indexer: llvm::sys::fs::UniqueID -> clang::FileID

Fix a minor issue that an empty included file is not recorded.

Note, we need to skip invalid uid2lid_and_path because
otherwise lid2path may contain invalid entries with lid: 0,
which will cause the file entry with file_id 0 to have an empty path.

src/clang_tu.cc

@@ -42,42 +42,30 @@ static Pos decomposed2LineAndCol(const SourceManager &sm,
 }
 
 Range fromCharSourceRange(const SourceManager &sm, const LangOptions &lang,
-                          CharSourceRange csr,
+                          CharSourceRange csr, FileID *fid) {
-                          llvm::sys::fs::UniqueID *uniqueID) {
   SourceLocation bloc = csr.getBegin(), eloc = csr.getEnd();
   std::pair<FileID, unsigned> binfo = sm.getDecomposedLoc(bloc),
                               einfo = sm.getDecomposedLoc(eloc);
   if (csr.isTokenRange())
     einfo.second += Lexer::MeasureTokenLength(eloc, sm, lang);
-  if (uniqueID) {
+  if (fid)
-    if (const FileEntry *F = sm.getFileEntryForID(binfo.first))
+    *fid = binfo.first;
-      *uniqueID = F->getUniqueID();
-    else
-      *uniqueID = llvm::sys::fs::UniqueID(0, 0);
-  }
   return {decomposed2LineAndCol(sm, binfo), decomposed2LineAndCol(sm, einfo)};
 }
 
-Range fromCharRange(const SourceManager &sm, const LangOptions &lang,
-                    SourceRange sr, llvm::sys::fs::UniqueID *uniqueID) {
-  return fromCharSourceRange(sm, lang, CharSourceRange::getCharRange(sr),
-                             uniqueID);
-}
-
 Range fromTokenRange(const SourceManager &sm, const LangOptions &lang,
-                     SourceRange sr, llvm::sys::fs::UniqueID *uniqueID) {
+                     SourceRange sr, FileID *fid) {
-  return fromCharSourceRange(sm, lang, CharSourceRange::getTokenRange(sr),
+  return fromCharSourceRange(sm, lang, CharSourceRange::getTokenRange(sr), fid);
-                             uniqueID);
 }
 
 Range fromTokenRangeDefaulted(const SourceManager &sm, const LangOptions &lang,
-                              SourceRange sr, const FileEntry *fe, Range range) {
+                              SourceRange sr, FileID fid, Range range) {
   auto decomposed = sm.getDecomposedLoc(sm.getExpansionLoc(sr.getBegin()));
-  if (sm.getFileEntryForID(decomposed.first) == fe)
+  if (decomposed.first == fid)
     range.start = decomposed2LineAndCol(sm, decomposed);
   SourceLocation sl = sm.getExpansionLoc(sr.getEnd());
   decomposed = sm.getDecomposedLoc(sl);
-  if (sm.getFileEntryForID(decomposed.first) == fe) {
+  if (decomposed.first == fid) {
     decomposed.second += Lexer::MeasureTokenLength(sl, sm, lang);
     range.end = decomposed2LineAndCol(sm, decomposed);
   }

src/clang_tu.hh

@@ -22,20 +22,16 @@ std::string pathFromFileEntry(const clang::FileEntry &file);
 
 Range fromCharSourceRange(const clang::SourceManager &sm,
                           const clang::LangOptions &lang,
-                          clang::CharSourceRange sr,
+                          clang::CharSourceRange csr,
-                          llvm::sys::fs::UniqueID *uniqueID = nullptr);
+                          clang::FileID *fid = nullptr);
-
-Range fromCharRange(const clang::SourceManager &sm,
-                    const clang::LangOptions &lang, clang::SourceRange sr,
-                    llvm::sys::fs::UniqueID *uniqueID = nullptr);
 
 Range fromTokenRange(const clang::SourceManager &sm,
                      const clang::LangOptions &lang, clang::SourceRange sr,
-                     llvm::sys::fs::UniqueID *uniqueID = nullptr);
+                     clang::FileID *fid = nullptr);
 
 Range fromTokenRangeDefaulted(const clang::SourceManager &sm,
                               const clang::LangOptions &lang,
-                              clang::SourceRange sr, const clang::FileEntry *fe,
+                              clang::SourceRange sr, clang::FileID fid,
                               Range range);
 
 std::unique_ptr<clang::CompilerInvocation>

src/indexer.cc

@@ -40,8 +40,8 @@ struct File {
 };
 
 struct IndexParam {
-  std::unordered_map<llvm::sys::fs::UniqueID, File> uid2file;
+  std::unordered_map<FileID, File> uid2file;
-  std::unordered_map<llvm::sys::fs::UniqueID, bool> uid2multi;
+  std::unordered_map<FileID, bool> uid2multi;
   struct DeclInfo {
     Usr usr;
     std::string short_name;
@@ -54,14 +54,17 @@ struct IndexParam {
   bool no_linkage;
   IndexParam(VFS &vfs, bool no_linkage) : vfs(vfs), no_linkage(no_linkage) {}
 
-  void seenFile(const FileEntry &file) {
+  void seenFile(FileID fid) {
     // If this is the first time we have seen the file (ignoring if we are
     // generating an index for it):
-    auto [it, inserted] = uid2file.try_emplace(file.getUniqueID());
+    auto [it, inserted] = uid2file.try_emplace(fid);
     if (inserted) {
-      std::string path = pathFromFileEntry(file);
+      const FileEntry *fe = ctx->getSourceManager().getFileEntryForID(fid);
+      if (!fe)
+        return;
+      std::string path = pathFromFileEntry(*fe);
       it->second.path = path;
-      it->second.mtime = file.getModificationTime();
+      it->second.mtime = fe->getModificationTime();
       if (!it->second.mtime)
         if (auto tim = lastWriteTime(path))
           it->second.mtime = *tim;
@@ -75,15 +78,16 @@ struct IndexParam {
     }
   }
 
-  IndexFile *consumeFile(const FileEntry &fe) {
+  IndexFile *consumeFile(FileID fid) {
-    seenFile(fe);
+    seenFile(fid);
-    return uid2file[fe.getUniqueID()].db.get();
+    return uid2file[fid].db.get();
   }
 
-  bool useMultiVersion(const FileEntry &fe) {
+  bool useMultiVersion(FileID fid) {
-    auto it = uid2multi.try_emplace(fe.getUniqueID());
+    auto it = uid2multi.try_emplace(fid);
     if (it.second)
-      it.first->second = multiVersionMatcher->matches(pathFromFileEntry(fe));
+      if (const FileEntry *fe = ctx->getSourceManager().getFileEntryForID(fid))
+        it.first->second = multiVersionMatcher->matches(pathFromFileEntry(*fe));
     return it.first->second;
   }
 };
@@ -609,27 +613,24 @@ public:
     }
   }
 
-  static int getFileLID(IndexFile *db, SourceManager &sm, const FileEntry &fe) {
+  static int getFileLID(IndexFile *db, SourceManager &sm, FileID fid) {
-    auto [it, inserted] = db->uid2lid_and_path.try_emplace(fe.getUniqueID());
+    auto [it, inserted] = db->uid2lid_and_path.try_emplace(fid);
     if (inserted) {
-      it->second.first = db->uid2lid_and_path.size() - 1;
+      const FileEntry *fe = sm.getFileEntryForID(fid);
-      SmallString<256> path = fe.tryGetRealPathName();
+      if (!fe) {
-      if (path.empty())
+        it->second.first = -1;
-        path = fe.getName();
-      if (!llvm::sys::path::is_absolute(path) &&
-          !sm.getFileManager().makeAbsolutePath(path))
         return -1;
-      it->second.second = llvm::sys::path::convert_to_slash(path.str());
+      }
+      it->second.first = db->uid2lid_and_path.size() - 1;
+      it->second.second = pathFromFileEntry(*fe);
     }
     return it->second.first;
   }
 
   void addMacroUse(IndexFile *db, SourceManager &sm, Usr usr, Kind kind,
                    SourceLocation sl) const {
-    const FileEntry *FE = sm.getFileEntryForID(sm.getFileID(sl));
+    FileID fid = sm.getFileID(sl);
-    if (!FE)
+    int lid = getFileLID(db, sm, fid);
-      return;
-    int lid = getFileLID(db, sm, *FE);
     if (lid < 0)
       return;
     Range spell = fromTokenRange(sm, ctx->getLangOpts(), SourceRange(sl, sl));
@@ -693,27 +694,25 @@ public:
     const LangOptions &lang = ctx->getLangOpts();
     FileID fid;
     SourceLocation spell = sm.getSpellingLoc(src_loc);
-    const FileEntry *fe;
     Range loc;
     auto r = sm.isMacroArgExpansion(src_loc)
                  ? CharSourceRange::getTokenRange(spell)
                  : sm.getExpansionRange(src_loc);
     loc = fromCharSourceRange(sm, lang, r);
     fid = sm.getFileID(r.getBegin());
-    fe = sm.getFileEntryForID(fid);
+    if (fid.isInvalid())
-    if (!fe)
       return true;
     int lid = -1;
     IndexFile *db;
-    if (g_config->index.multiVersion && param.useMultiVersion(*fe)) {
+    if (g_config->index.multiVersion && param.useMultiVersion(fid)) {
-      db = param.consumeFile(*sm.getFileEntryForID(sm.getMainFileID()));
+      db = param.consumeFile(sm.getMainFileID());
       if (!db)
         return true;
-      param.seenFile(*fe);
+      param.seenFile(fid);
       if (!sm.isWrittenInMainFile(r.getBegin()))
-        lid = getFileLID(db, sm, *fe);
+        lid = getFileLID(db, sm, fid);
     } else {
-      db = param.consumeFile(*fe);
+      db = param.consumeFile(fid);
       if (!db)
         return true;
     }
@@ -775,13 +774,13 @@ public:
       if (is_def) {
         SourceRange sr = origD->getSourceRange();
         entity->def.spell = {use,
-                             fromTokenRangeDefaulted(sm, lang, sr, fe, loc)};
+                             fromTokenRangeDefaulted(sm, lang, sr, fid, loc)};
         entity->def.parent_kind = SymbolKind::File;
         getKind(cast<Decl>(sem_dc), entity->def.parent_kind);
       } else if (is_decl) {
         SourceRange sr = origD->getSourceRange();
         entity->declarations.push_back(
-            {use, fromTokenRangeDefaulted(sm, lang, sr, fe, loc)});
+            {use, fromTokenRangeDefaulted(sm, lang, sr, fid, loc)});
       } else {
         entity->uses.push_back(use);
         return;
@@ -1077,6 +1076,11 @@ class IndexPPCallbacks : public PPCallbacks {
 public:
   IndexPPCallbacks(SourceManager &sm, IndexParam &param)
       : sm(sm), param(param) {}
+  void FileChanged(SourceLocation sl, FileChangeReason reason,
+                   SrcMgr::CharacteristicKind, FileID) override {
+    if (reason == FileChangeReason::EnterFile)
+      (void)param.consumeFile(sm.getFileID(sl));
+  }
   void InclusionDirective(SourceLocation hashLoc, const Token &tok,
                           StringRef included, bool isAngled,
                           CharSourceRange filenameRange, const FileEntry *file,
@@ -1087,11 +1091,8 @@ public:
       return;
     auto spell = fromCharSourceRange(sm, param.ctx->getLangOpts(),
                                      filenameRange, nullptr);
-    const FileEntry *fe =
+    FileID fid = sm.getFileID(filenameRange.getBegin());
-        sm.getFileEntryForID(sm.getFileID(filenameRange.getBegin()));
+    if (IndexFile *db = param.consumeFile(fid)) {
-    if (!fe)
-      return;
-    if (IndexFile *db = param.consumeFile(*fe)) {
       std::string path = pathFromFileEntry(*file);
       if (path.size())
         db->includes.push_back({spell.start.line, intern(path)});
@@ -1100,10 +1101,8 @@ public:
   void MacroDefined(const Token &tok, const MacroDirective *md) override {
     const LangOptions &lang = param.ctx->getLangOpts();
     SourceLocation sl = md->getLocation();
-    const FileEntry *fe = sm.getFileEntryForID(sm.getFileID(sl));
+    FileID fid = sm.getFileID(sl);
-    if (!fe)
+    if (IndexFile *db = param.consumeFile(fid)) {
-      return;
-    if (IndexFile *db = param.consumeFile(*fe)) {
       auto [name, usr] = getMacro(tok);
       IndexVar &var = db->toVar(usr);
       Range range = fromTokenRange(sm, lang, {sl, sl}, nullptr);
@@ -1128,15 +1127,12 @@ public:
   }
   void MacroExpands(const Token &tok, const MacroDefinition &, SourceRange sr,
                     const MacroArgs *) override {
-    llvm::sys::fs::UniqueID uniqueID;
     SourceLocation sl = sm.getSpellingLoc(sr.getBegin());
-    const FileEntry *fe = sm.getFileEntryForID(sm.getFileID(sl));
+    FileID fid = sm.getFileID(sl);
-    if (!fe)
+    if (IndexFile *db = param.consumeFile(fid)) {
-      return;
-    if (IndexFile *db = param.consumeFile(*fe)) {
       IndexVar &var = db->toVar(getMacro(tok).second);
       var.uses.push_back(
-          {{fromTokenRange(sm, param.ctx->getLangOpts(), {sl, sl}, &uniqueID),
+          {{fromTokenRange(sm, param.ctx->getLangOpts(), {sl, sl}, nullptr),
             Role::Dynamic}});
     }
   }
@@ -1150,8 +1146,9 @@ public:
   void SourceRangeSkipped(SourceRange sr, SourceLocation) override {
     Range range = fromCharSourceRange(sm, param.ctx->getLangOpts(),
                                       CharSourceRange::getCharRange(sr));
-    if (const FileEntry *fe = sm.getFileEntryForID(sm.getFileID(sr.getBegin())))
+    FileID fid = sm.getFileID(sr.getBegin());
-      if (IndexFile *db = param.consumeFile(*fe))
+    if (fid.isValid())
+      if (IndexFile *db = param.consumeFile(fid))
         db->skipped_ranges.push_back(range);
   }
 };
@@ -1321,7 +1318,8 @@ index(SemaManager *manager, WorkingFiles *wfiles, VFS *vfs,
     entry->import_file = main;
     entry->args = args;
     for (auto &[_, it] : entry->uid2lid_and_path)
-      entry->lid2path.emplace_back(it.first, std::move(it.second));
+      if (it.first >= 0)
+        entry->lid2path.emplace_back(it.first, std::move(it.second));
     entry->uid2lid_and_path.clear();
     for (auto &it : entry->usr2func) {
       // e.g. declaration + out-of-line definition
@@ -1341,6 +1339,8 @@ index(SemaManager *manager, WorkingFiles *wfiles, VFS *vfs,
     // Update dependencies for the file.
     for (auto &[_, file] : param.uid2file) {
       const std::string &path = file.path;
+      if (path.empty())
+        continue;
       if (path == entry->path)
         entry->mtime = file.mtime;
       else if (path != entry->import_file)

src/indexer.hh

@@ -8,7 +8,7 @@
 #include "serializer.hh"
 #include "utils.hh"
 
-#include <clang/Basic/FileManager.h>
+#include <clang/Basic/SourceLocation.h>
 #include <clang/Basic/Specifiers.h>
 #include <llvm/ADT/CachedHashString.h>
 #include <llvm/ADT/DenseMap.h>
@@ -19,12 +19,8 @@
 #include <vector>
 
 namespace std {
-template <> struct hash<llvm::sys::fs::UniqueID> {
+template <> struct hash<clang::FileID> {
-  std::size_t operator()(llvm::sys::fs::UniqueID ID) const {
+  std::size_t operator()(clang::FileID fid) const { return fid.getHashValue(); }
-    size_t ret = ID.getDevice();
-    ccls::hash_combine(ret, ID.getFile());
-    return ret;
-  }
 };
 } // namespace std
 
@@ -294,7 +290,7 @@ struct IndexFile {
   bool no_linkage;
 
   // uid2lid_and_path is used to generate lid2path, but not serialized.
-  std::unordered_map<llvm::sys::fs::UniqueID, std::pair<int, std::string>>
+  std::unordered_map<clang::FileID, std::pair<int, std::string>>
       uid2lid_and_path;
   std::vector<std::pair<int, std::string>> lid2path;