Adding new candidates is now faster; + more

Also fixed a memory leak and started work on adding filepath-aware candidate
repos.

cpp/Completer.cpp

@@ -17,33 +17,51 @@
 
 #include "standard.h"
 #include "Completer.h"
+#include "Utils.h"
 
 namespace YouCompleteMe
 {
 
 Completer::Completer( const Pylist &candidates )
 {
-  AddCandidatesToDatabase( candidates );
+  AddCandidatesToDatabase( candidates, "" );
 }
 
+
+Completer::Completer( const Pylist &candidates, const std::string &filepath)
+{
+  AddCandidatesToDatabase( candidates, filepath );
+}
+
+
 Completer::~Completer()
 {
-  foreach ( Candidate* candidate, candidates_ )
+  foreach ( const CandidateRepository::value_type &pair,
+            candidate_repository_ )
   {
-    delete candidate;
+    delete pair.second;
   }
 }
 
-void Completer::AddCandidatesToDatabase(
+
-		const Pylist &candidates )
+void Completer::AddCandidatesToDatabase( const Pylist &candidates,
+                                         const std::string &filepath )
 {
+  std::string candidate_text;
   for (int i = 0; i < boost::python::len( candidates ); ++i)
   {
-    candidates_.insert( new Candidate(
+    candidate_text = boost::python::extract< std::string >( candidates[ i ] );
-      boost::python::extract< std::string >( candidates[ i ] ) ) );
+    Candidate *&candidate = GetValueElseInsert( candidate_repository_,
+                                                candidate_text, NULL );
+    if ( !candidate )
+    {
+      candidate = new Candidate( candidate_text );
+      candidates_.insert( candidate );
+    }
   }
 }
 
+
 void Completer::GetCandidatesForQuery(
     const std::string &query, Pylist &candidates ) const
 {
@@ -70,4 +88,5 @@ void Completer::GetCandidatesForQuery(
   }
 }
 
+
 } // namespace YouCompleteMe

cpp/Completer.h

@@ -22,6 +22,7 @@
 
 #include <boost/utility.hpp>
 #include <boost/python.hpp>
+#include <boost/unordered_map.hpp>
 
 #include <set>
 #include <vector>
@@ -31,6 +32,7 @@ namespace YouCompleteMe
 {
 
 typedef boost::python::list Pylist;
+typedef boost::unordered_map< std::string, Candidate* > CandidateRepository;
 
 // class Completer : boost::noncopyable
 class Completer
@@ -38,9 +40,11 @@ class Completer
 public:
   Completer() {}
   Completer( const Pylist &candidates );
+  Completer( const Pylist &candidates, const std::string &filepath );
   ~Completer();
 
-  void AddCandidatesToDatabase( const Pylist &candidates );
+  void AddCandidatesToDatabase( const Pylist &candidates,
+                                const std::string &filepath );
 
 	void GetCandidatesForQuery(
 			const std::string &query, Pylist &candidates ) const;
@@ -54,6 +58,7 @@ private:
     }
   };
 
+  CandidateRepository candidate_repository_;
 	std::set< Candidate*, CandidatePointerLess > candidates_;
 };
 

cpp/Utils.h

@@ -27,6 +27,22 @@ namespace YouCompleteMe
 
 bool AlmostEqual( double a, double b );
 
+template <class Container, class Key>
+typename Container::mapped_type &
+GetValueElseInsert( Container &container,
+                    Key const& key,
+                    typename Container::mapped_type const& value )
+{
+  return container.insert( typename Container::value_type( key, value ) )
+    .first->second;
+}
+
+template <class Container, class Key>
+bool ContainsKey( Container &container, Key const& key)
+{
+  return container.find( key ) != container.end();
+}
+
 } // namespace YouCompleteMe
 
 #endif /* end of include guard: UTILS_H_KEPMRPBH */

python/ycm.py

@@ -35,10 +35,11 @@ class CompletionSystem( object ):
 
   def AddBufferIdentifiers( self ):
     text = "\n".join( vim.current.buffer )
-    text = RemoveCppComments( text )
+    text = RemoveIdentFreeText( text )
 
     idents = re.findall( self.pattern, text )
-    self.completer.AddCandidatesToDatabase( idents )
+    filepath = vim.eval( "expand('%:p')" )
+    self.completer.AddCandidatesToDatabase( idents, filepath )
 
 def CurrentColumn():
   # vim's columns start at 1 while vim.current.line columns start at 0
@@ -74,7 +75,10 @@ def CurrentCursorText():
 def SanitizeQuery( query ):
   return query.strip()
 
-def RemoveCppComments( text ):
+def RemoveIdentFreeText( text ):
+  """Removes commented-out code and code in quotes."""
+
+  # TODO: do we still need this sub-func?
   def replacer( match ):
     s = match.group( 0 )
     if s.startswith( '/' ):
@@ -83,7 +87,7 @@ def RemoveCppComments( text ):
       return s
 
   pattern = re.compile(
-    r'//.*?$|/\*.*?\*/|\'(?:\\.|[^\\\'])*\'|"(?:\\.|[^\\"])*"',
+    r'//.*?$|#.*?$|/\*.*?\*/|\'(?:\\.|[^\\\'])*\'|"(?:\\.|[^\\"])*"',
     re.DOTALL | re.MULTILINE )
 
   return re.sub( pattern, replacer, text )