]>
Commit | Line | Data |
---|---|---|
a2a8927a XL |
1 | use clippy_utils::diagnostics::span_lint_and_then; |
2 | use rustc_ast::ast::{Expr, ExprKind}; | |
3 | use rustc_ast::token::{Lit, LitKind}; | |
4 | use rustc_errors::Applicability; | |
5099ac24 | 5 | use rustc_lint::{EarlyContext, EarlyLintPass, LintContext}; |
a2a8927a XL |
6 | use rustc_middle::lint::in_external_macro; |
7 | use rustc_session::{declare_lint_pass, declare_tool_lint}; | |
8 | use rustc_span::Span; | |
9 | use std::fmt::Write; | |
10 | ||
11 | declare_clippy_lint! { | |
12 | /// ### What it does | |
13 | /// Checks for `\0` escapes in string and byte literals that look like octal | |
14 | /// character escapes in C. | |
15 | /// | |
16 | /// ### Why is this bad? | |
17 | /// | |
18 | /// C and other languages support octal character escapes in strings, where | |
19 | /// a backslash is followed by up to three octal digits. For example, `\033` | |
20 | /// stands for the ASCII character 27 (ESC). Rust does not support this | |
21 | /// notation, but has the escape code `\0` which stands for a null | |
22 | /// byte/character, and any following digits do not form part of the escape | |
23 | /// sequence. Therefore, `\033` is not a compiler error but the result may | |
24 | /// be surprising. | |
25 | /// | |
26 | /// ### Known problems | |
27 | /// The actual meaning can be the intended one. `\x00` can be used in these | |
04454e1e | 28 | /// cases to be unambiguous. |
a2a8927a XL |
29 | /// |
30 | /// The lint does not trigger for format strings in `print!()`, `write!()` | |
31 | /// and friends since the string is already preprocessed when Clippy lints | |
32 | /// can see it. | |
33 | /// | |
064997fb | 34 | /// ### Example |
a2a8927a | 35 | /// ```rust |
a2a8927a XL |
36 | /// let one = "\033[1m Bold? \033[0m"; // \033 intended as escape |
37 | /// let two = "\033\0"; // \033 intended as null-3-3 | |
923072b8 | 38 | /// ``` |
a2a8927a | 39 | /// |
923072b8 FG |
40 | /// Use instead: |
41 | /// ```rust | |
a2a8927a XL |
42 | /// let one = "\x1b[1mWill this be bold?\x1b[0m"; |
43 | /// let two = "\x0033\x00"; | |
44 | /// ``` | |
923072b8 | 45 | #[clippy::version = "1.59.0"] |
a2a8927a XL |
46 | pub OCTAL_ESCAPES, |
47 | suspicious, | |
48 | "string escape sequences looking like octal characters" | |
49 | } | |
50 | ||
51 | declare_lint_pass!(OctalEscapes => [OCTAL_ESCAPES]); | |
52 | ||
53 | impl EarlyLintPass for OctalEscapes { | |
5099ac24 FG |
54 | fn check_expr(&mut self, cx: &EarlyContext<'_>, expr: &Expr) { |
55 | if in_external_macro(cx.sess(), expr.span) { | |
a2a8927a XL |
56 | return; |
57 | } | |
58 | ||
59 | if let ExprKind::Lit(lit) = &expr.kind { | |
60 | if matches!(lit.token.kind, LitKind::Str) { | |
61 | check_lit(cx, &lit.token, lit.span, true); | |
62 | } else if matches!(lit.token.kind, LitKind::ByteStr) { | |
63 | check_lit(cx, &lit.token, lit.span, false); | |
64 | } | |
65 | } | |
66 | } | |
67 | } | |
68 | ||
5099ac24 | 69 | fn check_lit(cx: &EarlyContext<'_>, lit: &Lit, span: Span, is_string: bool) { |
a2a8927a XL |
70 | let contents = lit.symbol.as_str(); |
71 | let mut iter = contents.char_indices().peekable(); | |
72 | let mut found = vec![]; | |
73 | ||
74 | // go through the string, looking for \0[0-7][0-7]? | |
75 | while let Some((from, ch)) = iter.next() { | |
76 | if ch == '\\' { | |
77 | if let Some((_, '0')) = iter.next() { | |
78 | // collect up to two further octal digits | |
79 | if let Some((mut to, '0'..='7')) = iter.next() { | |
80 | if let Some((_, '0'..='7')) = iter.peek() { | |
81 | to += 1; | |
82 | } | |
83 | found.push((from, to + 1)); | |
84 | } | |
85 | } | |
86 | } | |
87 | } | |
88 | ||
89 | if found.is_empty() { | |
90 | return; | |
91 | } | |
92 | ||
93 | // construct two suggestion strings, one with \x escapes with octal meaning | |
94 | // as in C, and one with \x00 for null bytes. | |
95 | let mut suggest_1 = if is_string { "\"" } else { "b\"" }.to_string(); | |
96 | let mut suggest_2 = suggest_1.clone(); | |
97 | let mut index = 0; | |
98 | for (from, to) in found { | |
99 | suggest_1.push_str(&contents[index..from]); | |
100 | suggest_2.push_str(&contents[index..from]); | |
101 | ||
102 | // construct a replacement escape | |
103 | // the maximum value is \077, or \x3f, so u8 is sufficient here | |
104 | if let Ok(n) = u8::from_str_radix(&contents[from + 1..to], 8) { | |
5099ac24 | 105 | write!(suggest_1, "\\x{:02x}", n).unwrap(); |
a2a8927a XL |
106 | } |
107 | ||
108 | // append the null byte as \x00 and the following digits literally | |
109 | suggest_2.push_str("\\x00"); | |
110 | suggest_2.push_str(&contents[from + 2..to]); | |
111 | ||
112 | index = to; | |
113 | } | |
114 | suggest_1.push_str(&contents[index..]); | |
115 | suggest_1.push('"'); | |
116 | suggest_2.push_str(&contents[index..]); | |
117 | suggest_2.push('"'); | |
118 | ||
119 | span_lint_and_then( | |
120 | cx, | |
121 | OCTAL_ESCAPES, | |
122 | span, | |
123 | &format!( | |
124 | "octal-looking escape in {} literal", | |
125 | if is_string { "string" } else { "byte string" } | |
126 | ), | |
127 | |diag| { | |
128 | diag.help(&format!( | |
129 | "octal escapes are not supported, `\\0` is always a null {}", | |
130 | if is_string { "character" } else { "byte" } | |
131 | )); | |
132 | // suggestion 1: equivalent hex escape | |
133 | diag.span_suggestion( | |
134 | span, | |
135 | "if an octal escape was intended, use the hexadecimal representation instead", | |
136 | suggest_1, | |
137 | Applicability::MaybeIncorrect, | |
138 | ); | |
139 | // suggestion 2: unambiguous null byte | |
140 | diag.span_suggestion( | |
141 | span, | |
142 | &format!( | |
143 | "if the null {} is intended, disambiguate using", | |
144 | if is_string { "character" } else { "byte" } | |
145 | ), | |
146 | suggest_2, | |
147 | Applicability::MaybeIncorrect, | |
148 | ); | |
149 | }, | |
150 | ); | |
151 | } |