inefficient semanticTokens/range on the end of the document

[Techatrix]

Apply the following patch to semanticTokens/range and print which ast nodes will be traversed to generate semantic tokens.

diff --git a/src/Server.zig b/src/Server.zig
index 567de571..fec62c81 100644
--- a/src/Server.zig
+++ b/src/Server.zig
@@ -640,7 +640,7 @@ fn initializeHandler(server: *Server, arena: std.mem.Allocator, request: types.I
             },
             .semanticTokensProvider = .{
                 .SemanticTokensOptions = .{
-                    .full = .{ .bool = true },
+                    // .full = .{ .bool = true },
                     .range = .{ .bool = true },
                     .legend = .{
                         .tokenTypes = std.meta.fieldNames(semantic_tokens.TokenType),
diff --git a/src/features/semantic_tokens.zig b/src/features/semantic_tokens.zig
index dfd7b1a0..69fd55d8 100644
--- a/src/features/semantic_tokens.zig
+++ b/src/features/semantic_tokens.zig
@@ -1048,6 +1048,7 @@ pub fn writeSemanticTokens(
 
     // reverse the ast from the root declarations
     for (nodes) |child| {
+        std.debug.print("node: {d}\n", .{child});
         try writeNodeTokens(&builder, child);
     }

Then open a large file like Sema.zig or AstGen.zig and the move to the end of document. ZLS will generate tokens for the entire document.

node: 3
node: 5
node: 7
node: 9

...

node: 140721
node: 140880
node: 140903
node: 140907
node: 140911
node: 140915
node: 140919
node: 140923
debug (server): Took 6527ms to process request-15-textDocument/semanticTokens/range on Thread 79111