Add slice_as_bytes lint

CHANGELOG.md

@@ -5205,6 +5205,7 @@ 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
+[`slice_as_bytes`]: https://rust-lang.github.io/rust-clippy/master/index.html#slice_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

@@ -398,6 +398,7 @@ pub(crate) static LINTS: &[&crate::LintInfo] = &[
     crate::methods::SINGLE_CHAR_ADD_STR_INFO,
     crate::methods::SINGLE_CHAR_PATTERN_INFO,
     crate::methods::SKIP_WHILE_NEXT_INFO,
+    crate::methods::SLICE_AS_BYTES_INFO,
     crate::methods::STABLE_SORT_PRIMITIVE_INFO,
     crate::methods::STRING_EXTEND_CHARS_INFO,
     crate::methods::SUSPICIOUS_COMMAND_ARG_SPACE_INFO,

clippy_lints/src/methods/mod.rs

@@ -81,6 +81,7 @@ mod single_char_insert_string;
 mod single_char_pattern;
 mod single_char_push_string;
 mod skip_while_next;
+mod slice_as_bytes;
 mod stable_sort_primitive;
 mod str_splitn;
 mod string_extend_chars;
@@ -3316,6 +3317,26 @@ declare_clippy_lint! {
     "checks for usage of `Iterator::fold` with a type that implements `Try`"
 }
 
+declare_clippy_lint! {
+    /// 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.
+    /// ### 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 = "1.72.0"]
+    pub SLICE_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,
@@ -3448,6 +3469,7 @@ impl_lint_pass!(Methods => [
     UNNECESSARY_LITERAL_UNWRAP,
     DRAIN_COLLECT,
     MANUAL_TRY_FOLD,
+    SLICE_AS_BYTES,
 ]);
 
 /// Extracts a method call name, args, and `Span` of the method name.
@@ -3656,6 +3678,9 @@ impl Methods {
                 ("arg", [arg]) => {
                     suspicious_command_arg_space::check(cx, recv, arg, span);
                 }
+                ("as_bytes",[]) => {
+                    slice_as_bytes::check(cx, expr, recv);
+                }
                 ("as_deref" | "as_deref_mut", []) => {
                     needless_option_as_deref::check(cx, expr, recv, name);
                 },

clippy_lints/src/methods/slice_as_bytes.rs

@@ -0,0 +1,33 @@
+use clippy_utils::{diagnostics::span_lint_and_sugg, source::snippet_with_applicability, ty::is_type_lang_item};
+use rustc_errors::Applicability;
+use rustc_hir::{is_range_literal, Expr, ExprKind, LangItem};
+use rustc_lint::LateContext;
+
+use super::SLICE_AS_BYTES;
+
+pub(super) fn check(cx: &LateContext<'_>, expr: &Expr<'_>, recv: &Expr<'_>) {
+    if let ExprKind::Index(indexed, index) = recv.kind {
+        if is_range_literal(index) {
+            let ty = cx.typeck_results().expr_ty(indexed).peel_refs();
+            let is_str = ty.is_str();
+            let is_string = is_type_lang_item(cx, ty, LangItem::String);
+            if is_str || is_string {
+                let mut applicability = Applicability::MachineApplicable;
+                let stringish = snippet_with_applicability(cx, indexed.span, "..", &mut applicability);
+                let range = snippet_with_applicability(cx, index.span, "..", &mut applicability);
+                let type_name = if is_str { "str" } else { "String" };
+                span_lint_and_sugg(
+                    cx,
+                    SLICE_AS_BYTES,
+                    expr.span,
+                    &(format!(
+                        "slicing a {type_name} before calling `as_bytes` results in needless UTF-8 alignment checks, and has the possiblity of panicking"
+                    )),
+                    "try",
+                    format!("&{stringish}.as_bytes()[{range}]"),
+                    applicability,
+                );
+            }
+        }
+    }
+}

tests/ui/bytes_nth.fixed

@@ -1,6 +1,7 @@
 //@run-rustfix
 
 #![allow(clippy::unnecessary_operation)]
+#![allow(clippy::slice_as_bytes)]
 #![warn(clippy::bytes_nth)]
 
 fn main() {

tests/ui/bytes_nth.rs

@@ -1,6 +1,7 @@
 //@run-rustfix
 
 #![allow(clippy::unnecessary_operation)]
+#![allow(clippy::slice_as_bytes)]
 #![warn(clippy::bytes_nth)]
 
 fn main() {

tests/ui/bytes_nth.stderr

@@ -1,19 +1,19 @@
 error: called `.bytes().nth()` on a `String`
-  --> $DIR/bytes_nth.rs:8:13
+  --> $DIR/bytes_nth.rs:9:13
    |
 LL |     let _ = s.bytes().nth(3);
    |             ^^^^^^^^^^^^^^^^ help: try: `s.as_bytes().get(3).copied()`
    |
    = note: `-D clippy::bytes-nth` implied by `-D warnings`
 
 error: called `.bytes().nth().unwrap()` on a `String`
-  --> $DIR/bytes_nth.rs:9:14
+  --> $DIR/bytes_nth.rs:10:14
    |
 LL |     let _ = &s.bytes().nth(3).unwrap();
    |              ^^^^^^^^^^^^^^^^^^^^^^^^^ help: try: `s.as_bytes()[3]`
 
 error: called `.bytes().nth()` on a `str`
-  --> $DIR/bytes_nth.rs:10:13
+  --> $DIR/bytes_nth.rs:11:13
    |
 LL |     let _ = s[..].bytes().nth(3);
    |             ^^^^^^^^^^^^^^^^^^^^ help: try: `s[..].as_bytes().get(3).copied()`

tests/ui/slice_as_bytes.fixed

@@ -0,0 +1,35 @@
+//@run-rustfix
+#![allow(unused)]
+#![warn(clippy::slice_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/slice_as_bytes.rs

@@ -0,0 +1,35 @@
+//@run-rustfix
+#![allow(unused)]
+#![warn(clippy::slice_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/slice_as_bytes.stderr

@@ -0,0 +1,22 @@
+error: slicing a str before calling `as_bytes` results in needless UTF-8 alignment checks, and has the possiblity of panicking
+  --> $DIR/slice_as_bytes.rs:29:17
+   |
+LL |     let bytes = s[1..5].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^ help: try: `&s.as_bytes()[1..5]`
+   |
+   = note: `-D clippy::slice-as-bytes` implied by `-D warnings`
+
+error: slicing a String before calling `as_bytes` results in needless UTF-8 alignment checks, and has the possiblity of panicking
+  --> $DIR/slice_as_bytes.rs:30:17
+   |
+LL |     let bytes = string[1..].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^^^^^ help: try: `&string.as_bytes()[1..]`
+
+error: slicing a str before calling `as_bytes` results in needless UTF-8 alignment checks, and has the possiblity of panicking
+  --> $DIR/slice_as_bytes.rs:31:17
+   |
+LL |     let bytes = "consectetur adipiscing"[..=5].as_bytes();
+   |                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try: `&"consectetur adipiscing".as_bytes()[..=5]`
+
+error: aborting due to 3 previous errors
+