Remove <regex> from lex_utils.h and separate language_server_api.h

% time clang++ -fsyntax-only -std=c++11 a.cc

<iostream> => 0.35s
<regex> => 0.68s

src/cache_manager.cc

@@ -2,7 +2,7 @@
 
 #include "config.h"
 #include "indexer.h"
-#include "language_server_api.h"
+#include "lsp.h"
 #include "platform.h"
 
 #include <loguru/loguru.hpp>

src/clang_complete.h

@@ -3,7 +3,8 @@
 #include "atomic_object.h"
 #include "clang_index.h"
 #include "clang_translation_unit.h"
-#include "language_server_api_completion.h"
+#include "lsp_completion.h"
+#include "lsp_diagnostic.h"
 #include "lru_cache.h"
 #include "project.h"
 #include "threaded_queue.h"

src/clang_format.h

@@ -2,7 +2,7 @@
 
 #if USE_CLANG_CXX
 
-#include "language_server_api.h"
+#include "lsp.h"
 #include "working_files.h"
 
 #include <clang/Format/Format.h>

src/clang_utils.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "language_server_api.h"
+#include "lsp_diagnostic.h"
 
 #include <clang-c/Index.h>
 #if USE_CLANG_CXX

src/code_complete_cache.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "language_server_api_completion.h"
+#include "lsp_completion.h"
 
 #include <optional.h>
 

src/command_line.cc

@@ -7,7 +7,7 @@
 #include "import_pipeline.h"
 #include "include_complete.h"
 #include "indexer.h"
-#include "language_server_api.h"
+#include "lsp_diagnostic.h"
 #include "lex_utils.h"
 #include "lru_cache.h"
 #include "match.h"

src/import_pipeline.cc

@@ -4,7 +4,7 @@
 #include "config.h"
 #include "iindexer.h"
 #include "import_manager.h"
-#include "language_server_api.h"
+#include "lsp.h"
 #include "message_handler.h"
 #include "platform.h"
 #include "project.h"

src/include_complete.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "language_server_api_completion.h"
+#include "lsp_completion.h"
 
 #include <atomic>
 #include <mutex>

src/indexer.h

@@ -7,7 +7,7 @@
 #include "file_consumer.h"
 #include "file_contents.h"
 #include "language.h"
-#include "language_server_api.h"
+#include "lsp.h"
 #include "maybe.h"
 #include "nt_string.h"
 #include "performance.h"

src/lex_utils.cc

@@ -46,23 +46,6 @@ lsPosition CharPos(std::string_view search,
   return result;
 }
 
-ParseIncludeLineResult ParseIncludeLine(const std::string& line) {
-  static const std::regex pattern(
-      "(\\s*)"        // [1]: spaces before '#'
-      "#"             //
-      "(\\s*)"        // [2]: spaces after '#'
-      "([^\\s\"<]*)"  // [3]: "include"
-      "(\\s*)"        // [4]: spaces before quote
-      "([\"<])?"      // [5]: the first quote char
-      "([^\\s\">]*)"  // [6]: path of file
-      "[\">]?"        //
-      "(.*)");        // [7]: suffix after quote char
-  std::smatch match;
-  bool ok = std::regex_match(line, match, pattern);
-  std::string text = match[3].str() + match[6].str();
-  return {ok, text, match};
-}
-
 // TODO: eliminate |line_number| param.
 optional<lsRange> ExtractQuotedRange(int line_number, const std::string& line) {
   // Find starting and ending quote.

src/lex_utils.h

@@ -1,10 +1,9 @@
 #pragma once
 
-#include "language_server_api.h"
+#include "lsp.h"
 
 #include <string_view.h>
 
-#include <regex>
 #include <string>
 #include <tuple>
 
@@ -15,14 +14,6 @@ lsPosition CharPos(std::string_view search,
                    char character,
                    int character_offset = 0);
 
-struct ParseIncludeLineResult {
-  bool ok;
-  std::string text;  // include the "include" part
-  std::smatch match;
-};
-
-ParseIncludeLineResult ParseIncludeLine(const std::string& line);
-
 // TODO: eliminate |line_number| param.
 optional<lsRange> ExtractQuotedRange(int line_number, const std::string& line);
 

src/lsp.cc

@@ -1,4 +1,4 @@
-#include "language_server_api.h"
+#include "lsp.h"
 
 #include "recorder.h"
 #include "serializers/json.h"

src/lsp.h

@@ -292,110 +292,12 @@ struct lsFormattingOptions {
 };
 MAKE_REFLECT_STRUCT(lsFormattingOptions, tabSize, insertSpaces);
 
-enum class lsDiagnosticSeverity {
-  // Reports an error.
-  Error = 1,
-  // Reports a warning.
-  Warning = 2,
-  // Reports an information.
-  Information = 3,
-  // Reports a hint.
-  Hint = 4
-};
-MAKE_REFLECT_TYPE_PROXY(lsDiagnosticSeverity);
-
-struct lsDiagnostic {
-  // The range at which the message applies.
-  lsRange range;
-
-  // The diagnostic's severity. Can be omitted. If omitted it is up to the
-  // client to interpret diagnostics as error, warning, info or hint.
-  optional<lsDiagnosticSeverity> severity;
-
-  // The diagnostic's code. Can be omitted.
-  int code = 0;
-
-  // A human-readable string describing the source of this
-  // diagnostic, e.g. 'typescript' or 'super lint'.
-  std::string source = "cquery";
-
-  // The diagnostic's message.
-  std::string message;
-
-  // Non-serialized set of fixits.
-  std::vector<lsTextEdit> fixits_;
-};
-MAKE_REFLECT_STRUCT(lsDiagnostic, range, severity, source, message);
-
-enum class lsErrorCodes {
-  // Defined by JSON RPC
-  ParseError = -32700,
-  InvalidRequest = -32600,
-  MethodNotFound = -32601,
-  InvalidParams = -32602,
-  InternalError = -32603,
-  serverErrorStart = -32099,
-  serverErrorEnd = -32000,
-  ServerNotInitialized = -32002,
-  UnknownErrorCode = -32001,
-
-  // Defined by the protocol.
-  RequestCancelled = -32800,
-};
-MAKE_REFLECT_TYPE_PROXY(lsErrorCodes);
-struct Out_Error : public lsOutMessage<Out_Error> {
-  struct lsResponseError {
-    // A number indicating the error type that occurred.
-    lsErrorCodes code;
-
-    // A string providing a short description of the error.
-    std::string message;
-
-    // A Primitive or Structured value that contains additional
-    // information about the error. Can be omitted.
-    // optional<D> data;
-  };
-
-  lsRequestId id;
-
-  // The error object in case a request fails.
-  lsResponseError error;
-};
-MAKE_REFLECT_STRUCT(Out_Error::lsResponseError, code, message);
-MAKE_REFLECT_STRUCT(Out_Error, jsonrpc, id, error);
-
 // Cancel an existing request.
 struct Ipc_CancelRequest : public RequestMessage<Ipc_CancelRequest> {
   static const IpcId kIpcId = IpcId::CancelRequest;
 };
 MAKE_REFLECT_STRUCT(Ipc_CancelRequest, id);
 
-// Diagnostics
-struct Out_TextDocumentPublishDiagnostics
-    : public lsOutMessage<Out_TextDocumentPublishDiagnostics> {
-  struct Params {
-    // The URI for which diagnostic information is reported.
-    lsDocumentUri uri;
-
-    // An array of diagnostic information items.
-    std::vector<lsDiagnostic> diagnostics;
-  };
-
-  Params params;
-};
-template <typename TVisitor>
-void Reflect(TVisitor& visitor, Out_TextDocumentPublishDiagnostics& value) {
-  std::string method = "textDocument/publishDiagnostics";
-  REFLECT_MEMBER_START();
-  REFLECT_MEMBER(jsonrpc);
-  REFLECT_MEMBER2("method", method);
-  REFLECT_MEMBER(params);
-  REFLECT_MEMBER_END();
-}
-MAKE_REFLECT_STRUCT(Out_TextDocumentPublishDiagnostics::Params,
-                    uri,
-                    diagnostics);
-
 // MarkedString can be used to render human readable text. It is either a
 // markdown string or a code-block that provides a language and a code snippet.
 // The language identifier is sematically equal to the optional language

src/lsp_completion.h

@@ -1,5 +1,5 @@
 #pragma once
-#include "language_server_api.h"
+#include "lsp.h"
 
 // The kind of a completion entry.
 enum class lsCompletionItemKind {

src/lsp_diagnostic.cc

@@ -0,0 +1,20 @@
+#include "lsp_diagnostic.h"
+
+#include "queue_manager.h"
+#include "working_files.h"
+
+void EmitDiagnostics(WorkingFiles* working_files,
+                     std::string path,
+                     std::vector<lsDiagnostic> diagnostics) {
+  // Emit diagnostics.
+  Out_TextDocumentPublishDiagnostics out;
+  out.params.uri = lsDocumentUri::FromPath(path);
+  out.params.diagnostics = diagnostics;
+  QueueManager::WriteStdout(IpcId::TextDocumentPublishDiagnostics, out);
+
+  // Cache diagnostics so we can show fixits.
+  working_files->DoActionOnFile(path, [&](WorkingFile* working_file) {
+                                        if (working_file)
+                                          working_file->diagnostics_ = diagnostics;
+                                      });
+}

src/lsp_diagnostic.h

@@ -0,0 +1,106 @@
+#pragma once
+
+#include "lsp.h"
+
+enum class lsDiagnosticSeverity {
+  // Reports an error.
+  Error = 1,
+  // Reports a warning.
+  Warning = 2,
+  // Reports an information.
+  Information = 3,
+  // Reports a hint.
+  Hint = 4
+};
+MAKE_REFLECT_TYPE_PROXY(lsDiagnosticSeverity);
+
+struct lsDiagnostic {
+  // The range at which the message applies.
+  lsRange range;
+
+  // The diagnostic's severity. Can be omitted. If omitted it is up to the
+  // client to interpret diagnostics as error, warning, info or hint.
+  optional<lsDiagnosticSeverity> severity;
+
+  // The diagnostic's code. Can be omitted.
+  int code = 0;
+
+  // A human-readable string describing the source of this
+  // diagnostic, e.g. 'typescript' or 'super lint'.
+  std::string source = "cquery";
+
+  // The diagnostic's message.
+  std::string message;
+
+  // Non-serialized set of fixits.
+  std::vector<lsTextEdit> fixits_;
+};
+MAKE_REFLECT_STRUCT(lsDiagnostic, range, severity, source, message);
+
+enum class lsErrorCodes {
+  // Defined by JSON RPC
+  ParseError = -32700,
+  InvalidRequest = -32600,
+  MethodNotFound = -32601,
+  InvalidParams = -32602,
+  InternalError = -32603,
+  serverErrorStart = -32099,
+  serverErrorEnd = -32000,
+  ServerNotInitialized = -32002,
+  UnknownErrorCode = -32001,
+
+  // Defined by the protocol.
+  RequestCancelled = -32800,
+};
+MAKE_REFLECT_TYPE_PROXY(lsErrorCodes);
+struct Out_Error : public lsOutMessage<Out_Error> {
+  struct lsResponseError {
+    // A number indicating the error type that occurred.
+    lsErrorCodes code;
+
+    // A string providing a short description of the error.
+    std::string message;
+
+    // A Primitive or Structured value that contains additional
+    // information about the error. Can be omitted.
+    // optional<D> data;
+  };
+
+  lsRequestId id;
+
+  // The error object in case a request fails.
+  lsResponseError error;
+};
+MAKE_REFLECT_STRUCT(Out_Error::lsResponseError, code, message);
+MAKE_REFLECT_STRUCT(Out_Error, jsonrpc, id, error);
+
+// Diagnostics
+struct Out_TextDocumentPublishDiagnostics
+    : public lsOutMessage<Out_TextDocumentPublishDiagnostics> {
+  struct Params {
+    // The URI for which diagnostic information is reported.
+    lsDocumentUri uri;
+
+    // An array of diagnostic information items.
+    std::vector<lsDiagnostic> diagnostics;
+  };
+
+  Params params;
+};
+template <typename TVisitor>
+void Reflect(TVisitor& visitor, Out_TextDocumentPublishDiagnostics& value) {
+  std::string method = "textDocument/publishDiagnostics";
+  REFLECT_MEMBER_START();
+  REFLECT_MEMBER(jsonrpc);
+  REFLECT_MEMBER2("method", method);
+  REFLECT_MEMBER(params);
+  REFLECT_MEMBER_END();
+}
+MAKE_REFLECT_STRUCT(Out_TextDocumentPublishDiagnostics::Params,
+                    uri,
+                    diagnostics);
+
+struct WorkingFiles;
+void EmitDiagnostics(WorkingFiles* working_files,
+                     std::string path,
+                     std::vector<lsDiagnostic> diagnostics);

src/match.cc

@@ -1,6 +1,6 @@
 #include "match.h"
 
-#include "language_server_api.h"
+#include "lsp.h"
 #include "queue_manager.h"
 
 #include <doctest/doctest.h>

src/message_handler.h

@@ -1,7 +1,7 @@
 #pragma once
 
 #include "ipc.h"
-#include "language_server_api.h"
+#include "lsp.h"
 #include "query.h"
 
 #include <optional.h>

src/messages/cquery_random.cc

@@ -128,7 +128,7 @@ struct CqueryRandomHandler : BaseMessageHandler<Ipc_CqueryRandom> {
     for (int i = 0; i < n; i++) {
       sum += x[i];
       if (sum >= roulette) {
-        Maybe<Use> use = GetDefinitionExtentOfSymbol(db, syms[i]);
+        Maybe<Use> use = GetDefinitionExtent(db, syms[i]);
         if (!use)
           continue;
         if (auto ls_loc = GetLsLocationEx(db, working_files, *use,

src/messages/initialize.cc

@@ -498,7 +498,7 @@ struct InitializeHandler : BaseMessageHandler<Ipc_InitializeRequest> {
             Reflect(json_reader, *config);
           } catch (std::invalid_argument&) {
             // This will not trigger because parse error is handled in
-            // MessageRegistry::Parse in language_server_api.cc
+            // MessageRegistry::Parse in lsp.cc
           }
         }
 

src/messages/text_document_code_lens.cc

@@ -177,7 +177,7 @@ struct TextDocumentCodeLensHandler
           // extent since that is better for outline. This tries to convert the
           // extent location to the spelling location.
           auto try_ensure_spelling = [&](Use use) {
-            Maybe<Use> def = GetDefinitionSpellingOfSymbol(db, use);
+            Maybe<Use> def = GetDefinitionSpell(db, use);
             if (!def || def->range.start.line != use.range.start.line) {
               return use;
             }
@@ -214,7 +214,7 @@ struct TextDocumentCodeLensHandler
 
           // "Base"
           if (def->base.size() == 1) {
-            Maybe<Use> base_loc = GetDefinitionSpellingOfSymbol(
+            Maybe<Use> base_loc = GetDefinitionSpell(
                 db, SymbolIdx{def->base[0], SymbolKind::Func});
             if (base_loc) {
               optional<lsLocation> ls_base =

src/messages/text_document_completion.cc

@@ -10,6 +10,8 @@
 
 #include <loguru.hpp>
 
+#include <regex>
+
 namespace {
 
 // How a completion was triggered
@@ -106,6 +108,29 @@ void DecorateIncludePaths(const std::smatch& match,
   }
 }
 
+struct ParseIncludeLineResult {
+  bool ok;
+  std::string text;  // include the "include" part
+  std::smatch match;
+};
+
+ParseIncludeLineResult ParseIncludeLine(const std::string& line) {
+  static const std::regex pattern(
+      "(\\s*)"        // [1]: spaces before '#'
+      "#"             //
+      "(\\s*)"        // [2]: spaces after '#'
+      "([^\\s\"<]*)"  // [3]: "include"
+      "(\\s*)"        // [4]: spaces before quote
+      "([\"<])?"      // [5]: the first quote char
+      "([^\\s\">]*)"  // [6]: path of file
+      "[\">]?"        //
+      "(.*)");        // [7]: suffix after quote char
+  std::smatch match;
+  bool ok = std::regex_match(line, match, pattern);
+  std::string text = match[3].str() + match[6].str();
+  return {ok, text, match};
+}
+
 template <typename T>
 char* tofixedbase64(T input, char* out) {
   const char* digits =

src/messages/text_document_definition.cc

@@ -24,7 +24,7 @@ struct Out_TextDocumentDefinition
 };
 MAKE_REFLECT_STRUCT(Out_TextDocumentDefinition, jsonrpc, id, result);
 
-std::vector<Use> GetGotoDefinitionTargets(QueryDatabase* db, SymbolRef sym) {
+std::vector<Use> GetNonDefDeclarationTargets(QueryDatabase* db, SymbolRef sym) {
   switch (sym.kind) {
     case SymbolKind::Var: {
       std::vector<Use> ret = GetNonDefDeclarations(db, sym);
@@ -32,7 +32,7 @@ std::vector<Use> GetGotoDefinitionTargets(QueryDatabase* db, SymbolRef sym) {
       if (ret.empty()) {
         for (auto& def : db->GetVar(sym).def)
           if (def.type) {
-            if (Maybe<Use> use = GetDefinitionSpellingOfSymbol(
+            if (Maybe<Use> use = GetDefinitionSpell(
                     db, SymbolIdx{*def.type, SymbolKind::Type})) {
               ret.push_back(*use);
               break;
@@ -99,7 +99,7 @@ struct TextDocumentDefinitionHandler
 
       if (uses.empty()) {
         // The symbol has no definition or the cursor is on a definition.
-        uses = GetGotoDefinitionTargets(db, sym);
+        uses = GetNonDefDeclarationTargets(db, sym);
         // There is no declaration but the cursor is on a definition.
         if (uses.empty() && on_def)
           uses.push_back(*on_def);
@@ -143,7 +143,7 @@ struct TextDocumentDefinitionHandler
           auto pos = name.find(query);
           if (pos == std::string::npos)
             continue;
-          Maybe<Use> use = GetDefinitionSpellingOfSymbol(db, db->symbols[i]);
+          Maybe<Use> use = GetDefinitionSpell(db, db->symbols[i]);
           if (!use)
             continue;
 
@@ -157,7 +157,7 @@ struct TextDocumentDefinitionHandler
         }
         if (best_i != -1) {
           Maybe<Use> use =
-              GetDefinitionSpellingOfSymbol(db, db->symbols[best_i]);
+              GetDefinitionSpell(db, db->symbols[best_i]);
           assert(use);
           if (auto ls_loc = GetLsLocationEx(db, working_files, *use,
                                             config->xref.container))

src/messages/workspace_symbol.cc

@@ -23,7 +23,7 @@ bool InsertSymbolIntoResult(QueryDatabase* db,
   if (!info)
     return false;
 
-  Maybe<Use> location = GetDefinitionExtentOfSymbol(db, symbol);
+  Maybe<Use> location = GetDefinitionExtent(db, symbol);
   Use loc;
   if (location)
     loc = *location;

src/query_utils.cc

@@ -40,13 +40,13 @@ std::vector<Use> GetDeclarations(std::vector<Q>& entities,
 
 }  // namespace
 
-Maybe<Use> GetDefinitionSpellingOfSymbol(QueryDatabase* db, SymbolIdx sym) {
+Maybe<Use> GetDefinitionSpell(QueryDatabase* db, SymbolIdx sym) {
   Maybe<Use> ret;
   EachEntityDef(db, sym, [&](const auto& def) { return !(ret = def.spell); });
   return ret;
 }
 
-Maybe<Use> GetDefinitionExtentOfSymbol(QueryDatabase* db, SymbolIdx sym) {
+Maybe<Use> GetDefinitionExtent(QueryDatabase* db, SymbolIdx sym) {
   // Used to jump to file.
   if (sym.kind == SymbolKind::File)
     return Use(Range(Position(0, 0), Position(0, 0)), sym.id, sym.kind,
@@ -383,19 +383,3 @@ std::vector<SymbolRef> FindSymbolsAtLocation(WorkingFile* working_file,
 
   return symbols;
 }
-
-void EmitDiagnostics(WorkingFiles* working_files,
-                     std::string path,
-                     std::vector<lsDiagnostic> diagnostics) {
-  // Emit diagnostics.
-  Out_TextDocumentPublishDiagnostics out;
-  out.params.uri = lsDocumentUri::FromPath(path);
-  out.params.diagnostics = diagnostics;
-  QueueManager::WriteStdout(IpcId::TextDocumentPublishDiagnostics, out);
-
-  // Cache diagnostics so we can show fixits.
-  working_files->DoActionOnFile(path, [&](WorkingFile* working_file) {
-    if (working_file)
-      working_file->diagnostics_ = diagnostics;
-  });
-}

src/query_utils.h

@@ -5,8 +5,8 @@
 
 #include <optional.h>
 
-Maybe<Use> GetDefinitionSpellingOfSymbol(QueryDatabase* db, SymbolIdx sym);
-Maybe<Use> GetDefinitionExtentOfSymbol(QueryDatabase* db, SymbolIdx sym);
+Maybe<Use> GetDefinitionSpell(QueryDatabase* db, SymbolIdx sym);
+Maybe<Use> GetDefinitionExtent(QueryDatabase* db, SymbolIdx sym);
 Maybe<QueryFileId> GetDeclarationFileForSymbol(QueryDatabase* db,
                                                SymbolIdx sym);
 
@@ -53,10 +53,6 @@ std::vector<SymbolRef> FindSymbolsAtLocation(WorkingFile* working_file,
                                              QueryFile* file,
                                              lsPosition position);
 
-void EmitDiagnostics(WorkingFiles* working_files,
-                     std::string path,
-                     std::vector<lsDiagnostic> diagnostics);
-
 template <typename Fn>
 void WithEntity(QueryDatabase* db, SymbolIdx sym, Fn&& fn) {
   switch (sym.kind) {

src/queue_manager.cc

@@ -1,7 +1,7 @@
 #include "queue_manager.h"
 
 #include "cache_manager.h"
-#include "language_server_api.h"
+#include "lsp.h"
 #include "query.h"
 
 #include <sstream>

src/working_files.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "language_server_api.h"
+#include "lsp_diagnostic.h"
 #include "utils.h"
 
 #include <clang-c/Index.h>