chore: Add APIs to get text from AST nodes

Author: robertoraggi

src/frontend/cxx/cxx_document.cc

@@ -21,13 +21,8 @@
 #include "cxx_document.h"
 
 #include <cxx/ast.h>
-#include <cxx/ast_slot.h>
-#include <cxx/ast_visitor.h>
 #include <cxx/control.h>
 #include <cxx/gcc_linux_toolchain.h>
-#include <cxx/lexer.h>
-#include <cxx/lsp/enums.h>
-#include <cxx/lsp/requests.h>
 #include <cxx/lsp/types.h>
 #include <cxx/macos_toolchain.h>
 #include <cxx/preprocessor.h>
@@ -209,43 +204,28 @@ auto CxxDocument::translationUnit() const -> TranslationUnit* {
   return &d->unit;
 }
 
-namespace {
-
-struct Visit {
-  CxxDocument* doc_;
-  std::function<bool(AST*)> visitor;
-  ASTSlot slotInfo;
-
-  void preVisit(AST* ast) {
-    if (!ast) return;
-
-    if (visitor(ast)) return;
-
-    // do a pre-visit using the low-level AST API
+auto CxxDocument::textOf(AST* ast) -> std::optional<std::string_view> {
+  return textInRange(ast->firstSourceLocation(), ast->lastSourceLocation());
+}
 
-    const auto slotCount = slotInfo(ast, 0).slotCount;
+auto CxxDocument::textInRange(SourceLocation start, SourceLocation end)
+    -> std::optional<std::string_view> {
+  auto& unit = d->unit;
+  auto preprocessor = unit.preprocessor();
 
-    for (int index = 0; index < slotCount; ++index) {
-      const auto childInfo = slotInfo(ast, index);
+  const auto startToken = unit.tokenAt(start);
+  const auto endToken = unit.tokenAt(end.previous());
 
-      if (childInfo.kind == ASTSlotKind::kNode) {
-        auto child = reinterpret_cast<AST*>(childInfo.handle);
-        if (child) preVisit(child);
-      } else if (childInfo.kind == ASTSlotKind::kNodeList) {
-        auto list = reinterpret_cast<List<AST*>*>(childInfo.handle);
-        for (auto node : ListView{list}) {
-          preVisit(node);
-        }
-      }
-    }
+  if (startToken.fileId() != endToken.fileId()) {
+    return std::nullopt;
   }
-};
 
-}  // namespace
+  std::string_view source = preprocessor->source(startToken.fileId());
+
+  const auto offset = startToken.offset();
+  const auto length = endToken.offset() + endToken.length() - offset;
 
-void CxxDocument::preVisit(std::function<bool(AST*)> visitor) {
-  auto ast = d->unit.ast();
-  Visit{this, std::move(visitor)}.preVisit(ast);
+  return source.substr(offset, length);
 }
 
 }  // namespace cxx::lsp

src/frontend/cxx/cxx_document.h

@@ -23,8 +23,7 @@
 #include <cxx/lsp/fwd.h>
 #include <cxx/translation_unit.h>
 
-#include <functional>
-#include <memory>
+#include <optional>
 #include <string>
 
 #include "cli.h"
@@ -43,7 +42,10 @@ class CxxDocument {
 
   [[nodiscard]] auto translationUnit() const -> TranslationUnit*;
 
-  void preVisit(std::function<bool(AST*)> visitor);
+  [[nodiscard]] auto textOf(AST* ast) -> std::optional<std::string_view>;
+
+  [[nodiscard]] auto textInRange(SourceLocation start, SourceLocation end)
+      -> std::optional<std::string_view>;
 
  private:
   struct Private;

src/frontend/cxx/lsp_server.cc

@@ -20,9 +20,14 @@
 
 #include "lsp_server.h"
 
+#include <cxx/ast.h>
 #include <cxx/lsp/enums.h>
 #include <cxx/lsp/requests.h>
 #include <cxx/lsp/types.h>
+#include <cxx/name_printer.h>
+#include <cxx/preprocessor.h>
+#include <cxx/symbols.h>
+#include <cxx/type_printer.h>
 #include <utf8/unchecked.h>
 
 #include <format>
@@ -433,12 +438,15 @@ void Server::operator()(DocumentSymbolRequest request) {
 
   auto uri = request.params().textDocument().uri();
   auto doc = latestDocument(uri);
+  auto id = request.id();
 
-  withUnsafeJson([&](json storage) {
-    DocumentSymbolResponse response(storage);
-    response.id(request.id());
-    (void)response.result();
-    sendToClient(response);
+  run([=, this] {
+    withUnsafeJson([&](json storage) {
+      DocumentSymbolResponse response(storage);
+      response.id(id);
+      (void)response.result();
+      sendToClient(response);
+    });
   });
 }
 
@@ -462,7 +470,7 @@ void Server::operator()(SetTraceNotification notification) {
   trace_ = notification.params().value();
 
   if (trace_ != TraceValue::kOff) {
-    logTrace("Trace level set to {}", to_string(trace_));
+    logTrace(std::format("Trace level set to {}", to_string(trace_)));
     return;
   }
 }

src/parser/cxx/ast_cursor.cc

@@ -22,13 +22,10 @@
 #include <cxx/ast_cursor.h>
 #include <cxx/ast_slot.h>
 
-#include <format>
-#include <iostream>
-
 namespace cxx {
 
 ASTCursor::ASTCursor(AST* root, std::string_view name) {
-  stack_.emplace_back(root, name);
+  stack_.push_back(Node{root, name});
 }
 
 ASTCursor::~ASTCursor() {}
@@ -57,12 +54,13 @@ void ASTCursor::step() {
         auto name = to_string(slot.nameIndex);
         if (slot.kind == ASTSlotKind::kNode) {
           auto node = reinterpret_cast<AST*>(slot.handle);
-          stack_.emplace_back(node, name);
+          stack_.push_back(Node{node, name});
         } else if (slot.kind == ASTSlotKind::kNodeList) {
           auto list = reinterpret_cast<List<AST*>*>(slot.handle);
-          stack_.emplace_back(list, name);
+          stack_.push_back(Node{list, name});
         }
       }
+
       break;
     } else if (auto list = std::get_if<List<AST*>*>(&node)) {
       if (!*list) continue;
@@ -74,8 +72,9 @@ void ASTCursor::step() {
       }
 
       for (const auto& ast : children | std::views::reverse) {
-        stack_.emplace_back(ast, kind);
+        stack_.push_back(Node{ast, kind});
       }
+
       break;
     }
   }

tests/unit_tests/lsp/document_symbols.yml

@@ -0,0 +1,21 @@
+# RUN: %cxx -lsp-test < %s | %filecheck %s
+
+{ "method": "initialize", "id": 0 }
+
+# CHECK: "id": 0
+
+{ "method": "textDocument/didOpen", "id": 1, "params": { "textDocument": { "uri": "test:///source.cc", "version": 0, "text": "#include <cstdio>\nauto main() -> int;" } } }
+
+{ "method": "$/setTrace", "id": 2, "params": { "value": "verbose" } }
+
+{ "method": "textDocument/documentSymbol", "id": 3, "params": { "textDocument": { "uri": "test:///source.cc" } } }
+
+# CHECK: "message": "Did receive DocumentSymbolRequest"
+# CHECK: "id": 3
+
+{ "method": "shutdown", "id": 4 }
+
+{ "method": "exit" }
+
+
+