2 files changed, 41 insertions, 10 deletions

diff --git a/helix-lsp/src/client.rs b/helix-lsp/src/client.rs
index cb6e5d81..2f5b498d 100644
--- a/helix-lsp/src/client.rs
+++ b/helix-lsp/src/client.rs
@@ -6,6 +6,7 @@ use crate::{
 
 use helix_core::{find_root, ChangeSet, Rope};
 use helix_loader::{self, VERSION_AND_GIT_HASH};
+use lsp::PositionEncodingKind;
 use lsp_types as lsp;
 use serde::Deserialize;
 use serde_json::Value;
@@ -32,7 +33,6 @@ pub struct Client {
     server_tx: UnboundedSender<Payload>,
     request_counter: AtomicU64,
     pub(crate) capabilities: OnceCell<lsp::ServerCapabilities>,
-    offset_encoding: OffsetEncoding,
     config: Option<Value>,
     root_path: std::path::PathBuf,
     root_uri: Option<lsp::Url>,
@@ -104,7 +104,6 @@ impl Client {
             server_tx,
             request_counter: AtomicU64::new(0),
             capabilities: OnceCell::new(),
-            offset_encoding: OffsetEncoding::Utf16,
             config,
             req_timeout,
 
@@ -147,7 +146,19 @@ impl Client {
     }
 
     pub fn offset_encoding(&self) -> OffsetEncoding {
-        self.offset_encoding
+        self.capabilities()
+            .position_encoding
+            .as_ref()
+            .and_then(|encoding| match encoding.as_str() {
+                "utf-8" => Some(OffsetEncoding::Utf8),
+                "utf-16" => Some(OffsetEncoding::Utf16),
+                "utf-32" => Some(OffsetEncoding::Utf32),
+                encoding => {
+                    log::error!("Server provided invalid position encording {encoding}, defaulting to utf-16");
+                    None
+                },
+            })
+            .unwrap_or_default()
     }
 
     pub fn config(&self) -> Option<&Value> {
@@ -377,6 +388,14 @@ impl Client {
                     work_done_progress: Some(true),
                     ..Default::default()
                 }),
+                general: Some(lsp::GeneralClientCapabilities {
+                    position_encodings: Some(vec![
+                        PositionEncodingKind::UTF32,
+                        PositionEncodingKind::UTF8,
+                        PositionEncodingKind::UTF16,
+                    ]),
+                    ..Default::default()
+                }),
                 ..Default::default()
             },
             trace: None,
@@ -577,7 +596,7 @@ impl Client {
                 }]
             }
             lsp::TextDocumentSyncKind::INCREMENTAL => {
-                Self::changeset_to_changes(old_text, new_text, changes, self.offset_encoding)
+                Self::changeset_to_changes(old_text, new_text, changes, self.offset_encoding())
             }
             lsp::TextDocumentSyncKind::NONE => return None,
             kind => unimplemented!("{:?}", kind),
@@ -1027,7 +1046,7 @@ impl Client {
             partial_result_params: lsp::PartialResultParams::default(),
         };
 
-        Some(self.call::<lsp::request::WorkspaceSymbol>(params))
+        Some(self.call::<lsp::request::WorkspaceSymbolRequest>(params))
     }
 
     pub fn code_actions(
diff --git a/helix-lsp/src/lib.rs b/helix-lsp/src/lib.rs
index bcaff10a..72456b37 100644
--- a/helix-lsp/src/lib.rs
+++ b/helix-lsp/src/lib.rs
@@ -20,7 +20,6 @@ use std::{
     },
 };
 
-use serde::{Deserialize, Serialize};
 use thiserror::Error;
 use tokio_stream::wrappers::UnboundedReceiverStream;
 
@@ -45,13 +44,14 @@ pub enum Error {
     Other(#[from] anyhow::Error),
 }
 
-#[derive(Clone, Copy, Debug, Serialize, Deserialize)]
+#[derive(Clone, Copy, Debug, Default)]
 pub enum OffsetEncoding {
     /// UTF-8 code units aka bytes
-    #[serde(rename = "utf-8")]
     Utf8,
+    /// UTF-32 code units aka chars
+    Utf32,
     /// UTF-16 code units
-    #[serde(rename = "utf-16")]
+    #[default]
     Utf16,
 }
 
@@ -168,6 +168,11 @@ pub mod util {
                 let line_end = line_end_char_index(&doc.slice(..), pos_line);
                 doc.char_to_utf16_cu(line_start)..doc.char_to_utf16_cu(line_end)
             }
+            OffsetEncoding::Utf32 => {
+                let line_start = doc.line_to_char(pos_line);
+                let line_end = line_end_char_index(&doc.slice(..), pos_line);
+                line_start..line_end
+            }
         };
 
         // The LSP spec demands that the offset is capped to the end of the line
@@ -177,10 +182,10 @@ pub mod util {
             .unwrap_or(line.end)
             .min(line.end);
 
-        // TODO prefer UTF32/char indices to avoid this step
         match offset_encoding {
             OffsetEncoding::Utf8 => doc.try_byte_to_char(pos).ok(),
             OffsetEncoding::Utf16 => doc.try_utf16_cu_to_char(pos).ok(),
+            OffsetEncoding::Utf32 => Some(pos),
         }
     }
 
@@ -207,6 +212,13 @@ pub mod util {
 
                 lsp::Position::new(line as u32, col as u32)
             }
+            OffsetEncoding::Utf32 => {
+                let line = doc.char_to_line(pos);
+                let line_start = doc.line_to_char(line);
+                let col = pos - line_start;
+
+                lsp::Position::new(line as u32, col as u32)
+            }
         }
     }