Change default formatter for any language (#2942)

* Change default formatter for any language

* Fix clippy error

* Close stdin for Stdio formatters

* Better indentation and pattern matching

* Return Result<Option<...>> for fn format instead of Option

* Remove unwrap for stdin

* Handle FormatterErrors instead of Result<Option<...>>

* Use Transaction instead of LspFormatting

* Use Transaction directly in Document::format

* Perform stdin type formatting asynchronously

* Rename formatter.type values to kebab-case

* Debug format for displaying io::ErrorKind (msrv fix)

* Solve conflict?

* Use only stdio type formatters

* Remove FormatterType enum

* Remove old comment

* Check if the formatter exited correctly

* Add formatter configuration to the book

* Avoid allocations when writing to stdin and formatting errors

* Remove unused import

Co-authored-by: Gokul Soumya <gokulps15@gmail.com>

book/src/languages.md

@@ -40,6 +40,7 @@ file-types = ["mylang", "myl"]
 comment-token = "#"
 indent = { tab-width = 2, unit = "  " }
 language-server = { command = "mylang-lsp", args = ["--stdio"] }
+formatter = { command = "mylang-formatter" , args = ["--stdin"] }
 ```
 
 These configuration keys are available:
@@ -59,6 +60,7 @@ These configuration keys are available:
 | `language-server`     | The Language Server to run. See the Language Server configuration section below. |
 | `config`              | Language Server configuration                                 |
 | `grammar`             | The tree-sitter grammar to use (defaults to the value of `name`) |
+| `formatter`           | The formatter for the language, it will take precedence over the lsp when defined. The formatter must be able to take the original file as input from stdin and write the formatted file to stdout |
 
 ### Language Server configuration
 

helix-core/src/syntax.rs

@@ -79,6 +79,9 @@ pub struct LanguageConfiguration {
     #[serde(default)]
     pub auto_format: bool,
 
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub formatter: Option<FormatterConfiguration>,
+
     #[serde(default)]
     pub diagnostic_severity: Severity,
 
@@ -126,6 +129,15 @@ pub struct LanguageServerConfiguration {
     pub language_id: Option<String>,
 }
 
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "kebab-case")]
+pub struct FormatterConfiguration {
+    pub command: String,
+    #[serde(default)]
+    #[serde(skip_serializing_if = "Vec::is_empty")]
+    pub args: Vec<String>,
+}
+
 #[derive(Debug, PartialEq, Clone, Deserialize, Serialize)]
 #[serde(rename_all = "kebab-case")]
 pub struct AdvancedCompletion {

helix-lsp/src/lib.rs

@@ -213,22 +213,6 @@ pub mod util {
             }),
         )
     }
-
-    /// The result of asking the language server to format the document. This can be turned into a
-    /// `Transaction`, but the advantage of not doing that straight away is that this one is
-    /// `Send` and `Sync`.
-    #[derive(Clone, Debug)]
-    pub struct LspFormatting {
-        pub doc: Rope,
-        pub edits: Vec<lsp::TextEdit>,
-        pub offset_encoding: OffsetEncoding,
-    }
-
-    impl From<LspFormatting> for Transaction {
-        fn from(fmt: LspFormatting) -> Transaction {
-            generate_transaction_from_edits(&fmt.doc, fmt.edits, fmt.offset_encoding)
-        }
-    }
 }
 
 #[derive(Debug, PartialEq, Clone)]

helix-term/src/commands.rs

@@ -28,7 +28,7 @@ use helix_core::{
 };
 use helix_view::{
     clipboard::ClipboardType,
-    document::{Mode, SCRATCH_BUFFER_NAME},
+    document::{FormatterError, Mode, SCRATCH_BUFFER_NAME},
     editor::{Action, Motion},
     info::Info,
     input::KeyEvent,
@@ -2471,14 +2471,14 @@ async fn make_format_callback(
     doc_id: DocumentId,
     doc_version: i32,
     modified: Modified,
-    format: impl Future<Output = helix_lsp::util::LspFormatting> + Send + 'static,
+    format: impl Future<Output = Result<Transaction, FormatterError>> + Send + 'static,
 ) -> anyhow::Result<job::Callback> {
-    let format = format.await;
+    let format = format.await?;
     let call: job::Callback = Box::new(move |editor, _compositor| {
         let view_id = view!(editor).id;
         if let Some(doc) = editor.document_mut(doc_id) {
             if doc.version() == doc_version {
-                doc.apply(&Transaction::from(format), view_id);
+                doc.apply(&format, view_id);
                 doc.append_changes_to_history(view_id);
                 doc.detect_indent_and_line_ending();
                 if let Modified::SetUnmodified = modified {

helix-view/src/document.rs

@@ -1,4 +1,6 @@
 use anyhow::{anyhow, bail, Context, Error};
+use futures_util::future::BoxFuture;
+use futures_util::FutureExt;
 use helix_core::auto_pairs::AutoPairs;
 use helix_core::Range;
 use serde::de::{self, Deserialize, Deserializer};
@@ -20,7 +22,6 @@ use helix_core::{
     ChangeSet, Diagnostic, LineEnding, Rope, RopeBuilder, Selection, State, Syntax, Transaction,
     DEFAULT_LINE_ENDING,
 };
-use helix_lsp::util::LspFormatting;
 
 use crate::{DocumentId, Editor, ViewId};
 
@@ -397,7 +398,7 @@ impl Document {
 
     /// The same as [`format`], but only returns formatting changes if auto-formatting
     /// is configured.
-    pub fn auto_format(&self) -> Option<impl Future<Output = LspFormatting> + 'static> {
+    pub fn auto_format(&self) -> Option<BoxFuture<'static, Result<Transaction, FormatterError>>> {
         if self.language_config()?.auto_format {
             self.format()
         } else {
@@ -407,7 +408,56 @@ impl Document {
 
     /// If supported, returns the changes that should be applied to this document in order
     /// to format it nicely.
-    pub fn format(&self) -> Option<impl Future<Output = LspFormatting> + 'static> {
+    // We can't use anyhow::Result here since the output of the future has to be
+    // clonable to be used as shared future. So use a custom error type.
+    pub fn format(&self) -> Option<BoxFuture<'static, Result<Transaction, FormatterError>>> {
+        if let Some(formatter) = self.language_config().and_then(|c| c.formatter.clone()) {
+            use std::process::Stdio;
+            let text = self.text().clone();
+            let mut process = tokio::process::Command::new(&formatter.command);
+            process
+                .args(&formatter.args)
+                .stdin(Stdio::piped())
+                .stdout(Stdio::piped())
+                .stderr(Stdio::piped());
+
+            let formatting_future = async move {
+                let mut process = process
+                    .spawn()
+                    .map_err(|e| FormatterError::SpawningFailed {
+                        command: formatter.command.clone(),
+                        error: e.kind(),
+                    })?;
+                {
+                    let mut stdin = process.stdin.take().ok_or(FormatterError::BrokenStdin)?;
+                    to_writer(&mut stdin, encoding::UTF_8, &text)
+                        .await
+                        .map_err(|_| FormatterError::BrokenStdin)?;
+                }
+
+                let output = process
+                    .wait_with_output()
+                    .await
+                    .map_err(|_| FormatterError::WaitForOutputFailed)?;
+
+                if !output.stderr.is_empty() {
+                    return Err(FormatterError::Stderr(
+                        String::from_utf8_lossy(&output.stderr).to_string(),
+                    ));
+                }
+
+                if !output.status.success() {
+                    return Err(FormatterError::NonZeroExitStatus);
+                }
+
+                let str = String::from_utf8(output.stdout)
+                    .map_err(|_| FormatterError::InvalidUtf8Output)?;
+
+                Ok(helix_core::diff::compare_ropes(&text, &Rope::from(str)))
+            };
+            return Some(formatting_future.boxed());
+        };
+
         let language_server = self.language_server()?;
         let text = self.text.clone();
         let offset_encoding = language_server.offset_encoding();
@@ -427,13 +477,13 @@ impl Document {
                 log::warn!("LSP formatting failed: {}", e);
                 Default::default()
             });
-            LspFormatting {
+            Ok(helix_lsp::util::generate_transaction_from_edits(
-                doc: text,
+                &text,
                 edits,
                 offset_encoding,
-            }
+            ))
         };
-        Some(fut)
+        Some(fut.boxed())
     }
 
     pub fn save(&mut self, force: bool) -> impl Future<Output = Result<(), anyhow::Error>> {
@@ -442,7 +492,7 @@ impl Document {
 
     pub fn format_and_save(
         &mut self,
-        formatting: Option<impl Future<Output = LspFormatting>>,
+        formatting: Option<impl Future<Output = Result<Transaction, FormatterError>>>,
         force: bool,
     ) -> impl Future<Output = anyhow::Result<()>> {
         self.save_impl(formatting, force)
@@ -454,7 +504,7 @@ impl Document {
     /// at its `path()`.
     ///
     /// If `formatting` is present, it supplies some changes that we apply to the text before saving.
-    fn save_impl<F: Future<Output = LspFormatting>>(
+    fn save_impl<F: Future<Output = Result<Transaction, FormatterError>>>(
         &mut self,
         formatting: Option<F>,
         force: bool,
@@ -488,7 +538,8 @@ impl Document {
             }
 
             if let Some(fmt) = formatting {
-                let success = Transaction::from(fmt.await).changes().apply(&mut text);
+                let transaction = fmt.await?;
+                let success = transaction.changes().apply(&mut text);
                 if !success {
                     // This shouldn't happen, because the transaction changes were generated
                     // from the same text we're saving.
@@ -1034,6 +1085,38 @@ impl Default for Document {
     }
 }
 
+#[derive(Clone, Debug)]
+pub enum FormatterError {
+    SpawningFailed {
+        command: String,
+        error: std::io::ErrorKind,
+    },
+    BrokenStdin,
+    WaitForOutputFailed,
+    Stderr(String),
+    InvalidUtf8Output,
+    DiskReloadError(String),
+    NonZeroExitStatus,
+}
+
+impl std::error::Error for FormatterError {}
+
+impl Display for FormatterError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::SpawningFailed { command, error } => {
+                write!(f, "Failed to spawn formatter {}: {:?}", command, error)
+            }
+            Self::BrokenStdin => write!(f, "Could not write to formatter stdin"),
+            Self::WaitForOutputFailed => write!(f, "Waiting for formatter output failed"),
+            Self::Stderr(output) => write!(f, "Formatter error: {}", output),
+            Self::InvalidUtf8Output => write!(f, "Invalid UTF-8 formatter output"),
+            Self::DiskReloadError(error) => write!(f, "Error reloading file from disk: {}", error),
+            Self::NonZeroExitStatus => write!(f, "Formatter exited with non zero exit status:"),
+        }
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;

helix-view/src/editor.rs

@@ -613,7 +613,6 @@ impl Editor {
         syn_loader: Arc<syntax::Loader>,
         config: Box<dyn DynAccess<Config>>,
     ) -> Self {
-        let language_servers = helix_lsp::Registry::new();
         let conf = config.load();
         let auto_pairs = (&conf.auto_pairs).into();
 
@@ -629,7 +628,7 @@ impl Editor {
             macro_recording: None,
             macro_replaying: Vec::new(),
             theme: theme_loader.default(),
-            language_servers,
+            language_servers: helix_lsp::Registry::new(),
             diagnostics: BTreeMap::new(),
             debugger: None,
             debugger_events: SelectAll::new(),

languages.toml

@@ -706,6 +706,8 @@ auto-format = true
 comment-token = "//"
 language-server = { command = "zls" }
 indent = { tab-width = 4, unit = "    " }
+formatter = { command = "zig" , args = ["fmt", "--stdin"] }
+
 
 [[grammar]]
 name = "zig"