Fix unaligned load/store; add index.multiVersion prototype, rename index.onParse to index.OnOpen

Don't call getFieldOffset() on RD->isInvalidDecl()

src/config.h

@@ -137,8 +137,8 @@ struct Config {
     //   xxx: at most every xxx milliseconds
     int frequencyMs = 0;
 
-    // If true, diagnostics from a full document parse will be reported.
+    // If true, diagnostics will be reported in textDocument/didOpen.
-    bool onParse = true;
+    bool onOpen = true;
 
     // If true, diagnostics from typing will be reported.
     bool onType = true;
@@ -159,14 +159,6 @@ struct Config {
   } highlight;
 
   struct Index {
-    // Attempt to convert calls of make* functions to constructors based on
-    // hueristics.
-    //
-    // For example, this will show constructor calls for std::make_unique
-    // invocations. Specifically, ccls will try to attribute a ctor call
-    // whenever the function name starts with make (ignoring case).
-    bool attributeMakeCallsToCtor = true;
-
     // If a translation unit's absolute path matches any EMCAScript regex in the
     // whitelist, or does not match any regex in the blacklist, it will be
     // indexed. To only index files in the whitelist, add ".*" to the blacklist.
@@ -184,6 +176,9 @@ struct Config {
     // If false, the indexer will be disabled.
     bool enabled = true;
 
+    // If not 0, a file will be indexed in each tranlation unit that includes it.
+    int multiVersion = 0;
+
     // Allow indexing on textDocument/didChange.
     // May be too slow for big projects, so it is off by default.
     bool onDidChange = false;
@@ -226,12 +221,11 @@ MAKE_REFLECT_STRUCT(Config::Completion, caseSensitivity, dropOldRequests,
                     detailedLabel, filterAndSort, includeBlacklist,
                     includeMaxPathSize, includeSuffixWhitelist,
                     includeWhitelist);
-MAKE_REFLECT_STRUCT(Config::Diagnostics, blacklist, frequencyMs, onParse,
+MAKE_REFLECT_STRUCT(Config::Diagnostics, blacklist, frequencyMs, onOpen,
                     onType, whitelist)
 MAKE_REFLECT_STRUCT(Config::Highlight, lsRanges, blacklist, whitelist)
-MAKE_REFLECT_STRUCT(Config::Index, attributeMakeCallsToCtor, blacklist,
+MAKE_REFLECT_STRUCT(Config::Index, blacklist, comments, enabled, multiVersion,
-                    comments, enabled, onDidChange, reparseForDependency,
+                    onDidChange, reparseForDependency, threads, whitelist);
-                    threads, whitelist);
 MAKE_REFLECT_STRUCT(Config::WorkspaceSymbol, caseSensitivity, maxNum, sort);
 MAKE_REFLECT_STRUCT(Config::Xref, container, maxNum);
 MAKE_REFLECT_STRUCT(Config, compilationDatabaseCommand,

src/indexer.cc

@@ -48,9 +48,7 @@ struct IndexParam {
 
   IndexParam(FileConsumer *file_consumer) : file_consumer(file_consumer) {}
 
-  IndexFile *ConsumeFile(const FileEntry &File) {
+  void SeenFile(const FileEntry &File) {
-    IndexFile *db = file_consumer->TryConsumeFile(File, &file_contents);
-
     // If this is the first time we have seen the file (ignoring if we are
     // generating an index for it):
     auto [it, inserted] = SeenUniqueID.try_emplace(File.getUniqueID());
@@ -65,8 +63,11 @@ struct IndexParam {
       if (write_time)
         file2write_time[file_name] = *write_time;
     }
+  }
 
-    return db;
+  IndexFile *ConsumeFile(const FileEntry &File) {
+    SeenFile(File);
+    return file_consumer->TryConsumeFile(File, &file_contents);
   }
 };
 
@@ -392,20 +393,20 @@ public:
     return it->second.usr;
   }
 
-  Use GetUse(IndexFile *db, Range range, const DeclContext *DC,
+  Use GetUse(IndexFile *db, int lid, Range range, const DeclContext *DC,
              Role role) const {
     if (!DC)
-      return Use{{range, 0, SymbolKind::File, role}};
+      return {{range, 0, SymbolKind::File, role}, lid};
     const Decl *D = cast<Decl>(DC);
     switch (GetSymbolKind(D)) {
     case SymbolKind::Func:
-      return Use{{range, db->ToFunc(GetUsr(D)).usr, SymbolKind::Func, role}};
+      return {{range, db->ToFunc(GetUsr(D)).usr, SymbolKind::Func, role}, lid};
     case SymbolKind::Type:
-      return Use{{range, db->ToType(GetUsr(D)).usr, SymbolKind::Type, role}};
+      return {{range, db->ToType(GetUsr(D)).usr, SymbolKind::Type, role}, lid};
     case SymbolKind::Var:
-      return Use{{range, db->ToVar(GetUsr(D)).usr, SymbolKind::Var, role}};
+      return {{range, db->ToVar(GetUsr(D)).usr, SymbolKind::Var, role}, lid};
     default:
-      return Use{{range, 0, SymbolKind::File, role}};
+      return {{range, 0, SymbolKind::File, role}, lid};
     }
   }
 
@@ -535,26 +536,32 @@ public:
     }
   }
 
+  static int GetFileLID(IndexFile *db, SourceManager &SM, const FileEntry &FE) {
+    auto [it, inserted] = db->uid2lid_and_path.try_emplace(FE.getUniqueID());
+    if (inserted) {
+      it->second.first = db->uid2lid_and_path.size() - 1;
+      SmallString<256> Path = FE.tryGetRealPathName();
+      if (Path.empty())
+        Path = FE.getName();
+      if (!llvm::sys::path::is_absolute(Path) &&
+          !SM.getFileManager().makeAbsolutePath(Path))
+        return -1;
+      it->second.second = Path.str();
+    }
+    return it->second.first;
+  }
+
   void AddMacroUse(IndexFile *db, SourceManager &SM, Usr usr, SymbolKind kind,
                    SourceLocation Spell) const {
     const FileEntry *FE = SM.getFileEntryForID(SM.getFileID(Spell));
     if (!FE)
       return;
-    auto UID = FE->getUniqueID();
+    int lid = GetFileLID(db, SM, *FE);
-    auto [it, inserted] = db->uid2lid_and_path.try_emplace(UID);
+    if (lid < 0)
-    if (inserted) {
-      it->second.first = db->uid2lid_and_path.size() - 1;
-      SmallString<256> Path = FE->tryGetRealPathName();
-      if (Path.empty())
-        Path = FE->getName();
-      if (!llvm::sys::path::is_absolute(Path) &&
-          !SM.getFileManager().makeAbsolutePath(Path))
       return;
-      it->second.second = Path.str();
-    }
     Range spell =
         FromTokenRange(SM, Ctx->getLangOpts(), SourceRange(Spell, Spell));
-    Use use{{spell, 0, SymbolKind::File, Role::Dynamic}, it->second.first};
+    Use use{{spell, 0, SymbolKind::File, Role::Dynamic}, lid};
     switch (kind) {
     case SymbolKind::Func:
       db->ToFunc(usr).uses.push_back(use);
@@ -615,15 +622,26 @@ public:
     FE = SM.getFileEntryForID(LocFID);
     if (!FE)
       return true;
-    IndexFile *db = param.ConsumeFile(*FE);
+    int lid = -1;
+    IndexFile *db;
+    if (g_config->index.multiVersion) {
+      db = param.ConsumeFile(*SM.getFileEntryForID(SM.getMainFileID()));
       if (!db)
         return true;
+      param.SeenFile(*FE);
+      if (!SM.isInMainFile(R.getBegin()))
+        lid = GetFileLID(db, SM, *FE);
+    } else {
+      db = param.ConsumeFile(*FE);
+      if (!db)
+        return true;
+    }
 
     const Decl *OrigD = ASTNode.OrigD;
     const DeclContext *SemDC = OrigD->getDeclContext();
     const DeclContext *LexDC = ASTNode.ContainerDC;
     Role role = static_cast<Role>(Roles);
-    db->language = std::max(db->language, GetDeclLanguage(OrigD));
+    db->language = LanguageId((int)db->language | (int)GetDeclLanguage(OrigD));
 
     bool is_decl = Roles & uint32_t(index::SymbolRole::Declaration);
     bool is_def = Roles & uint32_t(index::SymbolRole::Definition);
@@ -638,18 +656,18 @@ public:
 
     auto do_def_decl = [&](auto *entity) {
       if (is_def) {
-        entity->def.spell = GetUse(db, loc, SemDC, role);
+        entity->def.spell = GetUse(db, lid, loc, SemDC, role);
         SourceRange R = OrigD->getSourceRange();
         entity->def.extent =
-            GetUse(db,
+            GetUse(db, lid,
                    R.getBegin().isFileID()
                        ? FromTokenRange(SM, Lang, OrigD->getSourceRange())
                        : loc,
                    LexDC, Role::None);
       } else if (is_decl) {
-        entity->declarations.push_back(GetUse(db, loc, LexDC, role));
+        entity->declarations.push_back(GetUse(db, lid, loc, LexDC, role));
       } else {
-        entity->uses.push_back(GetUse(db, loc, LexDC, role));
+        entity->uses.push_back(GetUse(db, lid, loc, LexDC, role));
         return;
       }
       if (entity->def.comments[0] == '\0' && g_config->index.comments)
@@ -735,10 +753,11 @@ public:
               if (SM.getFileID(R1.getBegin()) == LocFID) {
                 IndexType &type1 = db->ToType(usr1);
                 SourceLocation L1 = D1->getLocation();
-                type1.def.spell = GetUse(db, FromTokenRange(SM, Lang, {L1, L1}),
+                type1.def.spell =
-                                         SemDC, Role::Definition);
+                    GetUse(db, lid, FromTokenRange(SM, Lang, {L1, L1}), SemDC,
-                type1.def.extent =
+                           Role::Definition);
-                    GetUse(db, FromTokenRange(SM, Lang, R1), LexDC, Role::None);
+                type1.def.extent = GetUse(db, lid, FromTokenRange(SM, Lang, R1),
+                                          LexDC, Role::None);
                 type1.def.detailed_name = Intern(info1->short_name);
                 type1.def.short_name_size = int16_t(info1->short_name.size());
                 type1.def.kind = lsSymbolKind::TypeParameter;
@@ -753,10 +772,10 @@ public:
         // e.g. lambda parameter
         SourceLocation L = OrigD->getLocation();
         if (SM.getFileID(L) == LocFID) {
-          var->def.spell = GetUse(db, FromTokenRange(SM, Lang, {L, L}), SemDC,
+          var->def.spell = GetUse(db, lid, FromTokenRange(SM, Lang, {L, L}),
-                                  Role::Definition);
+                                  SemDC, Role::Definition);
           var->def.extent =
-              GetUse(db, FromTokenRange(SM, Lang, OrigD->getSourceRange()),
+              GetUse(db, lid, FromTokenRange(SM, Lang, OrigD->getSourceRange()),
                      LexDC, Role::None);
         }
       }
@@ -853,10 +872,10 @@ public:
             std::tie(RD, offset) = Stack.back();
             Stack.pop_back();
             if (!RD->isCompleteDefinition() || RD->isDependentType() ||
-                !ValidateRecord(RD))
+                RD->isInvalidDecl() || !ValidateRecord(RD))
               offset = -1;
             for (FieldDecl *FD : RD->fields()) {
-              int offset1 = offset >= 0 ? offset + Ctx->getFieldOffset(FD) : -1;
+              int offset1 = offset < 0 ? -1 : offset + Ctx->getFieldOffset(FD);
               if (FD->getIdentifier())
                 type->def.vars.emplace_back(GetUsr(FD), offset1);
               else if (const auto *RT1 = FD->getType()->getAs<RecordType>()) {
@@ -913,7 +932,8 @@ public:
             SourceLocation L1 = TSI->getTypeLoc().getBeginLoc();
             if (SM.getFileID(L1) == LocFID) {
               Range loc1 = FromTokenRange(SM, Lang, {L1, L1});
-              type1.uses.push_back(GetUse(db, loc1, LexDC, Role::Reference));
+              type1.uses.push_back(
+                  GetUse(db, lid, loc1, LexDC, Role::Reference));
             }
           }
         }
@@ -1066,8 +1086,7 @@ public:
     if (!FE)
       return;
     if (IndexFile *db = param.ConsumeFile(*FE)) {
-      auto [Name, usr] = GetMacro(Tok);
+      IndexVar &var = db->ToVar(GetMacro(Tok).second);
-      IndexVar &var = db->ToVar(usr);
       var.uses.push_back(
           {{FromTokenRange(SM, param.Ctx->getLangOpts(), {L, L}, &UniqueID), 0,
             SymbolKind::File, Role::Dynamic}});

src/indexer.h

@@ -229,7 +229,7 @@ struct IndexFile {
   std::string path;
   std::vector<std::string> args;
   int64_t last_write_time = 0;
-  LanguageId language = LanguageId::Unknown;
+  LanguageId language = LanguageId::C;
 
   // uid2lid_and_path is used to generate lid2path, but not serialized.
   std::unordered_map<llvm::sys::fs::UniqueID, std::pair<int, std::string>>

src/language.h

@@ -10,7 +10,7 @@
 // Used to identify the language at a file level. The ordering is important, as
 // a file previously identified as `C`, will be changed to `Cpp` if it
 // encounters a c++ declaration.
-enum class LanguageId { Unknown = 0, C = 1, Cpp = 2, ObjC = 3, ObjCpp = 4 };
+enum class LanguageId { Unknown = -1, C = 0, Cpp = 1, ObjC = 2, ObjCpp = 3 };
 MAKE_REFLECT_TYPE_PROXY(LanguageId);
 
 LanguageId SourceFileLanguage(std::string_view path);

src/messages/textDocument_didOpen.cc

@@ -66,7 +66,7 @@ struct Handler_TextDocumentDidOpen
     }
 
     clang_complete->NotifyView(path);
-    if (g_config->diagnostics.onParse)
+    if (g_config->diagnostics.onOpen)
       clang_complete->DiagnosticsUpdate({params.textDocument.uri});
   }
 };

src/query.cc

@@ -17,7 +17,7 @@
 
 namespace {
 
-void AssignFileId(const Lid2file_id &, int file_id, SymbolRef &ref) {
+void AssignFileId(const Lid2file_id &lid2file_id, int file_id, SymbolRef &ref) {
   if (ref.kind == SymbolKind::File)
     ref.usr = file_id;
 }

src/serializers/binary.h

@@ -5,13 +5,14 @@
 
 #include "serializer.h"
 
-#include <assert.h>
+#include <string.h>
 
 class BinaryReader : public Reader {
   const char *p_;
 
   template <typename T> T Get() {
-    auto ret = *reinterpret_cast<const T *>(p_);
+    T ret;
+    memcpy(&ret, p_, sizeof(T));
     p_ += sizeof(T);
     return ret;
   }
@@ -77,7 +78,7 @@ class BinaryWriter : public Writer {
   template <typename T> void Pack(T x) {
     auto i = buf_.size();
     buf_.resize(i + sizeof(x));
-    *reinterpret_cast<T *>(buf_.data() + i) = x;
+    memcpy(buf_.data() + i, &x, sizeof(x));
   }
 
   void VarUInt(uint64_t n) {