make tidy-alphabetical use a natural sort

src/bootstrap/src/core/build_steps/dist.rs

@@ -1036,18 +1036,18 @@ impl Step for PlainSourceTarball {
         let src_files = [
             // tidy-alphabetical-start
             ".gitmodules",
-            "bootstrap.example.toml",
-            "Cargo.lock",
-            "Cargo.toml",
-            "configure",
             "CONTRIBUTING.md",
             "COPYRIGHT",
+            "Cargo.lock",
+            "Cargo.toml",
             "LICENSE-APACHE",
-            "license-metadata.json",
             "LICENSE-MIT",
             "README.md",
             "RELEASES.md",
             "REUSE.toml",
+            "bootstrap.example.toml",
+            "configure",
+            "license-metadata.json",
             "x",
             "x.ps1",
             "x.py",

src/tools/tidy/src/alphabetical.rs

@@ -19,7 +19,9 @@
 //! If a line ends with an opening delimiter, we effectively join the following line to it before
 //! checking it. E.g. `foo(\nbar)` is treated like `foo(bar)`.
 
+use std::cmp::Ordering;
 use std::fmt::Display;
+use std::iter::Peekable;
 use std::path::Path;
 
 use crate::walk::{filter_dirs, walk};
@@ -99,9 +101,9 @@ fn check_section<'a>(
             continue;
         }
 
-        let prev_line_trimmed_lowercase = prev_line.trim_start_matches(' ').to_lowercase();
+        let prev_line_trimmed_lowercase = prev_line.trim_start_matches(' ');
 
-        if trimmed_line.to_lowercase() < prev_line_trimmed_lowercase {
+        if version_sort(&trimmed_line, &prev_line_trimmed_lowercase).is_lt() {
             tidy_error_ext!(err, bad, "{file}:{}: line not in alphabetical order", idx + 1);
         }
 
@@ -143,3 +145,56 @@ pub fn check(path: &Path, bad: &mut bool) {
         check_lines(file, lines, &mut crate::tidy_error, bad)
     });
 }
+
+fn consume_numeric_prefix<I: Iterator<Item = char>>(it: &mut Peekable<I>) -> String {
+    let mut result = String::new();
+
+    while let Some(&c) = it.peek() {
+        if !c.is_numeric() {
+            break;
+        }
+
+        result.push(c);
+        it.next();
+    }
+
+    result
+}
+
+// A sorting function that is case-sensitive, and sorts sequences of digits by their numeric value,
+// so that `9` sorts before `12`.
+fn version_sort(a: &str, b: &str) -> Ordering {
+    let mut it1 = a.chars().peekable();
+    let mut it2 = b.chars().peekable();
+
+    while let (Some(x), Some(y)) = (it1.peek(), it2.peek()) {
+        match (x.is_numeric(), y.is_numeric()) {
+            (true, true) => {
+                let num1: String = consume_numeric_prefix(it1.by_ref());
+                let num2: String = consume_numeric_prefix(it2.by_ref());
+
+                let int1: u64 = num1.parse().unwrap();
+                let int2: u64 = num2.parse().unwrap();
+
+                // Compare strings when the numeric value is equal to handle "00" versus "0".
+                match int1.cmp(&int2).then_with(|| num1.cmp(&num2)) {
+                    Ordering::Equal => continue,
+                    different => return different,
+                }
+            }
+            (false, false) => match x.cmp(y) {
+                Ordering::Equal => {
+                    it1.next();
+                    it2.next();
+                    continue;
+                }
+                different => return different,
+            },
+            (false, true) | (true, false) => {
+                return x.cmp(y);
+            }
+        }
+    }
+
+    it1.next().cmp(&it2.next())
+}

src/tools/tidy/src/alphabetical/tests.rs

@@ -3,6 +3,7 @@ use std::str::from_utf8;
 
 use super::*;
 
+#[track_caller]
 fn test(lines: &str, name: &str, expected_msg: &str, expected_bad: bool) {
     let mut actual_msg = Vec::new();
     let mut actual_bad = false;
@@ -15,10 +16,12 @@ fn test(lines: &str, name: &str, expected_msg: &str, expected_bad: bool) {
     assert_eq!(expected_bad, actual_bad);
 }
 
+#[track_caller]
 fn good(lines: &str) {
     test(lines, "good", "", false);
 }
 
+#[track_caller]
 fn bad(lines: &str, expected_msg: &str) {
     test(lines, "bad", expected_msg, true);
 }
@@ -187,3 +190,147 @@ fn test_double_end() {
     ";
     bad(lines, "bad:5 found `tidy-alphabetical-end` expecting `tidy-alphabetical-start`");
 }
+
+#[test]
+fn test_numeric_good() {
+    good(
+        "\
+        # tidy-alphabetical-start
+        rustc_ast = { path = \"../rustc_ast\" }
+        rustc_ast_lowering = { path = \"../rustc_ast_lowering\" }
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        fp-armv8
+        fp16
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        item1
+        item2
+        item10
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        foo
+        foo_
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        foo-bar
+        foo_bar
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        sme-lutv2
+        sme2
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        v5te
+        v6
+        v6k
+        v6t2
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        zve64d
+        zve64f
+        # tidy-alphabetical-end
+    ",
+    );
+
+    // Case is significant.
+    good(
+        "\
+        # tidy-alphabetical-start
+        _ZYXW
+        _abcd
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        v0
+        v00
+        v000
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        w005s09t
+        w5s009t
+        # tidy-alphabetical-end
+    ",
+    );
+
+    good(
+        "\
+        # tidy-alphabetical-start
+        v0s
+        v00t
+        # tidy-alphabetical-end
+    ",
+    );
+}
+
+#[test]
+fn test_numeric_bad() {
+    let lines = "\
+        # tidy-alphabetical-start
+        item1
+        item10
+        item2
+        # tidy-alphabetical-end
+    ";
+    bad(lines, "bad:4: line not in alphabetical order");
+
+    let lines = "\
+        # tidy-alphabetical-start
+        zve64f
+        zve64d
+        # tidy-alphabetical-end
+    ";
+    bad(lines, "bad:3: line not in alphabetical order");
+
+    let lines = "\
+        # tidy-alphabetical-start
+        000
+        00
+        # tidy-alphabetical-end
+    ";
+    bad(lines, "bad:3: line not in alphabetical order");
+}