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 //! Check license of third-party deps by inspecting src/vendor
13 use std::collections::{BTreeSet, HashSet, HashMap};
17 use std::process::Command;
21 static LICENSES: &'static [&'static str] = &[
32 /// These are exceptions to Rust's permissive licensing policy, and
33 /// should be considered bugs. Exceptions are only allowed in Rust
34 /// tooling. It is _crucial_ that no exception crates be dependencies
35 /// of the Rust runtime (std / test).
36 static EXCEPTIONS: &'static [&'static str] = &[
37 "mdbook", // MPL2, mdbook
38 "openssl", // BSD+advertising clause, cargo, mdbook
39 "pest", // MPL2, mdbook via handlebars
40 "thread-id", // Apache-2.0, mdbook
41 "toml-query", // MPL-2.0, mdbook
42 "is-match", // MPL-2.0, mdbook
43 "cssparser", // MPL-2.0, rustdoc
44 "smallvec", // MPL-2.0, rustdoc
45 "fuchsia-zircon-sys", // BSD-3-Clause, rustdoc, rustc, cargo
46 "fuchsia-zircon", // BSD-3-Clause, rustdoc, rustc, cargo (jobserver & tempdir)
47 "cssparser-macros", // MPL-2.0, rustdoc
48 "selectors", // MPL-2.0, rustdoc
49 "clippy_lints", // MPL-2.0, rls
50 "colored", // MPL-2.0, rustfmt
51 "ordslice", // Apache-2.0, rls
52 "cloudabi", // BSD-2-Clause, (rls -> crossbeam-channel 0.2 -> rand 0.5)
53 "ryu", // Apache-2.0, rls/cargo/... (b/c of serde)
56 /// Which crates to check against the whitelist?
57 static WHITELIST_CRATES: &'static [CrateVersion] = &[
58 CrateVersion("rustc", "0.0.0"),
59 CrateVersion("rustc_codegen_llvm", "0.0.0"),
62 /// Whitelist of crates rustc is allowed to depend on. Avoid adding to the list if possible.
63 static WHITELIST: &'static [Crate] = &[
64 Crate("aho-corasick"),
68 Crate("backtrace-sys"),
73 Crate("chalk-engine"),
74 Crate("chalk-macros"),
77 Crate("crossbeam-deque"),
78 Crate("crossbeam-epoch"),
79 Crate("crossbeam-utils"),
86 Crate("fuchsia-zircon"),
87 Crate("fuchsia-zircon-sys"),
91 Crate("kernel32-sys"),
95 Crate("log_settings"),
103 Crate("parking_lot"),
104 Crate("parking_lot_core"),
105 Crate("polonius-engine"),
107 Crate("quick-error"),
110 Crate("redox_syscall"),
111 Crate("redox_termios"),
113 Crate("regex-syntax"),
114 Crate("remove_dir_all"),
115 Crate("rustc-demangle"),
117 Crate("rustc-rayon"),
118 Crate("rustc-rayon-core"),
122 Crate("stable_deref_trait"),
127 Crate("thread_local"),
129 Crate("unicode-width"),
130 Crate("unreachable"),
131 Crate("utf8-ranges"),
132 Crate("version_check"),
135 Crate("winapi-build"),
136 Crate("winapi-i686-pc-windows-gnu"),
137 Crate("winapi-x86_64-pc-windows-gnu"),
141 // Some types for Serde to deserialize the output of `cargo metadata` to...
143 #[derive(Deserialize)]
148 #[derive(Deserialize)]
150 nodes: Vec<ResolveNode>,
153 #[derive(Deserialize)]
156 dependencies: Vec<String>,
159 /// A unique identifier for a crate
160 #[derive(Copy, Clone, PartialOrd, Ord, PartialEq, Eq, Debug, Hash)]
161 struct Crate<'a>(&'a str); // (name,)
163 #[derive(Copy, Clone, PartialOrd, Ord, PartialEq, Eq, Debug, Hash)]
164 struct CrateVersion<'a>(&'a str, &'a str); // (name, version)
167 pub fn id_str(&self) -> String {
168 format!("{} ", self.0)
172 impl<'a> CrateVersion<'a> {
173 /// Returns the struct and whether or not the dep is in-tree
174 pub fn from_str(s: &'a str) -> (Self, bool) {
175 let mut parts = s.split(' ');
176 let name = parts.next().unwrap();
177 let version = parts.next().unwrap();
178 let path = parts.next().unwrap();
180 let is_path_dep = path.starts_with("(path+");
182 (CrateVersion(name, version), is_path_dep)
185 pub fn id_str(&self) -> String {
186 format!("{} {}", self.0, self.1)
190 impl<'a> From<CrateVersion<'a>> for Crate<'a> {
191 fn from(cv: CrateVersion<'a>) -> Crate<'a> {
196 /// Checks the dependency at the given path. Changes `bad` to `true` if a check failed.
198 /// Specifically, this checks that the license is correct.
199 pub fn check(path: &Path, bad: &mut bool) {
201 let path = path.join("vendor");
202 assert!(path.exists(), "vendor directory missing");
203 let mut saw_dir = false;
204 for dir in t!(path.read_dir()) {
208 // skip our exceptions
209 if EXCEPTIONS.iter().any(|exception| {
213 .contains(&format!("src/vendor/{}", exception))
218 let toml = dir.path().join("Cargo.toml");
219 *bad = *bad || !check_license(&toml);
221 assert!(saw_dir, "no vendored source");
224 /// Checks the dependency of WHITELIST_CRATES at the given path. Changes `bad` to `true` if a check
227 /// Specifically, this checks that the dependencies are on the WHITELIST.
228 pub fn check_whitelist(path: &Path, cargo: &Path, bad: &mut bool) {
229 // Get dependencies from cargo metadata
230 let resolve = get_deps(path, cargo);
232 // Get the whitelist into a convenient form
233 let whitelist: HashSet<_> = WHITELIST.iter().cloned().collect();
235 // Check dependencies
236 let mut visited = BTreeSet::new();
237 let mut unapproved = BTreeSet::new();
238 for &krate in WHITELIST_CRATES.iter() {
239 let mut bad = check_crate_whitelist(&whitelist, &resolve, &mut visited, krate, false);
240 unapproved.append(&mut bad);
243 if unapproved.len() > 0 {
244 println!("Dependencies not on the whitelist:");
245 for dep in unapproved {
246 println!("* {}", dep.id_str());
251 check_crate_duplicate(&resolve, bad);
254 fn check_license(path: &Path) -> bool {
256 panic!("{} does not exist", path.display());
258 let mut contents = String::new();
259 t!(t!(File::open(path)).read_to_string(&mut contents));
261 let mut found_license = false;
262 for line in contents.lines() {
263 if !line.starts_with("license") {
266 let license = extract_license(line);
267 if !LICENSES.contains(&&*license) {
268 println!("invalid license {} in {}", license, path.display());
271 found_license = true;
275 println!("no license in {}", path.display());
282 fn extract_license(line: &str) -> String {
283 let first_quote = line.find('"');
284 let last_quote = line.rfind('"');
285 if let (Some(f), Some(l)) = (first_quote, last_quote) {
286 let license = &line[f + 1..l];
289 "bad-license-parse".into()
293 /// Get the dependencies of the crate at the given path using `cargo metadata`.
294 fn get_deps(path: &Path, cargo: &Path) -> Resolve {
295 // Run `cargo metadata` to get the set of dependencies
296 let output = Command::new(cargo)
298 .arg("--format-version")
300 .arg("--manifest-path")
301 .arg(path.join("Cargo.toml"))
303 .expect("Unable to run `cargo metadata`")
305 let output = String::from_utf8_lossy(&output);
306 let output: Output = serde_json::from_str(&output).unwrap();
311 /// Checks the dependencies of the given crate from the given cargo metadata to see if they are on
312 /// the whitelist. Returns a list of illegal dependencies.
313 fn check_crate_whitelist<'a, 'b>(
314 whitelist: &'a HashSet<Crate>,
315 resolve: &'a Resolve,
316 visited: &'b mut BTreeSet<CrateVersion<'a>>,
317 krate: CrateVersion<'a>,
318 must_be_on_whitelist: bool,
319 ) -> BTreeSet<Crate<'a>> {
320 // Will contain bad deps
321 let mut unapproved = BTreeSet::new();
323 // Check if we have already visited this crate
324 if visited.contains(&krate) {
328 visited.insert(krate);
330 // If this path is in-tree, we don't require it to be on the whitelist
331 if must_be_on_whitelist {
332 // If this dependency is not on the WHITELIST, add to bad set
333 if !whitelist.contains(&krate.into()) {
334 unapproved.insert(krate.into());
338 // Do a DFS in the crate graph (it's a DAG, so we know we have no cycles!)
339 let to_check = resolve
342 .find(|n| n.id.starts_with(&krate.id_str()))
343 .expect("crate does not exist");
345 for dep in to_check.dependencies.iter() {
346 let (krate, is_path_dep) = CrateVersion::from_str(dep);
348 let mut bad = check_crate_whitelist(whitelist, resolve, visited, krate, !is_path_dep);
349 unapproved.append(&mut bad);
355 fn check_crate_duplicate(resolve: &Resolve, bad: &mut bool) {
356 const FORBIDDEN_TO_HAVE_DUPLICATES: &[&str] = &[
357 // These two crates take quite a long time to build, let's not let two
358 // versions of them accidentally sneak into our dependency graph to
359 // ensure we keep our CI times under control
360 // "cargo", // FIXME(#53005)
363 let mut name_to_id: HashMap<_, Vec<_>> = HashMap::new();
364 for node in resolve.nodes.iter() {
365 name_to_id.entry(node.id.split_whitespace().next().unwrap())
370 for name in FORBIDDEN_TO_HAVE_DUPLICATES {
371 if name_to_id[name].len() <= 1 {
374 println!("crate `{}` is duplicated in `Cargo.lock`", name);
375 for id in name_to_id[name].iter() {
376 println!(" * {}", id);