lsp: support both utf-8 and utf-16 offsets.

Still need to implement the clangd encoding negotiation, but it's
a start. Should also manually override to utf8 for pyls.

2 files changed, 73 insertions, 21 deletions

diff --git a/helix-lsp/src/client.rs b/helix-lsp/src/client.rs
index 64070be0..c70e6e78 100644
--- a/helix-lsp/src/client.rs
+++ b/helix-lsp/src/client.rs
@@ -1,6 +1,6 @@
 use crate::{
     transport::{Payload, Transport},
-    Call, Error, Result,
+    Call, Error, OffsetEncoding, Result,
 };
 
 use helix_core::{ChangeSet, Rope};
@@ -29,8 +29,7 @@ pub struct Client {
     pub request_counter: AtomicU64,
 
     capabilities: Option<lsp::ServerCapabilities>,
-    // TODO: handle PublishDiagnostics Version
-    // diagnostics: HashMap<lsp::Url, Vec<lsp::Diagnostic>>,
+    offset_encoding: OffsetEncoding,
 }
 
 impl Client {
@@ -70,6 +69,7 @@ impl Client {
 
             capabilities: None,
             // diagnostics: HashMap::new(),
+            offset_encoding: OffsetEncoding::Utf8,
         };
 
         // TODO: async client.initialize()
@@ -100,6 +100,10 @@ impl Client {
             .expect("language server not yet initialized!")
     }
 
+    pub fn offset_encoding(&self) -> OffsetEncoding {
+        self.offset_encoding
+    }
+
     /// Execute a RPC request on the language server.
     pub async fn request<R: lsp::request::Request>(&self, params: R::Params) -> Result<R::Result>
     where
@@ -291,6 +295,7 @@ impl Client {
         old_text: &Rope,
         new_text: &Rope,
         changeset: &ChangeSet,
+        offset_encoding: OffsetEncoding,
     ) -> Vec<lsp::TextDocumentContentChangeEvent> {
         let mut iter = changeset.changes().iter().peekable();
         let mut old_pos = 0;
@@ -340,7 +345,7 @@ impl Client {
                     new_pos += i;
                 }
                 Delete(_) => {
-                    let start = pos_to_lsp_pos(new_text, new_pos);
+                    let start = pos_to_lsp_pos(new_text, new_pos, offset_encoding);
                     let end = traverse(start, old_text.slice(old_pos..old_end));
 
                     // deletion
@@ -351,7 +356,7 @@ impl Client {
                     });
                 }
                 Insert(s) => {
-                    let start = pos_to_lsp_pos(new_text, new_pos);
+                    let start = pos_to_lsp_pos(new_text, new_pos, offset_encoding);
 
                     new_pos += s.chars().count();
 
@@ -413,7 +418,7 @@ impl Client {
                 }]
             }
             lsp::TextDocumentSyncKind::Incremental => {
-                Self::changeset_to_changes(old_text, new_text, changes)
+                Self::changeset_to_changes(old_text, new_text, changes, self.offset_encoding)
             }
             lsp::TextDocumentSyncKind::None => return Ok(()),
         };
diff --git a/helix-lsp/src/lib.rs b/helix-lsp/src/lib.rs
index 40d2e477..dd925c14 100644
--- a/helix-lsp/src/lib.rs
+++ b/helix-lsp/src/lib.rs
@@ -16,6 +16,8 @@ use thiserror::Error;
 
 use std::{collections::HashMap, sync::Arc};
 
+use serde::{Deserialize, Serialize};
+
 #[derive(Error, Debug)]
 pub enum Error {
     #[error("protocol error: {0}")]
@@ -28,31 +30,76 @@ pub enum Error {
     Other(#[from] anyhow::Error),
 }
 
+#[derive(Clone, Copy, Debug, Serialize, Deserialize)]
+pub enum OffsetEncoding {
+    /// UTF-8 code units aka bytes
+    #[serde(rename = "utf-8")]
+    Utf8,
+    /// UTF-16 code units
+    #[serde(rename = "utf-16")]
+    Utf16,
+}
+
 pub mod util {
     use super::*;
     use helix_core::{Range, Rope, Transaction};
 
-    pub fn lsp_pos_to_pos(doc: &Rope, pos: lsp::Position) -> usize {
-        let line = doc.line_to_char(pos.line as usize);
-        let line_start = doc.char_to_utf16_cu(line);
-        doc.utf16_cu_to_char(line_start + pos.character as usize)
+    pub fn lsp_pos_to_pos(
+        doc: &Rope,
+        pos: lsp::Position,
+        offset_encoding: OffsetEncoding,
+    ) -> usize {
+        match offset_encoding {
+            OffsetEncoding::Utf8 => {
+                let line = doc.line_to_char(pos.line as usize);
+                line + pos.character as usize
+            }
+            OffsetEncoding::Utf16 => {
+                let line = doc.line_to_char(pos.line as usize);
+                let line_start = doc.char_to_utf16_cu(line);
+                doc.utf16_cu_to_char(line_start + pos.character as usize)
+            }
+        }
     }
-    pub fn pos_to_lsp_pos(doc: &Rope, pos: usize) -> lsp::Position {
-        let line = doc.char_to_line(pos);
-        let line_start = doc.char_to_utf16_cu(doc.line_to_char(line));
-        let col = doc.char_to_utf16_cu(pos) - line_start;
+    pub fn pos_to_lsp_pos(
+        doc: &Rope,
+        pos: usize,
+        offset_encoding: OffsetEncoding,
+    ) -> lsp::Position {
+        match offset_encoding {
+            OffsetEncoding::Utf8 => {
+                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)
+            }
+            OffsetEncoding::Utf16 => {
+                let line = doc.char_to_line(pos);
+                let line_start = doc.char_to_utf16_cu(doc.line_to_char(line));
+                let col = doc.char_to_utf16_cu(pos) - line_start;
 
-        lsp::Position::new(line as u32, col as u32)
+                lsp::Position::new(line as u32, col as u32)
+            }
+        }
     }
 
-    pub fn range_to_lsp_range(doc: &Rope, range: Range) -> lsp::Range {
-        let start = pos_to_lsp_pos(doc, range.from());
-        let end = pos_to_lsp_pos(doc, range.to());
+    pub fn range_to_lsp_range(
+        doc: &Rope,
+        range: Range,
+        offset_encoding: OffsetEncoding,
+    ) -> lsp::Range {
+        let start = pos_to_lsp_pos(doc, range.from(), offset_encoding);
+        let end = pos_to_lsp_pos(doc, range.to(), offset_encoding);
 
         lsp::Range::new(start, end)
     }
 
-    pub fn generate_transaction_from_edits(doc: &Rope, edits: Vec<lsp::TextEdit>) -> Transaction {
+    pub fn generate_transaction_from_edits(
+        doc: &Rope,
+        edits: Vec<lsp::TextEdit>,
+        offset_encoding: OffsetEncoding,
+    ) -> Transaction {
         Transaction::change(
             doc,
             edits.into_iter().map(|edit| {
@@ -63,8 +110,8 @@ pub mod util {
                     None
                 };
 
-                let start = lsp_pos_to_pos(doc, edit.range.start);
-                let end = lsp_pos_to_pos(doc, edit.range.end);
+                let start = lsp_pos_to_pos(doc, edit.range.start, offset_encoding);
+                let end = lsp_pos_to_pos(doc, edit.range.end, offset_encoding);
                 (start, end, replacement)
             }),
         )