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