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