filesystem.cc: deduplicate with UniqueID

Author: MaskRay

src/filesystem.cc

@@ -3,30 +3,55 @@ using namespace llvm;
 
 #include "utils.h"
 
+#include <set>
 #include <vector>
 
 void GetFilesInFolder(std::string folder,
                       bool recursive,
                       bool dir_prefix,
                       const std::function<void(const std::string&)>& handler) {
   EnsureEndsInSlash(folder);
-  std::vector<std::string> st{folder};
-  while (st.size()) {
-    std::error_code ec;
-    folder = st.back();
-    st.pop_back();
-    for (sys::fs::directory_iterator I(folder, ec), E; I != E && !ec;
-         I.increment(ec)) {
-      auto Status = I->status();
-      if (!Status) continue;
-      std::string path = I->path(), filename = sys::path::filename(path);
-      if (filename[0] != '.' || filename == ".ccls") {
-        if (sys::fs::is_regular_file(*Status)) {
+  sys::fs::file_status Status;
+  if (sys::fs::status(folder, Status, true))
+    return;
+  sys::fs::UniqueID ID;
+  std::vector<std::string> curr{folder};
+  std::vector<std::pair<std::string, sys::fs::file_status>> succ;
+  std::set<sys::fs::UniqueID> seen{Status.getUniqueID()};
+  while (curr.size() || succ.size()) {
+    if (curr.empty()) {
+      for (auto& it : succ)
+        if (!seen.count(it.second.getUniqueID()))
+          curr.push_back(std::move(it.first));
+      succ.clear();
+    } else {
+      std::error_code ec;
+      std::string folder1 = curr.back();
+      curr.pop_back();
+      for (sys::fs::directory_iterator I(folder1, ec, false), E; I != E && !ec;
+           I.increment(ec)) {
+        std::string path = I->path(), filename = sys::path::filename(path);
+        if ((filename[0] == '.' && filename != ".ccls") ||
+            sys::fs::status(path, Status, false))
+          continue;
+        if (sys::fs::is_symlink_file(Status)) {
+          if (sys::fs::status(path, Status, true))
+            continue;
+          if (sys::fs::is_directory(Status)) {
+            if (recursive)
+              succ.emplace_back(path, Status);
+            continue;
+          }
+        }
+        if (sys::fs::is_regular_file(Status)) {
           if (!dir_prefix)
             path = path.substr(folder.size());
           handler(path);
-        } else if (recursive && sys::fs::is_directory(*Status))
-          st.push_back(path);
+        } else if (recursive && sys::fs::is_directory(Status) &&
+                   !seen.count(ID = Status.getUniqueID())) {
+          curr.push_back(path);
+          seen.insert(ID);
+        }
       }
     }
   }