parsed_string_literals: new lint

This lint detects parsing of string literals into primitive types or
IP addresses when they are known correct.

Author: samueltardieu

CHANGELOG.md

@@ -6155,6 +6155,7 @@ Released 2018-09-13
 [`panic_params`]: https://rust-lang.github.io/rust-clippy/master/index.html#panic_params
 [`panicking_overflow_checks`]: https://rust-lang.github.io/rust-clippy/master/index.html#panicking_overflow_checks
 [`panicking_unwrap`]: https://rust-lang.github.io/rust-clippy/master/index.html#panicking_unwrap
+[`parsed_string_literals`]: https://rust-lang.github.io/rust-clippy/master/index.html#parsed_string_literals
 [`partial_pub_fields`]: https://rust-lang.github.io/rust-clippy/master/index.html#partial_pub_fields
 [`partialeq_ne_impl`]: https://rust-lang.github.io/rust-clippy/master/index.html#partialeq_ne_impl
 [`partialeq_to_none`]: https://rust-lang.github.io/rust-clippy/master/index.html#partialeq_to_none

clippy_lints/src/declared_lints.rs

@@ -437,6 +437,7 @@ pub static LINTS: &[&crate::LintInfo] = &[
     crate::methods::OPTION_MAP_OR_NONE_INFO,
     crate::methods::OR_FUN_CALL_INFO,
     crate::methods::OR_THEN_UNWRAP_INFO,
+    crate::methods::PARSED_STRING_LITERALS_INFO,
     crate::methods::PATH_BUF_PUSH_OVERWRITE_INFO,
     crate::methods::PATH_ENDS_WITH_EXT_INFO,
     crate::methods::RANGE_ZIP_WITH_LEN_INFO,

clippy_lints/src/lib.rs

@@ -1,9 +1,11 @@
 #![feature(array_windows)]
 #![feature(box_patterns)]
+#![feature(cow_is_borrowed)]
 #![feature(macro_metavar_expr_concat)]
 #![feature(f128)]
 #![feature(f16)]
 #![feature(if_let_guard)]
+#![feature(ip_as_octets)]
 #![feature(iter_intersperse)]
 #![feature(iter_partition_in_place)]
 #![feature(never_type)]

clippy_lints/src/methods/mod.rs

@@ -88,6 +88,7 @@ mod option_map_or_none;
 mod option_map_unwrap_or;
 mod or_fun_call;
 mod or_then_unwrap;
+mod parsed_string_literals;
 mod path_buf_push_overwrite;
 mod path_ends_with_ext;
 mod range_zip_with_len;
@@ -4528,6 +4529,36 @@ declare_clippy_lint! {
     "detect swap with a temporary value"
 }
 
+declare_clippy_lint! {
+    /// ### What it does
+    /// Checks for parsing string literals into types from the standard library
+    ///
+    /// ### Why is this bad?
+    /// Parsing known values at runtime consumes resources and forces to
+    /// unwrap the `Ok()` variant returned by `parse()`.
+    ///
+    /// ### Example
+    /// ```no_run
+    /// use std::net::Ipv4Addr;
+    ///
+    /// let number = "123".parse::<u32>().unwrap();
+    /// let addr1: Ipv4Addr = "10.2.3.4".parse().unwrap();
+    /// let addr2: Ipv4Addr = "127.0.0.1".parse().unwrap();
+    /// ```
+    /// Use instead:
+    /// ```no_run
+    /// use std::net::Ipv4Addr;
+    ///
+    /// let number = 123_u32;
+    /// let addr1: Ipv4Addr = Ipv4Addr::new(10, 2, 3, 4);
+    /// let addr2: Ipv4Addr = Ipv4Addr::LOCALHOST;
+    /// ```
+    #[clippy::version = "1.89.0"]
+    pub PARSED_STRING_LITERALS,
+    perf,
+    "known-correct literal IP address parsing"
+}
+
 #[expect(clippy::struct_excessive_bools)]
 pub struct Methods {
     avoid_breaking_exported_api: bool,
@@ -4706,6 +4737,7 @@ impl_lint_pass!(Methods => [
     MANUAL_CONTAINS,
     IO_OTHER_ERROR,
     SWAP_WITH_TEMPORARY,
+    PARSED_STRING_LITERALS,
 ]);
 
 /// Extracts a method call name, args, and `Span` of the method name.
@@ -5420,6 +5452,9 @@ impl Methods {
                         Some((sym::or, recv, [or_arg], or_span, _)) => {
                             or_then_unwrap::check(cx, expr, recv, or_arg, or_span);
                         },
+                        Some((sym::parse, recv, [], _, _)) => {
+                            parsed_string_literals::check(cx, expr, recv, self.msrv);
+                        },
                         _ => {},
                     }
                     unnecessary_literal_unwrap::check(cx, expr, recv, name, args);

clippy_lints/src/methods/parsed_string_literals.rs

@@ -0,0 +1,173 @@
+use std::borrow::Cow;
+use std::fmt::Display;
+use std::net::{Ipv4Addr, Ipv6Addr};
+use std::str::FromStr;
+
+use clippy_utils::diagnostics::span_lint_and_sugg;
+use clippy_utils::msrvs::{self, Msrv};
+use clippy_utils::source::{SpanRangeExt, str_literal_to_char_literal};
+use clippy_utils::sym;
+use clippy_utils::ty::is_type_diagnostic_item;
+use rustc_ast::LitKind;
+use rustc_errors::Applicability;
+use rustc_hir::{Expr, ExprKind};
+use rustc_lint::LateContext;
+use rustc_middle::ty::{self, Ty};
+use rustc_span::Symbol;
+
+use super::PARSED_STRING_LITERALS;
+
+pub(super) fn check(cx: &LateContext<'_>, expr: &Expr<'_>, recv: &Expr<'_>, msrv: Msrv) {
+    if let ExprKind::Lit(lit) = recv.kind
+        && let LitKind::Str(lit, _) = lit.node
+    {
+        let ty = cx.typeck_results().expr_ty(expr);
+        if !check_primitive(cx, expr, lit, ty, recv) {
+            check_ipaddr(cx, expr, lit, ty, msrv);
+        }
+    }
+}
+
+fn check_primitive(cx: &LateContext<'_>, expr: &Expr<'_>, lit: Symbol, ty: Ty<'_>, strlit: &Expr<'_>) -> bool {
+    macro_rules! number {
+        ($kind:ident, $expr:expr, $msg:expr, [$($subkind:ident => $ty:ident),*$(,)?]$(,)?) => {{
+            match $expr {
+                $(ty::$kind::$subkind => (try_parse::<$ty>(lit, Some(stringify!($ty))), $msg),)*
+                #[allow(unreachable_patterns)]
+                _ => return false,
+            }
+        }};
+    }
+
+    let mut app = Applicability::MachineApplicable;
+    if let (Some(subst), entity) = match ty.kind() {
+        ty::Int(int_ty) => number!(IntTy, int_ty, "a signed integer",
+                                     [Isize => isize, I8 => i8, I16 => i16, I32 => i32, I64 => i64, I128 => i128]),
+        ty::Uint(uint_ty) => number!(UintTy, uint_ty, "an unsigned integer",
+                                        [Usize => usize, U8 => u8, U16 => u16, U32 => u32, U64 => u64, U128 => u128]),
+        // FIXME: ignore `f16` and `f128` for now as they cannot use the default formatter
+        ty::Float(float_ty) => number!(FloatTy, float_ty, "a real number",
+                                           [F32 => f32, F64 => f64]),
+        ty::Bool => (try_parse::<bool>(lit, None), "a boolean"),
+        ty::Char => (str_literal_to_char_literal(cx, strlit, &mut app, false), "a character"),
+        _ => return false,
+    } {
+        maybe_emit_lint(cx, expr, false, entity, subst.into(), app);
+    }
+    true
+}
+
+fn check_ipaddr(cx: &LateContext<'_>, expr: &Expr<'_>, lit: Symbol, ty: Ty<'_>, msrv: Msrv) {
+    static IPV4_ENTITY: &str = "an IPv4 address";
+    static IPV6_ENTITY: &str = "an IPv6 address";
+    let ipaddr_consts_available = || msrv.meets(cx, msrvs::IPADDR_CONSTANTS);
+    if is_type_diagnostic_item(cx, ty, sym::Ipv4Addr)
+        && let Some(sugg) = ipv4_subst(lit, ipaddr_consts_available())
+    {
+        maybe_emit_lint(
+            cx,
+            expr,
+            sugg.is_borrowed(),
+            IPV4_ENTITY,
+            sugg,
+            Applicability::MaybeIncorrect,
+        );
+    } else if is_type_diagnostic_item(cx, ty, sym::Ipv6Addr)
+        && let Some(sugg) = ipv6_subst(lit, ipaddr_consts_available())
+    {
+        maybe_emit_lint(
+            cx,
+            expr,
+            sugg.is_borrowed(),
+            IPV6_ENTITY,
+            sugg,
+            Applicability::MaybeIncorrect,
+        );
+    } else if is_type_diagnostic_item(cx, ty, sym::IpAddr) {
+        let with_consts = ipaddr_consts_available();
+        if let Some(sugg) = ipv4_subst(lit, with_consts) {
+            maybe_emit_lint(
+                cx,
+                expr,
+                sugg.is_borrowed(),
+                IPV4_ENTITY,
+                format!("IpAddr::V4({sugg})").into(),
+                Applicability::MaybeIncorrect,
+            );
+        } else if let Some(sugg) = ipv6_subst(lit, with_consts) {
+            maybe_emit_lint(
+                cx,
+                expr,
+                sugg.is_borrowed(),
+                IPV6_ENTITY,
+                format!("IpAddr::V6({sugg})").into(),
+                Applicability::MaybeIncorrect,
+            );
+        }
+    }
+}
+
+/// Suggests a replacement if `addr` is a correct IPv4 address
+fn ipv4_subst(addr: Symbol, with_consts: bool) -> Option<Cow<'static, str>> {
+    addr.as_str().parse().ok().map(|ipv4: Ipv4Addr| {
+        if with_consts && ipv4.as_octets() == &[127, 0, 0, 1] {
+            "Ipv4Addr::LOCALHOST".into()
+        } else if with_consts && ipv4.is_broadcast() {
+            "Ipv4Addr::BROADCAST".into()
+        } else if with_consts && ipv4.is_unspecified() {
+            "Ipv4Addr::UNSPECIFIED".into()
+        } else {
+            let ipv4 = ipv4.as_octets();
+            format!("Ipv4Addr::new({}, {}, {}, {})", ipv4[0], ipv4[1], ipv4[2], ipv4[3]).into()
+        }
+    })
+}
+
+/// Suggests a replacement if `addr` is a correct IPv6 address
+fn ipv6_subst(addr: Symbol, with_consts: bool) -> Option<Cow<'static, str>> {
+    addr.as_str().parse().ok().map(|ipv6: Ipv6Addr| {
+        if with_consts && ipv6.is_loopback() {
+            "Ipv6Addr::LOCALHOST".into()
+        } else if with_consts && ipv6.is_unspecified() {
+            "Ipv6Addr::UNSPECIFIED".into()
+        } else {
+            format!(
+                "Ipv6Addr::new([{}])",
+                ipv6.segments()
+                    .map(|n| if n < 2 { n.to_string() } else { format!("{n:#x}") })
+                    .join(", ")
+            )
+            .into()
+        }
+    })
+}
+
+fn try_parse<T: FromStr + Display>(lit: Symbol, suffix: Option<&str>) -> Option<String> {
+    lit.as_str()
+        .parse::<T>()
+        .ok()
+        .map(|_| suffix.map_or_else(|| lit.to_string(), |suffix| format!("{lit}_{suffix}")))
+}
+
+/// Emit the lint if the length of `sugg` is no longer than the original `expr` span, or if `force`
+/// is set.
+fn maybe_emit_lint(
+    cx: &LateContext<'_>,
+    expr: &Expr<'_>,
+    force: bool,
+    entity: &str,
+    sugg: Cow<'_, str>,
+    applicability: Applicability,
+) {
+    if force || expr.span.check_source_text(cx, |snip| snip.len() >= sugg.len()) {
+        span_lint_and_sugg(
+            cx,
+            PARSED_STRING_LITERALS,
+            expr.span,
+            format!("unnecessary runtime parsing of {entity}"),
+            "use",
+            sugg.into(),
+            applicability,
+        );
+    }
+}

clippy_utils/src/msrvs.rs

@@ -68,7 +68,7 @@ msrv_aliases! {
     1,33,0 { UNDERSCORE_IMPORTS }
     1,32,0 { CONST_IS_POWER_OF_TWO }
     1,31,0 { OPTION_REPLACE }
-    1,30,0 { ITERATOR_FIND_MAP, TOOL_ATTRIBUTES }
+    1,30,0 { ITERATOR_FIND_MAP, TOOL_ATTRIBUTES, IPADDR_CONSTANTS }
     1,29,0 { ITER_FLATTEN }
     1,28,0 { FROM_BOOL, REPEAT_WITH, SLICE_FROM_REF }
     1,27,0 { ITERATOR_TRY_FOLD }

tests/ui/parsed_string_literals.fixed

@@ -0,0 +1,83 @@
+#![warn(clippy::parsed_string_literals)]
+
+use std::net::{IpAddr, Ipv4Addr, Ipv6Addr};
+
+fn main() {
+    _ = Ipv4Addr::new(137, 194, 161, 2);
+    //~^ parsed_string_literals
+
+    _ = Ipv4Addr::LOCALHOST;
+    //~^ parsed_string_literals
+
+    _ = Ipv4Addr::BROADCAST;
+    //~^ parsed_string_literals
+
+    _ = Ipv4Addr::UNSPECIFIED;
+    //~^ parsed_string_literals
+
+    // Wrong address family
+    _ = "::1".parse::<Ipv4Addr>().unwrap();
+    _ = "127.0.0.1".parse::<Ipv6Addr>().unwrap();
+
+    _ = Ipv6Addr::LOCALHOST;
+    //~^ parsed_string_literals
+
+    _ = Ipv6Addr::UNSPECIFIED;
+    //~^ parsed_string_literals
+
+    _ = IpAddr::V6(Ipv6Addr::LOCALHOST);
+    //~^ parsed_string_literals
+
+    _ = IpAddr::V6(Ipv6Addr::UNSPECIFIED);
+    //~^ parsed_string_literals
+
+    // The substition text would be larger than the original and wouldn't use constants
+    _ = "2a04:8ec0:0:47::131".parse::<Ipv6Addr>().unwrap();
+    _ = "2a04:8ec0:0:47::131".parse::<IpAddr>().unwrap();
+
+    _ = true;
+    //~^ parsed_string_literals
+    _ = false;
+    //~^ parsed_string_literals
+
+    let _: i64 = -17_i64;
+    //~^ parsed_string_literals
+    _ = 10_usize;
+    //~^ parsed_string_literals
+    _ = 1.23_f32;
+    //~^ parsed_string_literals
+    _ = 1.2300_f32;
+    //~^ parsed_string_literals
+    _ = 'c';
+    //~^ parsed_string_literals
+    _ = '"';
+    //~^ parsed_string_literals
+    _ = '\'';
+    //~^ parsed_string_literals
+
+    // Check that the original form is preserved ('🦀' == '\u{1f980}')
+    _ = '\u{1f980}';
+    //~^ parsed_string_literals
+    _ = '🦀';
+    //~^ parsed_string_literals
+
+    // Do not lint invalid values
+    _ = "-10".parse::<usize>().unwrap();
+
+    // Do not lint content or code coming from macros
+    macro_rules! mac {
+        (str) => {
+            "10"
+        };
+        (parse $l:literal) => {
+            $l.parse::<u32>().unwrap()
+        };
+    }
+    _ = mac!(str).parse::<u32>().unwrap();
+    _ = mac!(parse "10");
+}
+
+#[clippy::msrv = "1.29"]
+fn msrv_under() {
+    _ = "::".parse::<IpAddr>().unwrap();
+}