src/tools/tidy/src/deps.rs

   1 //! Checks the licenses of third-party dependencies by inspecting vendors.
   2
   3 use std::collections::{BTreeSet, HashSet, HashMap};
   4 use std::fs;
   5 use std::path::Path;
   6 use std::process::Command;
   7
   8 use serde::Deserialize;
   9 use serde_json;
  10
  11 const LICENSES: &[&str] = &[
  12     "MIT/Apache-2.0",
  13     "MIT / Apache-2.0",
  14     "Apache-2.0/MIT",
  15     "Apache-2.0 / MIT",
  16     "MIT OR Apache-2.0",
  17     "Apache-2.0 OR MIT",
  18     "Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT", // wasi license
  19     "MIT",
  20     "Unlicense/MIT",
  21     "Unlicense OR MIT",
  22 ];
  23
  24 /// These are exceptions to Rust's permissive licensing policy, and
  25 /// should be considered bugs. Exceptions are only allowed in Rust
  26 /// tooling. It is _crucial_ that no exception crates be dependencies
  27 /// of the Rust runtime (std/test).
  28 const EXCEPTIONS: &[&str] = &[
  29     "mdbook",             // MPL2, mdbook
  30     "openssl",            // BSD+advertising clause, cargo, mdbook
  31     "pest",               // MPL2, mdbook via handlebars
  32     "arrayref",           // BSD-2-Clause, mdbook via handlebars via pest
  33     "thread-id",          // Apache-2.0, mdbook
  34     "toml-query",         // MPL-2.0, mdbook
  35     "is-match",           // MPL-2.0, mdbook
  36     "cssparser",          // MPL-2.0, rustdoc
  37     "smallvec",           // MPL-2.0, rustdoc
  38     "rdrand",             // ISC, mdbook, rustfmt
  39     "fuchsia-cprng",      // BSD-3-Clause, mdbook, rustfmt
  40     "fuchsia-zircon-sys", // BSD-3-Clause, rustdoc, rustc, cargo
  41     "fuchsia-zircon",     // BSD-3-Clause, rustdoc, rustc, cargo (jobserver & tempdir)
  42     "cssparser-macros",   // MPL-2.0, rustdoc
  43     "selectors",          // MPL-2.0, rustdoc
  44     "clippy_lints",       // MPL-2.0, rls
  45     "colored",            // MPL-2.0, rustfmt
  46     "ordslice",           // Apache-2.0, rls
  47     "cloudabi",           // BSD-2-Clause, (rls -> crossbeam-channel 0.2 -> rand 0.5)
  48     "ryu",                // Apache-2.0, rls/cargo/... (because of serde)
  49     "bytesize",           // Apache-2.0, cargo
  50     "im-rc",              // MPL-2.0+, cargo
  51     "adler32",            // BSD-3-Clause AND Zlib, cargo dep that isn't used
  52     "constant_time_eq",   // CC0-1.0, rustfmt
  53     "utf8parse",          // Apache-2.0 OR MIT, cargo via strip-ansi-escapes
  54     "vte",                // Apache-2.0 OR MIT, cargo via strip-ansi-escapes
  55     "sized-chunks",       // MPL-2.0+, cargo via im-rc
  56     "bitmaps",            // MPL-2.0+, cargo via im-rc
  57     // FIXME: this dependency violates the documentation comment above:
  58     "fortanix-sgx-abi",   // MPL-2.0+, libstd but only for `sgx` target
  59     "dunce",              // CC0-1.0 mdbook-linkcheck
  60     "codespan-reporting", // Apache-2.0 mdbook-linkcheck
  61     "codespan",           // Apache-2.0 mdbook-linkcheck
  62 ];
  63
  64 /// Which crates to check against the whitelist?
  65 const WHITELIST_CRATES: &[CrateVersion<'_>] = &[
  66     CrateVersion("rustc", "0.0.0"),
  67     CrateVersion("rustc_codegen_llvm", "0.0.0"),
  68 ];
  69
  70 /// Whitelist of crates rustc is allowed to depend on. Avoid adding to the list if possible.
  71 const WHITELIST: &[Crate<'_>] = &[
  72     Crate("adler32"),
  73     Crate("aho-corasick"),
  74     Crate("annotate-snippets"),
  75     Crate("ansi_term"),
  76     Crate("arrayvec"),
  77     Crate("atty"),
  78     Crate("autocfg"),
  79     Crate("backtrace"),
  80     Crate("backtrace-sys"),
  81     Crate("bitflags"),
  82     Crate("build_const"),
  83     Crate("byteorder"),
  84     Crate("c2-chacha"),
  85     Crate("cc"),
  86     Crate("cfg-if"),
  87     Crate("chalk-engine"),
  88     Crate("chalk-macros"),
  89     Crate("cloudabi"),
  90     Crate("cmake"),
  91     Crate("compiler_builtins"),
  92     Crate("crc"),
  93     Crate("crc32fast"),
  94     Crate("crossbeam-deque"),
  95     Crate("crossbeam-epoch"),
  96     Crate("crossbeam-queue"),
  97     Crate("crossbeam-utils"),
  98     Crate("datafrog"),
  99     Crate("dlmalloc"),
 100     Crate("either"),
 101     Crate("ena"),
 102     Crate("env_logger"),
 103     Crate("filetime"),
 104     Crate("flate2"),
 105     Crate("fortanix-sgx-abi"),
 106     Crate("fuchsia-zircon"),
 107     Crate("fuchsia-zircon-sys"),
 108     Crate("getopts"),
 109     Crate("getrandom"),
 110     Crate("hashbrown"),
 111     Crate("humantime"),
 112     Crate("indexmap"),
 113     Crate("itertools"),
 114     Crate("jobserver"),
 115     Crate("kernel32-sys"),
 116     Crate("lazy_static"),
 117     Crate("libc"),
 118     Crate("libz-sys"),
 119     Crate("lock_api"),
 120     Crate("log"),
 121     Crate("log_settings"),
 122     Crate("measureme"),
 123     Crate("memchr"),
 124     Crate("memmap"),
 125     Crate("memoffset"),
 126     Crate("miniz-sys"),
 127     Crate("miniz_oxide"),
 128     Crate("miniz_oxide_c_api"),
 129     Crate("nodrop"),
 130     Crate("num_cpus"),
 131     Crate("owning_ref"),
 132     Crate("parking_lot"),
 133     Crate("parking_lot_core"),
 134     Crate("pkg-config"),
 135     Crate("polonius-engine"),
 136     Crate("ppv-lite86"),
 137     Crate("proc-macro2"),
 138     Crate("punycode"),
 139     Crate("quick-error"),
 140     Crate("quote"),
 141     Crate("rand"),
 142     Crate("rand_chacha"),
 143     Crate("rand_core"),
 144     Crate("rand_hc"),
 145     Crate("rand_isaac"),
 146     Crate("rand_pcg"),
 147     Crate("rand_xorshift"),
 148     Crate("redox_syscall"),
 149     Crate("redox_termios"),
 150     Crate("regex"),
 151     Crate("regex-syntax"),
 152     Crate("remove_dir_all"),
 153     Crate("rustc-demangle"),
 154     Crate("rustc-hash"),
 155     Crate("rustc-rayon"),
 156     Crate("rustc-rayon-core"),
 157     Crate("rustc_version"),
 158     Crate("scoped-tls"),
 159     Crate("scopeguard"),
 160     Crate("semver"),
 161     Crate("semver-parser"),
 162     Crate("serde"),
 163     Crate("serde_derive"),
 164     Crate("smallvec"),
 165     Crate("stable_deref_trait"),
 166     Crate("syn"),
 167     Crate("synstructure"),
 168     Crate("tempfile"),
 169     Crate("termcolor"),
 170     Crate("terminon"),
 171     Crate("termion"),
 172     Crate("term_size"),
 173     Crate("thread_local"),
 174     Crate("ucd-util"),
 175     Crate("unicode-width"),
 176     Crate("unicode-xid"),
 177     Crate("unreachable"),
 178     Crate("utf8-ranges"),
 179     Crate("vcpkg"),
 180     Crate("version_check"),
 181     Crate("void"),
 182     Crate("wasi"),
 183     Crate("winapi"),
 184     Crate("winapi-build"),
 185     Crate("winapi-i686-pc-windows-gnu"),
 186     Crate("winapi-util"),
 187     Crate("winapi-x86_64-pc-windows-gnu"),
 188     Crate("wincolor"),
 189     Crate("hermit-abi"),
 190 ];
 191
 192 // Some types for Serde to deserialize the output of `cargo metadata` to.
 193
 194 #[derive(Deserialize)]
 195 struct Output {
 196     resolve: Resolve,
 197 }
 198
 199 #[derive(Deserialize)]
 200 struct Resolve {
 201     nodes: Vec<ResolveNode>,
 202 }
 203
 204 #[derive(Deserialize)]
 205 struct ResolveNode {
 206     id: String,
 207     dependencies: Vec<String>,
 208 }
 209
 210 /// A unique identifier for a crate.
 211 #[derive(Copy, Clone, PartialOrd, Ord, PartialEq, Eq, Debug, Hash)]
 212 struct Crate<'a>(&'a str); // (name)
 213
 214 #[derive(Copy, Clone, PartialOrd, Ord, PartialEq, Eq, Debug, Hash)]
 215 struct CrateVersion<'a>(&'a str, &'a str); // (name, version)
 216
 217 impl Crate<'_> {
 218     pub fn id_str(&self) -> String {
 219         format!("{} ", self.0)
 220     }
 221 }
 222
 223 impl<'a> CrateVersion<'a> {
 224     /// Returns the struct and whether or not the dependency is in-tree.
 225     pub fn from_str(s: &'a str) -> (Self, bool) {
 226         let mut parts = s.split(' ');
 227         let name = parts.next().unwrap();
 228         let version = parts.next().unwrap();
 229         let path = parts.next().unwrap();
 230
 231         let is_path_dep = path.starts_with("(path+");
 232
 233         (CrateVersion(name, version), is_path_dep)
 234     }
 235
 236     pub fn id_str(&self) -> String {
 237         format!("{} {}", self.0, self.1)
 238     }
 239 }
 240
 241 impl<'a> From<CrateVersion<'a>> for Crate<'a> {
 242     fn from(cv: CrateVersion<'a>) -> Crate<'a> {
 243         Crate(cv.0)
 244     }
 245 }
 246
 247 /// Checks the dependency at the given path. Changes `bad` to `true` if a check failed.
 248 ///
 249 /// Specifically, this checks that the license is correct.
 250 pub fn check(path: &Path, bad: &mut bool) {
 251     // Check licences.
 252     let path = path.join("../vendor");
 253     assert!(path.exists(), "vendor directory missing");
 254     let mut saw_dir = false;
 255     for dir in t!(path.read_dir()) {
 256         saw_dir = true;
 257         let dir = t!(dir);
 258
 259         // Skip our exceptions.
 260         let is_exception = EXCEPTIONS.iter().any(|exception| {
 261             dir.path()
 262                 .to_str()
 263                 .unwrap()
 264                 .contains(&format!("vendor/{}", exception))
 265         });
 266         if is_exception {
 267             continue;
 268         }
 269
 270         let toml = dir.path().join("Cargo.toml");
 271         *bad = !check_license(&toml) || *bad;
 272     }
 273     assert!(saw_dir, "no vendored source");
 274 }
 275
 276 /// Checks the dependency of `WHITELIST_CRATES` at the given path. Changes `bad` to `true` if a
 277 /// check failed.
 278 ///
 279 /// Specifically, this checks that the dependencies are on the `WHITELIST`.
 280 pub fn check_whitelist(path: &Path, cargo: &Path, bad: &mut bool) {
 281     // Get dependencies from Cargo metadata.
 282     let resolve = get_deps(path, cargo);
 283
 284     // Get the whitelist in a convenient form.
 285     let whitelist: HashSet<_> = WHITELIST.iter().cloned().collect();
 286
 287     // Check dependencies.
 288     let mut visited = BTreeSet::new();
 289     let mut unapproved = BTreeSet::new();
 290     for &krate in WHITELIST_CRATES.iter() {
 291         let mut bad = check_crate_whitelist(&whitelist, &resolve, &mut visited, krate, false);
 292         unapproved.append(&mut bad);
 293     }
 294
 295     if !unapproved.is_empty() {
 296         println!("Dependencies not on the whitelist:");
 297         for dep in unapproved {
 298             println!("* {}", dep.id_str());
 299         }
 300         *bad = true;
 301     }
 302
 303     check_crate_duplicate(&resolve, bad);
 304 }
 305
 306 fn check_license(path: &Path) -> bool {
 307     if !path.exists() {
 308         panic!("{} does not exist", path.display());
 309     }
 310     let contents = t!(fs::read_to_string(&path));
 311
 312     let mut found_license = false;
 313     for line in contents.lines() {
 314         if !line.starts_with("license") {
 315             continue;
 316         }
 317         let license = extract_license(line);
 318         if !LICENSES.contains(&&*license) {
 319             println!("invalid license {} in {}", license, path.display());
 320             return false;
 321         }
 322         found_license = true;
 323         break;
 324     }
 325     if !found_license {
 326         println!("no license in {}", path.display());
 327         return false;
 328     }
 329
 330     true
 331 }
 332
 333 fn extract_license(line: &str) -> String {
 334     let first_quote = line.find('"');
 335     let last_quote = line.rfind('"');
 336     if let (Some(f), Some(l)) = (first_quote, last_quote) {
 337         let license = &line[f + 1..l];
 338         license.into()
 339     } else {
 340         "bad-license-parse".into()
 341     }
 342 }
 343
 344 /// Gets the dependencies of the crate at the given path using `cargo metadata`.
 345 fn get_deps(path: &Path, cargo: &Path) -> Resolve {
 346     // Run `cargo metadata` to get the set of dependencies.
 347     let output = Command::new(cargo)
 348         .arg("metadata")
 349         .arg("--format-version")
 350         .arg("1")
 351         .arg("--manifest-path")
 352         .arg(path.join("../Cargo.toml"))
 353         .output()
 354         .expect("Unable to run `cargo metadata`")
 355         .stdout;
 356     let output = String::from_utf8_lossy(&output);
 357     let output: Output = serde_json::from_str(&output).unwrap();
 358
 359     output.resolve
 360 }
 361
 362 /// Checks the dependencies of the given crate from the given cargo metadata to see if they are on
 363 /// the whitelist. Returns a list of illegal dependencies.
 364 fn check_crate_whitelist<'a>(
 365     whitelist: &'a HashSet<Crate<'_>>,
 366     resolve: &'a Resolve,
 367     visited: &mut BTreeSet<CrateVersion<'a>>,
 368     krate: CrateVersion<'a>,
 369     must_be_on_whitelist: bool,
 370 ) -> BTreeSet<Crate<'a>> {
 371     // This will contain bad deps.
 372     let mut unapproved = BTreeSet::new();
 373
 374     // Check if we have already visited this crate.
 375     if visited.contains(&krate) {
 376         return unapproved;
 377     }
 378
 379     visited.insert(krate);
 380
 381     // If this path is in-tree, we don't require it to be on the whitelist.
 382     if must_be_on_whitelist {
 383         // If this dependency is not on `WHITELIST`, add to bad set.
 384         if !whitelist.contains(&krate.into()) {
 385             unapproved.insert(krate.into());
 386         }
 387     }
 388
 389     // Do a DFS in the crate graph (it's a DAG, so we know we have no cycles!).
 390     let to_check = resolve
 391         .nodes
 392         .iter()
 393         .find(|n| n.id.starts_with(&krate.id_str()))
 394         .expect("crate does not exist");
 395
 396     for dep in to_check.dependencies.iter() {
 397         let (krate, is_path_dep) = CrateVersion::from_str(dep);
 398
 399         let mut bad = check_crate_whitelist(whitelist, resolve, visited, krate, !is_path_dep);
 400         unapproved.append(&mut bad);
 401     }
 402
 403     unapproved
 404 }
 405
 406 fn check_crate_duplicate(resolve: &Resolve, bad: &mut bool) {
 407     const FORBIDDEN_TO_HAVE_DUPLICATES: &[&str] = &[
 408         // These two crates take quite a long time to build, so don't allow two versions of them
 409         // to accidentally sneak into our dependency graph, in order to ensure we keep our CI times
 410         // under control.
 411
 412         "cargo",
 413         "rustc-ap-syntax",
 414     ];
 415     let mut name_to_id: HashMap<_, Vec<_>> = HashMap::new();
 416     for node in resolve.nodes.iter() {
 417         name_to_id.entry(node.id.split_whitespace().next().unwrap())
 418             .or_default()
 419             .push(&node.id);
 420     }
 421
 422     for name in FORBIDDEN_TO_HAVE_DUPLICATES {
 423         if name_to_id[name].len() <= 1 {
 424             continue
 425         }
 426         println!("crate `{}` is duplicated in `Cargo.lock`", name);
 427         for id in name_to_id[name].iter() {
 428             println!("  * {}", id);
 429         }
 430         *bad = true;
 431     }
 432 }