Returned offsets are now line and line index to be able to find correct offsets in UTF-8 files.

src/Diagnostic.cc

@@ -7,14 +7,13 @@ clang::Diagnostic::Diagnostic(CXTranslationUnit& cx_tu, CXDiagnostic& cx_diagnos
   severity=clang_getDiagnosticSeverity(cx_diagnostic);
   severity_spelling=get_severity_spelling(severity);
   spelling=clang::to_string(clang_getDiagnosticSpelling(cx_diagnostic));
-  clang::SourceLocation start_location(clang_getDiagnosticLocation(cx_diagnostic));
   
+  clang::SourceLocation start_location(clang_getDiagnosticLocation(cx_diagnostic));
   path=start_location.get_path();
-  unsigned start_offset=start_location.get_offset();
+  auto start_offset=start_location.get_offset();
   clang::Tokens tokens(cx_tu, SourceRange(start_location, start_location));
-  if(tokens.size()==1) {
+  if(tokens.size()==1)
-    offsets=std::pair<unsigned, unsigned>(start_offset, tokens.begin()->offsets.second);
+    offsets={start_offset, tokens.begin()->offsets.second};
-  }
 }
 
 const std::string clang::Diagnostic::get_severity_spelling(unsigned severity) {

src/Diagnostic.h

@@ -16,7 +16,7 @@ namespace clang {
     std::string severity_spelling;
     std::string spelling;
     std::string path;
-    std::pair<unsigned, unsigned> offsets;
+    std::pair<clang::Offset, clang::Offset> offsets;
   };
 }
 

src/SourceLocation.cc

@@ -14,10 +14,10 @@ std::string clang::SourceLocation::get_path() {
   get_data(&path, NULL, NULL, NULL);
   return path;
 }
-unsigned clang::SourceLocation::get_offset() {
+clang::Offset clang::SourceLocation::get_offset() {
-  unsigned offset;
+  unsigned line, index;
-  get_data(NULL, NULL, NULL, &offset);
+  get_data(NULL, &line, &index, NULL);
-  return offset;
+  return {line, index};
 }
 
 void clang::SourceLocation::get_data(std::string* path, unsigned *line, unsigned *column, unsigned *offset) {

src/SourceLocation.h

@@ -4,6 +4,15 @@
 #include <string>
 
 namespace clang {
+  class Offset {
+  public:
+    Offset() {}
+    Offset(unsigned line, unsigned index): line(line), index(index) {}
+    bool operator==(const clang::Offset &o) {return (line==o.line && index==o.index);}
+    bool operator!=(const clang::Offset &o) {return !(*this==o);}
+    unsigned line;
+    unsigned index; //byte index in line (not char number)
+  };
   
   class SourceLocation {
     friend class TranslationUnit;
@@ -13,7 +22,7 @@ namespace clang {
 
   public:
     std::string get_path();
-    unsigned get_offset();
+    clang::Offset get_offset();
 
     CXSourceLocation cx_location;
     

src/SourceRange.cc

@@ -5,10 +5,7 @@ SourceRange(clang::SourceLocation &start, clang::SourceLocation &end) {
   cx_range = clang_getRange(start.cx_location, end.cx_location);
 }
 
-std::pair<unsigned, unsigned> clang::SourceRange::get_offsets() {
+std::pair<clang::Offset, clang::Offset> clang::SourceRange::get_offsets() {
   SourceLocation start(clang_getRangeStart(cx_range)), end(clang_getRangeEnd(cx_range));
-  std::pair<unsigned, unsigned> offsets;
+  return {start.get_offset(), end.get_offset()};
-  offsets.first=start.get_offset();
-  offsets.second=end.get_offset();
-  return offsets;
 }

src/SourceRange.h

@@ -10,7 +10,7 @@ namespace clang {
   public:
     SourceRange(const CXSourceRange& cx_range) : cx_range(cx_range) {}
     SourceRange(SourceLocation &start, SourceLocation &end);
-    std::pair<unsigned, unsigned> get_offsets();
+    std::pair<clang::Offset, clang::Offset> get_offsets();
     CXSourceRange cx_range;
   };
 }  // namespace clang

src/Token.h

@@ -31,7 +31,7 @@ namespace clang {
     CXTranslationUnit &cx_tu;
     CXToken& cx_token;
     CXCursor& cx_cursor;
-    std::pair<unsigned, unsigned> offsets;
+    std::pair<clang::Offset, clang::Offset> offsets;
   };
 }  // namespace clang
 #endif  // TOKEN_H_

src/Tokens.cc

@@ -20,8 +20,8 @@ clang::Tokens::~Tokens() {
 //This works across TranslationUnits! However, to get rename refactoring to work, 
 //one have to open all the files that might include a similar token
 //Similar tokens defined as tokens with equal referenced cursors. 
-std::vector<std::pair<unsigned, unsigned> > clang::Tokens::get_similar_token_offsets(const std::string &usr) {
+std::vector<std::pair<clang::Offset, clang::Offset> > clang::Tokens::get_similar_token_offsets(const std::string &usr) {
-  std::vector<std::pair<unsigned, unsigned> > offsets;
+  std::vector<std::pair<clang::Offset, clang::Offset> > offsets;
   for(auto &token: *this) {
     if(token.get_kind()==clang::Token_Identifier) {
       auto referenced=token.get_cursor().get_referenced();
@@ -33,9 +33,9 @@ std::vector<std::pair<unsigned, unsigned> > clang::Tokens::get_similar_token_off
   return offsets;
 }
 
-std::vector<std::pair<std::string, unsigned> > clang::Tokens::get_cxx_methods() {
+std::vector<std::pair<std::string, clang::Offset> > clang::Tokens::get_cxx_methods() {
-  std::vector<std::pair<std::string, unsigned> > methods;
+  std::vector<std::pair<std::string, clang::Offset> > methods;
-  long last_offset=-1;
+  clang::Offset last_offset={(unsigned)-1,(unsigned) -1};
   for(auto &token: *this) {
     if(token.get_kind()==clang::Token_Identifier) {
       auto cursor=token.get_cursor();

src/Tokens.h

@@ -13,8 +13,8 @@ namespace clang {
     Tokens(CXTranslationUnit &cx_tu, const SourceRange &range);
   public:
     ~Tokens();
-    std::vector<std::pair<unsigned, unsigned> > get_similar_token_offsets(const std::string &usr);
+    std::vector<std::pair<clang::Offset, clang::Offset> > get_similar_token_offsets(const std::string &usr);
-    std::vector<std::pair<std::string, unsigned> > get_cxx_methods();
+    std::vector<std::pair<std::string, clang::Offset> > get_cxx_methods();
   private:
     CXToken *cx_tokens;
     unsigned num_tokens;

tests/SourceLocation_H_Test.cc

@@ -12,6 +12,6 @@ BOOST_AUTO_TEST_CASE(source_location) {
 
   auto offsets=(*tokens)[28].offsets;
   
-  BOOST_CHECK(offsets.first == 103);
+  BOOST_CHECK(offsets.first.line == 6 && offsets.first.index == 3);
-  BOOST_CHECK(offsets.second == 109);
+  BOOST_CHECK(offsets.second.line == 6 && offsets.second.index == 9);
 }