1use std::ffi::OsStr;
21use std::path::Path;
22use std::sync::LazyLock;
23
24use regex::RegexSetBuilder;
25use rustc_hash::FxHashMap;
26
27use crate::diagnostics::{CheckId, TidyCtx};
28use crate::walk::{filter_dirs, walk};
29
30#[cfg(test)]
31mod tests;
32
33const ERROR_CODE_COLS: usize = 80;
36const COLS: usize = 100;
37const GOML_COLS: usize = 120;
38
39const LINES: usize = 3000;
40
41const UNEXPLAINED_IGNORE_DOCTEST_INFO: &str = r#"unexplained "```ignore" doctest; try one:
42
43* make the test actually pass, by adding necessary imports and declarations, or
44* use "```text", if the code is not Rust code, or
45* use "```compile_fail,Ennnn", if the code is expected to fail at compile time, or
46* use "```should_panic", if the code is expected to fail at run time, or
47* use "```no_run", if the code should type-check but not necessary linkable/runnable, or
48* explain it like "```ignore (cannot-test-this-because-xxxx)", if the annotation cannot be avoided.
49
50"#;
51
52const LLVM_UNREACHABLE_INFO: &str = r"\
53C++ code used llvm_unreachable, which triggers undefined behavior
54when executed when assertions are disabled.
55Use llvm::report_fatal_error for increased robustness.";
56
57const DOUBLE_SPACE_AFTER_DOT: &str = r"\
58Use a single space after dots in comments.";
59
60const ANNOTATIONS_TO_IGNORE: &[&str] = &[
61 "// @!has",
62 "// @has",
63 "// @matches",
64 "// CHECK",
65 "// EMIT_MIR",
66 "// compile-flags",
67 "//@ compile-flags",
68 "// error-pattern",
69 "//@ error-pattern",
70 "//@ gdb",
71 "//@ lldb",
72 "//@ cdb",
73 "//@ normalize-stderr",
74];
75
76const LINELENGTH_CHECK: &str = "linelength";
77
78const CONFIGURABLE_CHECKS: [&str; 11] = [
80 "cr",
81 "undocumented-unsafe",
82 "tab",
83 LINELENGTH_CHECK,
84 "filelength",
85 "end-whitespace",
86 "trailing-newlines",
87 "leading-newlines",
88 "copyright",
89 "dbg",
90 "odd-backticks",
91];
92
93fn generate_problems<'a>(
94 consts: &'a [u32],
95 letter_digit: &'a FxHashMap<char, char>,
96) -> impl Iterator<Item = u32> + 'a {
97 consts.iter().flat_map(move |const_value| {
98 let problem = letter_digit.iter().fold(format!("{const_value:X}"), |acc, (key, value)| {
99 acc.replace(&value.to_string(), &key.to_string())
100 });
101 let indexes: Vec<usize> = problem
102 .chars()
103 .enumerate()
104 .filter_map(|(index, c)| if letter_digit.contains_key(&c) { Some(index) } else { None })
105 .collect();
106 (0..1 << indexes.len()).map(move |i| {
107 u32::from_str_radix(
108 &problem
109 .chars()
110 .enumerate()
111 .map(|(index, c)| {
112 if let Some(pos) = indexes.iter().position(|&x| x == index) {
113 if (i >> pos) & 1 == 1 { letter_digit[&c] } else { c }
114 } else {
115 c
116 }
117 })
118 .collect::<String>(),
119 0x10,
120 )
121 .unwrap()
122 })
123 })
124}
125
126const ROOT_PROBLEMATIC_CONSTS: &[u32] = &[
128 184594741, 2880289470, 2881141438, 2965027518, 2976579765, 3203381950, 3405691582, 3405697037,
129 3735927486, 3735932941, 4027431614, 4276992702, 195934910, 252707358, 762133, 179681982,
130 173390526, 721077,
131];
132
133const LETTER_DIGIT: &[(char, char)] = &[('A', '4'), ('B', '8'), ('E', '3')];
134
135fn generate_problematic_strings(
137 consts: &[u32],
138 letter_digit: &FxHashMap<char, char>,
139) -> Vec<String> {
140 generate_problems(consts, letter_digit)
141 .flat_map(|v| vec![v.to_string(), format!("{:X}", v)])
142 .collect()
143}
144
145static PROBLEMATIC_CONSTS_STRINGS: LazyLock<Vec<String>> = LazyLock::new(|| {
146 generate_problematic_strings(ROOT_PROBLEMATIC_CONSTS, &LETTER_DIGIT.iter().cloned().collect())
147});
148
149fn contains_problematic_const(trimmed: &str) -> bool {
150 PROBLEMATIC_CONSTS_STRINGS.iter().any(|s| trimmed.to_uppercase().contains(s))
151}
152
153const INTERNAL_COMPILER_DOCS_LINE: &str = "#### This error code is internal to the compiler and will not be emitted with normal Rust code.";
154
155#[derive(Clone, Copy, PartialEq)]
157#[allow(non_camel_case_types)]
158enum LIUState {
159 EXP_COMMENT_START,
160 EXP_LINK_LABEL_OR_URL,
161 EXP_URL,
162 EXP_END,
163}
164
165fn line_is_url(is_error_code: bool, columns: usize, line: &str) -> bool {
172 if is_error_code {
174 return line.starts_with('[') && line.contains("]:") && line.contains("http");
175 }
176
177 use self::LIUState::*;
178 let mut state: LIUState = EXP_COMMENT_START;
179 let is_url = |w: &str| w.starts_with("http://") || w.starts_with("https://");
180
181 for tok in line.split_whitespace() {
182 match (state, tok) {
183 (EXP_COMMENT_START, "//") | (EXP_COMMENT_START, "///") | (EXP_COMMENT_START, "//!") => {
184 state = EXP_LINK_LABEL_OR_URL
185 }
186
187 (EXP_LINK_LABEL_OR_URL, w)
188 if w.len() >= 4 && w.starts_with('[') && w.ends_with("]:") =>
189 {
190 state = EXP_URL
191 }
192
193 (EXP_LINK_LABEL_OR_URL, w) if is_url(w) => state = EXP_END,
194
195 (EXP_URL, w) if is_url(w) || w.starts_with("../") => state = EXP_END,
196
197 (_, w) if w.len() > columns && is_url(w) => state = EXP_END,
198
199 (_, _) => {}
200 }
201 }
202
203 state == EXP_END
204}
205
206fn should_ignore(line: &str) -> bool {
209 static_regex!("\\s*//(\\[.*\\])?~.*").is_match(line)
213 || ANNOTATIONS_TO_IGNORE.iter().any(|a| line.contains(a))
214
215 || static_regex!("\\s*//@(\\[.*\\]) (compile-flags|normalize-stderr|error-pattern).*")
219 .is_match(line)
220 || static_regex!(
223 "\\s*//@ \\!?(count|files|has|has-dir|hasraw|matches|matchesraw|snapshot)\\s.*"
224 ).is_match(line)
225 || static_regex!(
227 "\\s*// [a-zA-Z0-9-_]*:\\s.*"
228 ).is_match(line)
229}
230
231fn long_line_is_ok(extension: &str, is_error_code: bool, max_columns: usize, line: &str) -> bool {
233 match extension {
234 "md" if !is_error_code => true,
236 "md" if line == INTERNAL_COMPILER_DOCS_LINE => true,
238 _ => line_is_url(is_error_code, max_columns, line) || should_ignore(line),
239 }
240}
241
242#[derive(Clone, Copy)]
243enum Directive {
244 Deny,
246
247 Ignore(bool),
252}
253
254fn contains_ignore_directives<const N: usize>(
257 path_str: &str,
258 can_contain: bool,
259 contents: &str,
260 checks: [&str; N],
261) -> [Directive; N] {
262 let always_ignore_linelength = path_str.contains("rustdoc-json");
265
266 if !can_contain && !always_ignore_linelength {
267 return [Directive::Deny; N];
268 }
269
270 checks.map(|check| {
271 if check == LINELENGTH_CHECK && always_ignore_linelength {
272 return Directive::Ignore(false);
273 }
274
275 if contents.contains(&format!("// ignore-tidy-{check}"))
277 || contents.contains(&format!("# ignore-tidy-{check}"))
278 || contents.contains(&format!("/* ignore-tidy-{check} */"))
279 || contents.contains(&format!("<!-- ignore-tidy-{check} -->"))
280 {
281 Directive::Ignore(false)
282 } else {
283 Directive::Deny
284 }
285 })
286}
287
288macro_rules! suppressible_tidy_err {
289 ($err:ident, $skip:ident, $msg:literal) => {
290 if let Directive::Deny = $skip {
291 $err(&format!($msg));
292 } else {
293 $skip = Directive::Ignore(true);
294 }
295 };
296}
297
298pub fn is_in(full_path: &Path, parent_folder_to_find: &str, folder_to_find: &str) -> bool {
299 if let Some(parent) = full_path.parent() {
300 if parent.file_name().map_or_else(
301 || false,
302 |f| {
303 f == folder_to_find
304 && parent
305 .parent()
306 .and_then(|f| f.file_name())
307 .map_or_else(|| false, |f| f == parent_folder_to_find)
308 },
309 ) {
310 true
311 } else {
312 is_in(parent, parent_folder_to_find, folder_to_find)
313 }
314 } else {
315 false
316 }
317}
318
319fn skip_markdown_path(path: &Path) -> bool {
320 const SKIP_MD: &[&str] = &[
322 "src/doc/edition-guide",
323 "src/doc/embedded-book",
324 "src/doc/nomicon",
325 "src/doc/reference",
326 "src/doc/rust-by-example",
327 "src/doc/rustc-dev-guide",
328 ];
329 SKIP_MD.iter().any(|p| path.ends_with(p))
330}
331
332fn is_unexplained_ignore(extension: &str, line: &str) -> bool {
333 if !line.ends_with("```ignore") && !line.ends_with("```rust,ignore") {
334 return false;
335 }
336 if extension == "md" && line.trim().starts_with("//") {
337 return false;
340 }
341 true
342}
343
344pub fn check(path: &Path, tidy_ctx: TidyCtx) {
345 let mut check = tidy_ctx.start_check(CheckId::new("style").path(path));
346
347 fn skip(path: &Path, is_dir: bool) -> bool {
348 if path.file_name().is_some_and(|name| name.to_string_lossy().starts_with(".#")) {
349 return true;
351 }
352
353 if filter_dirs(path) || skip_markdown_path(path) {
354 return true;
355 }
356
357 if is_dir {
359 return false;
360 }
361
362 let extensions = ["rs", "py", "js", "sh", "c", "cpp", "h", "md", "css", "goml"];
363
364 if path.extension().is_none_or(|ext| !extensions.iter().any(|e| ext == OsStr::new(e))) {
366 return true;
367 }
368
369 path.extension().is_some_and(|e| e == "css") && !is_in(path, "src", "librustdoc")
371 }
372
373 let problematic_regex = RegexSetBuilder::new(PROBLEMATIC_CONSTS_STRINGS.as_slice())
376 .case_insensitive(true)
377 .build()
378 .unwrap();
379
380 let this_file = Path::new(file!());
383 let codegen_file = Path::new("src/tools/tidy/src/codegen.rs");
384
385 walk(path, skip, &mut |entry, contents| {
386 let file = entry.path();
387 let path_str = file.to_string_lossy();
388 let filename = file.file_name().unwrap().to_string_lossy();
389
390 let is_css_file = filename.ends_with(".css");
391 let under_rustfmt = filename.ends_with(".rs") &&
392 !file.ancestors().any(|a| {
395 (a.ends_with("tests") && a.join("COMPILER_TESTS.md").exists()) ||
396 a.ends_with("src/doc/book")
397 });
398
399 if contents.is_empty() {
400 check.error(format!("{}: empty file", file.display()));
401 }
402
403 let extension = file.extension().unwrap().to_string_lossy();
404 let is_error_code = extension == "md" && is_in(file, "src", "error_codes");
405 let is_goml_code = extension == "goml";
406
407 let max_columns = if is_error_code {
408 ERROR_CODE_COLS
409 } else if is_goml_code {
410 GOML_COLS
411 } else {
412 COLS
413 };
414
415 let can_contain = contents.contains("// ignore-tidy-")
417 || contents.contains("# ignore-tidy-")
418 || contents.contains("/* ignore-tidy-")
419 || contents.contains("<!-- ignore-tidy-");
420 if filename.contains("ignore-tidy") {
423 return;
424 }
425 if let Some(p) = file.parent()
427 && p.ends_with(Path::new("src/etc/completions"))
428 {
429 return;
430 }
431 let [
432 mut skip_cr,
433 mut skip_undocumented_unsafe,
434 mut skip_tab,
435 mut skip_line_length,
436 mut skip_file_length,
437 mut skip_end_whitespace,
438 mut skip_trailing_newlines,
439 mut skip_leading_newlines,
440 mut skip_copyright,
441 mut skip_dbg,
442 mut skip_odd_backticks,
443 ] = contains_ignore_directives(&path_str, can_contain, contents, CONFIGURABLE_CHECKS);
444 let mut leading_new_lines = false;
445 let mut trailing_new_lines = 0;
446 let mut lines = 0;
447 let mut last_safety_comment = false;
448 let mut comment_block: Option<(usize, usize)> = None;
449 let is_test = file.components().any(|c| c.as_os_str() == "tests")
450 || file.file_stem().unwrap() == "tests";
451 let is_codegen_test = is_test && file.components().any(|c| c.as_os_str() == "codegen-llvm");
452 let is_this_file = file.ends_with(this_file) || this_file.ends_with(file);
453 let is_test_for_this_file =
454 is_test && file.parent().unwrap().ends_with(this_file.with_extension(""));
455 let is_codegen_tidy_file = file.ends_with(codegen_file);
456 let any_problematic_line =
459 !is_this_file && !is_test_for_this_file && problematic_regex.is_match(contents);
460 for (i, line) in contents.split('\n').enumerate() {
461 if line.is_empty() {
462 if i == 0 {
463 leading_new_lines = true;
464 }
465 trailing_new_lines += 1;
466 continue;
467 } else {
468 trailing_new_lines = 0;
469 }
470
471 let trimmed = line.trim();
472
473 if !trimmed.starts_with("//") {
474 lines += 1;
475 }
476
477 let mut err = |msg: &str| {
478 check.error(format!("{}:{}: {msg}", file.display(), i + 1));
479 };
480
481 if trimmed.contains("dbg!")
482 && !trimmed.starts_with("//")
483 && !file.ancestors().any(|a| {
484 (a.ends_with("tests") && a.join("COMPILER_TESTS.md").exists())
485 || a.ends_with("library/alloctests")
486 })
487 && filename != "tests.rs"
488 {
489 suppressible_tidy_err!(
490 err,
491 skip_dbg,
492 "`dbg!` macro is intended as a debugging tool. It should not be in version control."
493 )
494 }
495
496 if is_codegen_test && trimmed.contains("CHECK") && trimmed.ends_with(": br") {
497 err("`CHECK: br` and `CHECK-NOT: br` in codegen tests are fragile to false \
498 positives in mangled symbols. Try using `br {{.*}}` instead.")
499 }
500
501 if !under_rustfmt
502 && line.chars().count() > max_columns
503 && !long_line_is_ok(&extension, is_error_code, max_columns, line)
504 {
505 suppressible_tidy_err!(
506 err,
507 skip_line_length,
508 "line longer than {max_columns} chars"
509 );
510 }
511 if !is_css_file && line.contains('\t') {
512 suppressible_tidy_err!(err, skip_tab, "tab character");
513 }
514 if line.ends_with(' ') || line.ends_with('\t') {
515 suppressible_tidy_err!(err, skip_end_whitespace, "trailing whitespace");
516 }
517 if is_css_file && line.starts_with(' ') {
518 err("CSS files use tabs for indent");
519 }
520 if line.contains('\r') {
521 suppressible_tidy_err!(err, skip_cr, "CR character");
522 }
523 if !is_this_file && !is_codegen_tidy_file {
524 let directive_line_starts = ["// ", "# ", "/* ", "<!-- "];
525 let possible_line_start =
526 directive_line_starts.into_iter().any(|s| line.starts_with(s));
527 let contains_potential_directive =
528 possible_line_start && (line.contains("-tidy") || line.contains("tidy-"));
529 let has_recognized_ignore_directive =
530 contains_ignore_directives(&path_str, can_contain, line, CONFIGURABLE_CHECKS)
531 .into_iter()
532 .any(|directive| matches!(directive, Directive::Ignore(_)));
533 let has_alphabetical_directive = line.contains("tidy-alphabetical-start")
534 || line.contains("tidy-alphabetical-end");
535 let has_other_tidy_ignore_directive =
536 line.contains("ignore-tidy-target-specific-tests");
537 let has_recognized_directive = has_recognized_ignore_directive
538 || has_alphabetical_directive
539 || has_other_tidy_ignore_directive;
540 if contains_potential_directive && (!has_recognized_directive) {
541 err("Unrecognized tidy directive")
542 }
543 if trimmed.contains("TODO") && !trimmed.contains("ignore-tidy-todo") {
546 err(
547 "TODO is used for tasks that should be done before merging a PR; If you want to leave a message in the codebase use FIXME",
548 )
549 }
550 if trimmed.contains("//") && trimmed.contains(" XXX") {
551 err("Instead of XXX use FIXME")
552 }
553 if any_problematic_line && contains_problematic_const(trimmed) {
554 err("Don't use magic numbers that spell things (consider 0x12345678)");
555 }
556 }
557 if trimmed.contains("unsafe {")
559 && !trimmed.starts_with("//")
560 && !last_safety_comment
561 && file.components().any(|c| c.as_os_str() == "core")
562 && !is_test
563 {
564 suppressible_tidy_err!(err, skip_undocumented_unsafe, "undocumented unsafe");
565 }
566 if trimmed.contains("// SAFETY:") {
567 last_safety_comment = true;
568 } else if trimmed.starts_with("//") || trimmed.is_empty() {
569 } else {
571 last_safety_comment = false;
572 }
573 if (line.starts_with("// Copyright")
574 || line.starts_with("# Copyright")
575 || line.starts_with("Copyright"))
576 && (trimmed.contains("Rust Developers")
577 || trimmed.contains("Rust Project Developers"))
578 {
579 suppressible_tidy_err!(
580 err,
581 skip_copyright,
582 "copyright notices attributed to the Rust Project Developers are deprecated"
583 );
584 }
585 if !file.components().any(|c| c.as_os_str() == "rustc_baked_icu_data")
586 && is_unexplained_ignore(&extension, line)
587 {
588 err(UNEXPLAINED_IGNORE_DOCTEST_INFO);
589 }
590
591 if filename.ends_with(".cpp") && line.contains("llvm_unreachable") {
592 err(LLVM_UNREACHABLE_INFO);
593 }
594
595 let is_compiler = || file.components().any(|c| c.as_os_str() == "compiler");
597
598 if is_compiler() {
599 if line.contains("//")
600 && line
601 .chars()
602 .collect::<Vec<_>>()
603 .windows(4)
604 .any(|cs| matches!(cs, ['.', ' ', ' ', last] if last.is_alphabetic()))
605 {
606 err(DOUBLE_SPACE_AFTER_DOT)
607 }
608
609 let likely_comment = |trimmed: &str| {
613 trimmed.contains("//")
615 || (trimmed.contains("cfg_attr") && trimmed.contains("doc"))
617 };
618
619 if likely_comment(trimmed) {
620 let (start_line, mut backtick_count) = comment_block.unwrap_or((i + 1, 0));
621 let line_backticks = trimmed.chars().filter(|ch| *ch == '`').count();
622
623 let comment_text = match trimmed.split("//").nth(1) {
626 Some(text) => text,
627 None => {
628 let (_doc, rest) =
630 trimmed.split_once("doc").expect("failed to find `doc` attribute");
631 rest
632 }
633 };
634
635 if line_backticks % 2 == 1 {
638 backtick_count += comment_text.chars().filter(|ch| *ch == '`').count();
639 }
640 comment_block = Some((start_line, backtick_count));
641 } else if let Some((start_line, backtick_count)) = comment_block.take()
642 && backtick_count % 2 == 1
643 {
644 let mut err = |msg: &str| {
645 check.error(format!("{}:{start_line}: {msg}", file.display()));
646 };
647 let block_len = (i + 1) - start_line;
648 if block_len == 1 {
649 suppressible_tidy_err!(
650 err,
651 skip_odd_backticks,
652 "comment with odd number of backticks"
653 );
654 } else {
655 suppressible_tidy_err!(
656 err,
657 skip_odd_backticks,
658 "{block_len}-line comment block with odd number of backticks"
659 );
660 }
661 }
662 }
663 }
664 if leading_new_lines {
665 let mut err = |_| {
666 check.error(format!("{}: leading newline", file.display()));
667 };
668 suppressible_tidy_err!(err, skip_leading_newlines, "missing leading newline");
669 }
670 let mut err = |msg: &str| {
671 check.error(format!("{}: {}", file.display(), msg));
672 };
673 match trailing_new_lines {
674 0 => suppressible_tidy_err!(err, skip_trailing_newlines, "missing trailing newline"),
675 1 => {}
676 n => suppressible_tidy_err!(
677 err,
678 skip_trailing_newlines,
679 "too many trailing newlines ({n})"
680 ),
681 };
682 if lines > LINES {
683 let mut err = |_| {
684 check.error(format!(
685 "{}: too many lines ({lines}) (add `// \
686 ignore-tidy-filelength` to the file to suppress this error)",
687 file.display(),
688 ));
689 };
690 suppressible_tidy_err!(err, skip_file_length, "");
691 }
692
693 if let Directive::Ignore(false) = skip_cr {
694 check.error(format!("{}: ignoring CR characters unnecessarily", file.display()));
695 }
696 if let Directive::Ignore(false) = skip_tab {
697 check.error(format!("{}: ignoring tab characters unnecessarily", file.display()));
698 }
699 if let Directive::Ignore(false) = skip_end_whitespace {
700 check.error(format!("{}: ignoring trailing whitespace unnecessarily", file.display()));
701 }
702 if let Directive::Ignore(false) = skip_trailing_newlines {
703 check.error(format!("{}: ignoring trailing newlines unnecessarily", file.display()));
704 }
705 if let Directive::Ignore(false) = skip_leading_newlines {
706 check.error(format!("{}: ignoring leading newlines unnecessarily", file.display()));
707 }
708 if let Directive::Ignore(false) = skip_copyright {
709 check.error(format!("{}: ignoring copyright unnecessarily", file.display()));
710 }
711 let _unused = skip_line_length;
714 let _unused = skip_file_length;
715 });
716}