src/tools/tidy/src/style.rs

   1 //! Tidy check to enforce various stylistic guidelines on the Rust codebase.
   2 //!
   3 //! Example checks are:
   4 //!
   5 //! * No lines over 100 characters (in non-Rust files).
   6 //! * No files with over 3000 lines (in non-Rust files).
   7 //! * No tabs.
   8 //! * No trailing whitespace.
   9 //! * No CR characters.
  10 //! * No `TODO` or `XXX` directives.
  11 //! * No unexplained ` ```ignore ` or ` ```rust,ignore ` doc tests.
  12 //!
  13 //! Note that some of these rules are excluded from Rust files because we enforce rustfmt. It is
  14 //! preferable to be formatted rather than tidy-clean.
  15 //!
  16 //! A number of these checks can be opted-out of with various directives of the form:
  17 //! `// ignore-tidy-CHECK-NAME`.
  18
  19 use regex::Regex;
  20 use std::path::Path;
  21
  22 /// Error code markdown is restricted to 80 columns because they can be
  23 /// displayed on the console with --example.
  24 const ERROR_CODE_COLS: usize = 80;
  25 const COLS: usize = 100;
  26
  27 const LINES: usize = 3000;
  28
  29 const UNEXPLAINED_IGNORE_DOCTEST_INFO: &str = r#"unexplained "```ignore" doctest; try one:
  30
  31 * make the test actually pass, by adding necessary imports and declarations, or
  32 * use "```text", if the code is not Rust code, or
  33 * use "```compile_fail,Ennnn", if the code is expected to fail at compile time, or
  34 * use "```should_panic", if the code is expected to fail at run time, or
  35 * use "```no_run", if the code should type-check but not necessary linkable/runnable, or
  36 * explain it like "```ignore (cannot-test-this-because-xxxx)", if the annotation cannot be avoided.
  37
  38 "#;
  39
  40 const LLVM_UNREACHABLE_INFO: &str = r"\
  41 C++ code used llvm_unreachable, which triggers undefined behavior
  42 when executed when assertions are disabled.
  43 Use llvm::report_fatal_error for increased robustness.";
  44
  45 const ANNOTATIONS_TO_IGNORE: &[&str] = &[
  46     "// @!has",
  47     "// @has",
  48     "// @matches",
  49     "// CHECK",
  50     "// EMIT_MIR",
  51     "// compile-flags",
  52     "// error-pattern",
  53     "// gdb",
  54     "// lldb",
  55     "// cdb",
  56     "// normalize-stderr-test",
  57 ];
  58
  59 /// Parser states for `line_is_url`.
  60 #[derive(Clone, Copy, PartialEq)]
  61 #[allow(non_camel_case_types)]
  62 enum LIUState {
  63     EXP_COMMENT_START,
  64     EXP_LINK_LABEL_OR_URL,
  65     EXP_URL,
  66     EXP_END,
  67 }
  68
  69 /// Returns `true` if `line` appears to be a line comment containing a URL,
  70 /// possibly with a Markdown link label in front, and nothing else.
  71 /// The Markdown link label, if present, may not contain whitespace.
  72 /// Lines of this form are allowed to be overlength, because Markdown
  73 /// offers no way to split a line in the middle of a URL, and the lengths
  74 /// of URLs to external references are beyond our control.
  75 fn line_is_url(is_error_code: bool, columns: usize, line: &str) -> bool {
  76     // more basic check for markdown, to avoid complexity in implementing two state machines
  77     if is_error_code {
  78         return line.starts_with('[') && line.contains("]:") && line.contains("http");
  79     }
  80
  81     use self::LIUState::*;
  82     let mut state: LIUState = EXP_COMMENT_START;
  83     let is_url = |w: &str| w.starts_with("http://") || w.starts_with("https://");
  84
  85     for tok in line.split_whitespace() {
  86         match (state, tok) {
  87             (EXP_COMMENT_START, "//") | (EXP_COMMENT_START, "///") | (EXP_COMMENT_START, "//!") => {
  88                 state = EXP_LINK_LABEL_OR_URL
  89             }
  90
  91             (EXP_LINK_LABEL_OR_URL, w)
  92                 if w.len() >= 4 && w.starts_with('[') && w.ends_with("]:") =>
  93             {
  94                 state = EXP_URL
  95             }
  96
  97             (EXP_LINK_LABEL_OR_URL, w) if is_url(w) => state = EXP_END,
  98
  99             (EXP_URL, w) if is_url(w) || w.starts_with("../") => state = EXP_END,
 100
 101             (_, w) if w.len() > columns && is_url(w) => state = EXP_END,
 102
 103             (_, _) => {}
 104         }
 105     }
 106
 107     state == EXP_END
 108 }
 109
 110 /// Returns `true` if `line` can be ignored. This is the case when it contains
 111 /// an annotation that is explicitly ignored.
 112 fn should_ignore(line: &str) -> bool {
 113     // Matches test annotations like `//~ ERROR text`.
 114     // This mirrors the regex in src/tools/compiletest/src/runtest.rs, please
 115     // update both if either are changed.
 116     let re = Regex::new("\\s*//(\\[.*\\])?~.*").unwrap();
 117     re.is_match(line) || ANNOTATIONS_TO_IGNORE.iter().any(|a| line.contains(a))
 118 }
 119
 120 /// Returns `true` if `line` is allowed to be longer than the normal limit.
 121 fn long_line_is_ok(extension: &str, is_error_code: bool, max_columns: usize, line: &str) -> bool {
 122     if extension != "md" || is_error_code {
 123         if line_is_url(is_error_code, max_columns, line) || should_ignore(line) {
 124             return true;
 125         }
 126     } else if extension == "md" {
 127         // non-error code markdown is allowed to be any length
 128         return true;
 129     }
 130
 131     false
 132 }
 133
 134 enum Directive {
 135     /// By default, tidy always warns against style issues.
 136     Deny,
 137
 138     /// `Ignore(false)` means that an `ignore-tidy-*` directive
 139     /// has been provided, but is unnecessary. `Ignore(true)`
 140     /// means that it is necessary (i.e. a warning would be
 141     /// produced if `ignore-tidy-*` was not present).
 142     Ignore(bool),
 143 }
 144
 145 fn contains_ignore_directive(can_contain: bool, contents: &str, check: &str) -> Directive {
 146     if !can_contain {
 147         return Directive::Deny;
 148     }
 149     // Update `can_contain` when changing this
 150     if contents.contains(&format!("// ignore-tidy-{}", check))
 151         || contents.contains(&format!("# ignore-tidy-{}", check))
 152         || contents.contains(&format!("/* ignore-tidy-{} */", check))
 153     {
 154         Directive::Ignore(false)
 155     } else {
 156         Directive::Deny
 157     }
 158 }
 159
 160 macro_rules! suppressible_tidy_err {
 161     ($err:ident, $skip:ident, $msg:expr) => {
 162         if let Directive::Deny = $skip {
 163             $err($msg);
 164         } else {
 165             $skip = Directive::Ignore(true);
 166         }
 167     };
 168 }
 169
 170 pub fn is_in(full_path: &Path, parent_folder_to_find: &str, folder_to_find: &str) -> bool {
 171     if let Some(parent) = full_path.parent() {
 172         if parent.file_name().map_or_else(
 173             || false,
 174             |f| {
 175                 f.to_string_lossy() == folder_to_find
 176                     && parent
 177                         .parent()
 178                         .and_then(|f| f.file_name())
 179                         .map_or_else(|| false, |f| f == parent_folder_to_find)
 180             },
 181         ) {
 182             true
 183         } else {
 184             is_in(parent, parent_folder_to_find, folder_to_find)
 185         }
 186     } else {
 187         false
 188     }
 189 }
 190
 191 fn skip_markdown_path(path: &Path) -> bool {
 192     // These aren't ready for tidy.
 193     const SKIP_MD: &[&str] = &[
 194         "src/doc/edition-guide",
 195         "src/doc/embedded-book",
 196         "src/doc/nomicon",
 197         "src/doc/reference",
 198         "src/doc/rust-by-example",
 199         "src/doc/rustc-dev-guide",
 200     ];
 201     SKIP_MD.iter().any(|p| path.ends_with(p))
 202 }
 203
 204 fn is_unexplained_ignore(extension: &str, line: &str) -> bool {
 205     if !line.ends_with("```ignore") && !line.ends_with("```rust,ignore") {
 206         return false;
 207     }
 208     if extension == "md" && line.trim().starts_with("//") {
 209         // Markdown examples may include doc comments with ignore inside a
 210         // code block.
 211         return false;
 212     }
 213     true
 214 }
 215
 216 pub fn check(path: &Path, bad: &mut bool) {
 217     fn skip(path: &Path) -> bool {
 218         super::filter_dirs(path) || skip_markdown_path(path)
 219     }
 220     super::walk(path, &mut skip, &mut |entry, contents| {
 221         let file = entry.path();
 222         let filename = file.file_name().unwrap().to_string_lossy();
 223         let extensions = [".rs", ".py", ".js", ".sh", ".c", ".cpp", ".h", ".md", ".css"];
 224         if extensions.iter().all(|e| !filename.ends_with(e)) || filename.starts_with(".#") {
 225             return;
 226         }
 227
 228         let is_style_file = filename.ends_with(".css");
 229         let under_rustfmt = filename.ends_with(".rs") &&
 230             // This list should ideally be sourced from rustfmt.toml but we don't want to add a toml
 231             // parser to tidy.
 232             !file.ancestors().any(|a| {
 233                 a.ends_with("src/test") ||
 234                     a.ends_with("src/doc/book")
 235             });
 236
 237         if is_style_file && !is_in(file, "src", "librustdoc") {
 238             // We only check CSS files in rustdoc.
 239             return;
 240         }
 241
 242         if contents.is_empty() {
 243             tidy_error!(bad, "{}: empty file", file.display());
 244         }
 245
 246         let extension = file.extension().unwrap().to_string_lossy();
 247         let is_error_code = extension == "md" && is_in(file, "src", "error_codes");
 248
 249         let max_columns = if is_error_code { ERROR_CODE_COLS } else { COLS };
 250
 251         let can_contain = contents.contains("// ignore-tidy-")
 252             || contents.contains("# ignore-tidy-")
 253             || contents.contains("/* ignore-tidy-");
 254         // Enable testing ICE's that require specific (untidy)
 255         // file formats easily eg. `issue-1234-ignore-tidy.rs`
 256         if filename.contains("ignore-tidy") {
 257             return;
 258         }
 259         let mut skip_cr = contains_ignore_directive(can_contain, &contents, "cr");
 260         let mut skip_undocumented_unsafe =
 261             contains_ignore_directive(can_contain, &contents, "undocumented-unsafe");
 262         let mut skip_tab = contains_ignore_directive(can_contain, &contents, "tab");
 263         let mut skip_line_length = contains_ignore_directive(can_contain, &contents, "linelength");
 264         let mut skip_file_length = contains_ignore_directive(can_contain, &contents, "filelength");
 265         let mut skip_end_whitespace =
 266             contains_ignore_directive(can_contain, &contents, "end-whitespace");
 267         let mut skip_trailing_newlines =
 268             contains_ignore_directive(can_contain, &contents, "trailing-newlines");
 269         let mut skip_copyright = contains_ignore_directive(can_contain, &contents, "copyright");
 270         let mut leading_new_lines = false;
 271         let mut trailing_new_lines = 0;
 272         let mut lines = 0;
 273         let mut last_safety_comment = false;
 274         for (i, line) in contents.split('\n').enumerate() {
 275             let mut err = |msg: &str| {
 276                 tidy_error!(bad, "{}:{}: {}", file.display(), i + 1, msg);
 277             };
 278             if !under_rustfmt
 279                 && line.chars().count() > max_columns
 280                 && !long_line_is_ok(&extension, is_error_code, max_columns, line)
 281             {
 282                 suppressible_tidy_err!(
 283                     err,
 284                     skip_line_length,
 285                     &format!("line longer than {} chars", max_columns)
 286                 );
 287             }
 288             if !is_style_file && line.contains('\t') {
 289                 suppressible_tidy_err!(err, skip_tab, "tab character");
 290             }
 291             if line.ends_with(' ') || line.ends_with('\t') {
 292                 suppressible_tidy_err!(err, skip_end_whitespace, "trailing whitespace");
 293             }
 294             if is_style_file && line.starts_with(' ') {
 295                 err("CSS files use tabs for indent");
 296             }
 297             if line.contains('\r') {
 298                 suppressible_tidy_err!(err, skip_cr, "CR character");
 299             }
 300             if filename != "style.rs" {
 301                 if line.contains("TODO") {
 302                     err("TODO is deprecated; use FIXME")
 303                 }
 304                 if line.contains("//") && line.contains(" XXX") {
 305                     err("XXX is deprecated; use FIXME")
 306                 }
 307             }
 308             let is_test = || file.components().any(|c| c.as_os_str() == "tests");
 309             // for now we just check libcore
 310             if line.contains("unsafe {") && !line.trim().starts_with("//") && !last_safety_comment {
 311                 if file.components().any(|c| c.as_os_str() == "core") && !is_test() {
 312                     suppressible_tidy_err!(err, skip_undocumented_unsafe, "undocumented unsafe");
 313                 }
 314             }
 315             if line.contains("// SAFETY:") {
 316                 last_safety_comment = true;
 317             } else if line.trim().starts_with("//") || line.trim().is_empty() {
 318                 // keep previous value
 319             } else {
 320                 last_safety_comment = false;
 321             }
 322             if (line.starts_with("// Copyright")
 323                 || line.starts_with("# Copyright")
 324                 || line.starts_with("Copyright"))
 325                 && (line.contains("Rust Developers") || line.contains("Rust Project Developers"))
 326             {
 327                 suppressible_tidy_err!(
 328                     err,
 329                     skip_copyright,
 330                     "copyright notices attributed to the Rust Project Developers are deprecated"
 331                 );
 332             }
 333             if is_unexplained_ignore(&extension, line) {
 334                 err(UNEXPLAINED_IGNORE_DOCTEST_INFO);
 335             }
 336             if filename.ends_with(".cpp") && line.contains("llvm_unreachable") {
 337                 err(LLVM_UNREACHABLE_INFO);
 338             }
 339             if line.is_empty() {
 340                 if i == 0 {
 341                     leading_new_lines = true;
 342                 }
 343                 trailing_new_lines += 1;
 344             } else {
 345                 trailing_new_lines = 0;
 346             }
 347
 348             if !line.trim().starts_with("//") {
 349                 lines += 1;
 350             }
 351         }
 352         if leading_new_lines {
 353             tidy_error!(bad, "{}: leading newline", file.display());
 354         }
 355         let mut err = |msg: &str| {
 356             tidy_error!(bad, "{}: {}", file.display(), msg);
 357         };
 358         match trailing_new_lines {
 359             0 => suppressible_tidy_err!(err, skip_trailing_newlines, "missing trailing newline"),
 360             1 => {}
 361             n => suppressible_tidy_err!(
 362                 err,
 363                 skip_trailing_newlines,
 364                 &format!("too many trailing newlines ({})", n)
 365             ),
 366         };
 367         if lines > LINES {
 368             let mut err = |_| {
 369                 tidy_error!(
 370                     bad,
 371                     "{}: too many lines ({}) (add `// \
 372                      ignore-tidy-filelength` to the file to suppress this error)",
 373                     file.display(),
 374                     lines
 375                 );
 376             };
 377             suppressible_tidy_err!(err, skip_file_length, "");
 378         }
 379
 380         if let Directive::Ignore(false) = skip_cr {
 381             tidy_error!(bad, "{}: ignoring CR characters unnecessarily", file.display());
 382         }
 383         if let Directive::Ignore(false) = skip_tab {
 384             tidy_error!(bad, "{}: ignoring tab characters unnecessarily", file.display());
 385         }
 386         if let Directive::Ignore(false) = skip_line_length {
 387             tidy_error!(bad, "{}: ignoring line length unnecessarily", file.display());
 388         }
 389         if let Directive::Ignore(false) = skip_file_length {
 390             tidy_error!(bad, "{}: ignoring file length unnecessarily", file.display());
 391         }
 392         if let Directive::Ignore(false) = skip_end_whitespace {
 393             tidy_error!(bad, "{}: ignoring trailing whitespace unnecessarily", file.display());
 394         }
 395         if let Directive::Ignore(false) = skip_trailing_newlines {
 396             tidy_error!(bad, "{}: ignoring trailing newlines unnecessarily", file.display());
 397         }
 398         if let Directive::Ignore(false) = skip_copyright {
 399             tidy_error!(bad, "{}: ignoring copyright unnecessarily", file.display());
 400         }
 401     })
 402 }