feat: support untitled url scheme for unsaved documents (#120)

crates/tinymist-query/Cargo.toml

@@ -37,6 +37,7 @@ reflexo.workspace = true
 
 lsp-types.workspace = true
 if_chain = "1"
+percent-encoding = "2"
 
 [dev-dependencies]
 once_cell.workspace = true

crates/tinymist-query/src/diagnostics.rs

@@ -1,4 +1,4 @@
-use crate::prelude::*;
+use crate::{path_to_url, prelude::*};
 
 /// Stores diagnostics for files.
 pub type DiagnosticsMap = HashMap<Url, Vec<LspDiagnostic>>;
@@ -28,20 +28,11 @@ fn convert_diagnostic(
     let uri;
     let lsp_range;
     if let Some((id, span)) = diagnostic_span_id(typst_diagnostic) {
-        uri = Url::from_file_path(ctx.path_for_id(id)?).map_err(|e| {
-            let _: () = e;
-            anyhow::anyhow!(
-                "could not convert path to URI: id: {id:?}, context: {:?}",
-                ctx.analysis.root
-            )
-        })?;
+        uri = path_to_url(&ctx.path_for_id(id)?)?;
         let source = ctx.world().source(id)?;
         lsp_range = diagnostic_range(&source, span, ctx.position_encoding());
     } else {
-        uri = Url::from_file_path(ctx.analysis.root.clone()).map_err(|e| {
-            let _: () = e;
-            anyhow::anyhow!("could not convert path to URI: {:?}", ctx.analysis.root)
-        })?;
+        uri = path_to_url(&ctx.analysis.root)?;
         lsp_range = LspRange::default();
     };
 
@@ -72,13 +63,7 @@ fn tracepoint_to_relatedinformation(
     position_encoding: PositionEncoding,
 ) -> anyhow::Result<Option<DiagnosticRelatedInformation>> {
     if let Some(id) = tracepoint.span.id() {
-        let uri = Url::from_file_path(project.path_for_id(id)?).map_err(|e| {
-            let _: () = e;
-            anyhow::anyhow!(
-                "could not convert path to URI: id: {id:?}, context: {:?}",
-                project.analysis.root
-            )
-        })?;
+        let uri = path_to_url(&project.path_for_id(id)?)?;
         let source = project.world().source(id)?;
 
         if let Some(typst_range) = source.range(tracepoint.span) {

crates/tinymist-query/src/goto_declaration.rs

@@ -60,7 +60,7 @@ impl SemanticRequest for GotoDeclarationRequest {
             let span_path = ctx.path_for_id(ref_id).ok()?;
             let range = ctx.to_lsp_range(ref_range, ref_source);
 
-            let uri = Url::from_file_path(span_path).ok()?;
+            let uri = path_to_url(&span_path).ok()?;
 
             links.push(LocationLink {
                 origin_selection_range: Some(origin_selection_range),

crates/tinymist-query/src/goto_definition.rs

@@ -54,7 +54,7 @@ impl SemanticRequest for GotoDefinitionRequest {
         let def = find_definition(ctx, source.clone(), deref_target)?;
 
         let span_path = ctx.path_for_id(def.fid).ok()?;
-        let uri = Url::from_file_path(span_path).ok()?;
+        let uri = path_to_url(&span_path).ok()?;
 
         let span_source = ctx.source_by_id(def.fid).ok()?;
         let range = ctx.to_lsp_range(def.def_range, &span_source);

crates/tinymist-query/src/lsp_typst_boundary.rs

@@ -3,7 +3,10 @@
 // todo: remove this
 #![allow(missing_docs)]
 
-use lsp_types;
+use std::path::{Path, PathBuf};
+
+use lsp_types::{self, Url};
+use reflexo::path::PathClean;
 
 pub type LspPosition = lsp_types::Position;
 /// The interpretation of an `LspCharacterOffset` depends on the
@@ -64,6 +67,41 @@ pub type LspCompletionKind = lsp_types::CompletionItemKind;
 pub type TypstCompletion = typst_ide::Completion;
 pub type TypstCompletionKind = typst_ide::CompletionKind;
 
+const UNTITLED_ROOT: &str = "/untitled";
+
+pub fn path_to_url(path: &Path) -> anyhow::Result<Url> {
+    if let Ok(untitled) = path.strip_prefix(UNTITLED_ROOT) {
+        return Ok(Url::parse(&format!("untitled:{}", untitled.display()))?);
+    }
+
+    Url::from_file_path(path).map_err(|e| {
+        let _: () = e;
+        anyhow::anyhow!("could not convert path to URI: path: {path:?}",)
+    })
+}
+
+pub fn url_to_path(uri: Url) -> PathBuf {
+    if uri.scheme() == "file" {
+        return uri.to_file_path().unwrap();
+    }
+
+    if uri.scheme() == "untitled" {
+        let mut bytes = UNTITLED_ROOT.as_bytes().to_vec();
+
+        // This is rust-url's path_segments, but vscode's untitle doesn't like it.
+        let path = uri.path();
+        let segs = path.strip_prefix('/').unwrap_or(path).split('/');
+        for segment in segs {
+            bytes.push(b'/');
+            bytes.extend(percent_encoding::percent_decode(segment.as_bytes()));
+        }
+
+        return Path::new(String::from_utf8_lossy(&bytes).as_ref()).clean();
+    }
+
+    uri.to_file_path().unwrap()
+}
+
 pub mod lsp_to_typst {
     use typst::syntax::Source;
 
@@ -326,6 +364,17 @@ mod test {
 
     use super::*;
 
+    #[test]
+    fn test_untitled() {
+        let path = Path::new("/untitled/test");
+        let uri = path_to_url(path).unwrap();
+        assert_eq!(uri.scheme(), "untitled");
+        assert_eq!(uri.path(), "test");
+
+        let path = url_to_path(uri);
+        assert_eq!(path, Path::new("/untitled/test").clean());
+    }
+
     const ENCODING_TEST_STRING: &str = "test 🥺 test";
 
     #[test]

crates/tinymist-query/src/prelude.rs

@@ -30,7 +30,7 @@ pub use typst::World;
 
 pub use crate::analysis::{analyze_expr, AnalysisContext};
 pub use crate::lsp_typst_boundary::{
-    lsp_to_typst, typst_to_lsp, LspDiagnostic, LspRange, LspSeverity, PositionEncoding,
-    TypstDiagnostic, TypstSeverity, TypstSpan,
+    lsp_to_typst, path_to_url, typst_to_lsp, LspDiagnostic, LspRange, LspSeverity,
+    PositionEncoding, TypstDiagnostic, TypstSeverity, TypstSpan,
 };
 pub use crate::VersionedDocument;

crates/tinymist-query/src/references.rs

@@ -116,7 +116,7 @@ pub(crate) fn find_references_root(
 ) -> Option<Vec<LspLocation>> {
     let def_source = ctx.source_by_id(def_fid).ok()?;
     let def_path = ctx.path_for_id(def_fid).ok()?;
-    let uri = Url::from_file_path(def_path).ok()?;
+    let uri = path_to_url(&def_path).ok()?;
 
     // todo: reuse uri, range to location
     let mut references = def_use
@@ -140,7 +140,7 @@ pub(crate) fn find_references_root(
             let def_use = ctx.ctx.def_use(ref_source.clone())?;
 
             let uri = ctx.ctx.path_for_id(ref_fid).ok()?;
-            let uri = Url::from_file_path(uri).ok()?;
+            let uri = path_to_url(&uri).ok()?;
 
             let mut redefines = vec![];
             if let Some((id, _def)) = def_use.get_def(def_fid, &def_ident) {
@@ -170,7 +170,7 @@ mod tests {
     use typst_ts_core::path::unix_slash;
 
     use super::*;
-    use crate::tests::*;
+    use crate::{tests::*, url_to_path};
 
     #[test]
     fn test() {
@@ -196,7 +196,7 @@ mod tests {
             let result = result.map(|v| {
                 v.into_iter()
                     .map(|l| {
-                        let fp = unix_slash(&l.uri.to_file_path().unwrap());
+                        let fp = unix_slash(&url_to_path(l.uri));
                         let fp = fp.strip_prefix("C:").unwrap_or(&fp);
                         format!(
                             "{fp}@{}:{}:{}:{}",

crates/tinymist-query/src/rename.rs

@@ -48,7 +48,7 @@ impl SemanticRequest for RenameRequest {
             let def_source = ctx.source_by_id(lnk.fid).ok()?;
 
             let span_path = ctx.path_for_id(lnk.fid).ok()?;
-            let uri = Url::from_file_path(span_path).ok()?;
+            let uri = path_to_url(&span_path).ok()?;
 
             let Some(range) = lnk.name_range else {
                 log::warn!("rename: no name range");

crates/tinymist-query/src/symbol.rs

@@ -40,7 +40,7 @@ impl SemanticRequest for SymbolRequest {
             let Ok(source) = ctx.source_by_path(path) else {
                 return;
             };
-            let uri = Url::from_file_path(path).unwrap();
+            let uri = path_to_url(path).unwrap();
             let res = get_lexical_hierarchy(source.clone(), LexicalScopeKind::Symbol).and_then(
                 |symbols| {
                     self.pattern.as_ref().map(|pattern| {