src/tools/clippy/clippy_lints/src/unicode.rs

   1 use clippy_utils::diagnostics::span_lint_and_sugg;
   2 use clippy_utils::is_lint_allowed;
   3 use clippy_utils::macros::span_is_local;
   4 use clippy_utils::source::snippet;
   5 use rustc_ast::ast::LitKind;
   6 use rustc_errors::Applicability;
   7 use rustc_hir::{Expr, ExprKind, HirId};
   8 use rustc_lint::{LateContext, LateLintPass};
   9 use rustc_session::declare_lint_pass;
  10 use rustc_span::Span;
  11 use unicode_normalization::UnicodeNormalization;
  12
  13 declare_clippy_lint! {
  14     /// ### What it does
  15     /// Checks for invisible Unicode characters in the code.
  16     ///
  17     /// ### Why is this bad?
  18     /// Having an invisible character in the code makes for all
  19     /// sorts of April fools, but otherwise is very much frowned upon.
  20     ///
  21     /// ### Example
  22     /// You don't see it, but there may be a zero-width space or soft hyphen
  23     /// somewhere in this text.
  24     #[clippy::version = "1.49.0"]
  25     pub INVISIBLE_CHARACTERS,
  26     correctness,
  27     "using an invisible character in a string literal, which is confusing"
  28 }
  29
  30 declare_clippy_lint! {
  31     /// ### What it does
  32     /// Checks for non-ASCII characters in string and char literals.
  33     ///
  34     /// ### Why is this bad?
  35     /// Yeah, we know, the 90's called and wanted their charset
  36     /// back. Even so, there still are editors and other programs out there that
  37     /// don't work well with Unicode. So if the code is meant to be used
  38     /// internationally, on multiple operating systems, or has other portability
  39     /// requirements, activating this lint could be useful.
  40     ///
  41     /// ### Example
  42     /// ```no_run
  43     /// let x = String::from("€");
  44     /// ```
  45     ///
  46     /// Use instead:
  47     /// ```no_run
  48     /// let x = String::from("\u{20ac}");
  49     /// ```
  50     #[clippy::version = "pre 1.29.0"]
  51     pub NON_ASCII_LITERAL,
  52     restriction,
  53     "using any literal non-ASCII chars in a string literal instead of using the `\\u` escape"
  54 }
  55
  56 declare_clippy_lint! {
  57     /// ### What it does
  58     /// Checks for string literals that contain Unicode in a form
  59     /// that is not equal to its
  60     /// [NFC-recomposition](http://www.unicode.org/reports/tr15/#Norm_Forms).
  61     ///
  62     /// ### Why is this bad?
  63     /// If such a string is compared to another, the results
  64     /// may be surprising.
  65     ///
  66     /// ### Example
  67     /// You may not see it, but "à"" and "à"" aren't the same string. The
  68     /// former when escaped is actually `"a\u{300}"` while the latter is `"\u{e0}"`.
  69     #[clippy::version = "pre 1.29.0"]
  70     pub UNICODE_NOT_NFC,
  71     pedantic,
  72     "using a Unicode literal not in NFC normal form (see [Unicode tr15](http://www.unicode.org/reports/tr15/) for further information)"
  73 }
  74
  75 declare_lint_pass!(Unicode => [INVISIBLE_CHARACTERS, NON_ASCII_LITERAL, UNICODE_NOT_NFC]);
  76
  77 impl LateLintPass<'_> for Unicode {
  78     fn check_expr(&mut self, cx: &LateContext<'_>, expr: &'_ Expr<'_>) {
  79         if let ExprKind::Lit(lit) = expr.kind {
  80             if let LitKind::Str(_, _) | LitKind::Char(_) = lit.node {
  81                 check_str(cx, lit.span, expr.hir_id);
  82             }
  83         }
  84     }
  85 }
  86
  87 fn escape<T: Iterator<Item = char>>(s: T) -> String {
  88     let mut result = String::new();
  89     for c in s {
  90         if c as u32 > 0x7F {
  91             for d in c.escape_unicode() {
  92                 result.push(d);
  93             }
  94         } else {
  95             result.push(c);
  96         }
  97     }
  98     result
  99 }
 100
 101 fn check_str(cx: &LateContext<'_>, span: Span, id: HirId) {
 102     if !span_is_local(span) {
 103         return;
 104     }
 105
 106     let string = snippet(cx, span, "");
 107     if string.chars().any(|c| ['\u{200B}', '\u{ad}', '\u{2060}'].contains(&c)) {
 108         span_lint_and_sugg(
 109             cx,
 110             INVISIBLE_CHARACTERS,
 111             span,
 112             "invisible character detected",
 113             "consider replacing the string with",
 114             string
 115                 .replace('\u{200B}', "\\u{200B}")
 116                 .replace('\u{ad}', "\\u{AD}")
 117                 .replace('\u{2060}', "\\u{2060}"),
 118             Applicability::MachineApplicable,
 119         );
 120     }
 121
 122     if string.chars().any(|c| c as u32 > 0x7F) {
 123         span_lint_and_sugg(
 124             cx,
 125             NON_ASCII_LITERAL,
 126             span,
 127             "literal non-ASCII character detected",
 128             "consider replacing the string with",
 129             if is_lint_allowed(cx, UNICODE_NOT_NFC, id) {
 130                 escape(string.chars())
 131             } else {
 132                 escape(string.nfc())
 133             },
 134             Applicability::MachineApplicable,
 135         );
 136     }
 137
 138     if is_lint_allowed(cx, NON_ASCII_LITERAL, id) && string.chars().zip(string.nfc()).any(|(a, b)| a != b) {
 139         span_lint_and_sugg(
 140             cx,
 141             UNICODE_NOT_NFC,
 142             span,
 143             "non-NFC Unicode sequence detected",
 144             "consider replacing the string with",
 145             string.nfc().collect::<String>(),
 146             Applicability::MachineApplicable,
 147         );
 148     }
 149 }