crates/cli/ents-web/src/pages/files.rs
files.rshistorycomment on this file
| 1 | //! `GET /files`, `GET /files/{*path}`: a read-only directory listing and |
| 2 | //! blob viewer over the `HEAD` tree of the repository `git ents serve` is |
| 3 | //! serving. A `.md` blob renders via `crate::markdown`, a |
| 4 | //! `.adoc`/`.asciidoc`/`.asc`/`.adc` blob via `crate::asciidoc`, and |
| 5 | //! everything else as a line-numbered source view, syntax-highlighted via |
| 6 | //! [`arborium`] when its filename maps to a known grammar (ported from |
| 7 | //! `pre-redo:crates/git-ents-server/src/web/pages.rs`'s own `highlight`; |
| 8 | //! see `highlight`'s own doc), escaped plain text otherwise. |
| 9 | //! |
| 10 | //! Tree/blob reads go through `gix`'s high-level `Repository`/`Tree`/`Blob` |
| 11 | //! types (`repo.head_tree()`, `Tree::lookup_entry_by_path`, |
| 12 | //! `Entry::object`), opened fresh per request from `state.path` -- the |
| 13 | //! same `gix::open(repo_path)` pattern `ents_forge::comment::add`/`show` |
| 14 | //! already use to browse a live working tree, not the |
| 15 | //! `facet-git-tree`/`gix_object::Find` convention the rest of this crate's |
| 16 | //! pages use to read typed meta-ref entities (`facet-git-tree` is for |
| 17 | //! structured meta-ref data; browsing arbitrary repository content is not |
| 18 | //! that). |
| 19 | //! |
| 20 | //! `crumbs` renders only the path trail now -- every action that used to |
| 21 | //! live at its trailing edge (jump into history, jump to the first |
| 22 | //! comment, add a comment) moved into `blob_header`'s own right-aligned |
| 23 | //! action group, rendered above every blob view regardless of how it |
| 24 | //! renders (raw source, a rendered document, or a binary placeholder); a |
| 25 | //! directory listing carries neither the actions nor a header, since a |
| 26 | //! comment anchors to a file, never a tree. |
| 27 | //! |
| 28 | //! A blob view also loads and renders the comments anchored to it |
| 29 | //! (`crate::pages::comments::for_path`). A raw-source view (not a |
| 30 | //! rendered document or a binary placeholder) interleaves each comment's |
| 31 | //! card directly after the row naming its anchored range's last line, |
| 32 | //! full width across the blob's line-number and code columns |
| 33 | //! (`source_view`); a comment with no current line range (a whole-file |
| 34 | //! anchor, or `ents_anchor::Projection::Outdated`) has nowhere to |
| 35 | //! interleave, and renders in a below-the-blob "outdated comments" |
| 36 | //! section instead (`outdated_comments_section`). Doc-rendered and |
| 37 | //! binary views keep every comment below the blob, unconditionally |
| 38 | //! (`crate::pages::comments::comments_section`), since there is no source |
| 39 | //! line to interleave at. |
| 40 | //! |
| 41 | //! A raw-source view additionally carries the client-side hooks |
| 42 | //! `crate::assets`'s `ents.js` progressively enhances: `div.blob` names its |
| 43 | //! own `path`/`rev` (`data-path`/`data-rev`, the latter the resolved `HEAD` |
| 44 | //! commit oid, not the string `"HEAD"`, so a captured selection names the |
| 45 | //! exact commit being viewed) so a click on a gutter line number can select |
| 46 | //! a line or a shift-extended range and open an inline comment composer |
| 47 | //! cloned from a server-rendered `<template id="composer-template">` |
| 48 | //! (`composer_template`) -- with JS disabled the page stays fully usable |
| 49 | //! via `blob_header`'s "comment on this file" link and the plain `#L<n>` |
| 50 | //! anchors `crate::pages::comments::comment_card` already emits. |
| 51 | |
| 52 | use std::sync::Arc; |
| 53 | |
| 54 | use arborium::{Config, Highlighter, HtmlFormat}; |
| 55 | use axum::extract::{Path, State}; |
| 56 | use gix::bstr::ByteSlice as _; |
| 57 | use gix_object::{Find, Write}; |
| 58 | use maud::{Markup, PreEscaped, html}; |
| 59 | |
| 60 | use crate::assets; |
| 61 | use crate::error::{Error, Result}; |
| 62 | use crate::session::Session; |
| 63 | use crate::state::AppState; |
| 64 | |
| 65 | /// `GET /files`: the repository root directory listing. |
| 66 | /// |
| 67 | /// # Errors |
| 68 | /// |
| 69 | /// Propagates a `gix::open`/tree-read failure. |
| 70 | pub async fn root<O>( |
| 71 | State(state): State<Arc<AppState<O>>>, |
| 72 | axum::Extension(session): axum::Extension<Session>, |
| 73 | ) -> Result<Markup> |
| 74 | where |
| 75 | O: Find + Write + Send + 'static, |
| 76 | { |
| 77 | at(&state, "", &session) |
| 78 | } |
| 79 | |
| 80 | /// `GET /files/{*path}`: a directory listing or blob view at `path`. |
| 81 | /// |
| 82 | /// # Errors |
| 83 | /// |
| 84 | /// [`Error::NotFound`] if `path` does not name a tree or blob entry (or |
| 85 | /// contains a `.`/`..` component); otherwise propagates a |
| 86 | /// `gix::open`/tree-read failure. |
| 87 | pub async fn show<O>( |
| 88 | State(state): State<Arc<AppState<O>>>, |
| 89 | axum::Extension(session): axum::Extension<Session>, |
| 90 | Path(path): Path<String>, |
| 91 | ) -> Result<Markup> |
| 92 | where |
| 93 | O: Find + Write + Send + 'static, |
| 94 | { |
| 95 | at(&state, &path, &session) |
| 96 | } |
| 97 | |
| 98 | /// The shared implementation behind [`root`] and [`show`]: resolve `path` |
| 99 | /// against `HEAD`'s tree and render whichever of a directory listing or a |
| 100 | /// blob view it names. `session` is only ever needed on a blob view, to |
| 101 | /// render [`composer_template`]'s csrf input -- threaded down from the |
| 102 | /// route handler rather than reached for a second time here. |
| 103 | fn at<O>(state: &AppState<O>, path: &str, session: &Session) -> Result<Markup> |
| 104 | where |
| 105 | O: Find + Write, |
| 106 | { |
| 107 | if !is_safe_path(path) { |
| 108 | return Err(Error::NotFound { |
| 109 | what: path.to_owned(), |
| 110 | }); |
| 111 | } |
| 112 | |
| 113 | let repo = gix::open(&state.path).map_err(|source| Error::Repo(source.to_string()))?; |
| 114 | let head_tree = match repo.head_tree() { |
| 115 | Ok(tree) => tree, |
| 116 | // An unborn HEAD (a freshly initialized, still-empty repository) |
| 117 | // reads as an empty root directory, not a failure -- mirrors |
| 118 | // `pre-redo:crates/git-ents-server/src/web/git.rs`'s `root_tree`, |
| 119 | // which returned an empty entry list rather than erroring when the |
| 120 | // repository had no `HEAD` yet. |
| 121 | Err(_) if path.is_empty() => { |
| 122 | return Ok(super::layout( |
| 123 | &super::RepoHeader::from_state(state), |
| 124 | &super::identity_label(state), |
| 125 | super::Tab::Files, |
| 126 | "Files", |
| 127 | html! { |
| 128 | (dir_listing(path, Vec::new())) |
| 129 | }, |
| 130 | )); |
| 131 | } |
| 132 | Err(_) => { |
| 133 | return Err(Error::NotFound { |
| 134 | what: path.to_owned(), |
| 135 | }); |
| 136 | } |
| 137 | }; |
| 138 | |
| 139 | if path.is_empty() { |
| 140 | let entries = tree_entries(&head_tree)?; |
| 141 | return Ok(super::layout_split( |
| 142 | &super::RepoHeader::from_state(state), |
| 143 | &super::identity_label(state), |
| 144 | super::Tab::Files, |
| 145 | "Files", |
| 146 | false, |
| 147 | tree_sidebar(&head_tree, "", ""), |
| 148 | html! { |
| 149 | (dir_listing(path, entries)) |
| 150 | (readme_card(&head_tree)) |
| 151 | }, |
| 152 | )); |
| 153 | } |
| 154 | |
| 155 | let entry = head_tree |
| 156 | .lookup_entry_by_path(path) |
| 157 | .map_err(|source| Error::Repo(source.to_string()))? |
| 158 | .ok_or_else(|| Error::NotFound { |
| 159 | what: path.to_owned(), |
| 160 | })?; |
| 161 | |
| 162 | if entry.mode().is_tree() { |
| 163 | let subtree = entry |
| 164 | .object() |
| 165 | .map_err(|source| Error::Repo(source.to_string()))? |
| 166 | .try_into_tree() |
| 167 | .map_err(|source| Error::Repo(source.to_string()))?; |
| 168 | let entries = tree_entries(&subtree)?; |
| 169 | Ok(super::layout_split( |
| 170 | &super::RepoHeader::from_state(state), |
| 171 | &super::identity_label(state), |
| 172 | super::Tab::Files, |
| 173 | path, |
| 174 | true, |
| 175 | tree_sidebar(&head_tree, path, path), |
| 176 | html! { |
| 177 | (crumbs(path)) |
| 178 | (dir_listing(path, entries)) |
| 179 | }, |
| 180 | )) |
| 181 | } else if entry.mode().is_blob() { |
| 182 | let blob = entry |
| 183 | .object() |
| 184 | .map_err(|source| Error::Repo(source.to_string()))? |
| 185 | .try_into_blob() |
| 186 | .map_err(|source| Error::Repo(source.to_string()))?; |
| 187 | let name = path.rsplit('/').next().unwrap_or(path); |
| 188 | let comments = super::comments::for_path(state, &repo, path); |
| 189 | let head_oid = repo |
| 190 | .head_id() |
| 191 | .map_err(|source| Error::Repo(source.to_string()))? |
| 192 | .to_string(); |
| 193 | let editor = super::editor_open(state, path, None); |
| 194 | let (body, below) = blob_view( |
| 195 | path, name, &head_oid, session, &blob.data, &comments, editor, |
| 196 | )?; |
| 197 | let parent = path.rsplit_once('/').map_or("", |(dir, _)| dir); |
| 198 | Ok(super::layout_split( |
| 199 | &super::RepoHeader::from_state(state), |
| 200 | &super::identity_label(state), |
| 201 | super::Tab::Files, |
| 202 | path, |
| 203 | true, |
| 204 | tree_sidebar(&head_tree, parent, path), |
| 205 | html! { |
| 206 | (crumbs(path)) |
| 207 | (body) |
| 208 | (below) |
| 209 | }, |
| 210 | )) |
| 211 | } else { |
| 212 | // A symlink or a submodule (gitlink) -- neither is a tree or a |
| 213 | // blob this browser can render. |
| 214 | Err(Error::NotFound { |
| 215 | what: path.to_owned(), |
| 216 | }) |
| 217 | } |
| 218 | } |
| 219 | |
| 220 | /// Whether `path` is safe to resolve against a tree: no empty, `.`, or |
| 221 | /// `..` component. The empty root path is itself safe. |
| 222 | fn is_safe_path(path: &str) -> bool { |
| 223 | path.is_empty() |
| 224 | || path |
| 225 | .split('/') |
| 226 | .all(|s| !s.is_empty() && s != "." && s != "..") |
| 227 | } |
| 228 | |
| 229 | /// One `(name, is_directory, size)` triple per direct child of `tree`, in |
| 230 | /// tree order (not yet sorted -- [`dir_listing`] sorts for display). `size` |
| 231 | /// is a blob entry's byte length, read from its odb header |
| 232 | /// ([`gix::Repository::find_header`], a header-only lookup -- never a |
| 233 | /// full blob read just to size it) and best-effort (`None` |
| 234 | /// on a header-read failure); always |
| 235 | /// `None` for a directory entry, which [`dir_listing`] renders with no |
| 236 | /// size cell at all. |
| 237 | fn tree_entries(tree: &gix::Tree<'_>) -> Result<Vec<(String, bool, Option<u64>)>> { |
| 238 | tree.iter() |
| 239 | .map(|entry| { |
| 240 | let entry = entry.map_err(|source| Error::Repo(source.to_string()))?; |
| 241 | let is_dir = entry.mode().is_tree(); |
| 242 | let size = (!is_dir) |
| 243 | .then(|| tree.repo.find_header(entry.oid()).ok()) |
| 244 | .flatten() |
| 245 | .map(|header| header.size()); |
| 246 | Ok((entry.filename().to_str_lossy().into_owned(), is_dir, size)) |
| 247 | }) |
| 248 | .collect() |
| 249 | } |
| 250 | |
| 251 | /// The Code split's `.tree` sidebar (`crate::pages::layout_split`): a |
| 252 | /// crumb trail back to the repository root (plain labels, no icon -- the |
| 253 | /// same bare style [`crumbs`] renders above the pane), then the entries of |
| 254 | /// the directory at `dir` (the viewed directory itself, or a viewed blob's |
| 255 | /// parent) each as an icon |
| 256 | /// ([`assets::icon_folder`]/[`assets::icon_file`]) beside its name, |
| 257 | /// directories first -- not a full recursive tree, just enough context to |
| 258 | /// move one level in any direction. `active` names the full path of the |
| 259 | /// entry (or trailing crumb) being viewed, and gets `.active` |
| 260 | /// ([`tree_class`]). Best-effort: a subtree that fails to read renders an |
| 261 | /// empty entry list rather than failing the page around it. |
| 262 | fn tree_sidebar(head_tree: &gix::Tree<'_>, dir: &str, active: &str) -> Markup { |
| 263 | let mut entries = if dir.is_empty() { |
| 264 | tree_entries(head_tree).unwrap_or_default() |
| 265 | } else { |
| 266 | head_tree |
| 267 | .lookup_entry_by_path(dir) |
| 268 | .ok() |
| 269 | .flatten() |
| 270 | .and_then(|entry| entry.object().ok()) |
| 271 | .and_then(|object| object.try_into_tree().ok()) |
| 272 | .map(|subtree| tree_entries(&subtree).unwrap_or_default()) |
| 273 | .unwrap_or_default() |
| 274 | }; |
| 275 | entries.sort_by(|(a_name, a_is_dir, _), (b_name, b_is_dir, _)| { |
| 276 | b_is_dir.cmp(a_is_dir).then_with(|| a_name.cmp(b_name)) |
| 277 | }); |
| 278 | |
| 279 | let crumb_parts: Vec<&str> = dir.split('/').filter(|s| !s.is_empty()).collect(); |
| 280 | let mut crumb_trail: Vec<(String, String)> = Vec::new(); |
| 281 | let mut acc = String::new(); |
| 282 | for part in &crumb_parts { |
| 283 | if !acc.is_empty() { |
| 284 | acc.push('/'); |
| 285 | } |
| 286 | acc.push_str(part); |
| 287 | crumb_trail.push(((*part).to_owned(), acc.clone())); |
| 288 | } |
| 289 | let entry_depth = crumb_parts.len().saturating_add(1); |
| 290 | |
| 291 | html! { |
| 292 | a class=(tree_class(true, 0, active.is_empty())) href="/files" { "/" } |
| 293 | @for (index, (label, crumb_path)) in crumb_trail.iter().enumerate() { |
| 294 | a class=(tree_class(true, index.saturating_add(1), crumb_path == active)) |
| 295 | href={ "/files/" (crumb_path) } { (label) "/" } |
| 296 | } |
| 297 | @for (name, is_dir, _) in &entries { |
| 298 | @let full = if dir.is_empty() { name.clone() } else { format!("{dir}/{name}") }; |
| 299 | a class=(tree_class(*is_dir, entry_depth, full == active)) |
| 300 | href=(child_href(dir, name)) { |
| 301 | @if *is_dir { (assets::icon_folder()) } @else { (assets::icon_file()) } |
| 302 | (name) @if *is_dir { "/" } |
| 303 | } |
| 304 | } |
| 305 | } |
| 306 | } |
| 307 | |
| 308 | /// The class list for one [`tree_sidebar`] link: `.dir` for a directory, |
| 309 | /// an `.i{1..3}` indent per crumb depth (capped -- the sidebar shows one |
| 310 | /// directory's entries, not an unbounded tree), `.active` for the viewed |
| 311 | /// entry. |
| 312 | fn tree_class(is_dir: bool, depth: usize, active: bool) -> String { |
| 313 | let mut classes = Vec::new(); |
| 314 | if is_dir { |
| 315 | classes.push("dir"); |
| 316 | } |
| 317 | match depth { |
| 318 | 0 => {} |
| 319 | 1 => classes.push("i1"), |
| 320 | 2 => classes.push("i2"), |
| 321 | _ => classes.push("i3"), |
| 322 | } |
| 323 | if active { |
| 324 | classes.push("active"); |
| 325 | } |
| 326 | classes.join(" ") |
| 327 | } |
| 328 | |
| 329 | /// The link to a child of the directory at `dir` (empty at the root). |
| 330 | fn child_href(dir: &str, name: &str) -> String { |
| 331 | if dir.is_empty() { |
| 332 | format!("/files/{name}") |
| 333 | } else { |
| 334 | format!("/files/{dir}/{name}") |
| 335 | } |
| 336 | } |
| 337 | |
| 338 | /// A directory listing at `dir`: entries sorted directories-first then |
| 339 | /// alphabetically, each an icon and a link one level deeper, plus a |
| 340 | /// right-aligned muted size for a blob entry (`span.entry-size`, |
| 341 | /// [`human_size`]) -- a directory entry carries no size cell, since a |
| 342 | /// tree's own byte length is not a meaningful measure of it. |
| 343 | fn dir_listing(dir: &str, mut entries: Vec<(String, bool, Option<u64>)>) -> Markup { |
| 344 | entries.sort_by(|(a_name, a_is_dir, _), (b_name, b_is_dir, _)| { |
| 345 | b_is_dir.cmp(a_is_dir).then_with(|| a_name.cmp(b_name)) |
| 346 | }); |
| 347 | html! { |
| 348 | div.card { |
| 349 | @if entries.is_empty() { |
| 350 | div.card-row.muted { "Empty directory." } |
| 351 | } |
| 352 | @for (name, is_dir, size) in &entries { |
| 353 | div.card-row.is-dir[*is_dir] { |
| 354 | a.row-link href=(child_href(dir, name)) { |
| 355 | @if *is_dir { (assets::icon_folder()) } @else { (assets::icon_file()) } |
| 356 | (name) |
| 357 | } |
| 358 | @if let Some(size) = size { |
| 359 | span.entry-size { (human_size(*size)) } |
| 360 | } |
| 361 | } |
| 362 | } |
| 363 | } |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | /// The rendered `README` card below the root listing -- re-homed here |
| 368 | /// from the old overview dashboard (`crate::pages::dashboard` is a work |
| 369 | /// surface now; the Code root is where the repository introduces itself). |
| 370 | /// Renders nothing at all when the root holds no renderable `README`. |
| 371 | fn readme_card(tree: &gix::Tree<'_>) -> Markup { |
| 372 | let Some((name, rendered)) = readme(tree) else { |
| 373 | return html! {}; |
| 374 | }; |
| 375 | html! { |
| 376 | div.card { |
| 377 | div.card-header { (assets::icon_file()) (name) } |
| 378 | div.doc-body { (rendered) } |
| 379 | } |
| 380 | } |
| 381 | } |
| 382 | |
| 383 | /// The first root-tree blob whose stem is `README` and whose extension |
| 384 | /// this crate renders (Markdown or AsciiDoc), converted to HTML and paired |
| 385 | /// with its filename; `None` when there is none or it fails to render |
| 386 | /// (mirrors `pre-redo:.../pages.rs`'s `readme`). |
| 387 | fn readme(tree: &gix::Tree<'_>) -> Option<(String, Markup)> { |
| 388 | let name = root_readme_name(tree)?; |
| 389 | let entry = tree.lookup_entry_by_path(&name).ok()??; |
| 390 | let blob = entry.object().ok()?.try_into_blob().ok()?; |
| 391 | let text = String::from_utf8_lossy(&blob.data); |
| 392 | render_doc(&name, &text).map(|rendered| (name, rendered)) |
| 393 | } |
| 394 | |
| 395 | /// The filename of the root's `README`, if it has a renderable one. |
| 396 | fn root_readme_name(tree: &gix::Tree<'_>) -> Option<String> { |
| 397 | for entry in tree.iter() { |
| 398 | let Ok(entry) = entry else { continue }; |
| 399 | if !entry.mode().is_blob() { |
| 400 | continue; |
| 401 | } |
| 402 | let name = entry.filename().to_str_lossy(); |
| 403 | let is_readme = name |
| 404 | .rsplit_once('.') |
| 405 | .is_some_and(|(stem, _)| stem.eq_ignore_ascii_case("readme")); |
| 406 | if is_readme && (crate::markdown::is_markdown(&name) || crate::asciidoc::is_asciidoc(&name)) |
| 407 | { |
| 408 | return Some(name.into_owned()); |
| 409 | } |
| 410 | } |
| 411 | None |
| 412 | } |
| 413 | |
| 414 | /// `text` rendered as its prose format (Markdown or AsciiDoc), or `None` |
| 415 | /// when it is neither or AsciiDoc rendering fails. |
| 416 | fn render_doc(name: &str, text: &str) -> Option<Markup> { |
| 417 | if crate::markdown::is_markdown(name) { |
| 418 | Some(crate::markdown::to_html(text)) |
| 419 | } else if crate::asciidoc::is_asciidoc(name) { |
| 420 | crate::asciidoc::to_html(text).ok() |
| 421 | } else { |
| 422 | None |
| 423 | } |
| 424 | } |
| 425 | |
| 426 | /// Breadcrumb navigation from the repository's files root down through |
| 427 | /// `path`, `chevron-right` icons separating segments -- pure navigation, |
| 428 | /// no trailing actions. The history/comment links that used to trail this |
| 429 | /// nav on a blob view now live in [`blob_header`]'s own action group |
| 430 | /// instead (see this module's own top-level doc for why). The files root |
| 431 | /// itself renders no crumbs at all: a lone self-referencing "files" crumb |
| 432 | /// under the page's own "Files" title (and above the listing card's own |
| 433 | /// "files" header) named the same place three times. |
| 434 | fn crumbs(path: &str) -> Markup { |
| 435 | let parts: Vec<&str> = path.split('/').filter(|s| !s.is_empty()).collect(); |
| 436 | let mut acc = String::new(); |
| 437 | let mut trail: Vec<(String, Option<String>)> = |
| 438 | vec![("files".to_owned(), Some("/files".to_owned()))]; |
| 439 | for (index, part) in parts.iter().enumerate() { |
| 440 | if !acc.is_empty() { |
| 441 | acc.push('/'); |
| 442 | } |
| 443 | acc.push_str(part); |
| 444 | let is_last = index.saturating_add(1) == parts.len(); |
| 445 | let href = (!is_last).then(|| format!("/files/{acc}")); |
| 446 | trail.push(((*part).to_owned(), href)); |
| 447 | } |
| 448 | html! { |
| 449 | nav.crumbs { |
| 450 | @for (index, (label, href)) in trail.iter().enumerate() { |
| 451 | @if index > 0 { span.sep { (assets::icon_chevron()) } } |
| 452 | @match href { |
| 453 | Some(href) => a href=(href) { (label) }, |
| 454 | None => span.here { (label) }, |
| 455 | } |
| 456 | } |
| 457 | } |
| 458 | } |
| 459 | } |
| 460 | |
| 461 | /// Format a byte count the way [`blob_header`] and [`dir_listing`] both |
| 462 | /// show a file's size: whole bytes under 1 KB, otherwise one decimal place |
| 463 | /// of KB or MB -- integer-only throughout (`checked_div`/`checked_rem`/ |
| 464 | /// `saturating_mul`, this crate's own arithmetic idiom) rather than a |
| 465 | /// float division, so there is no rounding-mode or precision question to |
| 466 | /// answer. |
| 467 | fn human_size(bytes: u64) -> String { |
| 468 | const KB: u64 = 1024; |
| 469 | const MB: u64 = 1024 * 1024; |
| 470 | if bytes < KB { |
| 471 | return format!("{bytes} B"); |
| 472 | } |
| 473 | let (scale, unit) = if bytes < MB { (KB, "KB") } else { (MB, "MB") }; |
| 474 | let whole = bytes.checked_div(scale).unwrap_or(0); |
| 475 | let remainder = bytes.checked_rem(scale).unwrap_or(0); |
| 476 | let tenths = remainder.saturating_mul(10).checked_div(scale).unwrap_or(0); |
| 477 | format!("{whole}.{tenths} {unit}") |
| 478 | } |
| 479 | |
| 480 | /// Whether `bytes` looks like binary content (a NUL byte in the leading |
| 481 | /// chunk -- the same heuristic git itself uses, carried over from |
| 482 | /// `pre-redo:crates/git-ents-server/src/web/pages.rs`'s own `is_binary`). |
| 483 | fn is_binary(bytes: &[u8]) -> bool { |
| 484 | bytes.iter().take(8000).any(|b| *b == 0) |
| 485 | } |
| 486 | |
| 487 | /// A single blob's contents, plus whatever comments belong below it: a |
| 488 | /// Markdown/AsciiDoc document rendered as such via |
| 489 | /// [`crate::markdown`]/[`crate::asciidoc`] or a binary-content placeholder |
| 490 | /// -- either way every `comment` renders below, unconditionally |
| 491 | /// ([`crate::pages::comments::comments_section`]), since there is no |
| 492 | /// source line to interleave a card at -- or [`source_view`]'s |
| 493 | /// line-per-row rendering, which interleaves a comment with a current line |
| 494 | /// range directly into the blob and returns the rest (no current line |
| 495 | /// range: a whole-file anchor, or `ents_anchor::Projection::Outdated`) as |
| 496 | /// a separate below-the-blob section ([`outdated_comments_section`]). |
| 497 | /// Every case renders a [`blob_header`] first -- above `div.card`/ |
| 498 | /// `div.binary` for a doc-rendered or binary view, as `div.blob`'s own |
| 499 | /// first child for a raw-source view (see [`source_view`]'s own doc). |
| 500 | /// |
| 501 | /// # Errors |
| 502 | /// |
| 503 | /// Propagates [`crate::asciidoc::to_html`]'s own [`Error::Asciidoc`]. |
| 504 | fn blob_view( |
| 505 | path: &str, |
| 506 | name: &str, |
| 507 | head_oid: &str, |
| 508 | session: &Session, |
| 509 | bytes: &[u8], |
| 510 | comments: &[super::comments::FileComment], |
| 511 | editor: Markup, |
| 512 | ) -> Result<(Markup, Markup)> { |
| 513 | let size = u64::try_from(bytes.len()).unwrap_or(u64::MAX); |
| 514 | let comment_count = comments.len(); |
| 515 | let no_line_count_header = || { |
| 516 | blob_header(&BlobHeaderMeta { |
| 517 | name, |
| 518 | path, |
| 519 | size, |
| 520 | line_count: None, |
| 521 | language: None, |
| 522 | comments: comment_count, |
| 523 | editor: editor.clone(), |
| 524 | }) |
| 525 | }; |
| 526 | // Every view kind carries the whole-file composer template -- a |
| 527 | // doc-rendered or binary view has no per-line gutter to open one with |
| 528 | // a specific line range, but "comment on this file" (empty `lines`, |
| 529 | // [`composer_template`]'s own default) is exactly as meaningful there |
| 530 | // as on a raw-source view. |
| 531 | let composer = composer_template(path, head_oid, session); |
| 532 | if is_binary(bytes) { |
| 533 | return Ok(( |
| 534 | html! { |
| 535 | (no_line_count_header()) |
| 536 | div.binary { "Binary file (" (bytes.len()) " bytes) not shown." } |
| 537 | (composer) |
| 538 | }, |
| 539 | super::comments::comments_section(comments), |
| 540 | )); |
| 541 | } |
| 542 | let Ok(text) = std::str::from_utf8(bytes) else { |
| 543 | return Ok(( |
| 544 | html! { |
| 545 | (no_line_count_header()) |
| 546 | div.binary { "Binary file (" (bytes.len()) " bytes) not shown." } |
| 547 | (composer) |
| 548 | }, |
| 549 | super::comments::comments_section(comments), |
| 550 | )); |
| 551 | }; |
| 552 | if crate::markdown::is_markdown(name) { |
| 553 | return Ok(( |
| 554 | html! { |
| 555 | (no_line_count_header()) |
| 556 | div.card { div.doc-body { (crate::markdown::to_html(text)) } } |
| 557 | (composer) |
| 558 | }, |
| 559 | super::comments::comments_section(comments), |
| 560 | )); |
| 561 | } |
| 562 | if crate::asciidoc::is_asciidoc(name) { |
| 563 | return Ok(( |
| 564 | html! { |
| 565 | (no_line_count_header()) |
| 566 | div.card { div.doc-body { (crate::asciidoc::to_html(text)?) } } |
| 567 | (composer) |
| 568 | }, |
| 569 | super::comments::comments_section(comments), |
| 570 | )); |
| 571 | } |
| 572 | let language = arborium::detect_language(name); |
| 573 | let highlighted = highlight(name, text); |
| 574 | let line_count = text.lines().count().max(1); |
| 575 | let header = blob_header(&BlobHeaderMeta { |
| 576 | name, |
| 577 | path, |
| 578 | size, |
| 579 | line_count: Some(line_count), |
| 580 | language, |
| 581 | comments: comment_count, |
| 582 | editor, |
| 583 | }); |
| 584 | let below: Vec<(usize, &super::comments::FileComment)> = comments |
| 585 | .iter() |
| 586 | .enumerate() |
| 587 | .filter(|(_, comment)| comment.lines.is_none()) |
| 588 | .collect(); |
| 589 | Ok(( |
| 590 | source_view( |
| 591 | path, |
| 592 | head_oid, |
| 593 | header, |
| 594 | composer, |
| 595 | text, |
| 596 | highlighted, |
| 597 | comments, |
| 598 | ), |
| 599 | outdated_comments_section(&below), |
| 600 | )) |
| 601 | } |
| 602 | |
| 603 | /// The metadata [`blob_header`] shows beside a blob's name -- gathered by |
| 604 | /// [`blob_view`], one per view kind (see that function's own doc for which |
| 605 | /// fields each kind fills in). |
| 606 | struct BlobHeaderMeta<'a> { |
| 607 | /// The file's own name (the last path segment), shown as the title. |
| 608 | name: &'a str, |
| 609 | /// The full repository-relative path -- only used to build the |
| 610 | /// "comment on this file" link's `?file=` query. |
| 611 | path: &'a str, |
| 612 | /// The blob's byte length, always known ([`human_size`]). |
| 613 | size: u64, |
| 614 | /// The raw-source view's line count; `None` for a doc-rendered or |
| 615 | /// binary view, which has no source line to count. |
| 616 | line_count: Option<usize>, |
| 617 | /// [`arborium::detect_language`]'s own identifier, shown as-is when it |
| 618 | /// recognized `name`'s grammar; `None` otherwise, or for a |
| 619 | /// doc-rendered/binary view (a rendered document is not "highlighted |
| 620 | /// as" a language, and a binary blob was never linted for one at all). |
| 621 | language: Option<&'static str>, |
| 622 | /// How many comments [`super::comments::for_path`] found for this |
| 623 | /// blob -- the "N comments" jump link renders only when this is above |
| 624 | /// zero (mirrors [`crumbs`]'s own former stance, now moved here). |
| 625 | comments: usize, |
| 626 | /// The pre-rendered open-in-editor affordance ([`super::editor_open`]; |
| 627 | /// empty when no editor is recognized), leading the actions so the |
| 628 | /// jump back to the desk sits first. |
| 629 | editor: Markup, |
| 630 | } |
| 631 | |
| 632 | /// The header bar above every blob view -- a leading file icon |
| 633 | /// ([`assets::icon_file`]), the file's name and metadata |
| 634 | /// (`span.blob-title`/`span.blob-meta`), the actions that used to |
| 635 | /// trail [`crumbs`] on the right (`span.blob-actions`): a jump into |
| 636 | /// `crate::pages::commits`'s `GET /commits` history (the file browser's |
| 637 | /// one entry point into commit history, since history is a view of the |
| 638 | /// code, not a tab of its own -- `crate::pages::mod`'s own doc), a jump |
| 639 | /// straight to the first comment card (`#comment-0`, in display order -- |
| 640 | /// see [`super::comments::comment_card`]'s own doc) when at least one |
| 641 | /// comment is already anchored here, and `crate::pages::comments`'s own |
| 642 | /// add form for this file ("comment on this file" -- the no-JS fallback |
| 643 | /// entry point into the composer [`composer_template`] otherwise opens |
| 644 | /// inline). Renders identically whether the view below it is raw source, a |
| 645 | /// rendered document, or a binary placeholder -- only [`BlobHeaderMeta`]'s |
| 646 | /// fields differ per kind. |
| 647 | fn blob_header(meta: &BlobHeaderMeta<'_>) -> Markup { |
| 648 | html! { |
| 649 | div.blob-header { |
| 650 | (assets::icon_file()) |
| 651 | span.blob-title { (meta.name) } |
| 652 | span.blob-meta { |
| 653 | @if let Some(lines) = meta.line_count { |
| 654 | (lines) @if lines == 1 { " line" } @else { " lines" } " \u{b7} " (human_size(meta.size)) |
| 655 | } @else { |
| 656 | (human_size(meta.size)) |
| 657 | } |
| 658 | @if let Some(language) = meta.language { |
| 659 | " \u{b7} " (language) |
| 660 | } |
| 661 | } |
| 662 | span.blob-actions { |
| 663 | (meta.editor) |
| 664 | a href="/commits" { "history" } |
| 665 | @if meta.comments > 0 { |
| 666 | a href="#comment-0" { |
| 667 | (meta.comments) @if meta.comments == 1 { " comment" } @else { " comments" } |
| 668 | } |
| 669 | } |
| 670 | a.composer-trigger data-composer="composer-template" href={ "/comments?file=" (meta.path) } { "comment on this file" } |
| 671 | } |
| 672 | } |
| 673 | } |
| 674 | } |
| 675 | |
| 676 | /// The composer's server-rendered `<template>`, cloned by `ents.js` when a |
| 677 | /// reader clicks the gutter's `+` affordance on a raw-source view (see this |
| 678 | /// module's own top-level doc). Its `form` posts to |
| 679 | /// `crate::pages::comments`'s own `POST /comments` handler |
| 680 | /// ([`super::comments::AddForm`]), pre-filled with this file's own |
| 681 | /// `path`/`rev` (`head_oid`, the resolved `HEAD` commit, exactly what |
| 682 | /// `div.blob`'s own `data-rev` names -- see [`source_view`]'s own doc) so |
| 683 | /// the only field `ents.js` ever needs to fill in before submit is the |
| 684 | /// hidden `lines` input, left empty here. With JS disabled this template |
| 685 | /// never becomes visible at all (a `<template>` element's contents are |
| 686 | /// inert, never rendered by a browser on their own), which is exactly why |
| 687 | /// [`blob_header`]'s "comment on this file" link remains the no-JS path to |
| 688 | /// the same form. |
| 689 | // @relation(roots.web-session, scope=function) |
| 690 | fn composer_template(path: &str, head_oid: &str, session: &Session) -> Markup { |
| 691 | html! { |
| 692 | template id="composer-template" { |
| 693 | form.composer-form method="post" action="/comments" { |
| 694 | (super::csrf_input(session)) |
| 695 | input type="hidden" name="path" value=(path); |
| 696 | input type="hidden" name="rev" value=(head_oid); |
| 697 | input type="hidden" name="lines" value=""; |
| 698 | textarea name="body" placeholder="Leave a comment (AsciiDoc)" {} |
| 699 | div.composer-buttons { |
| 700 | button type="submit" { "Comment" } |
| 701 | button.composer-cancel type="button" { "Cancel" } |
| 702 | } |
| 703 | } |
| 704 | } |
| 705 | } |
| 706 | } |
| 707 | |
| 708 | /// The raw-source view: `header` and `composer` (see [`blob_header`]/ |
| 709 | /// [`composer_template`]'s own docs) around one table row per line (a |
| 710 | /// `<tr>` pairing a `.blob-nums` line-number cell carrying the row's |
| 711 | /// `#L{n}` anchor with a `.blob-code` cell, no wrapper beyond those two |
| 712 | /// cells -- lean enough that thousands of lines stay cheap), highlighted |
| 713 | /// via [`highlight`] when `highlighted` is `Some` and falling back to |
| 714 | /// plain (still per-line, still auto-escaped by `maud`'s own |
| 715 | /// interpolation) text otherwise. Each |
| 716 | /// [`FileComment`](super::comments::FileComment) in `comments` whose |
| 717 | /// [`ents_anchor::LineRange`] is `Some` renders its card |
| 718 | /// ([`super::comments::comment_card`]) immediately after the row naming |
| 719 | /// its range's last line, full width across both columns |
| 720 | /// (`tr.blob-comment-row`, `colspan="2"`) -- multiple comments ending on |
| 721 | /// the same line stack in `comments`' own order (`comment::list`'s ref |
| 722 | /// order). A comment with no current line range is [`blob_view`]'s own |
| 723 | /// concern, not this function's: it never appears here. |
| 724 | /// |
| 725 | /// `div.blob` itself carries `data-path=(path)`/`data-rev=(head_oid)` -- |
| 726 | /// `ents.js`'s own activation check and the values it writes into |
| 727 | /// [`composer_template`]'s clone -- so a click on a gutter line number |
| 728 | /// selects it (and a shift-click extends the selection) with no further |
| 729 | /// server round trip needed until the reader actually submits a comment. |
| 730 | fn source_view( |
| 731 | path: &str, |
| 732 | head_oid: &str, |
| 733 | header: Markup, |
| 734 | composer: Markup, |
| 735 | text: &str, |
| 736 | highlighted: Option<String>, |
| 737 | comments: &[super::comments::FileComment], |
| 738 | ) -> Markup { |
| 739 | let physical_lines: Vec<&str> = text.lines().collect(); |
| 740 | let line_count = physical_lines.len().max(1); |
| 741 | |
| 742 | let mut code_lines: Vec<Markup> = match &highlighted { |
| 743 | Some(html) => split_highlighted_lines(html, line_count) |
| 744 | .into_iter() |
| 745 | .map(|fragment| html! { (PreEscaped(fragment)) }) |
| 746 | .collect(), |
| 747 | None => physical_lines |
| 748 | .iter() |
| 749 | .map(|line| html! { (*line) }) |
| 750 | .collect(), |
| 751 | }; |
| 752 | // Exactly `line_count` rows either way: arborium trims trailing |
| 753 | // newlines before highlighting (see `split_highlighted_lines`'s own |
| 754 | // doc), so a file ending in blank lines can highlight to fewer |
| 755 | // embedded newlines than `text.lines().count()` -- padding (never |
| 756 | // truncating in practice, since `split_highlighted_lines` never |
| 757 | // returns fewer than one fragment) keeps every gutter number paired |
| 758 | // with a code cell, with no per-row fallback indexing needed below. |
| 759 | code_lines.resize_with(line_count, Markup::default); |
| 760 | |
| 761 | let mut by_end_line: std::collections::BTreeMap<u64, Vec<usize>> = |
| 762 | std::collections::BTreeMap::new(); |
| 763 | for (index, comment) in comments.iter().enumerate() { |
| 764 | if let Some(range) = comment.lines { |
| 765 | by_end_line.entry(range.end).or_default().push(index); |
| 766 | } |
| 767 | } |
| 768 | |
| 769 | html! { |
| 770 | // `header` is a sibling before `.blob`, never nested inside it -- |
| 771 | // `.blob-header`'s own border and top-only radius are meant to sit |
| 772 | // flush atop `.blob`'s bottom-only radius as one continuous box |
| 773 | // (`ents.css`'s own note), which only holds when the two are |
| 774 | // siblings, not when the header sits inset inside `.blob`'s own |
| 775 | // padded, fully-bordered box. |
| 776 | (header) |
| 777 | div.blob data-path=(path) data-rev=(head_oid) { |
| 778 | table { |
| 779 | tbody { |
| 780 | @for (index, code) in code_lines.into_iter().enumerate() { |
| 781 | @let n = index.saturating_add(1); |
| 782 | tr { |
| 783 | td.blob-nums { a id={ "L" (n) } href={ "#L" (n) } { (n) } } |
| 784 | @if highlighted.is_some() { |
| 785 | td.blob-code { code.code { (code) } } |
| 786 | } @else { |
| 787 | td.blob-code { code { (code) } } |
| 788 | } |
| 789 | } |
| 790 | @if let Some(indices) = by_end_line.get(&u64::try_from(n).unwrap_or(u64::MAX)) { |
| 791 | @for &comment_index in indices { |
| 792 | @if let Some(comment) = comments.get(comment_index) { |
| 793 | tr.blob-comment-row { |
| 794 | td colspan="2" { |
| 795 | (super::comments::comment_card( |
| 796 | comment_index, |
| 797 | comment, |
| 798 | super::comments::LinkMode::SameFile, |
| 799 | )) |
| 800 | } |
| 801 | } |
| 802 | } |
| 803 | } |
| 804 | } |
| 805 | } |
| 806 | } |
| 807 | } |
| 808 | (composer) |
| 809 | } |
| 810 | } |
| 811 | } |
| 812 | |
| 813 | /// The below-the-blob section for comments with no current line range to |
| 814 | /// interleave at (a whole-file anchor, or |
| 815 | /// `ents_anchor::Projection::Outdated`) -- a warn-glyphed heading |
| 816 | /// (`div.outdated-head`, [`assets::icon_use`]'s `"i-warn"`) distinguishing |
| 817 | /// it from the inline cards [`source_view`] interleaves directly into the |
| 818 | /// blob, since every comment reaching here either predates line-level |
| 819 | /// anchoring or has literally gone stale, then each comment through the |
| 820 | /// same [`super::comments::comment_card`] every other blob-adjacent comment |
| 821 | /// renders through. Renders nothing at all when `comments` is empty |
| 822 | /// (mirrors [`super::comments::comments_section`]'s identical stance). |
| 823 | fn outdated_comments_section(comments: &[(usize, &super::comments::FileComment)]) -> Markup { |
| 824 | if comments.is_empty() { |
| 825 | return html! {}; |
| 826 | } |
| 827 | html! { |
| 828 | div.outdated-head { |
| 829 | (assets::icon_use("i-warn")) |
| 830 | span { "Outdated Comments" } |
| 831 | } |
| 832 | p.muted { "Anchored to lines that no longer map onto HEAD." } |
| 833 | @for &(index, comment) in comments { |
| 834 | (super::comments::comment_card(index, comment, super::comments::LinkMode::SameFile)) |
| 835 | } |
| 836 | } |
| 837 | } |
| 838 | |
| 839 | /// Split [`highlight`]'s single HTML string into one HTML fragment per |
| 840 | /// source line (`line_count` of them, padding with an empty string past |
| 841 | /// whatever [`tokenize`] actually produced -- arborium trims trailing |
| 842 | /// newlines from its input before highlighting, so a file ending in |
| 843 | /// several blank lines can highlight to fewer embedded newlines than |
| 844 | /// `text.lines().count()`; [`source_view`]'s own row loop indexes |
| 845 | /// defensively for the same reason). |
| 846 | /// |
| 847 | /// The hard part: a highlight span **can** cross a newline (a multiline |
| 848 | /// block comment, a triple-quoted string), so it is not enough to split on |
| 849 | /// `\n` -- a span open at a line boundary must be closed before the split |
| 850 | /// and reopened after it, or the two resulting fragments are not |
| 851 | /// independently well-formed HTML. This walks [`tokenize`]'s token stream |
| 852 | /// with an explicit stack of open span classes: a `Text` token's embedded |
| 853 | /// newlines close every open span, end the current line, and reopen them |
| 854 | /// (in the same order) at the start of the next. |
| 855 | fn split_highlighted_lines(html: &str, line_count: usize) -> Vec<String> { |
| 856 | let mut lines: Vec<String> = Vec::with_capacity(line_count.max(1)); |
| 857 | let mut current = String::new(); |
| 858 | let mut open: Vec<&str> = Vec::new(); |
| 859 | |
| 860 | for token in tokenize(html) { |
| 861 | match token { |
| 862 | Token::Open(class) => { |
| 863 | current.push_str("<span class=\""); |
| 864 | current.push_str(class); |
| 865 | current.push_str("\">"); |
| 866 | open.push(class); |
| 867 | } |
| 868 | Token::Close => { |
| 869 | current.push_str("</span>"); |
| 870 | open.pop(); |
| 871 | } |
| 872 | Token::Text(text) => { |
| 873 | let mut parts = text.split('\n'); |
| 874 | if let Some(first) = parts.next() { |
| 875 | current.push_str(first); |
| 876 | } |
| 877 | for rest in parts { |
| 878 | for _ in &open { |
| 879 | current.push_str("</span>"); |
| 880 | } |
| 881 | lines.push(std::mem::take(&mut current)); |
| 882 | for class in &open { |
| 883 | current.push_str("<span class=\""); |
| 884 | current.push_str(class); |
| 885 | current.push_str("\">"); |
| 886 | } |
| 887 | current.push_str(rest); |
| 888 | } |
| 889 | } |
| 890 | } |
| 891 | } |
| 892 | lines.push(current); |
| 893 | lines |
| 894 | } |
| 895 | |
| 896 | /// One tokenized fragment of arborium's `HtmlFormat::ClassNames` output |
| 897 | /// (`arborium_highlight::render::spans_to_html`'s own doc): an opening |
| 898 | /// `<span class="...">`, its matching `</span>`, or a run of |
| 899 | /// already-escaped text between tags. That renderer never emits any tag |
| 900 | /// but these two, and every text run it emits is already HTML-escaped |
| 901 | /// (`<`, `&`, ...) -- [`split_highlighted_lines`] never re-escapes |
| 902 | /// or splits an entity, since [`Token::Text`] is only ever split on |
| 903 | /// literal `\n` bytes, never re-parsed. |
| 904 | #[derive(Debug, Clone, Copy, PartialEq, Eq)] |
| 905 | enum Token<'a> { |
| 906 | /// `<span class="{0}">`. |
| 907 | Open(&'a str), |
| 908 | /// `</span>`. |
| 909 | Close, |
| 910 | /// Already-escaped text between tags. |
| 911 | Text(&'a str), |
| 912 | } |
| 913 | |
| 914 | /// Tokenize `html` into a stream of [`Token`]s -- see [`Token`]'s own doc |
| 915 | /// for why a simple `<span class="...">`/`</span>` scan is sufficient |
| 916 | /// (arborium's own HTML renderer emits no other tag, and every text run is |
| 917 | /// already escaped so it never contains a literal `<`). Malformed input |
| 918 | /// (which arborium's own renderer never produces) degrades to treating the |
| 919 | /// unrecognized byte as plain text rather than panicking or looping |
| 920 | /// forever. |
| 921 | fn tokenize(html: &str) -> Vec<Token<'_>> { |
| 922 | const OPEN_PREFIX: &str = "<span class=\""; |
| 923 | const CLOSE_TAG: &str = "</span>"; |
| 924 | |
| 925 | let mut tokens = Vec::new(); |
| 926 | let mut rest = html; |
| 927 | while !rest.is_empty() { |
| 928 | // `.get(..)`/`.get(n..)` rather than direct indexing throughout: |
| 929 | // every offset here comes from `find`/`strip_prefix`, always a |
| 930 | // valid char boundary, but this function still never indexes a |
| 931 | // `str` directly (`clippy::string_slice`) or performs raw |
| 932 | // arithmetic on an offset (`clippy::arithmetic_side_effects`) -- |
| 933 | // `.get(end..)` then `strip_prefix('"')` finds "just past the |
| 934 | // quote" without ever computing `end + 1`. |
| 935 | if let Some(after_prefix) = rest.strip_prefix(OPEN_PREFIX) |
| 936 | && let Some(end) = after_prefix.find('"') |
| 937 | && let Some(class) = after_prefix.get(..end) |
| 938 | && let Some(after_quote) = after_prefix.get(end..).and_then(|s| s.strip_prefix('"')) |
| 939 | && let Some(after_gt) = after_quote.strip_prefix('>') |
| 940 | { |
| 941 | tokens.push(Token::Open(class)); |
| 942 | rest = after_gt; |
| 943 | continue; |
| 944 | } |
| 945 | if let Some(after) = rest.strip_prefix(CLOSE_TAG) { |
| 946 | tokens.push(Token::Close); |
| 947 | rest = after; |
| 948 | continue; |
| 949 | } |
| 950 | let next_tag = [rest.find(OPEN_PREFIX), rest.find(CLOSE_TAG)] |
| 951 | .into_iter() |
| 952 | .flatten() |
| 953 | .min(); |
| 954 | match next_tag { |
| 955 | Some(0) | None => { |
| 956 | // No recognized tag anywhere ahead (or, defensively, right |
| 957 | // at the cursor despite the checks above not matching it |
| 958 | // -- malformed input arborium never actually produces): |
| 959 | // take the rest as one text run rather than looping. |
| 960 | tokens.push(Token::Text(rest)); |
| 961 | rest = ""; |
| 962 | } |
| 963 | Some(idx) => { |
| 964 | let text = rest.get(..idx).unwrap_or(rest); |
| 965 | rest = rest.get(idx..).unwrap_or_default(); |
| 966 | tokens.push(Token::Text(text)); |
| 967 | } |
| 968 | } |
| 969 | } |
| 970 | tokens |
| 971 | } |
| 972 | |
| 973 | /// Highlighted HTML for `source`, or `None` when `name`'s extension names |
| 974 | /// no grammar [`arborium::detect_language`] recognizes -- [`blob_view`] |
| 975 | /// then falls back to escaped plain text. Ported from |
| 976 | /// `pre-redo:crates/git-ents-server/src/web/pages.rs`'s own `highlight`, |
| 977 | /// its `HtmlFormat::ClassNames` output matched by |
| 978 | /// `crate::assets::OVERRIDES`'s `.code .keyword`-family rules. |
| 979 | /// |
| 980 | /// The [`Highlighter`] is built and used entirely within this synchronous |
| 981 | /// call -- its grammar store is not `Send`, so it must never be held |
| 982 | /// across an `.await` (this function itself is never `async`, and neither |
| 983 | /// is any caller between it and the request handler). |
| 984 | fn highlight(name: &str, source: &str) -> Option<String> { |
| 985 | let language = arborium::detect_language(name)?; |
| 986 | let config = Config { |
| 987 | html_format: HtmlFormat::ClassNames, |
| 988 | ..Default::default() |
| 989 | }; |
| 990 | Highlighter::with_config(config) |
| 991 | .highlight(language, source) |
| 992 | .ok() |
| 993 | } |
| 994 | |
| 995 | #[cfg(test)] |
| 996 | mod tests { |
| 997 | #![allow(clippy::expect_used, reason = "unit test")] |
| 998 | |
| 999 | use ents_anchor::LineRange; |
| 1000 | use rstest::rstest; |
| 1001 | |
| 1002 | use super::*; |
| 1003 | use crate::pages::comments::FileComment; |
| 1004 | |
| 1005 | /// A minimal [`FileComment`] fixture -- the `body`/`author`/`seconds` |
| 1006 | /// values never matter to a rendering-position assertion, only |
| 1007 | /// `lines`. |
| 1008 | fn comment(lines: Option<LineRange>) -> FileComment { |
| 1009 | FileComment { |
| 1010 | author: "commenter".to_owned(), |
| 1011 | seconds: 0, |
| 1012 | path: "src/main.rs".to_owned(), |
| 1013 | lines, |
| 1014 | outdated: false, |
| 1015 | body: html! { p { "worth a look" } }, |
| 1016 | editor: html! {}, |
| 1017 | } |
| 1018 | } |
| 1019 | |
| 1020 | /// A minimal [`Session`] fixture -- [`blob_view`]'s own tests only ever |
| 1021 | /// need a csrf token to render into [`composer_template`]'s hidden |
| 1022 | /// input, never a real [`crate::session::SessionStore`]-minted one. |
| 1023 | fn session() -> Session { |
| 1024 | Session { |
| 1025 | csrf: "test-csrf-token".to_owned(), |
| 1026 | member: None, |
| 1027 | } |
| 1028 | } |
| 1029 | |
| 1030 | #[rstest] |
| 1031 | #[case::empty("", true)] |
| 1032 | #[case::simple("src/main.rs", true)] |
| 1033 | #[case::nested("a/b/c", true)] |
| 1034 | #[case::dot(".", false)] |
| 1035 | #[case::dotdot("..", false)] |
| 1036 | #[case::traversal("a/../b", false)] |
| 1037 | #[case::trailing_slash("a/", false)] |
| 1038 | #[case::double_slash("a//b", false)] |
| 1039 | fn is_safe_path_rejects_dot_components_and_empty_segments( |
| 1040 | #[case] path: &str, |
| 1041 | #[case] expected: bool, |
| 1042 | ) { |
| 1043 | assert_eq!(is_safe_path(path), expected); |
| 1044 | } |
| 1045 | |
| 1046 | #[test] |
| 1047 | fn dir_listing_sorts_directories_first_then_alphabetically() { |
| 1048 | let entries = vec![ |
| 1049 | ("zeta.txt".to_owned(), false, Some(10)), |
| 1050 | ("alpha".to_owned(), true, None), |
| 1051 | ("beta.txt".to_owned(), false, Some(2048)), |
| 1052 | ("gamma".to_owned(), true, None), |
| 1053 | ]; |
| 1054 | let rendered = dir_listing("", entries).into_string(); |
| 1055 | let alpha = rendered.find("alpha").expect("alpha listed"); |
| 1056 | let gamma = rendered.find("gamma").expect("gamma listed"); |
| 1057 | let beta = rendered.find("beta.txt").expect("beta listed"); |
| 1058 | let zeta = rendered.find("zeta.txt").expect("zeta listed"); |
| 1059 | assert!(alpha < gamma, "directories sort among themselves"); |
| 1060 | assert!(gamma < beta, "every directory sorts before every file"); |
| 1061 | assert!(beta < zeta, "files sort among themselves"); |
| 1062 | } |
| 1063 | |
| 1064 | #[test] |
| 1065 | fn dir_listing_shows_a_size_for_a_file_and_none_for_a_directory() { |
| 1066 | let entries = vec![ |
| 1067 | ("src".to_owned(), true, None), |
| 1068 | ("main.rs".to_owned(), false, Some(2048)), |
| 1069 | ]; |
| 1070 | let rendered = dir_listing("", entries).into_string(); |
| 1071 | let dir_index = rendered.find("src").expect("directory entry renders"); |
| 1072 | let file_index = rendered.find("main.rs").expect("file entry renders"); |
| 1073 | assert!(dir_index < file_index, "directories sort before files"); |
| 1074 | assert!( |
| 1075 | !rendered |
| 1076 | .get(..file_index) |
| 1077 | .expect("slice up to the file entry") |
| 1078 | .contains("entry-size"), |
| 1079 | "the directory row carries no size cell" |
| 1080 | ); |
| 1081 | assert!( |
| 1082 | rendered.contains("entry-size"), |
| 1083 | "the file row carries a size span" |
| 1084 | ); |
| 1085 | assert!(rendered.contains("2.0 KB"), "the size is human-formatted"); |
| 1086 | } |
| 1087 | |
| 1088 | #[rstest] |
| 1089 | #[case::bytes(0, "0 B")] |
| 1090 | #[case::bytes_under_a_kb(1023, "1023 B")] |
| 1091 | #[case::exactly_one_kb(1024, "1.0 KB")] |
| 1092 | #[case::fractional_kb(1536, "1.5 KB")] |
| 1093 | #[case::just_under_a_mb(1_048_575, "1023.9 KB")] |
| 1094 | #[case::exactly_one_mb(1_048_576, "1.0 MB")] |
| 1095 | #[case::fractional_mb(1_572_864, "1.5 MB")] |
| 1096 | fn human_size_formats_bytes_kb_and_mb(#[case] bytes: u64, #[case] expected: &str) { |
| 1097 | assert_eq!(human_size(bytes), expected); |
| 1098 | } |
| 1099 | |
| 1100 | #[test] |
| 1101 | fn blob_view_renders_markdown_as_a_heading_not_raw_markup() { |
| 1102 | let (body, _below) = blob_view( |
| 1103 | "readme.md", |
| 1104 | "readme.md", |
| 1105 | "deadbeef", |
| 1106 | &session(), |
| 1107 | b"# Title\n", |
| 1108 | &[], |
| 1109 | maud::html! {}, |
| 1110 | ) |
| 1111 | .expect("markdown renders"); |
| 1112 | assert!(body.into_string().contains("<h1>Title</h1>")); |
| 1113 | } |
| 1114 | |
| 1115 | #[test] |
| 1116 | fn blob_view_renders_asciidoc_as_a_heading_not_raw_markup() { |
| 1117 | let (body, _below) = blob_view( |
| 1118 | "readme.adoc", |
| 1119 | "readme.adoc", |
| 1120 | "deadbeef", |
| 1121 | &session(), |
| 1122 | b"= Title\n\nBody.\n", |
| 1123 | &[], |
| 1124 | maud::html! {}, |
| 1125 | ) |
| 1126 | .expect("asciidoc renders"); |
| 1127 | assert!(body.into_string().contains("<h1>Title</h1>")); |
| 1128 | } |
| 1129 | |
| 1130 | #[test] |
| 1131 | fn blob_view_escapes_plain_text_into_a_line_numbered_code_block() { |
| 1132 | let (body, _below) = blob_view( |
| 1133 | "notes.txt", |
| 1134 | "notes.txt", |
| 1135 | "deadbeef", |
| 1136 | &session(), |
| 1137 | b"1 < 2 and true", |
| 1138 | &[], |
| 1139 | maud::html! {}, |
| 1140 | ) |
| 1141 | .expect("plain text renders"); |
| 1142 | let rendered = body.into_string(); |
| 1143 | assert!(rendered.contains("blob-nums")); |
| 1144 | assert!(rendered.contains("<td class=\"blob-code\"><code>")); |
| 1145 | assert!(rendered.contains("1 < 2")); |
| 1146 | } |
| 1147 | |
| 1148 | #[test] |
| 1149 | fn blob_view_highlights_a_recognized_language_with_syntax_token_classes() { |
| 1150 | let (body, _below) = blob_view( |
| 1151 | "src/main.rs", |
| 1152 | "main.rs", |
| 1153 | "deadbeef", |
| 1154 | &session(), |
| 1155 | b"fn main() { let x = 1; }", |
| 1156 | &[], |
| 1157 | maud::html! {}, |
| 1158 | ) |
| 1159 | .expect("rust renders"); |
| 1160 | let rendered = body.into_string(); |
| 1161 | assert!(rendered.contains("blob-nums")); |
| 1162 | assert!(rendered.contains("class=\"code\"")); |
| 1163 | assert!(rendered.contains("class=\"keyword\"")); |
| 1164 | } |
| 1165 | |
| 1166 | #[test] |
| 1167 | fn blob_view_shows_a_placeholder_for_binary_content() { |
| 1168 | let (body, _below) = blob_view( |
| 1169 | "data.bin", |
| 1170 | "data.bin", |
| 1171 | "deadbeef", |
| 1172 | &session(), |
| 1173 | b"\0\x01\x02binary", |
| 1174 | &[], |
| 1175 | maud::html! {}, |
| 1176 | ) |
| 1177 | .expect("binary placeholder renders"); |
| 1178 | assert!(body.into_string().contains("Binary file")); |
| 1179 | } |
| 1180 | |
| 1181 | #[test] |
| 1182 | fn blob_view_routes_a_doc_comment_below_the_blob_never_inline() { |
| 1183 | let comments = vec![comment(Some(LineRange { start: 1, end: 1 }))]; |
| 1184 | let (_body, below) = blob_view( |
| 1185 | "readme.md", |
| 1186 | "readme.md", |
| 1187 | "deadbeef", |
| 1188 | &session(), |
| 1189 | b"# Title\n", |
| 1190 | &comments, |
| 1191 | maud::html! {}, |
| 1192 | ) |
| 1193 | .expect("markdown renders"); |
| 1194 | // A doc view has no source line to interleave at: every comment, |
| 1195 | // even one with a current line range, renders in the below |
| 1196 | // section -- `comments_section`'s plain, untitled list, not |
| 1197 | // `outdated_comments_section`'s titled one. |
| 1198 | assert!(below.into_string().contains("worth a look")); |
| 1199 | } |
| 1200 | |
| 1201 | #[test] |
| 1202 | fn blob_view_shows_the_header_with_line_count_size_and_language() { |
| 1203 | let (body, _below) = blob_view( |
| 1204 | "src/main.rs", |
| 1205 | "main.rs", |
| 1206 | "deadbeef", |
| 1207 | &session(), |
| 1208 | b"fn main() {}\n", |
| 1209 | &[], |
| 1210 | maud::html! {}, |
| 1211 | ) |
| 1212 | .expect("rust renders"); |
| 1213 | let rendered = body.into_string(); |
| 1214 | assert!(rendered.contains("blob-header")); |
| 1215 | assert!(rendered.contains("1 line")); |
| 1216 | assert!(rendered.contains("13 B")); |
| 1217 | assert!(rendered.contains("rust")); |
| 1218 | assert!(rendered.contains("comment on this file")); |
| 1219 | } |
| 1220 | |
| 1221 | #[test] |
| 1222 | fn blob_view_carries_the_composer_hooks_on_every_view_kind() { |
| 1223 | let (body, _below) = blob_view( |
| 1224 | "src/main.rs", |
| 1225 | "main.rs", |
| 1226 | "cafef00dcafef00dcafef00dcafef00dcafef00d", |
| 1227 | &session(), |
| 1228 | b"fn main() {}\n", |
| 1229 | &[], |
| 1230 | maud::html! {}, |
| 1231 | ) |
| 1232 | .expect("rust renders"); |
| 1233 | let rendered = body.into_string(); |
| 1234 | assert!(rendered.contains("data-path=\"src/main.rs\"")); |
| 1235 | assert!(rendered.contains("data-rev=\"cafef00dcafef00dcafef00dcafef00dcafef00d\"")); |
| 1236 | assert!(rendered.contains("id=\"composer-template\"")); |
| 1237 | assert!(rendered.contains("name=\"csrf\"")); |
| 1238 | assert!(rendered.contains("test-csrf-token")); |
| 1239 | assert!(rendered.contains("name=\"path\" value=\"src/main.rs\"")); |
| 1240 | assert!( |
| 1241 | rendered.contains("name=\"rev\" value=\"cafef00dcafef00dcafef00dcafef00dcafef00d\"") |
| 1242 | ); |
| 1243 | |
| 1244 | let (doc_body, _below) = blob_view( |
| 1245 | "readme.md", |
| 1246 | "readme.md", |
| 1247 | "deadbeef", |
| 1248 | &session(), |
| 1249 | b"# Title\n", |
| 1250 | &[], |
| 1251 | maud::html! {}, |
| 1252 | ) |
| 1253 | .expect("markdown renders"); |
| 1254 | assert!( |
| 1255 | doc_body.into_string().contains("id=\"composer-template\""), |
| 1256 | "a doc-rendered view has no per-line gutter, but \"comment on \ |
| 1257 | this file\" (empty `lines`) is exactly as meaningful there" |
| 1258 | ); |
| 1259 | } |
| 1260 | |
| 1261 | #[test] |
| 1262 | fn source_view_interleaves_a_comment_directly_after_its_last_line() { |
| 1263 | let comments = vec![comment(Some(LineRange { start: 1, end: 2 }))]; |
| 1264 | let rendered = source_view( |
| 1265 | "src/main.rs", |
| 1266 | "deadbeef", |
| 1267 | Markup::default(), |
| 1268 | Markup::default(), |
| 1269 | "line 1\nline 2\nline 3\n", |
| 1270 | None, |
| 1271 | &comments, |
| 1272 | ) |
| 1273 | .into_string(); |
| 1274 | let line2 = rendered.find("id=\"L2\"").expect("line 2 renders"); |
| 1275 | let card = rendered.find("comment-meta").expect("card renders"); |
| 1276 | let line3 = rendered.find("id=\"L3\"").expect("line 3 renders"); |
| 1277 | assert!( |
| 1278 | line2 < card && card < line3, |
| 1279 | "the card lands strictly between line 2 and line 3: {rendered}" |
| 1280 | ); |
| 1281 | } |
| 1282 | |
| 1283 | #[test] |
| 1284 | fn source_view_stacks_multiple_comments_ending_on_the_same_line_in_order() { |
| 1285 | let comments = vec![ |
| 1286 | { |
| 1287 | let mut c = comment(Some(LineRange { start: 1, end: 1 })); |
| 1288 | c.body = html! { p { "first" } }; |
| 1289 | c |
| 1290 | }, |
| 1291 | { |
| 1292 | let mut c = comment(Some(LineRange { start: 1, end: 1 })); |
| 1293 | c.body = html! { p { "second" } }; |
| 1294 | c |
| 1295 | }, |
| 1296 | ]; |
| 1297 | let rendered = source_view( |
| 1298 | "src/main.rs", |
| 1299 | "deadbeef", |
| 1300 | Markup::default(), |
| 1301 | Markup::default(), |
| 1302 | "line 1\nline 2\n", |
| 1303 | None, |
| 1304 | &comments, |
| 1305 | ) |
| 1306 | .into_string(); |
| 1307 | let first = rendered.find("first").expect("first comment renders"); |
| 1308 | let second = rendered.find("second").expect("second comment renders"); |
| 1309 | assert!(first < second, "stacked comments keep ref order"); |
| 1310 | } |
| 1311 | |
| 1312 | #[test] |
| 1313 | fn source_view_omits_a_comment_with_no_current_line_range() { |
| 1314 | let comments = vec![comment(None)]; |
| 1315 | let rendered = source_view( |
| 1316 | "src/main.rs", |
| 1317 | "deadbeef", |
| 1318 | Markup::default(), |
| 1319 | Markup::default(), |
| 1320 | "line 1\nline 2\n", |
| 1321 | None, |
| 1322 | &comments, |
| 1323 | ) |
| 1324 | .into_string(); |
| 1325 | assert!( |
| 1326 | !rendered.contains("worth a look"), |
| 1327 | "a comment with no lines has nowhere to interleave -- blob_view routes it below instead" |
| 1328 | ); |
| 1329 | } |
| 1330 | |
| 1331 | #[test] |
| 1332 | fn child_href_nests_under_the_current_directory() { |
| 1333 | assert_eq!(child_href("", "src"), "/files/src"); |
| 1334 | assert_eq!(child_href("src", "main.rs"), "/files/src/main.rs"); |
| 1335 | } |
| 1336 | |
| 1337 | #[test] |
| 1338 | fn tokenize_splits_spans_and_text_without_touching_entities() { |
| 1339 | let tokens = tokenize("<span class=\"keyword\">fn</span> 1 < 2"); |
| 1340 | assert_eq!( |
| 1341 | tokens, |
| 1342 | vec![ |
| 1343 | Token::Open("keyword"), |
| 1344 | Token::Text("fn"), |
| 1345 | Token::Close, |
| 1346 | Token::Text(" 1 < 2"), |
| 1347 | ] |
| 1348 | ); |
| 1349 | } |
| 1350 | |
| 1351 | #[test] |
| 1352 | fn split_highlighted_lines_reopens_a_span_that_crosses_a_newline() { |
| 1353 | // A three-line block comment as one span, per arborium's own |
| 1354 | // `spans_to_html` shape (see that function's own tests): one |
| 1355 | // `<span>` whose text contains embedded newlines, followed by an |
| 1356 | // unrelated keyword span on the line after. |
| 1357 | let html = "<span class=\"comment\">/*\nfoo\nbar*/</span>\n<span class=\"keyword\">fn</span> main() {}"; |
| 1358 | let lines = split_highlighted_lines(html, 4); |
| 1359 | assert_eq!( |
| 1360 | lines, |
| 1361 | vec![ |
| 1362 | "<span class=\"comment\">/*</span>".to_owned(), |
| 1363 | "<span class=\"comment\">foo</span>".to_owned(), |
| 1364 | "<span class=\"comment\">bar*/</span>".to_owned(), |
| 1365 | "<span class=\"keyword\">fn</span> main() {}".to_owned(), |
| 1366 | ], |
| 1367 | "each fragment is independently well-formed and still classed" |
| 1368 | ); |
| 1369 | } |
| 1370 | |
| 1371 | #[test] |
| 1372 | fn split_highlighted_lines_never_re_escapes_or_splits_an_entity() { |
| 1373 | let html = "<span class=\"operator\"><</span>\nnext"; |
| 1374 | let lines = split_highlighted_lines(html, 2); |
| 1375 | assert_eq!( |
| 1376 | lines, |
| 1377 | vec![ |
| 1378 | "<span class=\"operator\"><</span>".to_owned(), |
| 1379 | "next".to_owned(), |
| 1380 | ] |
| 1381 | ); |
| 1382 | } |
| 1383 | |
| 1384 | #[test] |
| 1385 | fn split_highlighted_lines_handles_plain_unhighlighted_text() { |
| 1386 | let lines = split_highlighted_lines("a\nb\nc", 3); |
| 1387 | assert_eq!(lines, vec!["a".to_owned(), "b".to_owned(), "c".to_owned()]); |
| 1388 | } |
| 1389 | } |