Improved get_similar_token_offsets to also include virtual and override functions

cmake/Modules/FindLibClang.cmake

@@ -14,21 +14,20 @@
 #               Known LLVM release numbers.
 
 # most recent versions come first
-set(LIBCLANG_KNOWN_LLVM_VERSIONS 4.0.0
+set(LIBCLANG_KNOWN_LLVM_VERSIONS 5.0.0 5.0
-  4.0
+  4.0.1
+  4.0.0_1 4.0.0 4.0
   3.9.1
-  3.9.0
+  3.9.0 3.9
-  3.9
   3.8.1
-  3.8
+  3.8.0 3.8
   3.7.1
   3.7
   3.6.2
   3.6.1
   3.6
   3.5.1
-  3.5.0                   #Arch Linux
+  3.5.0 3.5
-  3.5                     #LLVM Debian/Ubuntu packages from http://llvm.org/apt/
   3.4.2 3.4.1 3.4 3.3 3.2 3.1)
 
 set(libclang_llvm_header_search_paths)
@@ -64,6 +63,7 @@ endforeach()
 find_path(LIBCLANG_INCLUDE_DIR clang-c/Index.h
   PATHS ${libclang_llvm_header_search_paths}
   PATH_SUFFIXES LLVM/include #Windows package from http://llvm.org/releases/
+    llvm50/include llvm41/include llvm40/include llvm39/include llvm38/include llvm37/include llvm36/include # FreeBSD
   DOC "The path to the directory that contains clang-c/Index.h")
 
 # On Windows with MSVC, the import library uses the ".imp" file extension
@@ -77,6 +77,7 @@ endif()
 find_library(LIBCLANG_LIBRARY NAMES libclang.imp libclang clang
   PATHS ${libclang_llvm_lib_search_paths}
   PATH_SUFFIXES LLVM/lib #Windows package from http://llvm.org/releases/
+    llvm50/lib llvm41/lib llvm40/lib llvm39/lib llvm38/lib llvm37/lib llvm36/lib # FreeBSD
   DOC "The file that corresponds to the libclang library.")
 
 get_filename_component(LIBCLANG_LIBRARY_DIR ${LIBCLANG_LIBRARY} PATH)

src/Cursor.cc

@@ -70,6 +70,30 @@ std::string clangmm::Cursor::get_usr_extended() const {
   return usr;
 }
 
+std::unordered_set<std::string> clangmm::Cursor::get_all_usr_extended() const {
+  std::unordered_set<std::string> usrs;
+  if(get_kind()==Kind::CXXMethod) {
+    class Recursive {
+    public:
+      static void overridden(std::unordered_set<std::string> &usrs, const Cursor &cursor) {
+        usrs.emplace(cursor.get_usr_extended());
+        CXCursor *cursors;
+        unsigned size;
+        clang_getOverriddenCursors(cursor.cx_cursor, &cursors, &size);
+        for(unsigned c=0;c<size;++c)
+          overridden(usrs, cursors[c]);
+        clang_disposeOverriddenCursors(cursors);
+      }
+    };
+    Recursive::overridden(usrs, *this);
+    return usrs;
+  }
+  else {
+    usrs.emplace(get_usr_extended());
+    return usrs;
+  }
+}
+
 clangmm::Cursor clangmm::Cursor::get_referenced() const {
   return Cursor(clang_getCursorReferenced(cx_cursor));
 }

src/Cursor.h

@@ -5,6 +5,7 @@
 #include "SourceRange.h"
 #include <string>
 #include <vector>
+#include <unordered_set>
 
 namespace clangmm {
   class Cursor {
@@ -197,6 +198,8 @@ namespace clangmm {
     std::string get_usr() const;
     /// Improved usr that is also template and argument invariant
     std::string get_usr_extended() const;
+    /// Also get overridden cursors
+    std::unordered_set<std::string> get_all_usr_extended() const;
     Cursor get_referenced() const;
     Cursor get_canonical() const;
     Cursor get_definition() const;

src/Tokens.cc

@@ -1,5 +1,6 @@
 #include "Tokens.h"
 #include "Utility.h"
+#include <unordered_set>
 
 clangmm::Tokens::Tokens(CXTranslationUnit &cx_tu, const SourceRange &range): cx_tu(cx_tu) {
   clang_tokenize(cx_tu, range.cx_range, &cx_tokens, &num_tokens);
@@ -19,14 +20,36 @@ clangmm::Tokens::~Tokens() {
 }
 
 //This works across TranslationUnits. Similar tokens defined as tokens with equal canonical cursors. 
-std::vector<std::pair<clangmm::Offset, clangmm::Offset> > clangmm::Tokens::get_similar_token_offsets(const std::string &spelling, const std::string &usr) {
+std::vector<std::pair<clangmm::Offset, clangmm::Offset> > clangmm::Tokens::get_similar_token_offsets(Cursor::Kind kind, const std::string &spelling,
+                                                                                                     const std::unordered_set<std::string> &usrs) {
+  // Special case for virtual and override functions
+  if(kind==Cursor::Kind::CXXMethod) {
     std::vector<std::pair<Offset, Offset> > offsets;
     for(auto &token: *this) {
       if(token.is_identifier()) {
         auto referenced=token.get_cursor().get_referenced();
-      if(referenced && spelling==token.get_spelling() && usr==referenced.get_usr_extended())
+        if(referenced && referenced.get_kind()==Cursor::Kind::CXXMethod && spelling==token.get_spelling()) {
+          auto referenced_usrs=referenced.get_all_usr_extended();
+          for(auto &usr: referenced_usrs) {
+            if(usrs.count(usr)) {
               offsets.emplace_back(token.offsets);
+              break;
+            }
+          }
+        }
       }
     }
     return offsets;
   }
+  else {
+    std::vector<std::pair<Offset, Offset> > offsets;
+    for(auto &token: *this) {
+      if(token.is_identifier()) {
+        auto referenced=token.get_cursor().get_referenced();
+        if(referenced && spelling==token.get_spelling() && usrs.count(referenced.get_usr_extended()))
+          offsets.emplace_back(token.offsets);
+      }
+    }
+    return offsets;
+  }
+}

src/Tokens.h

@@ -3,7 +3,7 @@
 #include <clang-c/Index.h>
 #include "SourceRange.h"
 #include "Token.h"
-#include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 namespace clangmm {
@@ -13,7 +13,8 @@ namespace clangmm {
     Tokens(CXTranslationUnit &cx_tu, const SourceRange &range);
   public:
     ~Tokens();
-    std::vector<std::pair<clangmm::Offset, clangmm::Offset> > get_similar_token_offsets(const std::string &spelling, const std::string &usr);
+    std::vector<std::pair<clangmm::Offset, clangmm::Offset> > get_similar_token_offsets(Cursor::Kind kind, const std::string &spelling,
+                                                                                        const std::unordered_set<std::string> &usrs);
   private:
     CXToken *cx_tokens;
     unsigned num_tokens;