feat: introspect and show complation statistics happening in the language server (#1958)

Adds capability to introspect complations happening in the language
server, to help improve efficiency. I expect most compilations are
caused by tracing for analyzing dynamic expressions, but I haven't
really profiled a document. Then introspection will help confirm or
refute the expectation.

crates/tinymist-query/src/analysis.rs

@@ -1,7 +1,6 @@
 //! Semantic static and dynamic analysis of the source code.
 
 mod bib;
-
 pub(crate) use bib::*;
 pub mod call;
 pub use call::*;
@@ -15,30 +14,30 @@ pub mod doc_highlight;
 pub use doc_highlight::*;
 pub mod link_expr;
 pub use link_expr::*;
-pub mod stats;
-pub use stats::*;
 pub mod definition;
 pub use definition::*;
 pub mod signature;
 pub use signature::*;
 pub mod semantic_tokens;
 pub use semantic_tokens::*;
-use tinymist_std::error::WithContextUntyped;
-mod post_tyck;
-mod tyck;
-pub(crate) use crate::ty::*;
-pub(crate) use post_tyck::*;
-pub(crate) use tyck::*;
-mod prelude;
 
 mod global;
+mod post_tyck;
+mod prelude;
+mod tyck;
+
+pub(crate) use crate::ty::*;
 pub use global::*;
+pub(crate) use post_tyck::*;
+pub(crate) use tinymist_analysis::stats::{AnalysisStats, QueryStatGuard};
+pub(crate) use tyck::*;
 
 use std::sync::Arc;
 
 use ecow::eco_format;
 use lsp_types::Url;
 use tinymist_project::LspComputeGraph;
+use tinymist_std::error::WithContextUntyped;
 use tinymist_std::{Result, bail};
 use tinymist_world::{EntryReader, EntryState, TaskInputs};
 use typst::diag::{FileError, FileResult, StrResult};

crates/tinymist-query/src/analysis/global.rs

@@ -1115,13 +1115,7 @@ impl SharedContext {
     }
 
     fn query_stat(&self, id: TypstFileId, query: &'static str) -> QueryStatGuard {
-        let stats = &self.analysis.stats.query_stats;
-        let entry = stats.entry(id).or_default();
-        let entry = entry.entry(query).or_default();
-        QueryStatGuard {
-            bucket: entry.clone(),
-            since: tinymist_std::time::Instant::now(),
-        }
+        self.analysis.stats.stat(id, query)
     }
 
     /// Check on a module before really needing them. But we likely use them

crates/tinymist-query/src/analysis/stats.rs

@@ -1,115 +0,0 @@
-//! Statistics about the analyzers
-
-use std::sync::Arc;
-
-use parking_lot::Mutex;
-use tinymist_std::hash::FxDashMap;
-use tinymist_std::time::Duration;
-use typst::syntax::FileId;
-
-#[derive(Clone)]
-pub(crate) struct QueryStatBucketData {
-    pub query: u64,
-    pub missing: u64,
-    pub total: Duration,
-    pub min: Duration,
-    pub max: Duration,
-}
-
-impl Default for QueryStatBucketData {
-    fn default() -> Self {
-        Self {
-            query: 0,
-            missing: 0,
-            total: Duration::from_secs(0),
-            min: Duration::from_secs(u64::MAX),
-            max: Duration::from_secs(0),
-        }
-    }
-}
-
-/// Statistics about some query
-#[derive(Default, Clone)]
-pub(crate) struct QueryStatBucket {
-    pub data: Arc<Mutex<QueryStatBucketData>>,
-}
-
-pub(crate) struct QueryStatGuard {
-    pub bucket: QueryStatBucket,
-    pub since: tinymist_std::time::Instant,
-}
-
-impl Drop for QueryStatGuard {
-    fn drop(&mut self) {
-        let elapsed = self.since.elapsed();
-        let mut data = self.bucket.data.lock();
-        data.query += 1;
-        data.total += elapsed;
-        data.min = data.min.min(elapsed);
-        data.max = data.max.max(elapsed);
-    }
-}
-
-impl QueryStatGuard {
-    pub(crate) fn miss(&self) {
-        let mut data = self.bucket.data.lock();
-        data.missing += 1;
-    }
-}
-
-/// Statistics about the analyzers
-#[derive(Default)]
-pub struct AnalysisStats {
-    pub(crate) query_stats: FxDashMap<FileId, FxDashMap<&'static str, QueryStatBucket>>,
-}
-
-impl AnalysisStats {
-    /// Report the statistics of the analysis.
-    pub fn report(&self) -> String {
-        let stats = &self.query_stats;
-        let mut data = Vec::new();
-        for refs in stats.iter() {
-            let id = refs.key();
-            let queries = refs.value();
-            for refs2 in queries.iter() {
-                let query = refs2.key();
-                let bucket = refs2.value().data.lock().clone();
-                let name = format!("{id:?}:{query}").replace('\\', "/");
-                data.push((name, bucket));
-            }
-        }
-
-        // sort by query duration
-        data.sort_by(|x, y| y.1.max.cmp(&x.1.max));
-
-        // format to html
-
-        let mut html = String::new();
-        html.push_str(r#"<div>
-<style>
-table.analysis-stats { width: 100%; border-collapse: collapse; }
-table.analysis-stats th, table.analysis-stats td { border: 1px solid black; padding: 8px; text-align: center; }
-table.analysis-stats th.name-column, table.analysis-stats td.name-column { text-align: left; }
-table.analysis-stats tr:nth-child(odd) { background-color: rgba(242, 242, 242, 0.8); }
-@media (prefers-color-scheme: dark) {
-    table.analysis-stats tr:nth-child(odd) { background-color: rgba(50, 50, 50, 0.8); }
-}
-</style>
-<table class="analysis-stats"><tr><th class="query-column">Query</th><th>Count</th><th>Missing</th><th>Total</th><th>Min</th><th>Max</th></tr>"#);
-
-        for (name, bucket) in data {
-            html.push_str("<tr>");
-            html.push_str(&format!(r#"<td class="query-column">{name}</td>"#));
-            html.push_str(&format!("<td>{}</td>", bucket.query));
-            html.push_str(&format!("<td>{}</td>", bucket.missing));
-            html.push_str(&format!("<td>{:?}</td>", bucket.total));
-            html.push_str(&format!("<td>{:?}</td>", bucket.min));
-            html.push_str(&format!("<td>{:?}</td>", bucket.max));
-            html.push_str("</tr>");
-        }
-        html.push_str("</table>");
-        html.push_str("</div>");
-
-        html
-    }
-}

crates/tinymist-query/src/code_action.rs

@@ -127,6 +127,7 @@ mod tests {
         source: &Source,
         request_range: &LspRange,
     ) -> CodeActionContext {
+        // todo: reuse world compute graph APIs.
         let Warned {
             output,
             warnings: compiler_warnings,
@@ -144,8 +145,8 @@ mod tests {
         CodeActionContext {
             // The filtering here matches the LSP specification and VS Code behavior;
             // see https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#codeActionContext:
-            //   `diagnostics`: An array of diagnostics known on the client side overlapping the range
-            //   provided to the textDocument/codeAction request [...]
+            //   `diagnostics`: An array of diagnostics known on the client side overlapping the
+            // range   provided to the textDocument/codeAction request [...]
             diagnostics: diagnostics
                 .filter(|diag| ranges_overlap(&diag.range, request_range))
                 .collect(),

crates/tinymist-query/src/lib.rs

@@ -48,7 +48,7 @@ pub mod index;
 pub mod package;
 pub mod syntax;
 pub mod testing;
-pub use tinymist_analysis::{ty, upstream};
+pub use tinymist_analysis::{stats::GLOBAL_STATS, ty, upstream};
 
 /// The physical position in a document.
 pub type FramePosition = typst::layout::Position;

crates/tinymist-query/src/syntax/docs.rs

@@ -1,6 +1,7 @@
 use std::{collections::BTreeMap, ops::Deref, sync::LazyLock};
 
 use ecow::eco_format;
+use tinymist_analysis::stats::GLOBAL_STATS;
 use typst::foundations::{IntoValue, Module, Str, Type};
 
 use crate::{StrRef, adt::interner::Interned};
@@ -20,6 +21,8 @@ pub(crate) fn do_compute_docstring(
     docs: String,
     kind: DefKind,
 ) -> Option<DocString> {
+    let _guard = GLOBAL_STATS.stat(fid, "compute_docstring");
+
     let checker = DocsChecker {
         fid,
         ctx,

crates/tinymist-query/src/tests.rs

@@ -460,19 +460,6 @@ pub(crate) fn file_path_(uri: &lsp_types::Url) -> String {
     unix_slash(&rel_path)
 }
 
-pub struct HashRepr<T>(pub T);
-
-// sha256
-impl fmt::Display for HashRepr<JsonRepr> {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        use sha2::{Digest, Sha256};
-
-        let res = self.0.to_string();
-        let hash = Sha256::digest(res).to_vec();
-        write!(f, "sha256:{}", hex::encode(hash))
-    }
-}
-
 /// Extension methods for `Regex` that operate on `Cow<str>` instead of `&str`.
 pub trait RegexCowExt {
     /// [`Regex::replace_all`], but taking text as `Cow<str>` instead of `&str`.