You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1210 lines
43KB

  1. extern crate glob;
  2. extern crate rayon;
  3. extern crate serde;
  4. extern crate tera;
  5. #[macro_use]
  6. extern crate serde_derive;
  7. extern crate sass_rs;
  8. #[macro_use]
  9. extern crate errors;
  10. extern crate config;
  11. extern crate front_matter;
  12. extern crate imageproc;
  13. extern crate library;
  14. extern crate link_checker;
  15. extern crate search;
  16. extern crate templates;
  17. extern crate utils;
  18. #[cfg(test)]
  19. extern crate tempfile;
  20. mod sitemap;
  21. use std::collections::HashMap;
  22. use std::fs::{copy, create_dir_all, remove_dir_all};
  23. use std::path::{Path, PathBuf};
  24. use std::sync::{Arc, Mutex, RwLock};
  25. use glob::glob;
  26. use rayon::prelude::*;
  27. use sass_rs::{compile_file, Options as SassOptions, OutputStyle};
  28. use tera::{Context, Tera};
  29. use config::{get_config, Config};
  30. use errors::{Error, ErrorKind, Result};
  31. use front_matter::InsertAnchor;
  32. use library::{
  33. find_taxonomies, sort_actual_pages_by_date, Library, Page, Paginator, Section, Taxonomy,
  34. };
  35. use link_checker::check_url;
  36. use templates::{global_fns, render_redirect_template, ZOLA_TERA};
  37. use utils::fs::{copy_directory, create_directory, create_file, ensure_directory_exists};
  38. use utils::net::get_available_port;
  39. use utils::templates::{render_template, rewrite_theme_paths};
  40. #[derive(Debug)]
  41. pub struct Site {
  42. /// The base path of the zola site
  43. pub base_path: PathBuf,
  44. /// The parsed config for the site
  45. pub config: Config,
  46. pub tera: Tera,
  47. imageproc: Arc<Mutex<imageproc::Processor>>,
  48. // the live reload port to be used if there is one
  49. pub live_reload: Option<u16>,
  50. pub output_path: PathBuf,
  51. content_path: PathBuf,
  52. pub static_path: PathBuf,
  53. pub taxonomies: Vec<Taxonomy>,
  54. /// A map of all .md files (section and pages) and their permalink
  55. /// We need that if there are relative links in the content that need to be resolved
  56. pub permalinks: HashMap<String, String>,
  57. /// Contains all pages and sections of the site
  58. pub library: Arc<RwLock<Library>>,
  59. }
  60. impl Site {
  61. /// Parse a site at the given path. Defaults to the current dir
  62. /// Passing in a path is only used in tests
  63. pub fn new<P: AsRef<Path>>(path: P, config_file: &str) -> Result<Site> {
  64. let path = path.as_ref();
  65. let mut config = get_config(path, config_file);
  66. config.load_extra_syntaxes(path)?;
  67. let tpl_glob =
  68. format!("{}/{}", path.to_string_lossy().replace("\\", "/"), "templates/**/*.*ml");
  69. // Only parsing as we might be extending templates from themes and that would error
  70. // as we haven't loaded them yet
  71. let mut tera =
  72. Tera::parse(&tpl_glob).map_err(|e| Error::chain("Error parsing templates", e))?;
  73. if let Some(theme) = config.theme.clone() {
  74. // Grab data from the extra section of the theme
  75. config.merge_with_theme(&path.join("themes").join(&theme).join("theme.toml"))?;
  76. // Test that the templates folder exist for that theme
  77. let theme_path = path.join("themes").join(&theme);
  78. if !theme_path.join("templates").exists() {
  79. bail!("Theme `{}` is missing a templates folder", theme);
  80. }
  81. let theme_tpl_glob = format!(
  82. "{}/{}",
  83. path.to_string_lossy().replace("\\", "/"),
  84. format!("themes/{}/templates/**/*.*ml", theme)
  85. );
  86. let mut tera_theme = Tera::parse(&theme_tpl_glob)
  87. .map_err(|e| Error::chain("Error parsing templates from themes", e))?;
  88. rewrite_theme_paths(&mut tera_theme, &theme);
  89. // TODO: we do that twice, make it dry?
  90. if theme_path.join("templates").join("robots.txt").exists() {
  91. tera_theme
  92. .add_template_file(theme_path.join("templates").join("robots.txt"), None)?;
  93. }
  94. tera_theme.build_inheritance_chains()?;
  95. tera.extend(&tera_theme)?;
  96. }
  97. tera.extend(&ZOLA_TERA)?;
  98. // the `extend` above already does it but hey
  99. tera.build_inheritance_chains()?;
  100. // TODO: Tera doesn't use globset right now so we can load the robots.txt as part
  101. // of the glob above, therefore we load it manually if it exists.
  102. if path.join("templates").join("robots.txt").exists() {
  103. tera.add_template_file(path.join("templates").join("robots.txt"), Some("robots.txt"))?;
  104. }
  105. let content_path = path.join("content");
  106. let static_path = path.join("static");
  107. let imageproc =
  108. imageproc::Processor::new(content_path.clone(), &static_path, &config.base_url);
  109. let site = Site {
  110. base_path: path.to_path_buf(),
  111. config,
  112. tera,
  113. imageproc: Arc::new(Mutex::new(imageproc)),
  114. live_reload: None,
  115. output_path: path.join("public"),
  116. content_path,
  117. static_path,
  118. taxonomies: Vec::new(),
  119. permalinks: HashMap::new(),
  120. // We will allocate it properly later on
  121. library: Arc::new(RwLock::new(Library::new(0, 0, false))),
  122. };
  123. Ok(site)
  124. }
  125. /// The index sections are ALWAYS at those paths
  126. /// There are one index section for the basic language + 1 per language
  127. fn index_section_paths(&self) -> Vec<(PathBuf, Option<String>)> {
  128. let mut res = vec![(self.content_path.join("_index.md"), None)];
  129. for language in &self.config.languages {
  130. res.push((
  131. self.content_path.join(format!("_index.{}.md", language.code)),
  132. Some(language.code.clone()),
  133. ));
  134. }
  135. res
  136. }
  137. /// We avoid the port the server is going to use as it's not bound yet
  138. /// when calling this function and we could end up having tried to bind
  139. /// both http and websocket server to the same port
  140. pub fn enable_live_reload(&mut self, port_to_avoid: u16) {
  141. self.live_reload = get_available_port(port_to_avoid);
  142. }
  143. /// Get the number of orphan (== without section) pages in the site
  144. pub fn get_number_orphan_pages(&self) -> usize {
  145. self.library.read().unwrap().get_all_orphan_pages().len()
  146. }
  147. pub fn set_base_url(&mut self, base_url: String) {
  148. let mut imageproc = self.imageproc.lock().expect("Couldn't lock imageproc (set_base_url)");
  149. imageproc.set_base_url(&base_url);
  150. self.config.base_url = base_url;
  151. }
  152. pub fn set_output_path<P: AsRef<Path>>(&mut self, path: P) {
  153. self.output_path = path.as_ref().to_path_buf();
  154. }
  155. /// Reads all .md files in the `content` directory and create pages/sections
  156. /// out of them
  157. pub fn load(&mut self) -> Result<()> {
  158. let base_path = self.base_path.to_string_lossy().replace("\\", "/");
  159. let content_glob = format!("{}/{}", base_path, "content/**/*.md");
  160. let (section_entries, page_entries): (Vec<_>, Vec<_>) = glob(&content_glob)
  161. .expect("Invalid glob")
  162. .filter_map(|e| e.ok())
  163. .filter(|e| !e.as_path().file_name().unwrap().to_str().unwrap().starts_with('.'))
  164. .partition(|entry| {
  165. entry.as_path().file_name().unwrap().to_str().unwrap().starts_with("_index.")
  166. });
  167. self.library = Arc::new(RwLock::new(Library::new(
  168. page_entries.len(),
  169. section_entries.len(),
  170. self.config.is_multilingual(),
  171. )));
  172. let sections = {
  173. let config = &self.config;
  174. section_entries
  175. .into_par_iter()
  176. .map(|entry| {
  177. let path = entry.as_path();
  178. Section::from_file(path, config, &self.base_path)
  179. })
  180. .collect::<Vec<_>>()
  181. };
  182. let pages = {
  183. let config = &self.config;
  184. page_entries
  185. .into_par_iter()
  186. .map(|entry| {
  187. let path = entry.as_path();
  188. Page::from_file(path, config, &self.base_path)
  189. })
  190. .collect::<Vec<_>>()
  191. };
  192. // Kinda duplicated code for add_section/add_page but necessary to do it that
  193. // way because of the borrow checker
  194. for section in sections {
  195. let s = section?;
  196. self.add_section(s, false)?;
  197. }
  198. self.create_default_index_sections()?;
  199. let mut pages_insert_anchors = HashMap::new();
  200. for page in pages {
  201. let p = page?;
  202. // Draft pages are not rendered in zola build so we just discard them
  203. if p.meta.draft && !self.config.is_in_serve_mode() {
  204. continue;
  205. }
  206. pages_insert_anchors.insert(
  207. p.file.path.clone(),
  208. self.find_parent_section_insert_anchor(&p.file.parent.clone(), &p.lang),
  209. );
  210. self.add_page(p, false)?;
  211. }
  212. // taxonomy Tera fns are loaded in `register_early_global_fns`
  213. // so we do need to populate it first.
  214. self.populate_taxonomies()?;
  215. self.register_early_global_fns();
  216. self.populate_sections();
  217. self.render_markdown()?;
  218. self.register_tera_global_fns();
  219. // Needs to be done after rendering markdown as we only get the anchors at that point
  220. self.check_internal_links_with_anchors()?;
  221. if self.config.is_in_check_mode() {
  222. self.check_external_links()?;
  223. }
  224. Ok(())
  225. }
  226. /// Very similar to check_external_links but can't be merged as far as I can see since we always
  227. /// want to check the internal links but only the external in zola check :/
  228. pub fn check_internal_links_with_anchors(&self) -> Result<()> {
  229. let library = self.library.write().expect("Get lock for check_internal_links_with_anchors");
  230. let page_links = library
  231. .pages()
  232. .values()
  233. .map(|p| {
  234. let path = &p.file.path;
  235. p.internal_links_with_anchors.iter().map(move |l| (path.clone(), l))
  236. })
  237. .flatten();
  238. let section_links = library
  239. .sections()
  240. .values()
  241. .map(|p| {
  242. let path = &p.file.path;
  243. p.internal_links_with_anchors.iter().map(move |l| (path.clone(), l))
  244. })
  245. .flatten();
  246. let all_links = page_links.chain(section_links).collect::<Vec<_>>();
  247. if self.config.is_in_check_mode() {
  248. println!("Checking {} internal link(s) with an anchor.", all_links.len());
  249. }
  250. if all_links.is_empty() {
  251. return Ok(());
  252. }
  253. let mut full_path = self.base_path.clone();
  254. full_path.push("content");
  255. let errors: Vec<_> = all_links
  256. .iter()
  257. .filter_map(|(page_path, (md_path, anchor))| {
  258. // There are a few `expect` here since the presence of the .md file will
  259. // already have been checked in the markdown rendering
  260. let mut p = full_path.clone();
  261. for part in md_path.split('/') {
  262. p.push(part);
  263. }
  264. if md_path.contains("_index.md") {
  265. let section = library
  266. .get_section(&p)
  267. .expect("Couldn't find section in check_internal_links_with_anchors");
  268. if section.has_anchor(&anchor) {
  269. None
  270. } else {
  271. Some((page_path, md_path, anchor))
  272. }
  273. } else {
  274. let page = library
  275. .get_page(&p)
  276. .expect("Couldn't find section in check_internal_links_with_anchors");
  277. if page.has_anchor(&anchor) {
  278. None
  279. } else {
  280. Some((page_path, md_path, anchor))
  281. }
  282. }
  283. })
  284. .collect();
  285. if self.config.is_in_check_mode() {
  286. println!(
  287. "> Checked {} internal link(s) with an anchor: {} error(s) found.",
  288. all_links.len(),
  289. errors.len()
  290. );
  291. }
  292. if errors.is_empty() {
  293. return Ok(());
  294. }
  295. let msg = errors
  296. .into_iter()
  297. .map(|(page_path, md_path, anchor)| {
  298. format!(
  299. "The anchor in the link `@/{}#{}` in {} does not exist.",
  300. md_path,
  301. anchor,
  302. page_path.to_string_lossy(),
  303. )
  304. })
  305. .collect::<Vec<_>>()
  306. .join("\n");
  307. Err(Error { kind: ErrorKind::Msg(msg), source: None })
  308. }
  309. pub fn check_external_links(&self) -> Result<()> {
  310. let library = self.library.write().expect("Get lock for check_external_links");
  311. let page_links = library
  312. .pages()
  313. .values()
  314. .map(|p| {
  315. let path = &p.file.path;
  316. p.external_links.iter().map(move |l| (path.clone(), l))
  317. })
  318. .flatten();
  319. let section_links = library
  320. .sections()
  321. .values()
  322. .map(|p| {
  323. let path = &p.file.path;
  324. p.external_links.iter().map(move |l| (path.clone(), l))
  325. })
  326. .flatten();
  327. let all_links = page_links.chain(section_links).collect::<Vec<_>>();
  328. println!("Checking {} external link(s).", all_links.len());
  329. if all_links.is_empty() {
  330. return Ok(());
  331. }
  332. // create thread pool with lots of threads so we can fetch
  333. // (almost) all pages simultaneously
  334. let threads = std::cmp::min(all_links.len(), 32);
  335. let pool = rayon::ThreadPoolBuilder::new()
  336. .num_threads(threads)
  337. .build()
  338. .map_err(|e| Error { kind: ErrorKind::Msg(e.to_string()), source: None })?;
  339. let errors: Vec<_> = pool.install(|| {
  340. all_links
  341. .par_iter()
  342. .filter_map(|(page_path, link)| {
  343. let res = check_url(&link);
  344. if res.is_valid() {
  345. None
  346. } else {
  347. Some((page_path, link, res))
  348. }
  349. })
  350. .collect()
  351. });
  352. println!(
  353. "> Checked {} external link(s): {} error(s) found.",
  354. all_links.len(),
  355. errors.len()
  356. );
  357. if errors.is_empty() {
  358. return Ok(());
  359. }
  360. let msg = errors
  361. .into_iter()
  362. .map(|(page_path, link, check_res)| {
  363. format!(
  364. "Dead link in {} to {}: {}",
  365. page_path.to_string_lossy(),
  366. link,
  367. check_res.message()
  368. )
  369. })
  370. .collect::<Vec<_>>()
  371. .join("\n");
  372. Err(Error { kind: ErrorKind::Msg(msg), source: None })
  373. }
  374. /// Insert a default index section for each language if necessary so we don't need to create
  375. /// a _index.md to render the index page at the root of the site
  376. pub fn create_default_index_sections(&mut self) -> Result<()> {
  377. for (index_path, lang) in self.index_section_paths() {
  378. if let Some(ref index_section) = self.library.read().unwrap().get_section(&index_path) {
  379. if self.config.build_search_index && !index_section.meta.in_search_index {
  380. bail!(
  381. "You have enabled search in the config but disabled it in the index section: \
  382. either turn off the search in the config or remote `in_search_index = true` from the \
  383. section front-matter."
  384. )
  385. }
  386. }
  387. let mut library = self.library.write().expect("Get lock for load");
  388. // Not in else because of borrow checker
  389. if !library.contains_section(&index_path) {
  390. let mut index_section = Section::default();
  391. index_section.file.parent = self.content_path.clone();
  392. index_section.file.filename =
  393. index_path.file_name().unwrap().to_string_lossy().to_string();
  394. if let Some(ref l) = lang {
  395. index_section.file.name = format!("_index.{}", l);
  396. index_section.permalink = self.config.make_permalink(l);
  397. let filename = format!("_index.{}.md", l);
  398. index_section.file.path = self.content_path.join(&filename);
  399. index_section.file.relative = filename;
  400. index_section.lang = index_section.file.find_language(&self.config)?;
  401. } else {
  402. index_section.file.name = "_index".to_string();
  403. index_section.permalink = self.config.make_permalink("");
  404. index_section.file.path = self.content_path.join("_index.md");
  405. index_section.file.relative = "_index.md".to_string();
  406. }
  407. library.insert_section(index_section);
  408. }
  409. }
  410. Ok(())
  411. }
  412. /// Render the markdown of all pages/sections
  413. /// Used in a build and in `serve` if a shortcode has changed
  414. pub fn render_markdown(&mut self) -> Result<()> {
  415. // Another silly thing needed to not borrow &self in parallel and
  416. // make the borrow checker happy
  417. let permalinks = &self.permalinks;
  418. let tera = &self.tera;
  419. let config = &self.config;
  420. // This is needed in the first place because of silly borrow checker
  421. let mut pages_insert_anchors = HashMap::new();
  422. for (_, p) in self.library.read().unwrap().pages() {
  423. pages_insert_anchors.insert(
  424. p.file.path.clone(),
  425. self.find_parent_section_insert_anchor(&p.file.parent.clone(), &p.lang),
  426. );
  427. }
  428. let mut library = self.library.write().expect("Get lock for render_markdown");
  429. library
  430. .pages_mut()
  431. .values_mut()
  432. .collect::<Vec<_>>()
  433. .par_iter_mut()
  434. .map(|page| {
  435. let insert_anchor = pages_insert_anchors[&page.file.path];
  436. page.render_markdown(permalinks, tera, config, insert_anchor)
  437. })
  438. .collect::<Result<()>>()?;
  439. library
  440. .sections_mut()
  441. .values_mut()
  442. .collect::<Vec<_>>()
  443. .par_iter_mut()
  444. .map(|section| section.render_markdown(permalinks, tera, config))
  445. .collect::<Result<()>>()?;
  446. Ok(())
  447. }
  448. /// Adds global fns that are to be available to shortcodes while
  449. /// markdown
  450. pub fn register_early_global_fns(&mut self) {
  451. self.tera.register_function(
  452. "get_url",
  453. global_fns::GetUrl::new(self.config.clone(), self.permalinks.clone()),
  454. );
  455. self.tera.register_function(
  456. "resize_image",
  457. global_fns::ResizeImage::new(self.imageproc.clone()),
  458. );
  459. self.tera.register_function(
  460. "get_image_metadata",
  461. global_fns::GetImageMeta::new(self.content_path.clone()),
  462. );
  463. self.tera.register_function("load_data", global_fns::LoadData::new(self.base_path.clone()));
  464. self.tera.register_function("trans", global_fns::Trans::new(self.config.clone()));
  465. self.tera.register_function(
  466. "get_taxonomy_url",
  467. global_fns::GetTaxonomyUrl::new(&self.taxonomies),
  468. );
  469. }
  470. pub fn register_tera_global_fns(&mut self) {
  471. self.tera.register_function(
  472. "get_page",
  473. global_fns::GetPage::new(self.base_path.clone(), self.library.clone()),
  474. );
  475. self.tera.register_function(
  476. "get_section",
  477. global_fns::GetSection::new(self.base_path.clone(), self.library.clone()),
  478. );
  479. self.tera.register_function(
  480. "get_taxonomy",
  481. global_fns::GetTaxonomy::new(self.taxonomies.clone(), self.library.clone()),
  482. );
  483. }
  484. /// Add a page to the site
  485. /// The `render` parameter is used in the serve command, when rebuilding a page.
  486. /// If `true`, it will also render the markdown for that page
  487. /// Returns the previous page struct if there was one at the same path
  488. pub fn add_page(&mut self, mut page: Page, render: bool) -> Result<Option<Page>> {
  489. self.permalinks.insert(page.file.relative.clone(), page.permalink.clone());
  490. if render {
  491. let insert_anchor =
  492. self.find_parent_section_insert_anchor(&page.file.parent, &page.lang);
  493. page.render_markdown(&self.permalinks, &self.tera, &self.config, insert_anchor)?;
  494. }
  495. let mut library = self.library.write().expect("Get lock for add_page");
  496. let prev = library.remove_page(&page.file.path);
  497. library.insert_page(page);
  498. Ok(prev)
  499. }
  500. /// Add a section to the site
  501. /// The `render` parameter is used in the serve command, when rebuilding a page.
  502. /// If `true`, it will also render the markdown for that page
  503. /// Returns the previous section struct if there was one at the same path
  504. pub fn add_section(&mut self, mut section: Section, render: bool) -> Result<Option<Section>> {
  505. self.permalinks.insert(section.file.relative.clone(), section.permalink.clone());
  506. if render {
  507. section.render_markdown(&self.permalinks, &self.tera, &self.config)?;
  508. }
  509. let mut library = self.library.write().expect("Get lock for add_section");
  510. let prev = library.remove_section(&section.file.path);
  511. library.insert_section(section);
  512. Ok(prev)
  513. }
  514. /// Finds the insert_anchor for the parent section of the directory at `path`.
  515. /// Defaults to `AnchorInsert::None` if no parent section found
  516. pub fn find_parent_section_insert_anchor(
  517. &self,
  518. parent_path: &PathBuf,
  519. lang: &str,
  520. ) -> InsertAnchor {
  521. let parent = if lang != self.config.default_language {
  522. parent_path.join(format!("_index.{}.md", lang))
  523. } else {
  524. parent_path.join("_index.md")
  525. };
  526. match self.library.read().unwrap().get_section(&parent) {
  527. Some(s) => s.meta.insert_anchor_links,
  528. None => InsertAnchor::None,
  529. }
  530. }
  531. /// Find out the direct subsections of each subsection if there are some
  532. /// as well as the pages for each section
  533. pub fn populate_sections(&mut self) {
  534. let mut library = self.library.write().expect("Get lock for populate_sections");
  535. library.populate_sections(&self.config);
  536. }
  537. /// Find all the tags and categories if it's asked in the config
  538. pub fn populate_taxonomies(&mut self) -> Result<()> {
  539. if self.config.taxonomies.is_empty() {
  540. return Ok(());
  541. }
  542. self.taxonomies = find_taxonomies(&self.config, &self.library.read().unwrap())?;
  543. Ok(())
  544. }
  545. /// Inject live reload script tag if in live reload mode
  546. fn inject_livereload(&self, html: String) -> String {
  547. if let Some(port) = self.live_reload {
  548. return html.replace(
  549. "</body>",
  550. &format!(
  551. r#"<script src="/livereload.js?port={}&mindelay=10"></script></body>"#,
  552. port
  553. ),
  554. );
  555. }
  556. html
  557. }
  558. /// Copy the main `static` folder and the theme `static` folder if a theme is used
  559. pub fn copy_static_directories(&self) -> Result<()> {
  560. // The user files will overwrite the theme files
  561. if let Some(ref theme) = self.config.theme {
  562. copy_directory(
  563. &self.base_path.join("themes").join(theme).join("static"),
  564. &self.output_path,
  565. false,
  566. )?;
  567. }
  568. // We're fine with missing static folders
  569. if self.static_path.exists() {
  570. copy_directory(&self.static_path, &self.output_path, self.config.hard_link_static)?;
  571. }
  572. Ok(())
  573. }
  574. pub fn num_img_ops(&self) -> usize {
  575. let imageproc = self.imageproc.lock().expect("Couldn't lock imageproc (num_img_ops)");
  576. imageproc.num_img_ops()
  577. }
  578. pub fn process_images(&self) -> Result<()> {
  579. let mut imageproc =
  580. self.imageproc.lock().expect("Couldn't lock imageproc (process_images)");
  581. imageproc.prune()?;
  582. imageproc.do_process()
  583. }
  584. /// Deletes the `public` directory if it exists
  585. pub fn clean(&self) -> Result<()> {
  586. if self.output_path.exists() {
  587. // Delete current `public` directory so we can start fresh
  588. remove_dir_all(&self.output_path)
  589. .map_err(|e| Error::chain("Couldn't delete output directory", e))?;
  590. }
  591. Ok(())
  592. }
  593. /// Renders a single content page
  594. pub fn render_page(&self, page: &Page) -> Result<()> {
  595. ensure_directory_exists(&self.output_path)?;
  596. // Copy the nesting of the content directory if we have sections for that page
  597. let mut current_path = self.output_path.to_path_buf();
  598. for component in page.path.split('/') {
  599. current_path.push(component);
  600. if !current_path.exists() {
  601. create_directory(&current_path)?;
  602. }
  603. }
  604. // Make sure the folder exists
  605. create_directory(&current_path)?;
  606. // Finally, create a index.html file there with the page rendered
  607. let output = page.render_html(&self.tera, &self.config, &self.library.read().unwrap())?;
  608. create_file(&current_path.join("index.html"), &self.inject_livereload(output))?;
  609. // Copy any asset we found previously into the same directory as the index.html
  610. for asset in &page.assets {
  611. let asset_path = asset.as_path();
  612. copy(
  613. &asset_path,
  614. &current_path
  615. .join(asset_path.file_name().expect("Couldn't get filename from page asset")),
  616. )?;
  617. }
  618. Ok(())
  619. }
  620. /// Deletes the `public` directory and builds the site
  621. pub fn build(&self) -> Result<()> {
  622. self.clean()?;
  623. // Generate/move all assets before rendering any content
  624. if let Some(ref theme) = self.config.theme {
  625. let theme_path = self.base_path.join("themes").join(theme);
  626. if theme_path.join("sass").exists() {
  627. self.compile_sass(&theme_path)?;
  628. }
  629. }
  630. if self.config.compile_sass {
  631. self.compile_sass(&self.base_path)?;
  632. }
  633. if self.config.build_search_index {
  634. self.build_search_index()?;
  635. }
  636. // Render aliases first to allow overwriting
  637. self.render_aliases()?;
  638. self.render_sections()?;
  639. self.render_orphan_pages()?;
  640. self.render_sitemap()?;
  641. let library = self.library.read().unwrap();
  642. if self.config.generate_rss {
  643. let pages = if self.config.is_multilingual() {
  644. library
  645. .pages_values()
  646. .iter()
  647. .filter(|p| p.lang == self.config.default_language)
  648. .cloned()
  649. .collect()
  650. } else {
  651. library.pages_values()
  652. };
  653. self.render_rss_feed(pages, None)?;
  654. }
  655. for lang in &self.config.languages {
  656. if !lang.rss {
  657. continue;
  658. }
  659. let pages =
  660. library.pages_values().iter().filter(|p| p.lang == lang.code).cloned().collect();
  661. self.render_rss_feed(pages, Some(&PathBuf::from(lang.code.clone())))?;
  662. }
  663. self.render_404()?;
  664. self.render_robots()?;
  665. self.render_taxonomies()?;
  666. // We process images at the end as we might have picked up images to process from markdown
  667. // or from templates
  668. self.process_images()?;
  669. // Processed images will be in static so the last step is to copy it
  670. self.copy_static_directories()?;
  671. Ok(())
  672. }
  673. pub fn build_search_index(&self) -> Result<()> {
  674. // index first
  675. create_file(
  676. &self.output_path.join(&format!("search_index.{}.js", self.config.default_language)),
  677. &format!(
  678. "window.searchIndex = {};",
  679. search::build_index(&self.config.default_language, &self.library.read().unwrap())?
  680. ),
  681. )?;
  682. // then elasticlunr.min.js
  683. create_file(&self.output_path.join("elasticlunr.min.js"), search::ELASTICLUNR_JS)?;
  684. Ok(())
  685. }
  686. pub fn compile_sass(&self, base_path: &Path) -> Result<()> {
  687. ensure_directory_exists(&self.output_path)?;
  688. let sass_path = {
  689. let mut sass_path = PathBuf::from(base_path);
  690. sass_path.push("sass");
  691. sass_path
  692. };
  693. let mut options = SassOptions::default();
  694. options.output_style = OutputStyle::Compressed;
  695. let mut compiled_paths = self.compile_sass_glob(&sass_path, "scss", &options.clone())?;
  696. options.indented_syntax = true;
  697. compiled_paths.extend(self.compile_sass_glob(&sass_path, "sass", &options)?);
  698. compiled_paths.sort();
  699. for window in compiled_paths.windows(2) {
  700. if window[0].1 == window[1].1 {
  701. bail!(
  702. "SASS path conflict: \"{}\" and \"{}\" both compile to \"{}\"",
  703. window[0].0.display(),
  704. window[1].0.display(),
  705. window[0].1.display(),
  706. );
  707. }
  708. }
  709. Ok(())
  710. }
  711. fn compile_sass_glob(
  712. &self,
  713. sass_path: &Path,
  714. extension: &str,
  715. options: &SassOptions,
  716. ) -> Result<Vec<(PathBuf, PathBuf)>> {
  717. let glob_string = format!("{}/**/*.{}", sass_path.display(), extension);
  718. let files = glob(&glob_string)
  719. .expect("Invalid glob for sass")
  720. .filter_map(|e| e.ok())
  721. .filter(|entry| {
  722. !entry.as_path().file_name().unwrap().to_string_lossy().starts_with('_')
  723. })
  724. .collect::<Vec<_>>();
  725. let mut compiled_paths = Vec::new();
  726. for file in files {
  727. let css = compile_file(&file, options.clone())?;
  728. let path_inside_sass = file.strip_prefix(&sass_path).unwrap();
  729. let parent_inside_sass = path_inside_sass.parent();
  730. let css_output_path = self.output_path.join(path_inside_sass).with_extension("css");
  731. if parent_inside_sass.is_some() {
  732. create_dir_all(&css_output_path.parent().unwrap())?;
  733. }
  734. create_file(&css_output_path, &css)?;
  735. compiled_paths.push((path_inside_sass.to_owned(), css_output_path));
  736. }
  737. Ok(compiled_paths)
  738. }
  739. fn render_alias(&self, alias: &str, permalink: &str) -> Result<()> {
  740. let mut output_path = self.output_path.to_path_buf();
  741. let mut split = alias.split('/').collect::<Vec<_>>();
  742. // If the alias ends with an html file name, use that instead of mapping
  743. // as a path containing an `index.html`
  744. let page_name = match split.pop() {
  745. Some(part) if part.ends_with(".html") => part,
  746. Some(part) => {
  747. split.push(part);
  748. "index.html"
  749. }
  750. None => "index.html",
  751. };
  752. for component in split {
  753. output_path.push(&component);
  754. if !output_path.exists() {
  755. create_directory(&output_path)?;
  756. }
  757. }
  758. create_file(
  759. &output_path.join(page_name),
  760. &render_redirect_template(&permalink, &self.tera)?,
  761. )
  762. }
  763. pub fn render_aliases(&self) -> Result<()> {
  764. ensure_directory_exists(&self.output_path)?;
  765. let library = self.library.read().unwrap();
  766. for (_, page) in library.pages() {
  767. for alias in &page.meta.aliases {
  768. self.render_alias(&alias, &page.permalink)?;
  769. }
  770. }
  771. for (_, section) in library.sections() {
  772. for alias in &section.meta.aliases {
  773. self.render_alias(&alias, &section.permalink)?;
  774. }
  775. }
  776. Ok(())
  777. }
  778. /// Renders 404.html
  779. pub fn render_404(&self) -> Result<()> {
  780. ensure_directory_exists(&self.output_path)?;
  781. let mut context = Context::new();
  782. context.insert("config", &self.config);
  783. let output = render_template("404.html", &self.tera, context, &self.config.theme)?;
  784. create_file(&self.output_path.join("404.html"), &self.inject_livereload(output))
  785. }
  786. /// Renders robots.txt
  787. pub fn render_robots(&self) -> Result<()> {
  788. ensure_directory_exists(&self.output_path)?;
  789. let mut context = Context::new();
  790. context.insert("config", &self.config);
  791. create_file(
  792. &self.output_path.join("robots.txt"),
  793. &render_template("robots.txt", &self.tera, context, &self.config.theme)?,
  794. )
  795. }
  796. /// Renders all taxonomies
  797. pub fn render_taxonomies(&self) -> Result<()> {
  798. for taxonomy in &self.taxonomies {
  799. self.render_taxonomy(taxonomy)?;
  800. }
  801. Ok(())
  802. }
  803. fn render_taxonomy(&self, taxonomy: &Taxonomy) -> Result<()> {
  804. if taxonomy.items.is_empty() {
  805. return Ok(());
  806. }
  807. ensure_directory_exists(&self.output_path)?;
  808. let output_path = if taxonomy.kind.lang != self.config.default_language {
  809. let mid_path = self.output_path.join(&taxonomy.kind.lang);
  810. create_directory(&mid_path)?;
  811. mid_path.join(&taxonomy.kind.name)
  812. } else {
  813. self.output_path.join(&taxonomy.kind.name)
  814. };
  815. let list_output =
  816. taxonomy.render_all_terms(&self.tera, &self.config, &self.library.read().unwrap())?;
  817. create_directory(&output_path)?;
  818. create_file(&output_path.join("index.html"), &self.inject_livereload(list_output))?;
  819. let library = self.library.read().unwrap();
  820. taxonomy
  821. .items
  822. .par_iter()
  823. .map(|item| {
  824. let path = output_path.join(&item.slug);
  825. if taxonomy.kind.is_paginated() {
  826. self.render_paginated(
  827. &path,
  828. &Paginator::from_taxonomy(&taxonomy, item, &library),
  829. )?;
  830. } else {
  831. let single_output =
  832. taxonomy.render_term(item, &self.tera, &self.config, &library)?;
  833. create_directory(&path)?;
  834. create_file(&path.join("index.html"), &self.inject_livereload(single_output))?;
  835. }
  836. if taxonomy.kind.rss {
  837. self.render_rss_feed(
  838. item.pages.iter().map(|p| library.get_page_by_key(*p)).collect(),
  839. Some(&PathBuf::from(format!("{}/{}", taxonomy.kind.name, item.slug))),
  840. )
  841. } else {
  842. Ok(())
  843. }
  844. })
  845. .collect::<Result<()>>()
  846. }
  847. /// What it says on the tin
  848. pub fn render_sitemap(&self) -> Result<()> {
  849. ensure_directory_exists(&self.output_path)?;
  850. let library = self.library.read().unwrap();
  851. let all_sitemap_entries = {
  852. let mut all_sitemap_entries =
  853. sitemap::find_entries(&library, &self.taxonomies[..], &self.config);
  854. all_sitemap_entries.sort();
  855. all_sitemap_entries
  856. };
  857. let sitemap_limit = 30000;
  858. if all_sitemap_entries.len() < sitemap_limit {
  859. // Create single sitemap
  860. let mut context = Context::new();
  861. context.insert("entries", &all_sitemap_entries);
  862. let sitemap = &render_template("sitemap.xml", &self.tera, context, &self.config.theme)?;
  863. create_file(&self.output_path.join("sitemap.xml"), sitemap)?;
  864. return Ok(());
  865. }
  866. // Create multiple sitemaps (max 30000 urls each)
  867. let mut sitemap_index = Vec::new();
  868. for (i, chunk) in
  869. all_sitemap_entries.iter().collect::<Vec<_>>().chunks(sitemap_limit).enumerate()
  870. {
  871. let mut context = Context::new();
  872. context.insert("entries", &chunk);
  873. let sitemap = &render_template("sitemap.xml", &self.tera, context, &self.config.theme)?;
  874. let file_name = format!("sitemap{}.xml", i + 1);
  875. create_file(&self.output_path.join(&file_name), sitemap)?;
  876. let mut sitemap_url: String = self.config.make_permalink(&file_name);
  877. sitemap_url.pop(); // Remove trailing slash
  878. sitemap_index.push(sitemap_url);
  879. }
  880. // Create main sitemap that reference numbered sitemaps
  881. let mut main_context = Context::new();
  882. main_context.insert("sitemaps", &sitemap_index);
  883. let sitemap = &render_template(
  884. "split_sitemap_index.xml",
  885. &self.tera,
  886. main_context,
  887. &self.config.theme,
  888. )?;
  889. create_file(&self.output_path.join("sitemap.xml"), sitemap)?;
  890. Ok(())
  891. }
  892. /// Renders a RSS feed for the given path and at the given path
  893. /// If both arguments are `None`, it will render only the RSS feed for the whole
  894. /// site at the root folder.
  895. pub fn render_rss_feed(
  896. &self,
  897. all_pages: Vec<&Page>,
  898. base_path: Option<&PathBuf>,
  899. ) -> Result<()> {
  900. ensure_directory_exists(&self.output_path)?;
  901. let mut context = Context::new();
  902. let mut pages = all_pages.into_iter().filter(|p| p.meta.date.is_some()).collect::<Vec<_>>();
  903. // Don't generate a RSS feed if none of the pages has a date
  904. if pages.is_empty() {
  905. return Ok(());
  906. }
  907. pages.par_sort_unstable_by(sort_actual_pages_by_date);
  908. context.insert("last_build_date", &pages[0].meta.date.clone());
  909. let library = self.library.read().unwrap();
  910. // limit to the last n elements if the limit is set; otherwise use all.
  911. let num_entries = self.config.rss_limit.unwrap_or_else(|| pages.len());
  912. let p = pages
  913. .iter()
  914. .take(num_entries)
  915. .map(|x| x.to_serialized_basic(&library))
  916. .collect::<Vec<_>>();
  917. context.insert("pages", &p);
  918. context.insert("config", &self.config);
  919. let rss_feed_url = if let Some(ref base) = base_path {
  920. self.config.make_permalink(&base.join("rss.xml").to_string_lossy().replace('\\', "/"))
  921. } else {
  922. self.config.make_permalink("rss.xml")
  923. };
  924. context.insert("feed_url", &rss_feed_url);
  925. let feed = &render_template("rss.xml", &self.tera, context, &self.config.theme)?;
  926. if let Some(ref base) = base_path {
  927. let mut output_path = self.output_path.clone();
  928. for component in base.components() {
  929. output_path.push(component);
  930. if !output_path.exists() {
  931. create_directory(&output_path)?;
  932. }
  933. }
  934. create_file(&output_path.join("rss.xml"), feed)?;
  935. } else {
  936. create_file(&self.output_path.join("rss.xml"), feed)?;
  937. }
  938. Ok(())
  939. }
  940. /// Renders a single section
  941. pub fn render_section(&self, section: &Section, render_pages: bool) -> Result<()> {
  942. ensure_directory_exists(&self.output_path)?;
  943. let mut output_path = self.output_path.clone();
  944. if section.lang != self.config.default_language {
  945. output_path.push(&section.lang);
  946. if !output_path.exists() {
  947. create_directory(&output_path)?;
  948. }
  949. }
  950. for component in &section.file.components {
  951. output_path.push(component);
  952. if !output_path.exists() {
  953. create_directory(&output_path)?;
  954. }
  955. }
  956. // Copy any asset we found previously into the same directory as the index.html
  957. for asset in &section.assets {
  958. let asset_path = asset.as_path();
  959. copy(
  960. &asset_path,
  961. &output_path.join(
  962. asset_path.file_name().expect("Failed to get asset filename for section"),
  963. ),
  964. )?;
  965. }
  966. if render_pages {
  967. section
  968. .pages
  969. .par_iter()
  970. .map(|k| self.render_page(self.library.read().unwrap().get_page_by_key(*k)))
  971. .collect::<Result<()>>()?;
  972. }
  973. if !section.meta.render {
  974. return Ok(());
  975. }
  976. if let Some(ref redirect_to) = section.meta.redirect_to {
  977. let permalink = self.config.make_permalink(redirect_to);
  978. create_file(
  979. &output_path.join("index.html"),
  980. &render_redirect_template(&permalink, &self.tera)?,
  981. )?;
  982. return Ok(());
  983. }
  984. if section.meta.is_paginated() {
  985. self.render_paginated(
  986. &output_path,
  987. &Paginator::from_section(&section, &self.library.read().unwrap()),
  988. )?;
  989. } else {
  990. let output =
  991. section.render_html(&self.tera, &self.config, &self.library.read().unwrap())?;
  992. create_file(&output_path.join("index.html"), &self.inject_livereload(output))?;
  993. }
  994. Ok(())
  995. }
  996. /// Used only on reload
  997. pub fn render_index(&self) -> Result<()> {
  998. self.render_section(
  999. &self
  1000. .library
  1001. .read()
  1002. .unwrap()
  1003. .get_section(&self.content_path.join("_index.md"))
  1004. .expect("Failed to get index section"),
  1005. false,
  1006. )
  1007. }
  1008. /// Renders all sections
  1009. pub fn render_sections(&self) -> Result<()> {
  1010. self.library
  1011. .read()
  1012. .unwrap()
  1013. .sections_values()
  1014. .into_par_iter()
  1015. .map(|s| self.render_section(s, true))
  1016. .collect::<Result<()>>()
  1017. }
  1018. /// Renders all pages that do not belong to any sections
  1019. pub fn render_orphan_pages(&self) -> Result<()> {
  1020. ensure_directory_exists(&self.output_path)?;
  1021. let library = self.library.read().unwrap();
  1022. for page in library.get_all_orphan_pages() {
  1023. self.render_page(page)?;
  1024. }
  1025. Ok(())
  1026. }
  1027. /// Renders a list of pages when the section/index is wanting pagination.
  1028. pub fn render_paginated(&self, output_path: &Path, paginator: &Paginator) -> Result<()> {
  1029. ensure_directory_exists(&self.output_path)?;
  1030. let folder_path = output_path.join(&paginator.paginate_path);
  1031. create_directory(&folder_path)?;
  1032. paginator
  1033. .pagers
  1034. .par_iter()
  1035. .map(|pager| {
  1036. let page_path = folder_path.join(&format!("{}", pager.index));
  1037. create_directory(&page_path)?;
  1038. let output = paginator.render_pager(
  1039. pager,
  1040. &self.config,
  1041. &self.tera,
  1042. &self.library.read().unwrap(),
  1043. )?;
  1044. if pager.index > 1 {
  1045. create_file(&page_path.join("index.html"), &self.inject_livereload(output))?;
  1046. } else {
  1047. create_file(&output_path.join("index.html"), &self.inject_livereload(output))?;
  1048. create_file(
  1049. &page_path.join("index.html"),
  1050. &render_redirect_template(&paginator.permalink, &self.tera)?,
  1051. )?;
  1052. }
  1053. Ok(())
  1054. })
  1055. .collect::<Result<()>>()
  1056. }
  1057. }