Simplify semantic highlighting

src/config.h

@@ -134,7 +134,7 @@ struct Config {
     // auto-completion. An example value is { ".h", ".hpp" }
     //
     // This is significantly faster than using a regex.
-    std::vector<std::string> includeSuffixWhitelist = {".h", ".hpp", ".hh"};
+    std::vector<std::string> includeSuffixWhitelist = {".h", ".hpp", ".hh", ".inc"};
 
     std::vector<std::string> includeWhitelist;
   } completion;

src/message_handler.cc

@@ -4,6 +4,7 @@
 #include "message_handler.h"
 
 #include "log.hh"
+#include "match.h"
 #include "pipeline.hh"
 #include "project.h"
 #include "query_utils.h"
@@ -47,74 +48,34 @@ struct ScanLineEvent {
 };
 } // namespace
 
-SemanticHighlightSymbolCache::Entry::Entry(
+int SemanticHighlight::GetStableId(SymbolKind kind, Usr usr) {
-    SemanticHighlightSymbolCache *all_caches, const std::string &path)
+  decltype(func2id) *map;
-    : all_caches_(all_caches), path(path) {}
-
-std::optional<int> SemanticHighlightSymbolCache::Entry::TryGetStableId(
-    SymbolKind kind, const std::string &detailed_name) {
-  TNameToId *map = GetMapForSymbol_(kind);
-  auto it = map->find(detailed_name);
-  if (it != map->end())
-    return it->second;
-
-  return std::nullopt;
-}
-
-int SemanticHighlightSymbolCache::Entry::GetStableId(
-    SymbolKind kind, const std::string &detailed_name) {
-  std::optional<int> id = TryGetStableId(kind, detailed_name);
-  if (id)
-    return *id;
-
-  // Create a new id. First try to find a key in another map.
-  all_caches_->cache_.IterateValues([&](const std::shared_ptr<Entry> &entry) {
-    std::optional<int> other_id = entry->TryGetStableId(kind, detailed_name);
-    if (other_id) {
-      id = other_id;
-      return false;
-    }
-    return true;
-  });
-
-  // Create a new id.
-  TNameToId *map = GetMapForSymbol_(kind);
-  if (!id)
-    id = all_caches_->next_stable_id_++;
-  return (*map)[detailed_name] = *id;
-}
-
-SemanticHighlightSymbolCache::Entry::TNameToId *
-SemanticHighlightSymbolCache::Entry::GetMapForSymbol_(SymbolKind kind) {
   switch (kind) {
-  case SymbolKind::Type:
-    return &detailed_type_name_to_stable_id;
   case SymbolKind::Func:
-    return &detailed_func_name_to_stable_id;
+    map = &func2id;
+    break;
+  case SymbolKind::Type:
+    map = &type2id;
+    break;
   case SymbolKind::Var:
-    return &detailed_var_name_to_stable_id;
+    map = &var2id;
+    break;
   case SymbolKind::File:
   case SymbolKind::Invalid:
-    break;
+    llvm_unreachable("");
   }
-  assert(false);
+
-  return nullptr;
+  auto it = map->try_emplace(usr, next_id);
+  if (it.second)
+    next_id++;
+  return it.first->second;
 }
 
-SemanticHighlightSymbolCache::SemanticHighlightSymbolCache()
+void SemanticHighlight::Init() {
-    : cache_(kCacheSize) {}
-
-void SemanticHighlightSymbolCache::Init() {
   match_ = std::make_unique<GroupMatch>(g_config->highlight.whitelist,
                                         g_config->highlight.blacklist);
 }
 
-std::shared_ptr<SemanticHighlightSymbolCache::Entry>
-SemanticHighlightSymbolCache::GetCacheForFile(const std::string &path) {
-  return cache_.Get(
-      path, [&, this]() { return std::make_shared<Entry>(this, path); });
-}
-
 MessageHandler::MessageHandler() {
   // Dynamically allocate |message_handlers|, otherwise there will be static
   // initialization order races.
@@ -184,15 +145,12 @@ void EmitSkippedRanges(WorkingFile *working_file,
   pipeline::WriteStdout(kMethodType_CclsPublishSkippedRanges, out);
 }
 
-void EmitSemanticHighlighting(DB *db,
+void EmitSemanticHighlighting(DB *db, SemanticHighlight *highlight,
-                              SemanticHighlightSymbolCache *semantic_cache,
                               WorkingFile *wfile, QueryFile *file) {
   assert(file->def);
   if (wfile->buffer_content.size() > g_config->largeFileSize ||
-      !semantic_cache->match_->IsMatch(file->def->path))
+      !highlight->match_->IsMatch(file->def->path))
     return;
-  auto semantic_cache_for_file =
-      semantic_cache->GetCacheForFile(file->def->path);
 
   // Group symbols together.
   std::unordered_map<SymbolIdx, Out_CclsPublishSemanticHighlighting::Symbol>
@@ -288,8 +246,7 @@ void EmitSemanticHighlighting(DB *db,
         it->second.lsRanges.push_back(*loc);
       } else {
         Out_CclsPublishSemanticHighlighting::Symbol symbol;
-        symbol.stableId = semantic_cache_for_file->GetStableId(
+        symbol.stableId = highlight->GetStableId(sym.kind, sym.usr);
-            sym.kind, std::string(detailed_name));
         symbol.parentKind = parent_kind;
         symbol.kind = kind;
         symbol.storage = storage;

src/message_handler.h

@@ -5,7 +5,6 @@
 
 #include "lru_cache.h"
 #include "lsp.h"
-#include "match.h"
 #include "method.h"
 #include "query.h"
 
@@ -17,6 +16,7 @@
 struct CompletionManager;
 struct Config;
 class DiagnosticsPublisher;
+struct GroupMatch;
 struct VFS;
 struct IncludeComplete;
 struct MultiQueueWaiter;
@@ -27,35 +27,13 @@ struct WorkingFiles;
 
 // Caches symbols for a single file for semantic highlighting to provide
 // relatively stable ids. Only supports xxx files at a time.
-struct SemanticHighlightSymbolCache {
+struct SemanticHighlight {
-  struct Entry {
+  llvm::DenseMap<Usr, int, DenseMapInfoForUsr> func2id, type2id, var2id;
-    SemanticHighlightSymbolCache *all_caches_ = nullptr;
+  uint32_t next_id = 0;
-
-    // The path this cache belongs to.
-    std::string path;
-    // Detailed symbol name to stable id.
-    using TNameToId = std::unordered_map<std::string, int>;
-    TNameToId detailed_type_name_to_stable_id;
-    TNameToId detailed_func_name_to_stable_id;
-    TNameToId detailed_var_name_to_stable_id;
-
-    Entry(SemanticHighlightSymbolCache *all_caches, const std::string &path);
-
-    std::optional<int> TryGetStableId(SymbolKind kind,
-                                      const std::string &detailed_name);
-    int GetStableId(SymbolKind kind, const std::string &detailed_name);
-
-    TNameToId *GetMapForSymbol_(SymbolKind kind);
-  };
-
-  constexpr static int kCacheSize = 10;
-  LruCache<std::string, Entry> cache_;
-  uint32_t next_stable_id_ = 0;
   std::unique_ptr<GroupMatch> match_;
 
-  SemanticHighlightSymbolCache();
   void Init();
-  std::shared_ptr<Entry> GetCacheForFile(const std::string &path);
+  int GetStableId(SymbolKind kind, Usr usr);
 };
 
 struct Out_CclsPublishSemanticHighlighting
@@ -102,7 +80,7 @@ struct MessageHandler {
   Project *project = nullptr;
   DiagnosticsPublisher *diag_pub = nullptr;
   VFS *vfs = nullptr;
-  SemanticHighlightSymbolCache *semantic_cache = nullptr;
+  SemanticHighlight *highlight = nullptr;
   WorkingFiles *working_files = nullptr;
   CompletionManager *clang_complete = nullptr;
   IncludeComplete *include_complete = nullptr;
@@ -132,6 +110,5 @@ bool FindFileOrFail(DB *db, Project *project, std::optional<lsRequestId> id,
 void EmitSkippedRanges(WorkingFile *working_file,
                        const std::vector<Range> &skipped_ranges);
 
-void EmitSemanticHighlighting(DB *db,
+void EmitSemanticHighlighting(DB *db, SemanticHighlight *highlight,
-                              SemanticHighlightSymbolCache *semantic_cache,
                               WorkingFile *working_file, QueryFile *file);

src/messages/initialize.cc

@@ -476,7 +476,7 @@ struct Handler_Initialize : BaseMessageHandler<In_InitializeRequest> {
 
     diag_pub->Init();
     idx::Init();
-    semantic_cache->Init();
+    highlight->Init();
 
     // Open up / load the project.
     project->Load(project_path);

src/messages/textDocument_didOpen.cc

@@ -48,7 +48,7 @@ struct Handler_TextDocumentDidOpen
     FindFileOrFail(db, project, std::nullopt, path, &file);
     if (file && file->def) {
       EmitSkippedRanges(working_file, file->def->skipped_ranges);
-      EmitSemanticHighlighting(db, semantic_cache, working_file, file);
+      EmitSemanticHighlighting(db, highlight, working_file, file);
     }
 
     include_complete->AddFile(working_file->filename);

src/pipeline.cc

@@ -8,6 +8,7 @@
 #include "include_complete.h"
 #include "log.hh"
 #include "lsp.h"
+#include "match.h"
 #include "message_handler.h"
 #include "pipeline.hh"
 #include "platform.h"
@@ -358,7 +359,7 @@ void Indexer_Main(CompletionManager *completion, VFS *vfs, Project *project,
       indexer_waiter->Wait(index_request);
 }
 
-void Main_OnIndexed(DB *db, SemanticHighlightSymbolCache *semantic_cache,
+void Main_OnIndexed(DB *db, SemanticHighlight *highlight,
                     WorkingFiles *working_files, IndexUpdate *update) {
   if (update->refresh) {
     LOG_S(INFO)
@@ -369,7 +370,7 @@ void Main_OnIndexed(DB *db, SemanticHighlightSymbolCache *semantic_cache,
       if (db->name2file_id.find(filename) == db->name2file_id.end())
         continue;
       QueryFile *file = &db->files[db->name2file_id[filename]];
-      EmitSemanticHighlighting(db, semantic_cache, f.get(), file);
+      EmitSemanticHighlighting(db, highlight, f.get(), file);
     }
     return;
   }
@@ -389,7 +390,7 @@ void Main_OnIndexed(DB *db, SemanticHighlightSymbolCache *semantic_cache,
       wfile->SetIndexContent(g_config->index.onChange ? wfile->buffer_content
                                                       : def_u.second);
       EmitSkippedRanges(wfile, def_u.first.skipped_ranges);
-      EmitSemanticHighlighting(db, semantic_cache, wfile,
+      EmitSemanticHighlighting(db, highlight, wfile,
                                &db->files[update->file_id]);
     }
   }
@@ -460,7 +461,7 @@ void LaunchStdout() {
 
 void MainLoop() {
   Project project;
-  SemanticHighlightSymbolCache semantic_cache;
+  SemanticHighlight highlight;
   WorkingFiles working_files;
   VFS vfs;
   DiagnosticsPublisher diag_pub;
@@ -491,7 +492,7 @@ void MainLoop() {
     handler->project = &project;
     handler->diag_pub = &diag_pub;
     handler->vfs = &vfs;
-    handler->semantic_cache = &semantic_cache;
+    handler->highlight = &highlight;
     handler->working_files = &working_files;
     handler->clang_complete = &clang_complete;
     handler->include_complete = &include_complete;
@@ -518,7 +519,7 @@ void MainLoop() {
       if (!update)
         break;
       did_work = true;
-      Main_OnIndexed(&db, &semantic_cache, &working_files, &*update);
+      Main_OnIndexed(&db, &highlight, &working_files, &*update);
     }
 
     if (!did_work) {

src/test.cc

@@ -11,6 +11,8 @@
 #include "utils.h"
 
 #include <llvm/Config/llvm-config.h>
+#include <llvm/ADT/StringRef.h>
+using namespace llvm;
 
 #include <rapidjson/document.h>
 #include <rapidjson/prettywriter.h>
@@ -68,6 +70,12 @@ struct TextReplacer {
   }
 };
 
+void TrimInPlace(std::string &s) {
+  auto f = [](char c) { return !isspace(c); };
+  s.erase(s.begin(), std::find_if(s.begin(), s.end(), f));
+  s.erase(std::find_if(s.rbegin(), s.rend(), f).base(), s.end());
+}
+
 void ParseTestExpectation(
     const std::string &filename,
     const std::vector<std::string> &lines_with_endings, TextReplacer *replacer,
@@ -77,7 +85,7 @@ void ParseTestExpectation(
   {
     bool in_output = false;
     for (std::string line : lines_with_endings) {
-      TrimInPlace(line);
+      line = StringRef(line).trim().str();
 
       if (StartsWith(line, "EXTRA_FLAGS:")) {
         assert(!in_output && "multiple EXTRA_FLAGS sections");
@@ -113,8 +121,7 @@ void ParseTestExpectation(
         // one token assume it is a filename.
         std::vector<std::string> tokens = SplitString(line_with_ending, " ");
         if (tokens.size() > 1) {
-          active_output_filename = tokens[1];
+          active_output_filename = StringRef(tokens[1]).trim().str();
-          TrimInPlace(active_output_filename);
         } else {
           active_output_filename = filename;
         }

src/utils.cc

@@ -21,16 +21,6 @@ using namespace llvm;
 #include <string.h>
 #include <unordered_map>
 
-void TrimInPlace(std::string &s) {
-  auto f = [](char c) { return !isspace(c); };
-  s.erase(s.begin(), std::find_if(s.begin(), s.end(), f));
-  s.erase(std::find_if(s.rbegin(), s.rend(), f).base(), s.end());
-}
-std::string Trim(std::string s) {
-  TrimInPlace(s);
-  return s;
-}
-
 uint64_t HashUsr(std::string_view s) {
   union {
     uint64_t ret;

src/utils.h

@@ -16,9 +16,6 @@ namespace llvm {
 class StringRef;
 }
 
-void TrimInPlace(std::string &s);
-std::string Trim(std::string s);
-
 uint64_t HashUsr(std::string_view s);
 uint64_t HashUsr(llvm::StringRef s);