| 1 | //! Post-edit formatting normalization for Rust files (#6205). |
| 2 | //! |
| 3 | //! Model-generated edits rarely match `rustfmt` output exactly. Once an |
| 4 | //! unformatted edit lands, the *next* turn's `old_string` or patch context was |
| 5 | //! written against text that `cargo fmt` is about to move, so anchors drift and |
| 6 | //! the follow-up edit fails to match. Normalizing at edit time keeps anchors |
| 7 | //! stable for the rest of the session, and the tool result returns the |
| 8 | //! normalized text, so what the model remembers writing is what is on disk. |
| 9 | //! |
| 10 | //! `rustfmt` is the formatter, not `prettyplease`: `cargo fmt --check` is this |
| 11 | //! repository's actual gate, and a second formatter with its own opinions would |
| 12 | //! produce files that pass the edit path and fail the gate. Shelling out also |
| 13 | //! picks up the project's own `rustfmt.toml` — an in-process pretty-printer |
| 14 | //! cannot. |
| 15 | //! |
| 16 | //! # Policy |
| 17 | //! |
| 18 | //! Normalization applies to the whole edited file, and **only when that file |
| 19 | //! was already `rustfmt`-clean before the edit**. A file whose formatting the |
| 20 | //! author has not handed to `rustfmt` is never rewritten. Because a clean file |
| 21 | //! is a formatting fixpoint, reformatting it after an edit can only change the |
| 22 | //! edited region — so "whole file" and "edited region" coincide, without |
| 23 | //! needing span arithmetic to prove it. |
| 24 | //! |
| 25 | //! # Known limitations |
| 26 | //! |
| 27 | //! - **Non-fatal, always.** A missing `rustfmt`, a parse failure, a timeout, or |
| 28 | //! a non-zero exit skips normalization; the edit still lands. Nothing here |
| 29 | //! can fail an edit. |
| 30 | //! - **Silent when skipped.** Only an applied normalization is announced. A |
| 31 | //! "formatting skipped" note on every edit in a project without `rustfmt` |
| 32 | //! would be noise the model cannot act on, and "skipped" is indistinguishable |
| 33 | //! from "would have changed nothing" without running the formatter anyway. |
| 34 | //! - **Edition 2024 is assumed.** A file that `rustfmt` cannot parse under that |
| 35 | //! edition fails the clean-before check and is skipped, so the assumption |
| 36 | //! degrades to "no normalization", never to a mangled file. |
| 37 | //! - **CRLF files are skipped.** `rustfmt` emits LF; rewriting every line |
| 38 | //! ending is exactly the unrelated-churn this policy exists to avoid. |
| 39 | |
| 40 | use std::path::{Path, PathBuf}; |
| 41 | use std::process::Stdio; |
| 42 | use std::time::Duration; |
| 43 | |
| 44 | use tokio::io::AsyncWriteExt; |
| 45 | use tokio::process::Command; |
| 46 | |
| 47 | /// Files larger than this are not normalized. Two formatter runs on a |
| 48 | /// multi-megabyte file cost more interactive latency than stable anchors are |
| 49 | /// worth. |
| 50 | const MAX_FORMATTED_BYTES: usize = 1024 * 1024; |
| 51 | |
| 52 | /// Wall-clock budget for one `rustfmt` run. |
| 53 | const FORMAT_TIMEOUT: Duration = Duration::from_secs(5); |
| 54 | |
| 55 | /// Suffix appended to an edit summary when the content was normalized, so the |
| 56 | /// model knows the returned text is not byte-identical to what it sent. |
| 57 | pub(super) const NORMALIZED_NOTE: &str = " (rustfmt-normalized)"; |
| 58 | |
| 59 | /// Normalize `after` when the edit landed in an already-`rustfmt`-clean file. |
| 60 | /// |
| 61 | /// Returns the formatted text to write instead of `after`, or `None` to leave |
| 62 | /// `after` exactly as the caller produced it. Every failure path returns |
| 63 | /// `None`: normalization is a convenience and may never break an edit. |
| 64 | pub(super) async fn normalize_edit(path: &Path, before: &str, after: &str) -> Option<String> { |
| 65 | if path.extension()?.to_str()? != "rs" { |
| 66 | return None; |
| 67 | } |
| 68 | if after.len() > MAX_FORMATTED_BYTES || before.len() > MAX_FORMATTED_BYTES { |
| 69 | return None; |
| 70 | } |
| 71 | // Preserving a CRLF file's line endings outranks normalizing its layout. |
| 72 | if after.contains('\r') || before.contains('\r') { |
| 73 | return None; |
| 74 | } |
| 75 | |
| 76 | let formatted = rustfmt(path, after).await?; |
| 77 | if formatted == after { |
| 78 | // Already canonical. The common case, and it costs one run, not two. |
| 79 | return None; |
| 80 | } |
| 81 | // Only now is the second run worth paying for: was this file the author's |
| 82 | // to format, or `rustfmt`'s? |
| 83 | if rustfmt(path, before).await? != before { |
| 84 | return None; |
| 85 | } |
| 86 | Some(formatted) |
| 87 | } |
| 88 | |
| 89 | /// Find the `rustfmt.toml` governing `path`, walking up to the filesystem root. |
| 90 | /// |
| 91 | /// `None` means no config file exists, which is the signal to omit |
| 92 | /// `--config-path` entirely and let `rustfmt` use its defaults. |
| 93 | async fn nearest_config(path: &Path) -> Option<PathBuf> { |
| 94 | let mut directory = path.parent()?; |
| 95 | loop { |
| 96 | for name in ["rustfmt.toml", ".rustfmt.toml"] { |
| 97 | let candidate = directory.join(name); |
| 98 | if tokio::fs::try_exists(&candidate).await.unwrap_or(false) { |
| 99 | return Some(candidate); |
| 100 | } |
| 101 | } |
| 102 | directory = directory.parent()?; |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | /// Run `rustfmt` over `source`, returning its output, or `None` on any failure. |
| 107 | async fn rustfmt(path: &Path, source: &str) -> Option<String> { |
| 108 | let mut command = Command::new("rustfmt"); |
| 109 | command |
| 110 | .arg("--emit") |
| 111 | .arg("stdout") |
| 112 | .arg("--edition") |
| 113 | .arg("2024") |
| 114 | .arg("--quiet"); |
| 115 | // Pick up the project's own `rustfmt.toml`: with stdin input there is no |
| 116 | // file path for `rustfmt` to search upward from. The flag must name a |
| 117 | // config file that actually exists — pointed at a directory without one, |
| 118 | // `rustfmt` exits 1 with "unable to find a config file", which would |
| 119 | // silently disable normalization everywhere. |
| 120 | if let Some(config) = nearest_config(path).await { |
| 121 | command.arg("--config-path").arg(config); |
| 122 | } |
| 123 | let mut child = command |
| 124 | .stdin(Stdio::piped()) |
| 125 | .stdout(Stdio::piped()) |
| 126 | .stderr(Stdio::null()) |
| 127 | // A hung formatter must not outlive the edit that spawned it. |
| 128 | .kill_on_drop(true) |
| 129 | .spawn() |
| 130 | .ok()?; |
| 131 | |
| 132 | let mut stdin = child.stdin.take()?; |
| 133 | let payload = source.to_string(); |
| 134 | // Write and wait concurrently: `rustfmt` streams its output, so writing the |
| 135 | // whole input before reading can deadlock on a full pipe buffer. |
| 136 | let writer = tokio::spawn(async move { |
| 137 | let _ = stdin.write_all(payload.as_bytes()).await; |
| 138 | let _ = stdin.shutdown().await; |
| 139 | }); |
| 140 | |
| 141 | let output = match tokio::time::timeout(FORMAT_TIMEOUT, child.wait_with_output()).await { |
| 142 | Ok(Ok(output)) => output, |
| 143 | // A timed-out child is already killed by dropping the future's handle |
| 144 | // on the `wait_with_output` path; either way the edit proceeds. |
| 145 | Ok(Err(_)) | Err(_) => { |
| 146 | writer.abort(); |
| 147 | return None; |
| 148 | } |
| 149 | }; |
| 150 | writer.abort(); |
| 151 | |
| 152 | if !output.status.success() { |
| 153 | return None; |
| 154 | } |
| 155 | String::from_utf8(output.stdout).ok() |
| 156 | } |
| 157 | |
| 158 | #[cfg(test)] |
| 159 | mod tests { |
| 160 | use super::*; |
| 161 | use std::path::PathBuf; |
| 162 | |
| 163 | fn rust_path() -> PathBuf { |
| 164 | PathBuf::from("src/lib.rs") |
| 165 | } |
| 166 | |
| 167 | /// `rustfmt` ships with the toolchain this repository pins (rustup's |
| 168 | /// default profile), and `cargo fmt --check` is a standing gate here, so |
| 169 | /// its absence is a broken environment, not a reason to skip. A test that |
| 170 | /// passes vacuously without the formatter proves nothing — that exact |
| 171 | /// hazard hid a `--config-path` bug which disabled normalization |
| 172 | /// everywhere while the suite stayed green. |
| 173 | async fn require_rustfmt() { |
| 174 | assert_eq!( |
| 175 | rustfmt(&rust_path(), "fn main( ) {}\n").await.as_deref(), |
| 176 | Some("fn main() {}\n"), |
| 177 | "rustfmt must be on PATH for the formatting tests" |
| 178 | ); |
| 179 | } |
| 180 | |
| 181 | #[tokio::test] |
| 182 | async fn misformatted_edit_in_a_clean_file_is_normalized() { |
| 183 | require_rustfmt().await; |
| 184 | let before = "fn main() {\n let x = 1;\n}\n"; |
| 185 | let after = "fn main() {\n let x = 1;\n let y=2;\n}\n"; |
| 186 | let normalized = normalize_edit(&rust_path(), before, after) |
| 187 | .await |
| 188 | .expect("a clean file must be renormalized after a sloppy edit"); |
| 189 | assert_eq!( |
| 190 | normalized, |
| 191 | "fn main() {\n let x = 1;\n let y = 2;\n}\n" |
| 192 | ); |
| 193 | } |
| 194 | |
| 195 | #[tokio::test] |
| 196 | async fn a_file_the_author_formats_by_hand_is_left_alone() { |
| 197 | require_rustfmt().await; |
| 198 | // `rustfmt` would rewrite this file wholesale, so it was never its to |
| 199 | // format: the edit lands verbatim. |
| 200 | let before = "fn main() {\n let x = 1;\n}\n"; |
| 201 | let after = "fn main() {\n let x = 1;\n let y=2;\n}\n"; |
| 202 | assert!(normalize_edit(&rust_path(), before, after).await.is_none()); |
| 203 | } |
| 204 | |
| 205 | #[tokio::test] |
| 206 | async fn already_canonical_content_needs_no_rewrite() { |
| 207 | require_rustfmt().await; |
| 208 | let before = "fn main() {\n let x = 1;\n}\n"; |
| 209 | let after = "fn main() {\n let x = 1;\n let y = 2;\n}\n"; |
| 210 | assert!(normalize_edit(&rust_path(), before, after).await.is_none()); |
| 211 | } |
| 212 | |
| 213 | #[tokio::test] |
| 214 | async fn crlf_files_keep_their_line_endings() { |
| 215 | let before = "fn main() {\r\n let x = 1;\r\n}\r\n"; |
| 216 | let after = "fn main() {\r\n let x = 1;\r\n}\r\n"; |
| 217 | assert!(normalize_edit(&rust_path(), before, after).await.is_none()); |
| 218 | } |
| 219 | |
| 220 | #[tokio::test] |
| 221 | async fn non_rust_files_are_not_formatted() { |
| 222 | assert!( |
| 223 | normalize_edit(Path::new("data.json"), "{}", "{ }") |
| 224 | .await |
| 225 | .is_none() |
| 226 | ); |
| 227 | } |
| 228 | |
| 229 | #[tokio::test] |
| 230 | async fn unparseable_content_degrades_to_no_normalization() { |
| 231 | // `rustfmt` cannot parse this; the edit must still be allowed to land. |
| 232 | assert!( |
| 233 | normalize_edit(&rust_path(), "fn main() {}\n", "fn main( {\n") |
| 234 | .await |
| 235 | .is_none() |
| 236 | ); |
| 237 | } |
| 238 | } |
| 239 |