Ensure Fluent messages are in alphabetical order

3 files changed, 74 insertions, 0 deletions

diff --git a/src/tools/tidy/src/fluent_alphabetical.rs b/src/tools/tidy/src/fluent_alphabetical.rs
new file mode 100644
index 00000000000..5f8eaebf531
--- /dev/null
+++ b/src/tools/tidy/src/fluent_alphabetical.rs
@@ -0,0 +1,72 @@
+//! Checks that all Flunt files have messages in alphabetical order
+
+use crate::walk::{filter_dirs, walk};
+use std::{fs::OpenOptions, io::Write, path::Path};
+
+use regex::Regex;
+
+lazy_static::lazy_static! {
+    static ref MESSAGE: Regex = Regex::new(r#"(?m)^([a-zA-Z0-9_]+)\s*=\s*"#).unwrap();
+}
+
+fn filter_fluent(path: &Path) -> bool {
+    if let Some(ext) = path.extension() { ext.to_str() != Some("ftl") } else { true }
+}
+
+fn check_alphabetic(filename: &str, fluent: &str, bad: &mut bool) {
+    let mut matches = MESSAGE.captures_iter(fluent).peekable();
+    while let Some(m) = matches.next() {
+        if let Some(next) = matches.peek() {
+            let name = m.get(1).unwrap();
+            let next = next.get(1).unwrap();
+            if name.as_str() > next.as_str() {
+                tidy_error!(
+                    bad,
+                    "{filename}: message `{}` appears before `{}`, but is alphabetically later than it
+run tidy with `--bless` to sort the file correctly",
+                    name.as_str(),
+                    next.as_str()
+                );
+            }
+        } else {
+            break;
+        }
+    }
+}
+
+fn sort_messages(fluent: &str) -> String {
+    let mut chunks = vec![];
+    let mut cur = String::new();
+    for line in fluent.lines() {
+        if MESSAGE.is_match(line) {
+            chunks.push(std::mem::take(&mut cur));
+        }
+        cur += line;
+        cur.push('\n');
+    }
+    chunks.push(cur);
+    chunks.sort();
+    let mut out = chunks.join("");
+    out = out.trim().to_string();
+    out.push('\n');
+    out
+}
+
+pub fn check(path: &Path, bless: bool, bad: &mut bool) {
+    walk(
+        path,
+        |path, is_dir| filter_dirs(path) || (!is_dir && filter_fluent(path)),
+        &mut |ent, contents| {
+            if bless {
+                let sorted = sort_messages(contents);
+                if sorted != contents {
+                    let mut f =
+                        OpenOptions::new().write(true).truncate(true).open(ent.path()).unwrap();
+                    f.write(sorted.as_bytes()).unwrap();
+                }
+            } else {
+                check_alphabetic(ent.path().to_str().unwrap(), contents, bad);
+            }
+        },
+    );
+}
diff --git a/src/tools/tidy/src/lib.rs b/src/tools/tidy/src/lib.rs
index 35000320d1a..e467514a7a3 100644
--- a/src/tools/tidy/src/lib.rs
+++ b/src/tools/tidy/src/lib.rs
@@ -59,6 +59,7 @@ pub mod edition;
 pub mod error_codes;
 pub mod extdeps;
 pub mod features;
+pub mod fluent_alphabetical;
 pub mod mir_opt_tests;
 pub mod pal;
 pub mod primitive_docs;
diff --git a/src/tools/tidy/src/main.rs b/src/tools/tidy/src/main.rs
index f59406c404b..1c4d96c321c 100644
--- a/src/tools/tidy/src/main.rs
+++ b/src/tools/tidy/src/main.rs
@@ -96,6 +96,7 @@ fn main() {
 
         // Checks that only make sense for the compiler.
         check!(error_codes, &root_path, &[&compiler_path, &librustdoc_path], verbose);
+        check!(fluent_alphabetical, &compiler_path, bless);
 
         // Checks that only make sense for the std libs.
         check!(pal, &library_path);