93c7afc · helix · git.iko.soy

git.iko.soy/helix

Negotiate LSP Position Encoding (#5894)

So far LSP always required that `PositionEncoding.characters` is an UTF-16 offset. Now that LSP 3.17 is available in `lsp-types` request the server to send char offsets (UTF-32) or byte offsets (UTF-8) instead. For compatability with old servers, UTF-16 remains as the fallback as required by the standard.

commit93c7afc4ed2b8d848fa2779f43202ba7f837263b authorPascal Kuthe <pascal.kuthe@semimod.de> dateSat 11 Feb 2023, 07:50 (4 years ago) parent6929a12

5 files changed+49−14

compare with master

unifiedsplit

Cargo.lock+2−2 helix-lsp/Cargo.toml+1−1 helix-lsp/src/client.rs+24−5 helix-lsp/src/lib.rs+17−5 helix-term/src/commands/lsp.rs+5−1

Cargo.lock+2−2

@@ -1481,9 +1481,9 @@ dependencies = [

14811481

14821482[[package]]

14831483name = "lsp-types"

1484−version = "0.93.2"

1484+version = "0.94.0"

14851485source = "registry+https://github.com/rust-lang/crates.io-index"

1486−checksum = "9be6e9c7e2d18f651974370d7aff703f9513e0df6e464fd795660edc77e6ca51"

1486+checksum = "0b63735a13a1f9cd4f4835223d828ed9c2e35c8c5e61837774399f558b6a1237"

14871487dependencies = [

14881488 "bitflags",

14891489 "serde",

helix-lsp/Cargo.toml+1−1

@@ -19,7 +19,7 @@ anyhow = "1.0"

1919futures-executor = "0.3"

2020futures-util = { version = "0.3", features = ["std", "async-await"], default-features = false }

2121log = "0.4"

22−lsp-types = { version = "0.93", features = ["proposed"] }

22+lsp-types = { version = "0.94" }

2323serde = { version = "1.0", features = ["derive"] }

2424serde_json = "1.0"

2525thiserror = "1.0"

helix-lsp/src/client.rs+24−5

@@ -6,6 +6,7 @@ use crate::{

77use helix_core::{find_root, ChangeSet, Rope};

88use helix_loader::{self, VERSION_AND_GIT_HASH};

9+use lsp::PositionEncodingKind;

910use lsp_types as lsp;

1011use serde::Deserialize;

1112use serde_json::Value;

@@ -32,7 +33,6 @@ pub struct Client {

3233 server_tx: UnboundedSender<Payload>,

3334 request_counter: AtomicU64,

3435 pub(crate) capabilities: OnceCell<lsp::ServerCapabilities>,

35− offset_encoding: OffsetEncoding,

3636 config: Option<Value>,

3737 root_path: std::path::PathBuf,

3838 root_uri: Option<lsp::Url>,

@@ -104,7 +104,6 @@ impl Client {

104104 server_tx,

105105 request_counter: AtomicU64::new(0),

106106 capabilities: OnceCell::new(),

107− offset_encoding: OffsetEncoding::Utf16,

108107 config,

109108 req_timeout,

110109

@@ -147,7 +146,19 @@ impl Client {

147146 }

148147

149148 pub fn offset_encoding(&self) -> OffsetEncoding {

150− self.offset_encoding

149+ self.capabilities()

150+ .position_encoding

151+ .as_ref()

152+ .and_then(|encoding| match encoding.as_str() {

153+ "utf-8" => Some(OffsetEncoding::Utf8),

154+ "utf-16" => Some(OffsetEncoding::Utf16),

155+ "utf-32" => Some(OffsetEncoding::Utf32),

156+ encoding => {

157+ log::error!("Server provided invalid position encording {encoding}, defaulting to utf-16");

158+ None

159+ },

160+ })

161+ .unwrap_or_default()

151162 }

152163

153164 pub fn config(&self) -> Option<&Value> {

@@ -377,6 +388,14 @@ impl Client {

377388 work_done_progress: Some(true),

378389 ..Default::default()

379390 }),

391+ general: Some(lsp::GeneralClientCapabilities {

392+ position_encodings: Some(vec![

393+ PositionEncodingKind::UTF32,

394+ PositionEncodingKind::UTF8,

395+ PositionEncodingKind::UTF16,

396+ ]),

397+ ..Default::default()

398+ }),

380399 ..Default::default()

381400 },

382401 trace: None,

@@ -577,7 +596,7 @@ impl Client {

577596 }]

578597 }

579598 lsp::TextDocumentSyncKind::INCREMENTAL => {

580− Self::changeset_to_changes(old_text, new_text, changes, self.offset_encoding)

599+ Self::changeset_to_changes(old_text, new_text, changes, self.offset_encoding())

581600 }

582601 lsp::TextDocumentSyncKind::NONE => return None,

583602 kind => unimplemented!("{:?}", kind),

@@ -1027,7 +1046,7 @@ impl Client {

10271046 partial_result_params: lsp::PartialResultParams::default(),

10281047 };

10291048

1030− Some(self.call::<lsp::request::WorkspaceSymbol>(params))

1049+ Some(self.call::<lsp::request::WorkspaceSymbolRequest>(params))

10311050 }

10321051

10331052 pub fn code_actions(

helix-lsp/src/lib.rs+17−5

@@ -20,7 +20,6 @@ use std::{

2020 },

2121};

2222

23−use serde::{Deserialize, Serialize};

2423use thiserror::Error;

2524use tokio_stream::wrappers::UnboundedReceiverStream;

2625

@@ -45,13 +44,14 @@ pub enum Error {

4544 Other(#[from] anyhow::Error),

4645}

4746

48−#[derive(Clone, Copy, Debug, Serialize, Deserialize)]

47+#[derive(Clone, Copy, Debug, Default)]

4948pub enum OffsetEncoding {

5049 /// UTF-8 code units aka bytes

51− #[serde(rename = "utf-8")]

5250 Utf8,

51+ /// UTF-32 code units aka chars

52+ Utf32,

5353 /// UTF-16 code units

54− #[serde(rename = "utf-16")]

54+ #[default]

5555 Utf16,

5656}

5757

@@ -168,6 +168,11 @@ pub mod util {

168168 let line_end = line_end_char_index(&doc.slice(..), pos_line);

169169 doc.char_to_utf16_cu(line_start)..doc.char_to_utf16_cu(line_end)

170170 }

171+ OffsetEncoding::Utf32 => {

172+ let line_start = doc.line_to_char(pos_line);

173+ let line_end = line_end_char_index(&doc.slice(..), pos_line);

174+ line_start..line_end

175+ }

171176 };

172177

173178 // The LSP spec demands that the offset is capped to the end of the line

@@ -177,10 +182,10 @@ pub mod util {

177182 .unwrap_or(line.end)

178183 .min(line.end);

179184

180− // TODO prefer UTF32/char indices to avoid this step

181185 match offset_encoding {

182186 OffsetEncoding::Utf8 => doc.try_byte_to_char(pos).ok(),

183187 OffsetEncoding::Utf16 => doc.try_utf16_cu_to_char(pos).ok(),

188+ OffsetEncoding::Utf32 => Some(pos),

184189 }

185190 }

186191

@@ -205,6 +210,13 @@ pub mod util {

205210 let line_start = doc.char_to_utf16_cu(doc.line_to_char(line));

206211 let col = doc.char_to_utf16_cu(pos) - line_start;

207212

213+ lsp::Position::new(line as u32, col as u32)

214+ }

215+ OffsetEncoding::Utf32 => {

216+ let line = doc.char_to_line(pos);

217+ let line_start = doc.line_to_char(line);

218+ let col = pos - line_start;

219+

208220 lsp::Position::new(line as u32, col as u32)

209221 }

210222 }

helix-term/src/commands/lsp.rs+5−1

@@ -1,7 +1,10 @@

11use futures_util::FutureExt;

22use helix_lsp::{

33 block_on,

4− lsp::{self, CodeAction, CodeActionOrCommand, DiagnosticSeverity, NumberOrString},

4+ lsp::{

5+ self, CodeAction, CodeActionOrCommand, CodeActionTriggerKind, DiagnosticSeverity,

6+ NumberOrString,

7+ },

58 util::{diagnostic_to_lsp_diagnostic, lsp_pos_to_pos, lsp_range_to_range, range_to_lsp_range},

69 OffsetEncoding,

710};

@@ -561,6 +564,7 @@ pub fn code_action(cx: &mut Context) {

561564 .map(|diag| diagnostic_to_lsp_diagnostic(doc.text(), diag, offset_encoding))

562565 .collect(),

563566 only: None,

567+ trigger_kind: Some(CodeActionTriggerKind::INVOKED),

564568 },

565569 ) {

566570 Some(future) => future,