Skip to content

Commit 51c5eee

Browse files
borsbors
committed
Auto merge of rust-lang#12945 - Jarcho:octal_escape, r=Alexendoo
Rework `octal_escapes` Main changes are not doing UTF-8 decoding, noting each occurrence as an individual lint emission, and narrowing the span to point to the escape itself. changelog: none
2 parents 0625183 + 4b16e26 commit 51c5eee

File tree

3 files changed

+175
-193
lines changed

3 files changed

+175
-193
lines changed

‎clippy_lints/src/octal_escapes.rs‎

Lines changed: 63 additions & 98 deletions
Original file line numberDiff line numberDiff line change
@@ -1,12 +1,12 @@
11
use clippy_utils::diagnostics::span_lint_and_then;
2-
use rustc_ast::ast::{Expr, ExprKind};
3-
use rustc_ast::token::{Lit, LitKind};
2+
use clippy_utils::source::get_source_text;
3+
use rustc_ast::token::LitKind;
4+
use rustc_ast::{Expr, ExprKind};
45
use rustc_errors::Applicability;
56
use rustc_lint::{EarlyContext, EarlyLintPass, LintContext};
67
use rustc_middle::lint::in_external_macro;
78
use rustc_session::declare_lint_pass;
8-
use rustc_span::Span;
9-
use std::fmt::Write;
9+
use rustc_span::{BytePos, Pos, SpanData};
1010

1111
declare_clippy_lint! {
1212
/// ### What it does
@@ -52,104 +52,69 @@ declare_lint_pass!(OctalEscapes => [OCTAL_ESCAPES]);
5252

5353
impl EarlyLintPass for OctalEscapes {
5454
fn check_expr(&mut self, cx: &EarlyContext<'_>, expr: &Expr) {
55-
if in_external_macro(cx.sess(), expr.span) {
56-
return;
57-
}
58-
59-
if let ExprKind::Lit(token_lit) = &expr.kind {
60-
if matches!(token_lit.kind, LitKind::Str) {
61-
check_lit(cx, token_lit, expr.span, true);
62-
} else if matches!(token_lit.kind, LitKind::ByteStr) {
63-
check_lit(cx, token_lit, expr.span, false);
64-
}
65-
}
66-
}
67-
}
68-
69-
fn check_lit(cx: &EarlyContext<'_>, lit: &Lit, span: Span, is_string: bool) {
70-
let contents = lit.symbol.as_str();
71-
let mut iter = contents.char_indices().peekable();
72-
let mut found = vec![];
55+
if let ExprKind::Lit(lit) = &expr.kind
56+
// The number of bytes from the start of the token to the start of literal's text.
57+
&& let start_offset = BytePos::from_u32(match lit.kind {
58+
LitKind::Str => 1,
59+
LitKind::ByteStr | LitKind::CStr => 2,
60+
_ => return,
61+
})
62+
&& !in_external_macro(cx.sess(), expr.span)
63+
{
64+
let s = lit.symbol.as_str();
65+
let mut iter = s.as_bytes().iter();
66+
while let Some(&c) = iter.next() {
67+
if c == b'\\'
68+
// Always move the iterator to read the escape char.
69+
&& let Some(b'0') = iter.next()
70+
{
71+
// C-style octal escapes read from one to three characters.
72+
// The first character (`0`) has already been read.
73+
let (tail, len, c_hi, c_lo) = match *iter.as_slice() {
74+
[c_hi @ b'0'..=b'7', c_lo @ b'0'..=b'7', ref tail @ ..] => (tail, 4, c_hi, c_lo),
75+
[c_lo @ b'0'..=b'7', ref tail @ ..] => (tail, 3, b'0', c_lo),
76+
_ => continue,
77+
};
78+
iter = tail.iter();
79+
let offset = start_offset + BytePos::from_usize(s.len() - tail.len());
80+
let data = expr.span.data();
81+
let span = SpanData {
82+
lo: data.lo + offset - BytePos::from_u32(len),
83+
hi: data.lo + offset,
84+
..data
85+
}
86+
.span();
7387

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, _)) = iter.next_if(|(_, ch)| matches!(ch, '0'..='7')) {
80-
if iter.next_if(|(_, ch)| matches!(ch, '0'..='7')).is_some() {
81-
to += 1;
88+
// Last check to make sure the source text matches what we read from the string.
89+
// Macros are involved somehow if this doesn't match.
90+
if let Some(src) = get_source_text(cx, span)
91+
&& let Some(src) = src.as_str()
92+
&& match *src.as_bytes() {
93+
[b'\\', b'0', lo] => lo == c_lo,
94+
[b'\\', b'0', hi, lo] => hi == c_hi && lo == c_lo,
95+
_ => false,
96+
}
97+
{
98+
span_lint_and_then(cx, OCTAL_ESCAPES, span, "octal-looking escape in a literal", |diag| {
99+
diag.help_once("octal escapes are not supported, `\\0` is always null")
100+
.span_suggestion(
101+
span,
102+
"if an octal escape is intended, use a hex escape instead",
103+
format!("\\x{:02x}", (((c_hi - b'0') << 3) | (c_lo - b'0'))),
104+
Applicability::MaybeIncorrect,
105+
)
106+
.span_suggestion(
107+
span,
108+
"if a null escape is intended, disambiguate using",
109+
format!("\\x00{}{}", c_hi as char, c_lo as char),
110+
Applicability::MaybeIncorrect,
111+
);
112+
});
113+
} else {
114+
break;
82115
}
83-
found.push((from, to + 1));
84116
}
85117
}
86118
}
87119
}
88-
89-
if found.is_empty() {
90-
return;
91-
}
92-
93-
span_lint_and_then(
94-
cx,
95-
OCTAL_ESCAPES,
96-
span,
97-
format!(
98-
"octal-looking escape in {} literal",
99-
if is_string { "string" } else { "byte string" }
100-
),
101-
|diag| {
102-
diag.help(format!(
103-
"octal escapes are not supported, `\\0` is always a null {}",
104-
if is_string { "character" } else { "byte" }
105-
));
106-
107-
// Generate suggestions if the string is not too long (~ 5 lines)
108-
if contents.len() < 400 {
109-
// construct two suggestion strings, one with \x escapes with octal meaning
110-
// as in C, and one with \x00 for null bytes.
111-
let mut suggest_1 = if is_string { "\"" } else { "b\"" }.to_string();
112-
let mut suggest_2 = suggest_1.clone();
113-
let mut index = 0;
114-
for (from, to) in found {
115-
suggest_1.push_str(&contents[index..from]);
116-
suggest_2.push_str(&contents[index..from]);
117-
118-
// construct a replacement escape
119-
// the maximum value is \077, or \x3f, so u8 is sufficient here
120-
if let Ok(n) = u8::from_str_radix(&contents[from + 1..to], 8) {
121-
write!(suggest_1, "\\x{n:02x}").unwrap();
122-
}
123-
124-
// append the null byte as \x00 and the following digits literally
125-
suggest_2.push_str("\\x00");
126-
suggest_2.push_str(&contents[from + 2..to]);
127-
128-
index = to;
129-
}
130-
suggest_1.push_str(&contents[index..]);
131-
suggest_2.push_str(&contents[index..]);
132-
133-
suggest_1.push('"');
134-
suggest_2.push('"');
135-
// suggestion 1: equivalent hex escape
136-
diag.span_suggestion(
137-
span,
138-
"if an octal escape was intended, use the hexadecimal representation instead",
139-
suggest_1,
140-
Applicability::MaybeIncorrect,
141-
);
142-
// suggestion 2: unambiguous null byte
143-
diag.span_suggestion(
144-
span,
145-
format!(
146-
"if the null {} is intended, disambiguate using",
147-
if is_string { "character" } else { "byte" }
148-
),
149-
suggest_2,
150-
Applicability::MaybeIncorrect,
151-
);
152-
}
153-
},
154-
);
155120
}

‎tests/ui/octal_escapes.rs‎

Lines changed: 11 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -2,25 +2,20 @@
22
#![warn(clippy::octal_escapes)]
33

44
fn main() {
5-
let _bad1 = "\033[0m";
6-
//~^ ERROR: octal-looking escape in string literal
7-
let _bad2 = b"\033[0m";
8-
//~^ ERROR: octal-looking escape in byte string literal
9-
let _bad3 = "\\\033[0m";
10-
//~^ ERROR: octal-looking escape in string literal
5+
let _bad1 = "\033[0m"; //~ octal_escapes
6+
let _bad2 = b"\033[0m"; //~ octal_escapes
7+
let _bad3 = "\\\033[0m"; //~ octal_escapes
118
// maximum 3 digits (\012 is the escape)
12-
let _bad4 = "\01234567";
13-
//~^ ERROR: octal-looking escape in string literal
14-
let _bad5 = "\0\03";
15-
//~^ ERROR: octal-looking escape in string literal
9+
let _bad4 = "\01234567"; //~ octal_escapes
10+
let _bad5 = "\0\03"; //~ octal_escapes
1611
let _bad6 = "Text-\055\077-MoreText";
17-
//~^ ERROR: octal-looking escape in string literal
12+
//~^ octal_escapes
13+
//~| octal_escapes
1814
let _bad7 = "EvenMoreText-\01\02-ShortEscapes";
19-
//~^ ERROR: octal-looking escape in string literal
20-
let _bad8 = "锈\01锈";
21-
//~^ ERROR: octal-looking escape in string literal
22-
let _bad9 = "锈\011锈";
23-
//~^ ERROR: octal-looking escape in string literal
15+
//~^ octal_escapes
16+
//~| octal_escapes
17+
let _bad8 = "锈\01锈"; //~ octal_escapes
18+
let _bad9 = "锈\011锈"; //~ octal_escapes
2419

2520
let _good1 = "\\033[0m";
2621
let _good2 = "\0\\0";

0 commit comments

Comments
 (0)