fix(vscode): make "Run test" codelenses appear in the correct place and run the correct test (#2395)

Also add performance issue reproes in playground-server/tests

The "Run test" codelenses were appearing in the wrong place because
`str.chars()` iterates over unicode scalar values, not bytes, and people
often have complex utf8 scalars in their code because LLM outputs use
them! It's not just emojis: even just things like apostrophes, quotes
are often modelled as their fancy utf8 analog; and of course for emojis
there are things like [x] checkbox variants that will show up.

Running the correct test was actually fixed drive-by in #2358, but we
left in a bug where a test case with coverage for N functions would
render N^2 codelenses - that's now fixed. The issue was that the runtime
code didn't make it clear in its API that iterating over test cases
actually iterated over (fn, test) pairs, and so the language server code
would then expand those using test.parent_functions(), resulting in N^2
entries in list_testcases(). The internal APIs have been renamed to make
it much more explicit what's actually going on there.


<!-- ELLIPSIS_HIDDEN -->


----

> [!IMPORTANT]
> Fixes "Run test" codelenses in VSCode to appear and function
correctly, resolves N^2 rendering issue, and adds performance tests in
playground server.
> 
>   - **Behavior**:
> - Fixes "Run test" codelenses to appear correctly by iterating over
`char_indices` instead of `chars` in `span.rs`.
> - Resolves issue where test cases with N functions rendered N^2
codelenses by clarifying API in `ir/repr.rs` and `ir_features.rs`.
>   - **API Changes**:
> - Renames `walk_tests()` to `walk_function_test_pairs()` in
`ir/repr.rs`, `ir_features.rs`, and `test_executor/mod.rs`.
> - Changes `BamlTestCase` to `BamlFunctionTestCasePair` in
`baml_lsp_types.rs` and `baml_project/mod.rs`.
>   - **Playground Server**:
>     - Adds performance test cases in `playground-server/tests`.
> - Implements `load_project_from_directory()` in
`playground_barebones.rs` to load `.baml` files.
>   - **Misc**:
> - Updates `Cargo.toml` files to include new dependencies `chrono` and
`walkdir`.
> 
> <sup>This description was created by </sup>[<img alt="Ellipsis"
src="https://img.shields.io/badge/Ellipsis-blue?color=175173">](https://www.ellipsis.dev?ref=BoundaryML%2Fbaml&utm_source=github&utm_medium=referral)<sup>
for 8d12933. You can
[customize](https://app.ellipsis.dev/BoundaryML/settings/summaries) this
summary. It will automatically update as commits are pushed.</sup>


<!-- ELLIPSIS_HIDDEN -->

Author: sxlijin

.gitattributes

@@ -4,3 +4,4 @@
 **/baml_client/inlinedbaml.go binary
 **/baml_client/** merge=baml_client
 **/baml_client/** linguist-generated=true
+**/*.lfs.baml filter=lfs diff=lfs merge=lfs -text

engine/Cargo.lock

@@ -772,7 +772,7 @@ impl IntermediateRepr {
         self.expr_fns.iter().map(|e| Walker { ir: self, item: e })
     }
 
-    pub fn walk_tests(
+    pub fn walk_function_test_pairs(
         &self,
     ) -> impl Iterator<Item = Walker<'_, (&Node<Function>, &Node<TestCase>)>> {
         self.functions.iter().flat_map(move |f| {

engine/baml-lib/baml-core/src/ir/repr.rs

@@ -119,7 +119,7 @@ flowchart TD
   n50 --> n51
   n49 --> n50
   n48 --> n49
-%%__BAML_SPANMAP__={"n0":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":3,"start":4,"end_line":4,"end":8},"n1":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":6,"start":12,"end_line":7,"end":15},"n2":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":9,"start":19,"end_line":14,"end":0},"n3":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":16,"start":9,"end_line":17,"end":4},"n4":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":17,"start":8,"end_line":18,"end":3},"n5":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":18,"start":7,"end_line":19,"end":2},"n6":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":19,"start":6,"end_line":20,"end":1},"n7":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":20,"start":5,"end_line":21,"end":0},"n8":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":21,"start":4,"end_line":21,"end":19},"n9":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":21,"start":23,"end_line":22,"end":19},"n10":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":23,"start":2,"end_line":23,"end":19},"n11":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":24,"start":1,"end_line":25,"end":14},"n12":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":29,"start":23,"end_line":30,"end":19},"n13":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":30,"start":23,"end_line":31,"end":19},"n14":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":32,"start":1,"end_line":33,"end":14},"n15":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":35,"start":23,"end_line":36,"end":19},"n16":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":36,"start":23,"end_line":41,"end":5},"n17":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":42,"start":23,"end_line":43,"end":19},"n18":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":48,"start":23,"end_line":50,"end":14},"n19":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":51,"start":23,"end_line":56,"end":5},"n20":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":60,"start":27,"end_line":62,"end":10},"n21":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":64,"start":0,"end_line":65,"end":6},"n22":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":66,"start":35,"end_line":67,"end":19},"n23":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":70,"start":31,"end_line":72,"end":2},"n24":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":74,"start":3,"end_line":74,"end":19},"n25":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":77,"start":27,"end_line":79,"end":5},"n26":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":81,"start":23,"end_line":85,"end":5},"n27":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":86,"start":23,"end_line":87,"end":19},"n28":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":87,"start":23,"end_line":88,"end":19},"n29":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":88,"start":23,"end_line":89,"end":19},"n30":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":89,"start":23,"end_line":90,"end":19},"n31":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":90,"start":23,"end_line":91,"end":19},"n32":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":91,"start":23,"end_line":93,"end":14},"n33":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":98,"start":4,"end_line":98,"end":19},"n34":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":99,"start":3,"end_line":100,"end":0},"n35":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":102,"start":3,"end_line":102,"end":19},"n36":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":103,"start":3,"end_line":103,"end":19},"n37":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":104,"start":3,"end_line":105,"end":14},"n38":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":114,"start":5,"end_line":115,"end":2},"n39":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":116,"start":4,"end_line":117,"end":9},"n40":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":119,"start":2,"end_line":121,"end":3},"n41":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":123,"start":1,"end_line":123,"end":14},"n42":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":124,"start":5,"end_line":124,"end":19},"n43":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":126,"start":4,"end_line":128,"end":3},"n44":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":130,"start":1,"end_line":130,"end":14},"n45":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":131,"start":3,"end_line":134,"end":10},"n46":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":135,"start":23,"end_line":137,"end":14},"n47":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":137,"start":23,"end_line":139,"end":14},"n48":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":144,"start":4,"end_line":145,"end":4},"n49":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":145,"start":8,"end_line":145,"end":19},"n50":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":145,"start":23,"end_line":146,"end":19},"n51":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":146,"start":23,"end_line":147,"end":19},"n52":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":147,"start":23,"end_line":148,"end":19},"n53":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":148,"start":23,"end_line":151,"end":0}}
+%%__BAML_SPANMAP__={"n0":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":3,"start":4,"end_line":4,"end":0},"n1":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":6,"start":4,"end_line":7,"end":0},"n2":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":9,"start":4,"end_line":10,"end":0},"n3":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":15,"start":4,"end_line":16,"end":0},"n4":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":16,"start":4,"end_line":17,"end":0},"n5":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":17,"start":4,"end_line":18,"end":0},"n6":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":18,"start":4,"end_line":19,"end":0},"n7":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":19,"start":4,"end_line":20,"end":0},"n8":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":20,"start":4,"end_line":21,"end":0},"n9":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":21,"start":4,"end_line":22,"end":0},"n10":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":22,"start":4,"end_line":23,"end":0},"n11":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":23,"start":4,"end_line":24,"end":0},"n12":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":29,"start":4,"end_line":30,"end":0},"n13":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":30,"start":4,"end_line":31,"end":0},"n14":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":31,"start":4,"end_line":32,"end":0},"n15":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":35,"start":4,"end_line":36,"end":0},"n16":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":36,"start":4,"end_line":37,"end":0},"n17":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":42,"start":4,"end_line":43,"end":0},"n18":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":48,"start":4,"end_line":49,"end":0},"n19":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":51,"start":4,"end_line":52,"end":0},"n20":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":60,"start":8,"end_line":61,"end":0},"n21":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":63,"start":12,"end_line":64,"end":0},"n22":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":66,"start":16,"end_line":67,"end":0},"n23":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":70,"start":12,"end_line":71,"end":0},"n24":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":73,"start":12,"end_line":74,"end":0},"n25":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":77,"start":8,"end_line":78,"end":0},"n26":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":81,"start":4,"end_line":82,"end":0},"n27":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":86,"start":4,"end_line":87,"end":0},"n28":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":87,"start":4,"end_line":88,"end":0},"n29":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":88,"start":4,"end_line":89,"end":0},"n30":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":89,"start":4,"end_line":90,"end":0},"n31":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":90,"start":4,"end_line":91,"end":0},"n32":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":91,"start":4,"end_line":92,"end":0},"n33":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":97,"start":4,"end_line":98,"end":0},"n34":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":98,"start":4,"end_line":99,"end":0},"n35":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":101,"start":4,"end_line":102,"end":0},"n36":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":102,"start":4,"end_line":103,"end":0},"n37":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":103,"start":4,"end_line":104,"end":0},"n38":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":113,"start":4,"end_line":114,"end":0},"n39":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":115,"start":8,"end_line":116,"end":0},"n40":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":118,"start":8,"end_line":119,"end":0},"n41":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":121,"start":4,"end_line":122,"end":0},"n42":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":123,"start":4,"end_line":124,"end":0},"n43":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":125,"start":8,"end_line":126,"end":0},"n44":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":128,"start":4,"end_line":129,"end":0},"n45":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":130,"start":4,"end_line":131,"end":0},"n46":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":135,"start":4,"end_line":136,"end":0},"n47":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":137,"start":4,"end_line":138,"end":0},"n48":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":143,"start":4,"end_line":144,"end":0},"n49":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":144,"start":4,"end_line":145,"end":0},"n50":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":145,"start":4,"end_line":146,"end":0},"n51":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":146,"start":4,"end_line":147,"end":0},"n52":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":147,"start":4,"end_line":148,"end":0},"n53":{"file_path":"tests/validation_files/headers/edge_cases.baml","start_line":148,"start":4,"end_line":149,"end":0}}
   click n0 call bamlMermaidNodeClick() "Go to source"
   click n1 call bamlMermaidNodeClick() "Go to source"
   click n2 call bamlMermaidNodeClick() "Go to source"

engine/baml-lib/baml/tests/validation_files/headers/edge_cases.mmd

@@ -1,7 +1,7 @@
-error: Error validating: This line is invalid. It does not start with any known Baml schema keyword.
-  -->  tests/validation_files/headers/invalid.baml:11
-   | 
-10 | 
-11 | fn MixedOnSameLine() -> string {
-12 |     let x = "hello" ## Inline Header // Should fail
-   | 
+error: Error validating: This line is invalid. It does not start with any known Baml schema keyword.
+  -->  tests/validation_files/headers/invalid.baml:11
+   | 
+10 | 
+11 | fn MixedOnSameLine() -> string {
+12 |     let x = "hello" ## Inline Header // Should fail
+   | 

engine/baml-lib/baml/tests/validation_files/headers/invalid.err

@@ -44,11 +44,11 @@ impl Span {
         let mut start = None;
         let mut end = None;
 
-        for (idx, c) in contents.chars().enumerate() {
-            if idx == self.start {
+        for (byte_pos, c) in contents.char_indices() {
+            if byte_pos == self.start {
                 start = Some((line, column));
             }
-            if idx == self.end {
+            if byte_pos == self.end {
                 end = Some((line, column));
                 break;
             }
@@ -114,3 +114,64 @@ impl SerializedSpan {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_line_number_basic() {
+        let source = SourceFile::new_static("test.baml".into(), "line0\nline1\nline2");
+
+        // First line
+        let span = Span::new(source.clone(), 0, 5);
+        assert_eq!(span.line_and_column(), ((0, 0), (0, 5)));
+
+        // Second line
+        let span = Span::new(source.clone(), 6, 11);
+        assert_eq!(span.line_and_column(), ((1, 0), (1, 5)));
+
+        // Spanning lines
+        let span = Span::new(source.clone(), 3, 8);
+        assert_eq!(span.line_and_column(), ((0, 3), (1, 2)));
+    }
+
+    #[test]
+    fn test_line_number_with_emoji() {
+        let source = SourceFile::new_static("test.baml".into(), "hi🎉\nworld");
+
+        // After emoji, before newline (byte 6 is the newline)
+        let span = Span::new(source.clone(), 0, 6);
+        assert_eq!(span.line_and_column(), ((0, 0), (0, 3)));
+
+        // Second line after emoji (newline at byte 6, "world" starts at byte 7)
+        let span = Span::new(source.clone(), 7, 12);
+        assert_eq!(span.line_and_column(), ((1, 0), (1, 5)));
+    }
+
+    #[test]
+    fn test_line_number_multiple_emojis() {
+        let source = SourceFile::new_static("test.baml".into(), "👋🎉🚀\ntest");
+
+        // Each emoji is 4 bytes, so first line is bytes 0-12
+        let span = Span::new(source.clone(), 0, 12);
+        assert_eq!(span.line_and_column(), ((0, 0), (0, 3)));
+
+        // Second line starts at byte 13
+        let span = Span::new(source.clone(), 13, 17);
+        assert_eq!(span.line_and_column(), ((1, 0), (1, 4)));
+    }
+
+    #[test]
+    fn test_line_number_middle_of_file() {
+        let source = SourceFile::new_static("test.baml".into(), "a\nb\nc\nd\ne\nf");
+
+        // Line 5 (0-indexed) - "f"
+        let span = Span::new(source.clone(), 10, 11);
+        assert_eq!(span.line_and_column(), ((5, 0), (5, 1)));
+
+        // Line 2 - "c"
+        let span = Span::new(source.clone(), 4, 5);
+        assert_eq!(span.line_and_column(), ((2, 0), (2, 1)));
+    }
+}

engine/baml-lib/diagnostics/src/span.rs

@@ -41,19 +41,23 @@ pub struct BamlParam {
 }
 
 #[derive(Clone, Debug, Serialize, Deserialize)]
-pub struct BamlTestCase {
+pub struct BamlFunctionTestCasePair {
     pub name: String,
     pub inputs: Vec<BamlParam>,
     pub error: Option<String>,
+    /// The span of `test TestFoo`
     pub span: BamlSpan,
-    pub parent_functions: Vec<BamlParentFunction>,
+    /// The span of `FnBar` in `functions [FnFoo, FnBar, FnBaz]`
+    /// This is wired through to allow re-sorting these as the caller desires.
+    pub function_name_span: Option<BamlSpan>,
+    pub function: BamlParentFunction,
 }
 
 #[derive(Clone, Debug, Serialize, Deserialize)]
 pub struct BamlFunction {
     pub name: String,
     pub span: BamlSpan,
-    pub test_cases: Vec<BamlTestCase>,
+    pub test_cases: Vec<BamlFunctionTestCasePair>,
     pub test_snippet: String,
     pub signature: String,
 }

engine/baml-lsp-types/src/baml_lsp_types.rs

@@ -729,7 +729,7 @@ impl WithRenderRawCurl for AwsClient {
         // pretty, multi-line JSON
         let input_json_str = serde_json::to_string_pretty(&serde_json::Value::Object(root))?;
         let input_json_escaped = escape(Cow::Borrowed(&input_json_str));
-        cmd.push(format!("--cli-input-json {}", input_json_escaped));
+        cmd.push(format!("--cli-input-json {input_json_escaped}"));
 
         Ok(cmd.join(" "))
     }

engine/baml-runtime/src/internal/llm_client/primitive/aws/aws_client.rs

@@ -18,6 +18,6 @@ impl WithInternal for InternalBamlRuntime {
     }
 
     fn walk_tests(&self) -> impl Iterator<Item = TestCaseWalker<'_>> {
-        self.ir().walk_tests()
+        self.ir().walk_function_test_pairs()
     }
 }

engine/baml-runtime/src/runtime/ir_features.rs

@@ -143,7 +143,7 @@ impl TestExecutor for BamlRuntime {
         let func_test_pairs = {
             let ir = &self.inner.ir;
             // Regular LLM function tests
-            let from_fn_tests = ir.walk_tests().filter_map(|node_pair| {
+            let from_fn_tests = ir.walk_function_test_pairs().filter_map(|node_pair| {
                 let (function_name, test_name) = node_pair.name();
                 if args.includes(function_name, test_name) {
                     Some((function_name.to_string(), test_name.to_string()))
@@ -206,7 +206,7 @@ impl TestExecutor for BamlRuntime {
         let selected_tests = {
             let ir = &self.inner.ir;
             // Regular LLM function tests
-            let from_fn_tests = ir.walk_tests().filter_map(|node_pair| {
+            let from_fn_tests = ir.walk_function_test_pairs().filter_map(|node_pair| {
                 let (function_name, test_name) = node_pair.name();
                 if args.includes(function_name, test_name) {
                     node_pair.span().map(|s| {