1 // Copyright 2016 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! Script to check the validity of `href` links in our HTML documentation.
13 //! In the past we've been quite error prone to writing in broken links as most
14 //! of them are manually rather than automatically added. As files move over
15 //! time or apis change old links become stale or broken. The purpose of this
16 //! script is to check all relative links in our documentation to make sure they
17 //! actually point to a valid place.
19 //! Currently this doesn't actually do any HTML parsing or anything fancy like
20 //! that, it just has a simple "regex" to search for `href` and `id` tags.
21 //! These values are then translated to file URLs if possible and then the
22 //! destination is asserted to exist.
24 //! A few whitelisted exceptions are allowed as there's known bugs in rustdoc,
25 //! but this should catch the majority of "broken link" cases.
29 use std::io::prelude::*;
30 use std::path::{Path, PathBuf, Component};
31 use std::collections::{HashMap, HashSet};
32 use std::collections::hash_map::Entry;
37 ($e:expr) => (match $e {
39 Err(e) => panic!("{} failed with {:?}", stringify!($e), e),
44 let docs = env::args_os().nth(1).unwrap();
45 let docs = env::current_dir().unwrap().join(docs);
46 let mut errors = false;
47 walk(&mut HashMap::new(), &docs, &docs, &mut errors);
49 panic!("found some broken links");
55 IOError(std::io::Error),
56 BrokenRedirect(PathBuf, std::io::Error),
70 type Cache = HashMap<PathBuf, FileEntry>;
72 fn small_url_encode(s: &str) -> String {
88 fn parse_ids(&mut self, file: &Path, contents: &str, errors: &mut bool) {
89 if self.ids.is_empty() {
90 with_attrs_in_source(contents, " id", |fragment, i, _| {
91 let frag = fragment.trim_left_matches("#").to_owned();
92 let encoded = small_url_encode(&frag);
93 if !self.ids.insert(frag) {
95 println!("{}:{}: id is not unique: `{}`", file.display(), i, fragment);
97 // Just in case, we also add the encoded id.
98 self.ids.insert(encoded);
104 fn walk(cache: &mut Cache, root: &Path, dir: &Path, errors: &mut bool) {
105 for entry in t!(dir.read_dir()).map(|e| t!(e)) {
106 let path = entry.path();
107 let kind = t!(entry.file_type());
109 walk(cache, root, &path, errors);
111 let pretty_path = check(cache, root, &path, errors);
112 if let Some(pretty_path) = pretty_path {
113 let entry = cache.get_mut(&pretty_path).unwrap();
114 // we don't need the source anymore,
115 // so drop to reduce memory-usage
116 entry.source = String::new();
122 fn check(cache: &mut Cache,
127 // Ignore none HTML files.
128 if file.extension().and_then(|s| s.to_str()) != Some("html") {
132 // Unfortunately we're not 100% full of valid links today to we need a few
133 // whitelists to get this past `make check` today.
135 if file.ends_with("std/string/struct.String.html") ||
136 file.ends_with("interpret/struct.ValTy.html") ||
137 file.ends_with("symbol/struct.InternedString.html") ||
138 file.ends_with("ast/struct.ThinVec.html") ||
139 file.ends_with("util/struct.ThinVec.html") ||
140 file.ends_with("util/struct.RcSlice.html") ||
141 file.ends_with("layout/struct.TyLayout.html") ||
142 file.ends_with("ty/struct.Slice.html") ||
143 file.ends_with("ty/enum.Attributes.html") ||
144 file.ends_with("ty/struct.SymbolName.html") {
148 if file.ends_with("string/struct.String.html") {
152 if file.ends_with("btree_set/struct.BTreeSet.html") ||
153 file.ends_with("struct.BTreeSet.html") ||
154 file.ends_with("btree_map/struct.BTreeMap.html") ||
155 file.ends_with("hash_map/struct.HashMap.html") ||
156 file.ends_with("hash_set/struct.HashSet.html") ||
157 file.ends_with("sync/struct.Lrc.html") ||
158 file.ends_with("sync/struct.RwLock.html") {
162 let res = load_file(cache, root, file, SkipRedirect);
163 let (pretty_file, contents) = match res {
165 Err(_) => return None,
168 cache.get_mut(&pretty_file)
170 .parse_ids(&pretty_file, &contents, errors);
173 // Search for anything that's the regex 'href[ ]*=[ ]*".*?"'
174 with_attrs_in_source(&contents, " href", |url, i, base| {
175 // Ignore external URLs
176 if url.starts_with("http:") || url.starts_with("https:") ||
177 url.starts_with("javascript:") || url.starts_with("ftp:") ||
178 url.starts_with("irc:") || url.starts_with("data:") {
181 let mut parts = url.splitn(2, "#");
182 let url = parts.next().unwrap();
183 let fragment = parts.next();
184 let mut parts = url.splitn(2, "?");
185 let url = parts.next().unwrap();
187 // Once we've plucked out the URL, parse it using our base url and
188 // then try to extract a file path.
189 let mut path = file.to_path_buf();
190 if !base.is_empty() || !url.is_empty() {
192 for part in Path::new(base).join(url).components() {
194 Component::Prefix(_) |
195 Component::RootDir => panic!(),
196 Component::CurDir => {}
197 Component::ParentDir => { path.pop(); }
198 Component::Normal(s) => { path.push(s); }
203 // Alright, if we've found a file name then this file had better
204 // exist! If it doesn't then we register and print an error.
207 // Links to directories show as directory listings when viewing
208 // the docs offline so it's best to avoid them.
210 let pretty_path = path.strip_prefix(root).unwrap_or(&path);
211 println!("{}:{}: directory link - {}",
212 pretty_file.display(),
214 pretty_path.display());
217 if let Some(extension) = path.extension() {
218 // Ignore none HTML files.
219 if extension != "html" {
223 let res = load_file(cache, root, &path, FromRedirect(false));
224 let (pretty_path, contents) = match res {
226 Err(LoadError::IOError(err)) => {
227 panic!("error loading {}: {}", path.display(), err);
229 Err(LoadError::BrokenRedirect(target, _)) => {
231 println!("{}:{}: broken redirect to {}",
232 pretty_file.display(),
237 Err(LoadError::IsRedirect) => unreachable!(),
240 if let Some(ref fragment) = fragment {
241 // Fragments like `#1-6` are most likely line numbers to be
242 // interpreted by javascript, so we're ignoring these
243 if fragment.splitn(2, '-')
244 .all(|f| f.chars().all(|c| c.is_numeric())) {
248 let entry = &mut cache.get_mut(&pretty_path).unwrap();
249 entry.parse_ids(&pretty_path, &contents, errors);
251 if !entry.ids.contains(*fragment) {
253 print!("{}:{}: broken link fragment ",
254 pretty_file.display(),
256 println!("`#{}` pointing to `{}`", fragment, pretty_path.display());
261 print!("{}:{}: broken link - ", pretty_file.display(), i + 1);
262 let pretty_path = path.strip_prefix(root).unwrap_or(&path);
263 println!("{}", pretty_path.display());
269 fn load_file(cache: &mut Cache,
273 -> Result<(PathBuf, String), LoadError> {
274 let mut contents = String::new();
275 let pretty_file = PathBuf::from(file.strip_prefix(root).unwrap_or(&file));
277 let maybe_redirect = match cache.entry(pretty_file.clone()) {
278 Entry::Occupied(entry) => {
279 contents = entry.get().source.clone();
282 Entry::Vacant(entry) => {
283 let mut fp = File::open(file).map_err(|err| {
284 if let FromRedirect(true) = redirect {
285 LoadError::BrokenRedirect(file.to_path_buf(), err)
287 LoadError::IOError(err)
290 fp.read_to_string(&mut contents).map_err(|err| LoadError::IOError(err))?;
292 let maybe = maybe_redirect(&contents);
294 if let SkipRedirect = redirect {
295 return Err(LoadError::IsRedirect);
298 entry.insert(FileEntry {
299 source: contents.clone(),
306 match maybe_redirect.map(|url| file.parent().unwrap().join(url)) {
307 Some(redirect_file) => {
308 load_file(cache, root, &redirect_file, FromRedirect(true))
310 None => Ok((pretty_file, contents)),
314 fn maybe_redirect(source: &str) -> Option<String> {
315 const REDIRECT: &'static str = "<p>Redirecting to <a href=";
317 let mut lines = source.lines();
318 let redirect_line = match lines.nth(6) {
323 redirect_line.find(REDIRECT).map(|i| {
324 let rest = &redirect_line[(i + REDIRECT.len() + 1)..];
325 let pos_quote = rest.find('"').unwrap();
326 rest[..pos_quote].to_owned()
330 fn with_attrs_in_source<F: FnMut(&str, usize, &str)>(contents: &str, attr: &str, mut f: F) {
332 for (i, mut line) in contents.lines().enumerate() {
333 while let Some(j) = line.find(attr) {
334 let rest = &line[j + attr.len()..];
335 // The base tag should always be the first link in the document so
336 // we can get away with using one pass.
337 let is_base = line[..j].ends_with("<base");
339 let pos_equals = match rest.find("=") {
343 if rest[..pos_equals].trim_left_matches(" ") != "" {
347 let rest = &rest[pos_equals + 1..];
349 let pos_quote = match rest.find(&['"', '\''][..]) {
353 let quote_delim = rest.as_bytes()[pos_quote] as char;
355 if rest[..pos_quote].trim_left_matches(" ") != "" {
358 let rest = &rest[pos_quote + 1..];
359 let url = match rest.find(quote_delim) {
360 Some(i) => &rest[..i],