 37 | pub async fn index( |
| 38 | State(state): State<AppState>, |
| 39 | UrlPath((owner, name)): UrlPath<(String, String)>, |
| 40 | CurrentUser(user): CurrentUser, |
| 41 | CsrfToken(csrf): CsrfToken, |
| 42 | Nonce(nonce): Nonce, |
| 43 | ) -> AppResult<Response> { |
| 44 | let ctx = RepoContext::load(&state, &owner, &name, user.as_deref()).await?; |
| 45 | |
| 46 | let empty = state |
| 47 | .store |
| 48 | .is_empty(ctx.store_id()) |
| 49 | .await |
| 50 | .unwrap_or(true); |
| 51 | |
| 52 | let body = if empty { |
| 53 | v::empty_repo( |
| 54 | &state.config.https_clone_url(&ctx.owner, &ctx.repo.name), |
| 55 | &state.config.ssh_clone_url(&ctx.owner, &ctx.repo.name), |
| 56 | &ctx.repo.default_bookmark, |
| 57 | ) |
| 58 | } else { |
| 59 | let rev_label = ctx.repo.default_bookmark.clone(); |
| 60 | let rev = state |
| 61 | .store |
| 62 | .resolve(ctx.store_id(), &rev_label) |
| 63 | .await |
| 64 | .map_err(store_err)?; |
| 65 | let entries = state |
| 66 | .store |
| 67 | .list_tree(ctx.store_id(), &rev, Path::new("")) |
| 68 | .await |
| 69 | .map_err(store_err)?; |
| 70 | |
| 71 | let readme = render_readme(&state, &ctx, &rev, &entries).await; |
| 72 | let tip = tip_commit(&state, &ctx, &rev).await; |
| 73 | let last_commits = last_commits_for(&state, &ctx, &rev, Path::new(""), &entries).await; |
| 74 | |
| 75 | let marks = bookmark_rows(&state, &ctx).await?; |
| 76 | let (contributors, size_bytes) = repo_vitals(&state, &ctx).await; |
| 77 | let https = state.config.https_clone_url(&ctx.owner, &ctx.repo.name); |
| 78 | let ssh = state.config.ssh_clone_url(&ctx.owner, &ctx.repo.name); |
| 79 | let sidebar = v::RepoSidebar { |
| 80 | https: &https, |
| 81 | ssh: &ssh, |
| 82 | open_changes: ctx.nav.open_changes, |
| 83 | conflicted: ctx.nav.conflicted, |
| 84 | open_issues: ctx.nav.open_issues, |
| 85 | contributors, |
| 86 | size_bytes, |
| 87 | bookmarks: &marks, |
| 88 | }; |
| 89 | |
| 90 | v::tree_listing( |
| 91 | &ctx, |
| 92 | v::Tree { |
| 93 | rev_label: &rev_label, |
| 94 | path: "", |
| 95 | entries: &entries, |
| 96 | tip: tip.as_ref(), |
| 97 | readme: readme.as_ref(), |
| 98 | history: &last_commits, |
| 99 | sidebar: Some(&sidebar), |
| 100 | }, |
| 101 | ) |
| 102 | }; |
| 103 | |
| 104 | Ok(render(&ctx, "code", &csrf, &nonce, user.as_deref(), body)) |
| 105 | } |
| 106 | |
| 107 | /// Bookmarks with the change each one points at. |
| 108 | /// |
| 109 | /// The `LATERAL` subquery is scoped to the bookmark's own repository: a |
| 110 | /// revision id is only unique *within* a repository, so joining `revisions` on |
| 111 | /// `rev` alone would let one repository's bookmark resolve to another's change. |
| 112 | /// `(name, protected, updated_at, change_id, number, title)` |
| 113 | type BookmarkTuple = ( |
| 114 | String, |
| 115 | bool, |
| 116 | chrono::DateTime<chrono::Utc>, |
| 117 | Option<String>, |
| 118 | Option<i64>, |
| 119 | Option<String>, |
| 120 | ); |
| 121 | |
| 122 | async fn bookmark_rows(state: &AppState, ctx: &RepoContext) -> AppResult<Vec<v::MarkRow>> { |
| 123 | let rows: Vec<BookmarkTuple> = sqlx::query_as( |
| 124 | r#" |
| 125 | SELECT b.name, |
| 126 | b.protected, |
| 127 | b.updated_at, |
| 128 | c.change_id, |
| 129 | c.number, |
| 130 | c.title |
| 131 | FROM bookmarks b |
| 132 | LEFT JOIN LATERAL ( |
| 133 | SELECT ch.change_id, ch.number, ch.title |
| 134 | FROM revisions r |
| 135 | JOIN changes ch ON ch.id = r.change_id_fk |
| 136 | WHERE ch.repo_id = b.repo_id AND r.rev = b.target |
| 137 | LIMIT 1 |
| 138 | ) c ON true |
| 139 | WHERE b.repo_id = $1 |
| 140 | ORDER BY b.updated_at DESC |
| 141 | "#, |
| 142 | ) |
| 143 | .bind(ctx.repo.id) |
| 144 | .fetch_all(&state.db) |
| 145 | .await?; |
| 146 | |
| 147 | Ok(rows |
| 148 | .into_iter() |
| 149 | .map( |
| 150 | |(name, protected, updated_at, change_id, number, title)| v::MarkRow { |
| 151 | name, |
| 152 | protected, |
| 153 | updated_at, |
| 154 | change_id, |
| 155 | number, |
| 156 | title, |
| 157 | }, |
| 158 | ) |
| 159 | .collect()) |
| 160 | } |
| 161 | |
| 162 | /// Contributor count and on-disk size for the sidebar. |
| 163 | /// |
| 164 | /// Best-effort on both counts: the sidebar is context, and neither number is |
| 165 | /// worth failing a page render over. Contributors are distinct commit *emails* |
| 166 | /// rather than linked accounts — somebody who has pushed but never signed in is |
| 167 | /// still a contributor. |
| 168 | async fn repo_vitals(state: &AppState, ctx: &RepoContext) -> (i64, u64) { |
| 169 | let contributors: i64 = sqlx::query_scalar( |
| 170 | r#" |
| 171 | SELECT count(DISTINCT r.author_email) |
| 172 | FROM revisions r |
| 173 | JOIN changes c ON c.id = r.change_id_fk |
| 174 | WHERE c.repo_id = $1 |
| 175 | "#, |
| 176 | ) |
| 177 | .bind(ctx.repo.id) |
| 178 | .fetch_one(&state.db) |
| 179 | .await |
| 180 | .unwrap_or(0); |
| 181 | |
| 182 | // The stored size is what the last indexer pass recorded; asking the store |
| 183 | // is exact but walks the directory, so it is the fallback rather than the |
| 184 | // first choice on a page that renders on every visit. |
| 185 | let size = if ctx.repo.size_bytes > 0 { |
| 186 | ctx.repo.size_bytes as u64 |
| 187 | } else { |
| 188 | state.store.size_bytes(ctx.store_id()).await.unwrap_or(0) |
| 189 | }; |
| 190 | |
| 191 | (contributors, size) |
| 192 | } |
| 193 | |
| 194 | #[derive(Deserialize)] |
| 195 | pub struct RevPath { |
| 196 | pub owner: String, |
| 197 | pub repo: String, |
| 198 | pub rev: String, |
| 199 | #[serde(default)] |
| 200 | pub path: String, |
| 201 | } |
| 202 | |
| 203 | /// `GET /{owner}/{repo}/tree/{rev}/{path...}` |
| 204 | pub async fn tree( |
| 205 | State(state): State<AppState>, |
| 206 | UrlPath(p): UrlPath<RevPath>, |
| 207 | CurrentUser(user): CurrentUser, |
| 208 | CsrfToken(csrf): CsrfToken, |
| 209 | Nonce(nonce): Nonce, |
| 210 | ) -> AppResult<Response> { |
| 211 | let ctx = RepoContext::load(&state, &p.owner, &p.repo, user.as_deref()).await?; |
| 212 | |
| 213 | let rev = state |
| 214 | .store |
| 215 | .resolve(ctx.store_id(), &p.rev) |
| 216 | .await |
| 217 | .map_err(store_err)?; |
| 218 | |
| 219 | let entries = state |
| 220 | .store |
| 221 | .list_tree(ctx.store_id(), &rev, Path::new(&p.path)) |
| 222 | .await |
| 223 | .map_err(store_err)?; |
| 224 | |
| 225 | let readme = if p.path.is_empty() { |
| 226 | render_readme(&state, &ctx, &rev, &entries).await |
| 227 | } else { |
| 228 | None |
| 229 | }; |
| 230 | |
| 231 | let tip = tip_commit(&state, &ctx, &rev).await; |
| 232 | let last_commits = |
| 233 | last_commits_for(&state, &ctx, &rev, Path::new(&p.path), &entries).await; |
| 234 | |
| 235 | // No sidebar below the root: the reader is looking at files, and repeating |
| 236 | // the clone commands beside every folder is noise. |
| 237 | let body = v::tree_listing( |
| 238 | &ctx, |
| 239 | v::Tree { |
| 240 | rev_label: &p.rev, |
| 241 | path: &p.path, |
| 242 | entries: &entries, |
| 243 | tip: tip.as_ref(), |
| 244 | readme: readme.as_ref(), |
| 245 | history: &last_commits, |
| 246 | sidebar: None, |
| 247 | }, |
| 248 | ); |
| 249 | Ok(render(&ctx, "code", &csrf, &nonce, user.as_deref(), body)) |
| 250 | } |
| 251 | |
| 252 | /// The revision at the tip of what is being browsed. |
| 253 | /// |
| 254 | /// Best-effort: the listing is the point of the page, so a store that cannot |
| 255 | /// produce a log still renders the files rather than failing the request. |
| 256 | async fn tip_commit( |
| 257 | state: &AppState, |
| 258 | ctx: &RepoContext, |
| 259 | rev: &df_store::RevId, |
| 260 | ) -> Option<v::TipCommit> { |
| 261 | let revisions = state.store.log(ctx.store_id(), rev, 1).await.ok()?; |
| 262 | let r = revisions.into_iter().next()?; |
| 263 | |
| 264 | Some(v::TipCommit { |
| 265 | author: r.author.name.clone(), |
| 266 | author_handle: handle_for_email(state, &r.author.email).await, |
| 267 | summary: r.summary().to_string(), |
| 268 | when: r.author.when, |
| 269 | change_id: r.change_id.clone(), |
| 270 | }) |
| 271 | } |
| 272 | |
| 273 | /// The last commit that touched each entry of a directory listing. |
| 274 | /// |
| 275 | /// One `last_commits_in_dir` call resolves the whole directory (see its docs |
| 276 | /// for why that beats a lookup per file). Best-effort like `tip_commit`: a |
| 277 | /// store that cannot walk history still renders the listing, just without |
| 278 | /// this column. |
| 279 | async fn last_commits_for( |
| 280 | state: &AppState, |
| 281 | ctx: &RepoContext, |
| 282 | rev: &df_store::RevId, |
| 283 | dir: &Path, |
| 284 | entries: &[df_store::TreeEntry], |
| 285 | ) -> HashMap<String, v::EntryHistory> { |
| 286 | let names: Vec<String> = entries.iter().map(|e| e.name.clone()).collect(); |
| 287 | |
| 288 | let revisions = state |
| 289 | .store |
| 290 | .last_commits_in_dir(ctx.store_id(), rev, dir, &names) |
| 291 | .await |
| 292 | .unwrap_or_default(); |
| 293 | |
| 294 | revisions |
| 295 | .into_iter() |
| 296 | .map(|(name, r)| { |
| 297 | ( |
| 298 | name, |
| 299 | v::EntryHistory { |
| 300 | summary: r.summary().to_string(), |
| 301 | when: r.author.when, |
| 302 | change_id: r.change_id, |
| 303 | }, |
| 304 | ) |
| 305 | }) |
| 306 | .collect() |
| 307 | } |
| 308 | |
| 309 | /// The account that owns a commit-author email, if any. |
| 310 | /// |
| 311 | /// The same rule the indexer attributes changes by, applied to a commit read |
| 312 | /// straight from the store rather than from the index. Best-effort: a lookup |
| 313 | /// failure renders the commit's own name rather than failing the page. |
| 314 | pub(crate) async fn handle_for_email(state: &AppState, email: &str) -> Option<String> { |
| 315 | let email = email.trim(); |
| 316 | if email.is_empty() { |
| 317 | return None; |
| 318 | } |
| 319 | sqlx::query_scalar::<_, String>("SELECT handle::text FROM users WHERE email = $1") |
| 320 | .bind(email) |
| 321 | .fetch_optional(&state.db) |
| 322 | .await |
| 323 | .ok() |
| 324 | .flatten() |
| 325 | } |
| 326 | |
| 327 | /// The same lookup for a whole listing, in one round trip. |
| 328 | /// |
| 329 | /// A page of 200 commits must not become 200 queries. Returns email → handle |
| 330 | /// for the ones that matched; callers fall back to the commit's own name. |
| 331 | pub(crate) async fn handles_for_emails<'a>( |
| 332 | state: &AppState, |
| 333 | emails: impl IntoIterator<Item = &'a str>, |
| 334 | ) -> HashMap<String, String> { |
| 335 | let mut wanted: Vec<String> = emails |
| 336 | .into_iter() |
| 337 | .map(str::trim) |
| 338 | .filter(|e| !e.is_empty()) |
| 339 | .map(str::to_owned) |
| 340 | .collect(); |
| 341 | wanted.sort(); |
| 342 | wanted.dedup(); |
| 343 | |
| 344 | if wanted.is_empty() { |
| 345 | return HashMap::new(); |
| 346 | } |
| 347 | |
| 348 | // `email` is citext, so the join is case-insensitive without lowering here. |
| 349 | sqlx::query_as::<_, (String, String)>( |
| 350 | "SELECT email::text, handle::text FROM users WHERE email = ANY($1)", |
| 351 | ) |
| 352 | .bind(&wanted) |
| 353 | .fetch_all(&state.db) |
| 354 | .await |
| 355 | .unwrap_or_else(|e| { |
| 356 | tracing::warn!("resolving commit authors failed: {e}"); |
| 357 | Vec::new() |
| 358 | }) |
| 359 | .into_iter() |
| 360 | .collect() |
| 361 | } |
| 362 | |
| 363 | /// Is this path a markdown document? |
| 364 | /// |
| 365 | /// Extension-based on purpose: sniffing content would mean a file that happens |
| 366 | /// to start with a `#` gets rendered as a heading, which is worse than a `.txt` |
| 367 | /// missing out on rendering. |
| 368 | fn is_markdown(path: &str) -> bool { |
| 369 | let lower = path.to_lowercase(); |
| 370 | lower.ends_with(".md") || lower.ends_with(".markdown") |
| 371 | } |
| 372 | |
| 373 | #[derive(Deserialize, Default)] |
| 374 | pub struct BlobQuery { |
| 375 | /// `source` shows the highlighted source of a markdown file instead of the |
| 376 | /// rendered document. Anything else — including absent — renders. |
| 377 | #[serde(default)] |
| 378 | pub view: Option<String>, |
| 379 | /// When `blame=1`, show per-line blame annotations. |
| 380 | #[serde(default)] |
| 381 | pub blame: Option<String>, |
| 382 | } |
| 383 | |
| 384 | /// `GET /{owner}/{repo}/blob/{rev}/{path...}` |
| 385 | pub async fn blob( |
| 386 | State(state): State<AppState>, |
| 387 | UrlPath(p): UrlPath<RevPath>, |
| 388 | Query(q): Query<BlobQuery>, |
| 389 | CurrentUser(user): CurrentUser, |
| 390 | CsrfToken(csrf): CsrfToken, |
| 391 | Nonce(nonce): Nonce, |
| 392 | ) -> AppResult<Response> { |
| 393 | let ctx = RepoContext::load(&state, &p.owner, &p.repo, user.as_deref()).await?; |
| 394 | |
| 395 | let rev = state |
| 396 | .store |
| 397 | .resolve(ctx.store_id(), &p.rev) |
| 398 | .await |
| 399 | .map_err(store_err)?; |
| 400 | |
| 401 | let wants_source = q.view.as_deref() == Some("source"); |
| 402 | let wants_blame = q.blame.as_deref() == Some("1"); |
| 403 | let markdown_path = is_markdown(&p.path); |
| 404 | |
| 405 | // Fetch the file tree for the sidebar (directory containing this file). |
| 406 | let tree_dir = match p.path.rsplit_once('/') { |
| 407 | Some((parent, _)) => parent.to_string(), |
| 408 | None => String::new(), |
| 409 | }; |
| 410 | let sidebar_entries = state |
| 411 | .store |
| 412 | .list_tree(ctx.store_id(), &rev, Path::new(&tree_dir)) |
| 413 | .await |
| 414 | .unwrap_or_default(); |
| 415 | |
| 416 | // Fetch the last commit that modified this file. |
| 417 | let last_commit = state |
| 418 | .store |
| 419 | .last_commit_for_path(ctx.store_id(), &rev, Path::new(&p.path)) |
| 420 | .await |
| 421 | .ok() |
| 422 | .flatten(); |
| 423 | |
| 424 | let last_commit_handle = match &last_commit { |
| 425 | Some(c) => handle_for_email(&state, &c.author.email).await, |
| 426 | None => None, |
| 427 | }; |
| 428 | |
| 429 | // Optionally fetch blame. |
| 430 | let blame_lines = if wants_blame { |
| 431 | state |
| 432 | .store |
| 433 | .blame(ctx.store_id(), &rev, Path::new(&p.path)) |
| 434 | .await |
| 435 | .ok() |
| 436 | } else { |
| 437 | None |
| 438 | }; |
| 439 | |
| 440 | let body = match state |
| 441 | .store |
| 442 | .read_blob(ctx.store_id(), &rev, Path::new(&p.path)) |
| 443 | .await |
| 444 | { |
| 445 | Ok(b) => { |
| 446 | let limit = state.config.max_blob_render_bytes as u64; |
| 447 | if b.size > limit { |
| 448 | v::blob_view( |
| 449 | &ctx, |
| 450 | &p.rev, |
| 451 | &p.path, |
| 452 | v::BlobBody::TooLarge { size: b.size, limit }, |
| 453 | None, |
| 454 | &v::BlobExtras::default(), |
| 455 | ) |
| 456 | } else if b.text().is_some() { |
| 457 | // Highlighting is cached by content address, so re-viewing a |
| 458 | // file — or viewing it at a revision that did not change it — |
| 459 | // never re-parses (spec §8). |
| 460 | let hl = crate::highlight_cache::render(&state.db, &b).await; |
| 461 | let text = b.text().unwrap_or(""); |
| 462 | let lines = text.lines().count(); |
| 463 | |
| 464 | // Always sanitised — never trust repository content (spec §9). |
| 465 | let rendered = (markdown_path && !wants_source) |
| 466 | .then(|| v::rendered_markdown(&df_render::markdown_to_html(text))); |
| 467 | |
| 468 | let markdown = match (markdown_path, &rendered) { |
| 469 | (true, Some(doc)) => Some(v::MarkdownView::Rendered(doc)), |
| 470 | (true, None) => Some(v::MarkdownView::Source), |
| 471 | (false, _) => None, |
| 472 | }; |
| 473 | |
| 474 | // Extract symbols for the outline panel. |
| 475 | let symbols = df_render::symbols::extract_symbols(&p.path, text); |
| 476 | |
| 477 | let extras = v::BlobExtras { |
| 478 | sidebar_entries: &sidebar_entries, |
| 479 | sidebar_dir: &tree_dir, |
| 480 | symbols: &symbols, |
| 481 | last_commit: last_commit.as_ref(), |
| 482 | last_commit_handle: last_commit_handle.as_deref(), |
| 483 | blame: blame_lines.as_deref(), |
| 484 | wants_blame, |
| 485 | }; |
| 486 | |
| 487 | v::blob_view( |
| 488 | &ctx, |
| 489 | &p.rev, |
| 490 | &p.path, |
| 491 | v::BlobBody::Text { |
| 492 | content: text, |
| 493 | lines, |
| 494 | highlighted: &hl.lines, |
| 495 | language: hl.language.as_deref(), |
| 496 | plain_reason: hl.skipped.and_then(plain_reason), |
| 497 | }, |
| 498 | markdown, |
| 499 | &extras, |
| 500 | ) |
| 501 | } else { |
| 502 | v::blob_view( |
| 503 | &ctx, |
| 504 | &p.rev, |
| 505 | &p.path, |
| 506 | v::BlobBody::Binary { size: b.size }, |
| 507 | None, |
| 508 | &v::BlobExtras::default(), |
| 509 | ) |
| 510 | } |
| 511 | } |
| 512 | // The store's own cap fired before ours. |
| 513 | Err(StoreError::TooLarge { size, limit }) => { |
| 514 | v::blob_view( |
| 515 | &ctx, |
| 516 | &p.rev, |
| 517 | &p.path, |
| 518 | v::BlobBody::TooLarge { size, limit }, |
| 519 | None, |
| 520 | &v::BlobExtras::default(), |
| 521 | ) |
| 522 | } |
| 523 | Err(e) => return Err(store_err(e)), |
| 524 | }; |
| 525 | |
| 526 | Ok(render(&ctx, "code", &csrf, &nonce, user.as_deref(), body)) |
| 527 | } |
| 528 | |
| 529 | /// `GET /{owner}/{repo}/raw/{rev}/{path...}` |
| 530 | /// |
| 531 | /// Spec §9: "Blob content served ... with `Content-Disposition: attachment` and |
| 532 | /// `X-Content-Type-Options: nosniff` — never serve user-controlled HTML on the |
| 533 | /// app origin." Everything here is `application/octet-stream` and downloaded, |
| 534 | /// so a repository containing an HTML page cannot execute against our origin. |
| 535 | pub async fn raw( |
| 536 | State(state): State<AppState>, |
| 537 | UrlPath(p): UrlPath<RevPath>, |
| 538 | CurrentUser(user): CurrentUser, |
| 539 | ) -> AppResult<Response> { |
| 540 | let ctx = RepoContext::load(&state, &p.owner, &p.repo, user.as_deref()).await?; |
| 541 | |
| 542 | let rev = state |
| 543 | .store |
| 544 | .resolve(ctx.store_id(), &p.rev) |
| 545 | .await |
| 546 | .map_err(store_err)?; |
| 547 | |
| 548 | let blob = state |
| 549 | .store |
| 550 | .read_blob(ctx.store_id(), &rev, Path::new(&p.path)) |
| 551 | .await |
| 552 | .map_err(store_err)?; |
| 553 | |
| 554 | let filename = p |
| 555 | .path |
| 556 | .rsplit('/') |
| 557 | .next() |
| 558 | .filter(|s| !s.is_empty()) |
| 559 | .unwrap_or("file"); |
| 560 | |
| 561 | // The filename reaches a header; strip anything that could break out of the |
| 562 | // quoted string or inject a second header. |
| 563 | let safe_name: String = filename |
| 564 | .chars() |
| 565 | .filter(|c| !matches!(c, '"' | '\\' | '\r' | '\n') && !c.is_control()) |
| 566 | .take(200) |
| 567 | .collect(); |
| 568 | |
| 569 | Ok(( |
| 570 | [ |
| 571 | ( |
| 572 | axum::http::header::CONTENT_TYPE, |
| 573 | "application/octet-stream".to_string(), |
| 574 | ), |
| 575 | ( |
| 576 | axum::http::header::CONTENT_DISPOSITION, |
| 577 | format!("attachment; filename=\"{safe_name}\""), |
| 578 | ), |
| 579 | ( |
| 580 | axum::http::header::X_CONTENT_TYPE_OPTIONS, |
| 581 | "nosniff".to_string(), |
| 582 | ), |
| 583 | ], |
| 584 | blob.content, |
| 585 | ) |
| 586 | .into_response()) |
| 587 | } |
| 588 | |
| 589 | #[derive(Deserialize)] |
| 590 | pub struct LogQuery { |
| 591 | pub rev: Option<String>, |
| 592 | pub limit: Option<usize>, |
| 593 | } |
| 594 | |
| 595 | /// `GET /{owner}/{repo}/log` |
| 596 | pub async fn log( |
| 597 | State(state): State<AppState>, |
| 598 | UrlPath((owner, name)): UrlPath<(String, String)>, |
| 599 | Query(q): Query<LogQuery>, |
| 600 | CurrentUser(user): CurrentUser, |
| 601 | CsrfToken(csrf): CsrfToken, |
| 602 | Nonce(nonce): Nonce, |
| 603 | ) -> AppResult<Response> { |
| 604 | let ctx = RepoContext::load(&state, &owner, &name, user.as_deref()).await?; |
| 605 | |
| 606 | let rev_label = q.rev.unwrap_or_else(|| ctx.repo.default_bookmark.clone()); |
| 607 | let rev = state |
| 608 | .store |
| 609 | .resolve(ctx.store_id(), &rev_label) |
| 610 | .await |
| 611 | .map_err(store_err)?; |
| 612 | |
| 613 | // Clamped: a URL must not be able to ask for unbounded history. |
| 614 | let limit = q.limit.unwrap_or(50).clamp(1, 200); |
| 615 | let revs = state |
| 616 | .store |
| 617 | .log(ctx.store_id(), &rev, limit) |
| 618 | .await |
| 619 | .map_err(store_err)?; |
| 620 | |
| 621 | // One query for every author in the page rather than one per commit. |
| 622 | let handles = |
| 623 | handles_for_emails(&state, revs.iter().map(|r| r.author.email.as_str())).await; |
| 624 | |
| 625 | let body = v::log_view(&ctx, &rev_label, &revs, &handles); |
| 626 | Ok(render(&ctx, "code", &csrf, &nonce, user.as_deref(), body)) |
| 627 | } |
| 628 | |
| 629 | #[derive(Deserialize, Default)] |
| 630 | pub struct CommitQuery { |
| 631 | /// Fold every file. A link rather than a script, so the folded view is a |
| 632 | /// URL and works with scripting off. |
| 633 | pub collapse: Option<String>, |
| 634 | } |
| 635 | |
| 636 | /// `GET /{owner}/{repo}/commit/{rev}` |
| 637 | /// |
| 638 | /// One commit and its patch against the first parent. A root commit has no |
| 639 | /// parent and diffs against an empty tree, so the first commit in a repository |
| 640 | /// renders as an all-additions patch rather than as an error. |
| 641 | pub async fn commit( |
| 642 | State(state): State<AppState>, |
| 643 | UrlPath((owner, name, rev_spec)): UrlPath<(String, String, String)>, |
| 644 | Query(q): Query<CommitQuery>, |
| 645 | CurrentUser(user): CurrentUser, |
| 646 | CsrfToken(csrf): CsrfToken, |
| 647 | Nonce(nonce): Nonce, |
| 648 | ) -> AppResult<Response> { |
| 649 | let ctx = RepoContext::load(&state, &owner, &name, user.as_deref()).await?; |
| 650 | |
| 651 | // Bookmark names resolve too: `/commit/main` is a reasonable thing to type, |
| 652 | // and it lands on whichever commit that name points at right now. |
| 653 | let rev = state |
| 654 | .store |
| 655 | .resolve(ctx.store_id(), &rev_spec) |
| 656 | .await |
| 657 | .map_err(store_err)?; |
| 658 | |
| 659 | let revision = state |
| 660 | .store |
| 661 | .revision(ctx.store_id(), &rev) |
| 662 | .await |
| 663 | .map_err(store_err)?; |
| 664 | |
| 665 | let opts = df_store::DiffOpts { |
| 666 | context_lines: 3, |
| 667 | max_files: state.config.max_diff_files, |
| 668 | max_lines: state.config.max_diff_lines, |
| 669 | }; |
| 670 | |
| 671 | // Best-effort: a commit whose patch cannot be rendered still has a message, |
| 672 | // an author and parents worth showing, and the view says so in place of the |
| 673 | // diff rather than 500-ing the page. |
| 674 | let diff = match state.store.diff_from_parent(ctx.store_id(), &rev, opts).await { |
| 675 | Ok(d) => Some(d), |
| 676 | Err(e) => { |
| 677 | tracing::warn!("diffing {rev} failed: {e}"); |
| 678 | None |
| 679 | } |
| 680 | }; |
| 681 | |
| 682 | let handles = handles_for_emails( |
| 683 | &state, |
| 684 | [revision.author.email.as_str(), revision.committer.email.as_str()], |
| 685 | ) |
| 686 | .await; |
| 687 | |
| 688 | let body = v::commit_view( |
| 689 | &ctx, |
| 690 | v::CommitPage { |
| 691 | rev: &revision, |
| 692 | author_handle: handles.get(&revision.author.email).map(String::as_str), |
| 693 | committer_handle: handles.get(&revision.committer.email).map(String::as_str), |
| 694 | diff: diff.as_ref(), |
| 695 | collapsed: q.collapse.is_some(), |
| 696 | }, |
| 697 | ); |
| 698 | |
| 699 | Ok(render(&ctx, "code", &csrf, &nonce, user.as_deref(), body)) |
| 700 | } |
| 701 | |
| 702 | /// `GET /{owner}/{repo}/bookmarks` |
| 703 | pub async fn bookmarks( |
| 704 | State(state): State<AppState>, |
| 705 | UrlPath((owner, name)): UrlPath<(String, String)>, |
| 706 | CurrentUser(user): CurrentUser, |
| 707 | CsrfToken(csrf): CsrfToken, |
| 708 | Nonce(nonce): Nonce, |
| 709 | ) -> AppResult<Response> { |
| 710 | let ctx = RepoContext::load(&state, &owner, &name, user.as_deref()).await?; |
| 711 | let marks = bookmark_rows(&state, &ctx).await?; |
| 712 | |
| 713 | let body = v::bookmarks_view(&ctx, &marks); |
| 714 | Ok(render(&ctx, "bookmarks", &csrf, &nonce, user.as_deref(), body)) |
| 715 | } |
| 716 | |
| 717 | /// Explain a plain-text fallback to the reader. |
| 718 | /// |
| 719 | /// "No grammar for this file type" is not worth saying — most files in a |
| 720 | /// repository have no grammar and the absence of colour is unremarkable. A file |
| 721 | /// that *would* have been highlighted but was skipped for size is worth saying, |
| 722 | /// because the reader can otherwise not tell it from a broken renderer. |
| 723 | fn plain_reason(s: df_render::highlight::Skipped) -> Option<&'static str> { |
| 724 | use df_render::highlight::Skipped; |
| 725 | match s { |
| 726 | Skipped::NoGrammar => None, |
| 727 | Skipped::TooLarge => Some("too large to highlight"), |
| 728 | Skipped::LineTooLong => Some("lines too long to highlight"), |
| 729 | Skipped::Failed => Some("could not be highlighted"), |
| 730 | } |
| 731 | } |
| 732 | |
| 733 | // ─── creation ──────────────────────────────────────────────────────────────── |
| 734 | |
| 735 | /// `GET /new` |
| 736 | pub async fn new_form( |
| 737 | State(state): State<AppState>, |
| 738 | CurrentUser(user): CurrentUser, |
| 739 | CsrfToken(csrf): CsrfToken, |
| 740 | Nonce(nonce): Nonce, |
| 741 | ) -> AppResult<Response> { |
| 742 | let Some(user) = user else { |
| 743 | return Err(AppError::Unauthorized); |
| 744 | }; |
| 745 | let owners = owner_choices(&state, &user).await?; |
| 746 | |
| 747 | Ok(views::page( |
| 748 | Chrome { title: "New repository", user: Some(&user), csrf: &csrf, nonce: &nonce }, |
| 749 | v::new_repo_form(&csrf, None, &owners), |
| 750 | ) |
| 751 | .into_response()) |
| 752 | } |
| 753 | |
| 754 | #[derive(Deserialize)] |
| 755 | pub struct CreateRepo { |
| 756 | pub owner: String, |
| 757 | pub name: String, |
| 758 | pub description: Option<String>, |
| 759 | pub default_bookmark: String, |
| 760 | pub private: Option<String>, |
| 761 | } |
| 762 | |
| 763 | /// `POST /repos` |
| 764 | pub async fn create( |
| 765 | State(state): State<AppState>, |
| 766 | CurrentUser(user): CurrentUser, |
| 767 | CsrfToken(csrf): CsrfToken, |
| 768 | Nonce(nonce): Nonce, |
| 769 | Form(form): Form<CreateRepo>, |
| 770 | ) -> AppResult<Response> { |
| 771 | let Some(user) = user else { |
| 772 | return Err(AppError::Unauthorized); |
| 773 | }; |
| 774 | |
| 775 | let owners = owner_choices(&state, &user).await?; |
| 776 | let reject = |msg: &str| -> Response { |
| 777 | views::page( |
| 778 | Chrome { title: "New repository", user: Some(&user), csrf: &csrf, nonce: &nonce }, |
| 779 | v::new_repo_form(&csrf, Some(msg), &owners), |
| 780 | ) |
| 781 | .into_response() |
| 782 | }; |
| 783 | |
| 784 | let name = form.name.trim(); |
| 785 | if !valid_repo_name(name) { |
| 786 | return Ok(reject( |
| 787 | "Repository names must start with a letter or digit and may contain \ |
| 788 | letters, digits, dots, hyphens and underscores.", |
| 789 | )); |
| 790 | } |
| 791 | |
| 792 | let bookmark = form.default_bookmark.trim(); |
| 793 | if !valid_bookmark_name(bookmark) { |
| 794 | return Ok(reject("That is not a valid bookmark name.")); |
| 795 | } |
| 796 | |
| 797 | // The namespace must be the user's own, or an org they administer. Checked |
| 798 | // against the database rather than against the form's own option list — the |
| 799 | // list is a convenience, not a control. |
| 800 | let owner_org: Option<Uuid> = if form.owner == user.handle { |
| 801 | None |
| 802 | } else { |
| 803 | let org: Option<(Uuid,)> = sqlx::query_as( |
| 804 | "SELECT o.id FROM orgs o |
| 805 | JOIN org_members m ON m.org_id = o.id |
| 806 | WHERE o.handle = $1 AND m.user_id = $2 AND m.role = 'admin'", |
| 807 | ) |
| 808 | .bind(&form.owner) |
| 809 | .bind(user.id) |
| 810 | .fetch_optional(&state.db) |
| 811 | .await?; |
| 812 | |
| 813 | match org { |
| 814 | Some((id,)) => Some(id), |
| 815 | None => return Err(AppError::Forbidden), |
| 816 | } |
| 817 | }; |
| 818 | |
| 819 | let repo_id = new_id(); |
| 820 | let visibility = if form.private.is_some() { "private" } else { "public" }; |
| 821 | |
| 822 | // Create the database row first. If storage creation then fails we can |
| 823 | // roll the row back; the reverse ordering would leave an orphan directory |
| 824 | // with no owner and no way to reach it. |
| 825 | let mut tx = state.db.begin().await?; |
| 826 | |
| 827 | let inserted = sqlx::query( |
| 828 | "INSERT INTO repos (id, owner_kind, owner_user_id, owner_org_id, name, description, |
| 829 | visibility, default_bookmark) |
| 830 | VALUES ($1, |
| 831 | CASE WHEN $7::uuid IS NULL THEN 'user' ELSE 'org' END::owner_kind, |
| 832 | CASE WHEN $7::uuid IS NULL THEN $2 ELSE NULL END, |
| 833 | $7, $3, $4, $5::visibility, $6) |
| 834 | ON CONFLICT DO NOTHING", |
| 835 | ) |
| 836 | .bind(repo_id) |
| 837 | .bind(user.id) |
| 838 | .bind(name) |
| 839 | .bind(form.description.as_deref().filter(|d| !d.trim().is_empty())) |
| 840 | .bind(visibility) |
| 841 | .bind(bookmark) |
| 842 | .bind(owner_org) |
| 843 | .execute(&mut *tx) |
| 844 | .await?; |
| 845 | |
| 846 | if inserted.rows_affected() == 0 { |
| 847 | return Ok(reject("That namespace already has a repository with that name.")); |
| 848 | } |
| 849 | |
| 850 | sqlx::query("INSERT INTO repo_counters (repo_id) VALUES ($1)") |
| 851 | .bind(repo_id) |
| 852 | .execute(&mut *tx) |
| 853 | .await?; |
| 854 | |
| 855 | // Default labels, so the issue form has something to offer on day one. |
| 856 | // Kept in step with the backfill in migration 0003. |
| 857 | for (label, color) in [ |
| 858 | ("bug", "#d06b6b"), |
| 859 | ("enhancement", "#6ba9b8"), |
| 860 | ("question", "#d9a441"), |
| 861 | ("documentation", "#8b7fd4"), |
| 862 | ] { |
| 863 | sqlx::query( |
| 864 | "INSERT INTO labels (id, repo_id, name, color) VALUES ($1, $2, $3, $4) |
| 865 | ON CONFLICT DO NOTHING", |
| 866 | ) |
| 867 | .bind(new_id()) |
| 868 | .bind(repo_id) |
| 869 | .bind(label) |
| 870 | .bind(color) |
| 871 | .execute(&mut *tx) |
| 872 | .await?; |
| 873 | } |
| 874 | |
| 875 | sqlx::query( |
| 876 | "INSERT INTO events (id, repo_id, actor_id, kind, subject_type, subject_id) |
| 877 | VALUES ($1, $2, $3, 'repo.created', 'repo', $2)", |
| 878 | ) |
| 879 | .bind(new_id()) |
| 880 | .bind(repo_id) |
| 881 | .bind(user.id) |
| 882 | .execute(&mut *tx) |
| 883 | .await?; |
| 884 | |
| 885 | if let Err(e) = state |
| 886 | .store |
| 887 | .create(df_store::RepoId(repo_id), bookmark) |
| 888 | .await |
| 889 | { |
| 890 | tracing::error!("creating repository storage failed: {e}"); |
| 891 | tx.rollback().await?; |
| 892 | return Ok(reject( |
| 893 | "The repository could not be created on disk. Please try again.", |
| 894 | )); |
| 895 | } |
| 896 | |
| 897 | // Push validation, before the repository can receive a push. A repository |
| 898 | // without this hook accepts anything: no ref-name allowlist, no protected |
| 899 | // bookmarks (spec §9). Failing to install it fails the creation rather than |
| 900 | // producing a repository that is quietly unvalidated. |
| 901 | if let Err(e) = |
| 902 | crate::hooks::install(state.store.as_ref(), repo_id, &state.config.hook_binary).await |
| 903 | { |
| 904 | tracing::error!(repo = %repo_id, "{e:#}"); |
| 905 | tx.rollback().await?; |
| 906 | let _ = state.store.delete(df_store::RepoId(repo_id)).await; |
| 907 | return Ok(reject( |
| 908 | "The repository could not be prepared to receive pushes. Please try again.", |
| 909 | )); |
| 910 | } |
| 911 | |
| 912 | tx.commit().await?; |
| 913 | tracing::info!(repo = %repo_id, owner = %form.owner, name = %name, "repository created"); |
| 914 | |
| 915 | Ok(Redirect::to(&format!("/{}/{}", form.owner, name)).into_response()) |
| 916 | } |
| 917 | |
| 918 | // ─── helpers ───────────────────────────────────────────────────────────────── |
| 919 | |
| 920 | /// Namespaces this user may create a repository in: their own, plus every org |
| 921 | /// they administer. |
| 922 | /// |
| 923 | /// Org *members* are deliberately excluded — membership grants read access to |
| 924 | /// the org's repositories, not the right to add one. |
| 925 | async fn owner_choices(state: &AppState, user: &df_db::models::User) -> AppResult<Vec<String>> { |
| 926 | let mut owners = vec![user.handle.clone()]; |
| 927 | let orgs: Vec<String> = sqlx::query_scalar( |
| 928 | "SELECT o.handle::text FROM orgs o |
| 929 | JOIN org_members m ON m.org_id = o.id |
| 930 | WHERE m.user_id = $1 AND m.role = 'admin' |
| 931 | ORDER BY o.handle", |
| 932 | ) |
| 933 | .bind(user.id) |
| 934 | .fetch_all(&state.db) |
| 935 | .await?; |
| 936 | owners.extend(orgs); |
| 937 | Ok(owners) |
| 938 | } |
| 939 | |
| 940 | /// Mirror of the `name_format` CHECK on `repos`. |
| 941 | fn valid_repo_name(n: &str) -> bool { |
| 942 | !n.is_empty() |
| 943 | && n.len() <= 100 |
| 944 | && !n.contains("..") |
| 945 | && n.chars().next().is_some_and(|c| c.is_ascii_alphanumeric()) |
| 946 | && n.chars() |
| 947 | .all(|c| c.is_ascii_alphanumeric() || matches!(c, '.' | '_' | '-')) |
| 948 | } |
| 949 | |
| 950 | /// Git ref-name rules, plus the stricter allowlist from spec §9. |
| 951 | pub fn valid_bookmark_name(n: &str) -> bool { |
| 952 | !n.is_empty() |
| 953 | && n.len() <= 100 |
| 954 | && !n.contains("..") |
| 955 | && !n.ends_with(".lock") |
| 956 | && !n.starts_with('-') |
| 957 | && !n.starts_with('/') |
| 958 | && !n.ends_with('/') |
| 959 | && !n.contains("//") |
| 960 | && !n.contains('\\') |
| 961 | && !n.contains("@{") |
| 962 | && n.chars() |
| 963 | .all(|c| c.is_ascii_alphanumeric() || matches!(c, '/' | '.' | '_' | '-')) |
| 964 | } |
| 965 | |
| 966 | fn render( |
| 967 | ctx: &RepoContext, |
| 968 | tab: &str, |
| 969 | csrf: &str, |
| 970 | nonce: &str, |
| 971 | user: Option<&df_db::models::User>, |
| 972 | body: maud::Markup, |
| 973 | ) -> Response { |
| 974 | views::page_with_bar( |
| 975 | Chrome { |
| 976 | title: &format!("{}/{}", ctx.owner, ctx.repo.name), |
| 977 | user, |
| 978 | csrf, |
| 979 | nonce, |
| 980 | }, |
| 981 | v::header(ctx, tab), |
| 982 | body, |
| 983 | ) |
| 984 | .into_response() |
| 985 | } |
| 986 | |
| 987 | /// Render a README if the directory has one. |
| 988 | /// |
| 989 | /// Returns the file's own name alongside the markup so the card can be labelled |
| 990 | /// with what was actually found — a repository with a bare `README` should not |
| 991 | /// be told it has a `README.md`. |
| 992 | async fn render_readme( |
| 993 | state: &AppState, |
| 994 | ctx: &RepoContext, |
| 995 | rev: &df_store::RevId, |
| 996 | entries: &[df_store::TreeEntry], |
| 997 | ) -> Option<(String, maud::Markup)> { |
| 998 | let readme = entries.iter().find(|e| { |
| 999 | !e.is_dir() |
| 1000 | && matches!( |
| 1001 | e.name.to_lowercase().as_str(), |
| 1002 | "readme.md" | "readme" | "readme.markdown" | "readme.txt" |
| 1003 | ) |
| 1004 | })?; |
| 1005 | |
| 1006 | let blob = state |
| 1007 | .store |
| 1008 | .read_blob(ctx.store_id(), rev, Path::new(&readme.path)) |
| 1009 | .await |
| 1010 | .ok()?; |
| 1011 | let text = blob.text()?; |
| 1012 | |
| 1013 | // Always sanitised — never trust repository content (spec §9). |
| 1014 | let html_str = df_render::markdown_to_html(text); |
| 1015 | Some((readme.name.clone(), v::rendered_markdown(&html_str))) |
| 1016 | } |
| 1017 | |
| 1018 | #[cfg(test)] |
| 1019 | mod tests { |
| 1020 | use super::*; |
| 1021 | |
| 1022 | #[test] |
| 1023 | fn repo_name_validation_mirrors_the_db_constraint() { |
| 1024 | assert!(valid_repo_name("dogfood")); |
| 1025 | assert!(valid_repo_name("my.repo_name-1")); |
| 1026 | assert!(valid_repo_name("9lives")); |
| 1027 | |
| 1028 | assert!(!valid_repo_name("")); |
| 1029 | assert!(!valid_repo_name(".hidden"), "must start alphanumeric"); |
| 1030 | assert!(!valid_repo_name("-dash")); |
| 1031 | assert!(!valid_repo_name("a..b"), "traversal must be rejected"); |
| 1032 | assert!(!valid_repo_name("with space")); |
| 1033 | assert!(!valid_repo_name("with/slash")); |
| 1034 | assert!(!valid_repo_name(&"a".repeat(101))); |
| 1035 | } |
| 1036 | |
| 1037 | #[test] |
| 1038 | fn bookmark_validation_follows_the_ref_rules() { |
| 1039 | assert!(valid_bookmark_name("main")); |
| 1040 | assert!(valid_bookmark_name("feature/thing")); |
| 1041 | assert!(valid_bookmark_name("v1.2.3")); |
| 1042 | |
| 1043 | // Spec §9: reject refs containing `..`, ending in `.lock`, starting |
| 1044 | // with `-`, or containing control characters. |
| 1045 | assert!(!valid_bookmark_name("a..b")); |
| 1046 | assert!(!valid_bookmark_name("main.lock")); |
| 1047 | assert!(!valid_bookmark_name("-main")); |
| 1048 | assert!(!valid_bookmark_name("main\n")); |
| 1049 | assert!(!valid_bookmark_name("main\x1b[31m")); |
| 1050 | assert!(!valid_bookmark_name("a//b")); |
| 1051 | assert!(!valid_bookmark_name("/main")); |
| 1052 | assert!(!valid_bookmark_name("main/")); |
| 1053 | assert!(!valid_bookmark_name("main@{1}")); |
| 1054 | assert!(!valid_bookmark_name("back\\slash")); |
| 1055 | assert!(!valid_bookmark_name("")); |
| 1056 | } |
| 1057 | } |
1057 lines · Rust