Auto merge of #8001 - Jarcho:unprefixed_strlen, r=giraffate

Improve `strlen_on_c_string`

fixes: #7436

changelog: lint `strlen_on_c_string` when used without a fully-qualified path
changelog: suggest removing the surrounding unsafe block for `strlen_on_c_string` when possible

7 files changed, 197 insertions, 47 deletions

diff --git a/clippy_lints/src/strlen_on_c_strings.rs b/clippy_lints/src/strlen_on_c_strings.rs
index be7431f11aa..fee01fb0bd1 100644
--- a/clippy_lints/src/strlen_on_c_strings.rs
+++ b/clippy_lints/src/strlen_on_c_strings.rs
@@ -1,13 +1,14 @@
 use clippy_utils::diagnostics::span_lint_and_sugg;
-use clippy_utils::paths;
-use clippy_utils::source::snippet_with_macro_callsite;
-use clippy_utils::ty::{is_type_diagnostic_item, is_type_ref_to_diagnostic_item};
+use clippy_utils::source::snippet_with_context;
+use clippy_utils::ty::is_type_diagnostic_item;
+use clippy_utils::visitors::is_expr_unsafe;
+use clippy_utils::{get_parent_node, match_libc_symbol};
 use if_chain::if_chain;
 use rustc_errors::Applicability;
-use rustc_hir as hir;
+use rustc_hir::{Block, BlockCheckMode, Expr, ExprKind, Node, UnsafeSource};
 use rustc_lint::{LateContext, LateLintPass};
 use rustc_session::{declare_lint_pass, declare_tool_lint};
-use rustc_span::symbol::{sym, Symbol};
+use rustc_span::symbol::sym;
 
 declare_clippy_lint! {
     /// ### What it does
@@ -39,29 +40,35 @@ declare_clippy_lint! {
 declare_lint_pass!(StrlenOnCStrings => [STRLEN_ON_C_STRINGS]);
 
 impl LateLintPass<'tcx> for StrlenOnCStrings {
-    fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &'tcx hir::Expr<'_>) {
-        if expr.span.from_expansion() {
-            return;
-        }
-
+    fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &'tcx Expr<'_>) {
         if_chain! {
-            if let hir::ExprKind::Call(func, [recv]) = expr.kind;
-            if let hir::ExprKind::Path(hir::QPath::Resolved(_, path)) = func.kind;
-
-            if (&paths::LIBC_STRLEN).iter().map(|x| Symbol::intern(x)).eq(
-                path.segments.iter().map(|seg| seg.ident.name));
-            if let hir::ExprKind::MethodCall(path, _, args, _) = recv.kind;
-            if args.len() == 1;
-            if !args.iter().any(|e| e.span.from_expansion());
+            if !expr.span.from_expansion();
+            if let ExprKind::Call(func, [recv]) = expr.kind;
+            if let ExprKind::Path(path) = &func.kind;
+            if let Some(did) = cx.qpath_res(path, func.hir_id).opt_def_id();
+            if match_libc_symbol(cx, did, "strlen");
+            if let ExprKind::MethodCall(path, _, [self_arg], _) = recv.kind;
+            if !recv.span.from_expansion();
             if path.ident.name == sym::as_ptr;
             then {
-                let cstring = &args[0];
-                let ty = cx.typeck_results().expr_ty(cstring);
-                let val_name = snippet_with_macro_callsite(cx, cstring.span, "..");
-                let sugg = if is_type_diagnostic_item(cx, ty, sym::cstring_type){
-                    format!("{}.as_bytes().len()", val_name)
-                } else if is_type_ref_to_diagnostic_item(cx, ty, sym::CStr){
-                    format!("{}.to_bytes().len()", val_name)
+                let ctxt = expr.span.ctxt();
+                let span = match get_parent_node(cx.tcx, expr.hir_id) {
+                    Some(Node::Block(&Block {
+                        rules: BlockCheckMode::UnsafeBlock(UnsafeSource::UserProvided), span, ..
+                    }))
+                    if span.ctxt() == ctxt && !is_expr_unsafe(cx, self_arg) => {
+                        span
+                    }
+                    _ => expr.span,
+                };
+
+                let ty = cx.typeck_results().expr_ty(self_arg).peel_refs();
+                let mut app = Applicability::MachineApplicable;
+                let val_name = snippet_with_context(cx, self_arg.span, ctxt, "..", &mut app).0;
+                let method_name = if is_type_diagnostic_item(cx, ty, sym::cstring_type) {
+                    "as_bytes"
+                } else if is_type_diagnostic_item(cx, ty, sym::CStr) {
+                    "to_bytes"
                 } else {
                     return;
                 };
@@ -69,11 +76,11 @@ impl LateLintPass<'tcx> for StrlenOnCStrings {
                 span_lint_and_sugg(
                     cx,
                     STRLEN_ON_C_STRINGS,
-                    expr.span,
+                    span,
                     "using `libc::strlen` on a `CString` or `CStr` value",
-                    "try this (you might also need to get rid of `unsafe` block in some cases):",
-                    sugg,
-                    Applicability::Unspecified // Sometimes unnecessary `unsafe` block
+                    "try this",
+                    format!("{}.{}().len()", val_name, method_name),
+                    app,
                 );
             }
         }
diff --git a/clippy_utils/src/lib.rs b/clippy_utils/src/lib.rs
index 3fdea55aaa1..f011380c127 100644
--- a/clippy_utils/src/lib.rs
+++ b/clippy_utils/src/lib.rs
@@ -1597,6 +1597,14 @@ pub fn match_def_path<'tcx>(cx: &LateContext<'tcx>, did: DefId, syms: &[&str]) -
     syms.iter().map(|x| Symbol::intern(x)).eq(path.iter().copied())
 }
 
+/// Checks if the given `DefId` matches the `libc` item.
+pub fn match_libc_symbol(cx: &LateContext<'_>, did: DefId, name: &str) -> bool {
+    let path = cx.get_def_path(did);
+    // libc is meant to be used as a flat list of names, but they're all actually defined in different
+    // modules based on the target platform. Ignore everything but crate name and the item name.
+    path.first().map_or(false, |s| s.as_str() == "libc") && path.last().map_or(false, |s| s.as_str() == name)
+}
+
 pub fn match_panic_call(cx: &LateContext<'_>, expr: &'tcx Expr<'_>) -> Option<&'tcx Expr<'tcx>> {
     if let ExprKind::Call(func, [arg]) = expr.kind {
         expr_path_res(cx, func)
diff --git a/clippy_utils/src/paths.rs b/clippy_utils/src/paths.rs
index b04d1736426..3ffa548e470 100644
--- a/clippy_utils/src/paths.rs
+++ b/clippy_utils/src/paths.rs
@@ -86,7 +86,6 @@ pub const ITERTOOLS_NEXT_TUPLE: [&str; 3] = ["itertools", "Itertools", "next_tup
 pub const KW_MODULE: [&str; 3] = ["rustc_span", "symbol", "kw"];
 #[cfg(feature = "internal-lints")]
 pub const LATE_CONTEXT: [&str; 2] = ["rustc_lint", "LateContext"];
-pub const LIBC_STRLEN: [&str; 2] = ["libc", "strlen"];
 #[cfg(any(feature = "internal-lints", feature = "metadata-collector-lint"))]
 pub const LINT: [&str; 2] = ["rustc_lint_defs", "Lint"];
 pub const MEM_DISCRIMINANT: [&str; 3] = ["core", "mem", "discriminant"];
diff --git a/clippy_utils/src/visitors.rs b/clippy_utils/src/visitors.rs
index 823df5cb751..4bfd3c64b9c 100644
--- a/clippy_utils/src/visitors.rs
+++ b/clippy_utils/src/visitors.rs
@@ -1,8 +1,10 @@
 use crate::path_to_local_id;
 use rustc_hir as hir;
 use rustc_hir::def::{DefKind, Res};
-use rustc_hir::intravisit::{self, walk_expr, NestedVisitorMap, Visitor};
-use rustc_hir::{Arm, Block, Body, BodyId, Expr, ExprKind, HirId, Stmt, UnOp};
+use rustc_hir::intravisit::{self, walk_block, walk_expr, NestedVisitorMap, Visitor};
+use rustc_hir::{
+    Arm, Block, BlockCheckMode, Body, BodyId, Expr, ExprKind, HirId, ItemId, ItemKind, Stmt, UnOp, Unsafety,
+};
 use rustc_lint::LateContext;
 use rustc_middle::hir::map::Map;
 use rustc_middle::ty;
@@ -317,3 +319,64 @@ pub fn is_const_evaluatable(cx: &LateContext<'tcx>, e: &'tcx Expr<'_>) -> bool {
     v.visit_expr(e);
     v.is_const
 }
+
+/// Checks if the given expression performs an unsafe operation outside of an unsafe block.
+pub fn is_expr_unsafe(cx: &LateContext<'tcx>, e: &'tcx Expr<'_>) -> bool {
+    struct V<'a, 'tcx> {
+        cx: &'a LateContext<'tcx>,
+        is_unsafe: bool,
+    }
+    impl<'tcx> Visitor<'tcx> for V<'_, 'tcx> {
+        type Map = Map<'tcx>;
+        fn nested_visit_map(&mut self) -> NestedVisitorMap<Self::Map> {
+            NestedVisitorMap::OnlyBodies(self.cx.tcx.hir())
+        }
+        fn visit_expr(&mut self, e: &'tcx Expr<'_>) {
+            if self.is_unsafe {
+                return;
+            }
+            match e.kind {
+                ExprKind::Unary(UnOp::Deref, e) if self.cx.typeck_results().expr_ty(e).is_unsafe_ptr() => {
+                    self.is_unsafe = true;
+                },
+                ExprKind::MethodCall(..)
+                    if self
+                        .cx
+                        .typeck_results()
+                        .type_dependent_def_id(e.hir_id)
+                        .map_or(false, |id| self.cx.tcx.fn_sig(id).unsafety() == Unsafety::Unsafe) =>
+                {
+                    self.is_unsafe = true;
+                },
+                ExprKind::Call(func, _) => match *self.cx.typeck_results().expr_ty(func).peel_refs().kind() {
+                    ty::FnDef(id, _) if self.cx.tcx.fn_sig(id).unsafety() == Unsafety::Unsafe => self.is_unsafe = true,
+                    ty::FnPtr(sig) if sig.unsafety() == Unsafety::Unsafe => self.is_unsafe = true,
+                    _ => walk_expr(self, e),
+                },
+                ExprKind::Path(ref p)
+                    if self
+                        .cx
+                        .qpath_res(p, e.hir_id)
+                        .opt_def_id()
+                        .map_or(false, |id| self.cx.tcx.is_mutable_static(id)) =>
+                {
+                    self.is_unsafe = true;
+                },
+                _ => walk_expr(self, e),
+            }
+        }
+        fn visit_block(&mut self, b: &'tcx Block<'_>) {
+            if !matches!(b.rules, BlockCheckMode::UnsafeBlock(_)) {
+                walk_block(self, b);
+            }
+        }
+        fn visit_nested_item(&mut self, id: ItemId) {
+            if let ItemKind::Impl(i) = &self.cx.tcx.hir().item(id).kind {
+                self.is_unsafe = i.unsafety == Unsafety::Unsafe;
+            }
+        }
+    }
+    let mut v = V { cx, is_unsafe: false };
+    v.visit_expr(e);
+    v.is_unsafe
+}
diff --git a/tests/ui/strlen_on_c_strings.fixed b/tests/ui/strlen_on_c_strings.fixed
new file mode 100644
index 00000000000..947a59bcc02
--- /dev/null
+++ b/tests/ui/strlen_on_c_strings.fixed
@@ -0,0 +1,34 @@
+// run-rustfix
+
+#![warn(clippy::strlen_on_c_strings)]
+#![allow(dead_code)]
+#![feature(rustc_private)]
+extern crate libc;
+
+#[allow(unused)]
+use libc::strlen;
+use std::ffi::{CStr, CString};
+
+fn main() {
+    // CString
+    let cstring = CString::new("foo").expect("CString::new failed");
+    let _ = cstring.as_bytes().len();
+
+    // CStr
+    let cstr = CStr::from_bytes_with_nul(b"foo\0").expect("CStr::from_bytes_with_nul failed");
+    let _ = cstr.to_bytes().len();
+
+    let _ = cstr.to_bytes().len();
+
+    let pcstr: *const &CStr = &cstr;
+    let _ = unsafe { (*pcstr).to_bytes().len() };
+
+    unsafe fn unsafe_identity<T>(x: T) -> T {
+        x
+    }
+    let _ = unsafe { unsafe_identity(cstr).to_bytes().len() };
+    let _ = unsafe { unsafe_identity(cstr) }.to_bytes().len();
+
+    let f: unsafe fn(_) -> _ = unsafe_identity;
+    let _ = unsafe { f(cstr).to_bytes().len() };
+}
diff --git a/tests/ui/strlen_on_c_strings.rs b/tests/ui/strlen_on_c_strings.rs
index 21902fa8483..1237f1ab03a 100644
--- a/tests/ui/strlen_on_c_strings.rs
+++ b/tests/ui/strlen_on_c_strings.rs
@@ -1,16 +1,34 @@
+// run-rustfix
+
 #![warn(clippy::strlen_on_c_strings)]
 #![allow(dead_code)]
 #![feature(rustc_private)]
 extern crate libc;
 
+#[allow(unused)]
+use libc::strlen;
 use std::ffi::{CStr, CString};
 
 fn main() {
     // CString
     let cstring = CString::new("foo").expect("CString::new failed");
-    let len = unsafe { libc::strlen(cstring.as_ptr()) };
+    let _ = unsafe { libc::strlen(cstring.as_ptr()) };
 
     // CStr
     let cstr = CStr::from_bytes_with_nul(b"foo\0").expect("CStr::from_bytes_with_nul failed");
-    let len = unsafe { libc::strlen(cstr.as_ptr()) };
+    let _ = unsafe { libc::strlen(cstr.as_ptr()) };
+
+    let _ = unsafe { strlen(cstr.as_ptr()) };
+
+    let pcstr: *const &CStr = &cstr;
+    let _ = unsafe { strlen((*pcstr).as_ptr()) };
+
+    unsafe fn unsafe_identity<T>(x: T) -> T {
+        x
+    }
+    let _ = unsafe { strlen(unsafe_identity(cstr).as_ptr()) };
+    let _ = unsafe { strlen(unsafe { unsafe_identity(cstr) }.as_ptr()) };
+
+    let f: unsafe fn(_) -> _ = unsafe_identity;
+    let _ = unsafe { strlen(f(cstr).as_ptr()) };
 }
diff --git a/tests/ui/strlen_on_c_strings.stderr b/tests/ui/strlen_on_c_strings.stderr
index e0ca511557c..296268a5f1d 100644
--- a/tests/ui/strlen_on_c_strings.stderr
+++ b/tests/ui/strlen_on_c_strings.stderr
@@ -1,25 +1,46 @@
 error: using `libc::strlen` on a `CString` or `CStr` value
-  --> $DIR/strlen_on_c_strings.rs:11:24
+  --> $DIR/strlen_on_c_strings.rs:15:13
    |
-LL |     let len = unsafe { libc::strlen(cstring.as_ptr()) };
-   |                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+LL |     let _ = unsafe { libc::strlen(cstring.as_ptr()) };
+   |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `cstring.as_bytes().len()`
    |
    = note: `-D clippy::strlen-on-c-strings` implied by `-D warnings`
-help: try this (you might also need to get rid of `unsafe` block in some cases):
+
+error: using `libc::strlen` on a `CString` or `CStr` value
+  --> $DIR/strlen_on_c_strings.rs:19:13
+   |
+LL |     let _ = unsafe { libc::strlen(cstr.as_ptr()) };
+   |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `cstr.to_bytes().len()`
+
+error: using `libc::strlen` on a `CString` or `CStr` value
+  --> $DIR/strlen_on_c_strings.rs:21:13
+   |
+LL |     let _ = unsafe { strlen(cstr.as_ptr()) };
+   |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `cstr.to_bytes().len()`
+
+error: using `libc::strlen` on a `CString` or `CStr` value
+  --> $DIR/strlen_on_c_strings.rs:24:22
    |
-LL |     let len = unsafe { cstring.as_bytes().len() };
-   |                        ~~~~~~~~~~~~~~~~~~~~~~~~
+LL |     let _ = unsafe { strlen((*pcstr).as_ptr()) };
+   |                      ^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `(*pcstr).to_bytes().len()`
 
 error: using `libc::strlen` on a `CString` or `CStr` value
-  --> $DIR/strlen_on_c_strings.rs:15:24
+  --> $DIR/strlen_on_c_strings.rs:29:22
    |
-LL |     let len = unsafe { libc::strlen(cstr.as_ptr()) };
-   |                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+LL |     let _ = unsafe { strlen(unsafe_identity(cstr).as_ptr()) };
+   |                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `unsafe_identity(cstr).to_bytes().len()`
+
+error: using `libc::strlen` on a `CString` or `CStr` value
+  --> $DIR/strlen_on_c_strings.rs:30:13
    |
-help: try this (you might also need to get rid of `unsafe` block in some cases):
+LL |     let _ = unsafe { strlen(unsafe { unsafe_identity(cstr) }.as_ptr()) };
+   |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `unsafe { unsafe_identity(cstr) }.to_bytes().len()`
+
+error: using `libc::strlen` on a `CString` or `CStr` value
+  --> $DIR/strlen_on_c_strings.rs:33:22
    |
-LL |     let len = unsafe { cstr.to_bytes().len() };
-   |                        ~~~~~~~~~~~~~~~~~~~~~
+LL |     let _ = unsafe { strlen(f(cstr).as_ptr()) };
+   |                      ^^^^^^^^^^^^^^^^^^^^^^^^ help: try this: `f(cstr).to_bytes().len()`
 
-error: aborting due to 2 previous errors
+error: aborting due to 7 previous errors