New lint sliced_string_as_bytes (#14002)

resurrection of #10984

fixes #10981

changelog: [`sliced_string_as_bytes`]: add new lint
`sliced_string_as_bytes`

Author: Manishearth

CHANGELOG.md

@@ -6066,6 +6066,7 @@ Released 2018-09-13
 [`size_of_in_element_count`]: https://rust-lang.github.io/rust-clippy/master/index.html#size_of_in_element_count
 [`size_of_ref`]: https://rust-lang.github.io/rust-clippy/master/index.html#size_of_ref
 [`skip_while_next`]: https://rust-lang.github.io/rust-clippy/master/index.html#skip_while_next
+[`sliced_string_as_bytes`]: https://rust-lang.github.io/rust-clippy/master/index.html#sliced_string_as_bytes
 [`slow_vector_initialization`]: https://rust-lang.github.io/rust-clippy/master/index.html#slow_vector_initialization
 [`stable_sort_primitive`]: https://rust-lang.github.io/rust-clippy/master/index.html#stable_sort_primitive
 [`std_instead_of_alloc`]: https://rust-lang.github.io/rust-clippy/master/index.html#std_instead_of_alloc

clippy_lints/src/declared_lints.rs

@@ -468,6 +468,7 @@ pub static LINTS: &[&crate::LintInfo] = &[
     crate::methods::SHOULD_IMPLEMENT_TRAIT_INFO,
     crate::methods::SINGLE_CHAR_ADD_STR_INFO,
     crate::methods::SKIP_WHILE_NEXT_INFO,
+    crate::methods::SLICED_STRING_AS_BYTES_INFO,
     crate::methods::STABLE_SORT_PRIMITIVE_INFO,
     crate::methods::STRING_EXTEND_CHARS_INFO,
     crate::methods::STRING_LIT_CHARS_ANY_INFO,

clippy_lints/src/methods/mod.rs

@@ -102,6 +102,7 @@ mod single_char_add_str;
 mod single_char_insert_string;
 mod single_char_push_string;
 mod skip_while_next;
+mod sliced_string_as_bytes;
 mod stable_sort_primitive;
 mod str_split;
 mod str_splitn;
@@ -4363,6 +4364,34 @@ declare_clippy_lint! {
     "detect `repeat().take()` that can be replaced with `repeat_n()`"
 }
 
+declare_clippy_lint! {
+    /// ### What it does
+    /// Checks for string slices immediantly followed by `as_bytes`.
+    ///
+    /// ### Why is this bad?
+    /// It involves doing an unnecessary UTF-8 alignment check which is less efficient, and can cause a panic.
+    ///
+    /// ### Known problems
+    /// In some cases, the UTF-8 validation and potential panic from string slicing may be required for
+    /// the code's correctness. If you need to ensure the slice boundaries fall on valid UTF-8 character
+    /// boundaries, the original form (`s[1..5].as_bytes()`) should be preferred.
+    ///
+    /// ### Example
+    /// ```rust
+    /// let s = "Lorem ipsum";
+    /// s[1..5].as_bytes();
+    /// ```
+    /// Use instead:
+    /// ```rust
+    /// let s = "Lorem ipsum";
+    /// &s.as_bytes()[1..5];
+    /// ```
+     #[clippy::version = "1.86.0"]
+     pub SLICED_STRING_AS_BYTES,
+     perf,
+     "slicing a string and immediately calling as_bytes is less efficient and can lead to panics"
+}
+
 pub struct Methods {
     avoid_breaking_exported_api: bool,
     msrv: Msrv,
@@ -4531,6 +4560,7 @@ impl_lint_pass!(Methods => [
     DOUBLE_ENDED_ITERATOR_LAST,
     USELESS_NONZERO_NEW_UNCHECKED,
     MANUAL_REPEAT_N,
+    SLICED_STRING_AS_BYTES,
 ]);
 
 /// Extracts a method call name, args, and `Span` of the method name.
@@ -4798,6 +4828,7 @@ impl Methods {
                     if let Some(("as_str", recv, [], as_str_span, _)) = method_call(recv) {
                         redundant_as_str::check(cx, expr, recv, as_str_span, span);
                     }
+                    sliced_string_as_bytes::check(cx, expr, recv);
                 },
                 ("as_mut", []) => useless_asref::check(cx, expr, "as_mut", recv),
                 ("as_ptr", []) => manual_c_str_literals::check_as_ptr(cx, expr, recv, &self.msrv),

clippy_lints/src/methods/sliced_string_as_bytes.rs

@@ -0,0 +1,29 @@
+use clippy_utils::diagnostics::span_lint_and_sugg;
+use clippy_utils::source::snippet_with_applicability;
+use clippy_utils::ty::is_type_lang_item;
+use rustc_errors::Applicability;
+use rustc_hir::{Expr, ExprKind, LangItem, is_range_literal};
+use rustc_lint::LateContext;
+
+use super::SLICED_STRING_AS_BYTES;
+
+pub(super) fn check(cx: &LateContext<'_>, expr: &Expr<'_>, recv: &Expr<'_>) {
+    if let ExprKind::Index(indexed, index, _) = recv.kind
+        && is_range_literal(index)
+        && let ty = cx.typeck_results().expr_ty(indexed).peel_refs()
+        && (ty.is_str() || is_type_lang_item(cx, ty, LangItem::String))
+    {
+        let mut applicability = Applicability::MaybeIncorrect;
+        let stringish = snippet_with_applicability(cx, indexed.span, "_", &mut applicability);
+        let range = snippet_with_applicability(cx, index.span, "_", &mut applicability);
+        span_lint_and_sugg(
+            cx,
+            SLICED_STRING_AS_BYTES,
+            expr.span,
+            "calling `as_bytes` after slicing a string",
+            "try",
+            format!("&{stringish}.as_bytes()[{range}]"),
+            applicability,
+        );
+    }
+}

tests/ui/bytes_nth.fixed

@@ -1,4 +1,5 @@
 #![allow(clippy::unnecessary_operation)]
+#![allow(clippy::sliced_string_as_bytes)]
 #![warn(clippy::bytes_nth)]
 
 fn main() {

tests/ui/bytes_nth.rs

@@ -1,4 +1,5 @@
 #![allow(clippy::unnecessary_operation)]
+#![allow(clippy::sliced_string_as_bytes)]
 #![warn(clippy::bytes_nth)]
 
 fn main() {

tests/ui/bytes_nth.stderr

@@ -1,5 +1,5 @@
 error: called `.bytes().nth()` on a `String`
-  --> tests/ui/bytes_nth.rs:6:13
+  --> tests/ui/bytes_nth.rs:7:13
    |
 LL |     let _ = s.bytes().nth(3);
    |             ^^^^^^^^^^^^^^^^ help: try: `s.as_bytes().get(3).copied()`
@@ -8,13 +8,13 @@ LL |     let _ = s.bytes().nth(3);
    = help: to override `-D warnings` add `#[allow(clippy::bytes_nth)]`
 
 error: called `.bytes().nth().unwrap()` on a `String`
-  --> tests/ui/bytes_nth.rs:7:14
+  --> tests/ui/bytes_nth.rs:8:14
    |
 LL |     let _ = &s.bytes().nth(3).unwrap();
    |              ^^^^^^^^^^^^^^^^^^^^^^^^^ help: try: `s.as_bytes()[3]`
 
 error: called `.bytes().nth()` on a `str`
-  --> tests/ui/bytes_nth.rs:8:13
+  --> tests/ui/bytes_nth.rs:9:13
    |
 LL |     let _ = s[..].bytes().nth(3);
    |             ^^^^^^^^^^^^^^^^^^^^ help: try: `s[..].as_bytes().get(3).copied()`

tests/ui/sliced_string_as_bytes.fixed

@@ -0,0 +1,34 @@
+#![allow(unused)]
+#![warn(clippy::sliced_string_as_bytes)]
+
+use std::ops::{Index, Range};
+
+struct Foo;
+
+struct Bar;
+
+impl Bar {
+    fn as_bytes(&self) -> &[u8] {
+        &[0, 1, 2, 3]
+    }
+}
+
+impl Index<Range<usize>> for Foo {
+    type Output = Bar;
+
+    fn index(&self, _: Range<usize>) -> &Self::Output {
+        &Bar
+    }
+}
+
+fn main() {
+    let s = "Lorem ipsum";
+    let string: String = "dolor sit amet".to_owned();
+
+    let bytes = &s.as_bytes()[1..5];
+    let bytes = &string.as_bytes()[1..];
+    let bytes = &"consectetur adipiscing".as_bytes()[..=5];
+
+    let f = Foo;
+    let bytes = f[0..4].as_bytes();
+}

tests/ui/sliced_string_as_bytes.rs

@@ -0,0 +1,34 @@
+#![allow(unused)]
+#![warn(clippy::sliced_string_as_bytes)]
+
+use std::ops::{Index, Range};
+
+struct Foo;
+
+struct Bar;
+
+impl Bar {
+    fn as_bytes(&self) -> &[u8] {
+        &[0, 1, 2, 3]
+    }
+}
+
+impl Index<Range<usize>> for Foo {
+    type Output = Bar;
+
+    fn index(&self, _: Range<usize>) -> &Self::Output {
+        &Bar
+    }
+}
+
+fn main() {
+    let s = "Lorem ipsum";
+    let string: String = "dolor sit amet".to_owned();
+
+    let bytes = s[1..5].as_bytes();
+    let bytes = string[1..].as_bytes();
+    let bytes = "consectetur adipiscing"[..=5].as_bytes();
+
+    let f = Foo;
+    let bytes = f[0..4].as_bytes();
+}

tests/ui/sliced_string_as_bytes.stderr

@@ -0,0 +1,23 @@
+error: calling `as_bytes` after slicing a string
+  --> tests/ui/sliced_string_as_bytes.rs:28:17
+   |
+LL |     let bytes = s[1..5].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^ help: try: `&s.as_bytes()[1..5]`
+   |
+   = note: `-D clippy::sliced-string-as-bytes` implied by `-D warnings`
+   = help: to override `-D warnings` add `#[allow(clippy::sliced_string_as_bytes)]`
+
+error: calling `as_bytes` after slicing a string
+  --> tests/ui/sliced_string_as_bytes.rs:29:17
+   |
+LL |     let bytes = string[1..].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^^^^^ help: try: `&string.as_bytes()[1..]`
+
+error: calling `as_bytes` after slicing a string
+  --> tests/ui/sliced_string_as_bytes.rs:30:17
+   |
+LL |     let bytes = "consectetur adipiscing"[..=5].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try: `&"consectetur adipiscing".as_bytes()[..=5]`
+
+error: aborting due to 3 previous errors
+