ClangCompleter now async and caches Clang data

First off, we don't block the GUI thread anymore for ClangCompleter (that was
always temporary). Secondly, now ClangCompleter will cache the data coming from
clang so that query-based filtering of members is fast.
This commit is contained in:
Strahinja Val Markovic 2012-07-15 20:49:56 -07:00
parent a04c3322cc
commit c9e1706fa1
9 changed files with 335 additions and 130 deletions

View File

@ -77,6 +77,7 @@ endfunction
function! s:OnCursorHold()
" TODO: make this async, it's causing lag
py identcomp.AddBufferIdentifiers()
endfunction
@ -178,8 +179,10 @@ EOF
let l:results = []
py << EOF
results = identcomp.CandidatesFromStoredRequest()
result_string = ycm.StringVectorToString( results )
if results:
vim.command( 'let l:results = ' + str( results ) )
vim.command( 'let l:results = ' + result_string )
EOF
let s:searched_and_no_results_found = len( l:results ) == 0
@ -196,8 +199,29 @@ endfunction
function! s:ClangCompletion( query )
" TODO: don't trigger on a dot inside a string constant
py vim.command( 'let l:results = ' +
\ str( clangcomp.CandidatesForQuery( vim.eval( 'a:query' ) ) ) )
py clangcomp.CandidatesForQueryAsync( vim.eval('a:query') )
let l:results_ready = 0
while !l:results_ready
py << EOF
results_ready = clangcomp.AsyncCandidateRequestReady()
if results_ready:
vim.command( 'let l:results_ready = 1' )
EOF
if complete_check()
return { 'words' : [], 'refresh' : 'always'}
endif
endwhile
let l:results = []
py << EOF
results = clangcomp.CandidatesFromStoredRequest()
result_string = ycm.StringVectorToString( results )
if results:
vim.command( 'let l:results = ' + result_string )
EOF
let s:searched_and_no_results_found = len( l:results ) == 0
return { 'words' : l:results, 'refresh' : 'always' }

View File

@ -19,15 +19,29 @@
#include "Candidate.h"
#include "standard.h"
#include "CandidateRepository.h"
#include "ConcurrentLatestValue.h"
#include <clang-c/Index.h>
#include <boost/make_shared.hpp>
using boost::packaged_task;
using boost::bind;
using boost::unique_future;
using boost::make_shared;
using boost::shared_ptr;
using boost::bind;
using boost::thread;
namespace YouCompleteMe
{
extern const unsigned int MAX_ASYNC_THREADS;
extern const unsigned int MIN_ASYNC_THREADS;
namespace
{
std::vector< CXUnsavedFile > ToCXUnsavedFiles(
const std::vector< UnsavedFile > &unsaved_files )
{
@ -91,7 +105,9 @@ std::vector< std::string > ToStringVector( CXCodeCompleteResults *results )
ClangCompleter::ClangCompleter()
: candidate_repository_( CandidateRepository::Instance() )
: candidate_repository_( CandidateRepository::Instance() ),
threading_enabled_( false ),
clang_data_ready_( false )
{
clang_index_ = clang_createIndex( 0, 0 );
}
@ -109,6 +125,15 @@ ClangCompleter::~ClangCompleter()
}
// We need this mostly so that we can not use it in tests. Apparently the
// GoogleTest framework goes apeshit on us if we enable threads by default.
void ClangCompleter::EnableThreading()
{
threading_enabled_ = true;
InitThreads();
}
void ClangCompleter::SetGlobalCompileFlags(
const std::vector< std::string > &flags )
{
@ -152,7 +177,6 @@ void ClangCompleter::UpdateTranslationUnit(
std::vector< std::string > ClangCompleter::CandidatesForLocationInFile(
const std::string &query,
const std::string &filename,
int line,
int column,
@ -166,7 +190,7 @@ std::vector< std::string > ClangCompleter::CandidatesForLocationInFile(
// If there are unsaved files, then codeCompleteAt will parse the in-memory
// file contents we are giving it. In short, it is NEVER a good idea to call
// clang_reparseTranslationUnit right before a call to clang_codeCompleteAt.
// The only makes clang reparse the whole file TWICE, which has a huge impact
// This only makes clang reparse the whole file TWICE, which has a huge impact
// on latency. At the time of writing, it seems that most users of libclang
// in the open-source world don't realize this (I checked). Some don't even
// call reparse*, but parse* which is even less efficient.
@ -180,12 +204,64 @@ std::vector< std::string > ClangCompleter::CandidatesForLocationInFile(
cxunsaved_files.size(),
clang_defaultCodeCompleteOptions());
std::vector< std::string > completions = ToStringVector( results );
if ( !query.empty() )
completions = SortCandidatesForQuery( query, completions );
std::vector< std::string > candidates = ToStringVector( results );
clang_disposeCodeCompleteResults( results );
return completions;
return candidates;
}
Future< AsyncResults > ClangCompleter::CandidatesForQueryAndLocationInFileAsync(
const std::string &query,
const std::string &filename,
int line,
int column,
const std::vector< UnsavedFile > &unsaved_files )
{
// TODO: throw exception when threading is not enabled and this is called
if ( !threading_enabled_ )
return Future< AsyncResults >();
if ( query.empty() )
{
{
boost::lock_guard< boost::mutex > lock( clang_data_ready_mutex_ );
clang_data_ready_ = false;
}
sorting_threads_.interrupt_all();
}
// the sorting task needs to be set before the clang task (if any) just in
// case the clang task finishes (and therefore notifies a sorting thread to
// consume a sorting task) before the sorting task is set
shared_ptr< packaged_task< AsyncResults > > task =
make_shared< packaged_task< AsyncResults > >(
bind( ReturnValueAsShared< std::vector< std::string > >,
static_cast< FunctionReturnsStringVector >(
bind( &ClangCompleter::SortCandidatesForQuery,
boost::ref( *this ),
query,
boost::cref( latest_clang_results_ ) ) ) ) );
unique_future< AsyncResults > future = task->get_future();
sorting_task_.Set( task );
if ( query.empty() )
{
shared_ptr< packaged_task< AsyncResults > > task =
make_shared< packaged_task< AsyncResults > >(
bind( ReturnValueAsShared< std::vector< std::string > >,
static_cast< FunctionReturnsStringVector >(
bind( &ClangCompleter::CandidatesForLocationInFile,
boost::ref( *this ),
filename,
line,
column,
unsaved_files ) ) ) );
clang_task_.Set( task );
}
return Future< AsyncResults >( move( future ) );
}
@ -281,4 +357,82 @@ std::vector< std::string > ClangCompleter::SortCandidatesForQuery(
return sorted_candidates;
}
void ClangCompleter::InitThreads()
{
int threads_to_create =
std::max( MIN_ASYNC_THREADS,
std::min( MAX_ASYNC_THREADS, thread::hardware_concurrency() ) );
for ( int i = 0; i < threads_to_create; ++i )
{
sorting_threads_.create_thread(
bind( &ClangCompleter::SortingThreadMain,
boost::ref( *this ),
boost::ref( sorting_task_ ) ) );
}
clang_thread_ = boost::thread( &ClangCompleter::ClangThreadMain,
boost::ref( *this ),
boost::ref( clang_task_ ) );
}
void ClangCompleter::ClangThreadMain( LatestTask &clang_task )
{
while ( true )
{
shared_ptr< packaged_task< AsyncResults > > task = clang_task.Get();
( *task )();
unique_future< AsyncResults > future = task->get_future();
{
boost::unique_lock< boost::shared_mutex > writer_lock(
latest_clang_results_shared_mutex_ );
latest_clang_results_ = *future.get();
}
{
boost::lock_guard< boost::mutex > lock( clang_data_ready_mutex_ );
clang_data_ready_ = true;
}
clang_data_ready_condition_variable_.notify_all();
}
}
void ClangCompleter::SortingThreadMain( LatestTask &sorting_task )
{
while ( true )
{
try
{
{
boost::unique_lock< boost::mutex > lock( clang_data_ready_mutex_ );
while ( !clang_data_ready_ )
{
clang_data_ready_condition_variable_.wait( lock );
}
}
shared_ptr< packaged_task< AsyncResults > > task = sorting_task.Get();
{
boost::shared_lock< boost::shared_mutex > reader_lock(
latest_clang_results_shared_mutex_ );
( *task )();
}
}
catch ( boost::thread_interrupted& )
{
// Do nothing and re-enter the loop
}
}
}
} // namespace YouCompleteMe

View File

@ -18,6 +18,9 @@
#ifndef CLANGCOMPLETE_H_WLKDU0ZV
#define CLANGCOMPLETE_H_WLKDU0ZV
#include "ConcurrentLatestValue.h"
#include "Future.h"
#include <boost/utility.hpp>
#include <boost/unordered_map.hpp>
@ -55,6 +58,7 @@ struct UnsavedFile
typedef boost::unordered_map< std::string, std::vector< std::string > >
FlagsForFile;
typedef boost::unordered_map< std::string, CXTranslationUnit >
TranslationUnitForFilename;
@ -65,6 +69,8 @@ public:
ClangCompleter();
~ClangCompleter();
void EnableThreading();
void SetGlobalCompileFlags( const std::vector< std::string > &flags );
void SetFileCompileFlags( const std::string &filename,
@ -73,8 +79,13 @@ public:
void UpdateTranslationUnit( const std::string &filename,
const std::vector< UnsavedFile > &unsaved_files );
// TODO: rename this
std::vector< std::string > CandidatesForLocationInFile(
const std::string &filename,
int line,
int column,
const std::vector< UnsavedFile > &unsaved_files );
Future< AsyncResults > CandidatesForQueryAndLocationInFileAsync(
const std::string &query,
const std::string &filename,
int line,
@ -82,6 +93,9 @@ public:
const std::vector< UnsavedFile > &unsaved_files );
private:
typedef ConcurrentLatestValue<
boost::shared_ptr<
boost::packaged_task< AsyncResults > > > LatestTask;
// caller takes ownership of translation unit
CXTranslationUnit CreateTranslationUnit(
@ -99,17 +113,47 @@ private:
const std::string &query,
const std::vector< std::string > &candidates );
void InitThreads();
void ClangThreadMain( LatestTask &clang_task );
void SortingThreadMain( LatestTask &sorting_task );
/////////////////////////////
// PRIVATE MEMBER VARIABLES
/////////////////////////////
CXIndex clang_index_;
FlagsForFile flags_for_file_;
TranslationUnitForFilename filename_to_translation_unit_;
std::vector< std::string > global_flags_;
CandidateRepository &candidate_repository_;
mutable LatestTask clang_task_;
mutable LatestTask sorting_task_;
bool threading_enabled_;
// TODO: use boost.atomic for clang_data_ready_
bool clang_data_ready_;
boost::mutex clang_data_ready_mutex_;
boost::condition_variable clang_data_ready_condition_variable_;
std::vector< std::string > latest_clang_results_;
boost::shared_mutex latest_clang_results_shared_mutex_;
// Unfortunately clang is not thread-safe so we can only ever use one thread
// to access it. So this one background thread will be the only thread that
// can access libclang.
boost::thread clang_thread_;
boost::thread_group sorting_threads_;
};
} // namespace YouCompleteMe

View File

@ -1,58 +0,0 @@
// Copyright (C) 2011, 2012 Strahinja Val Markovic <val@markovic.io>
//
// This file is part of YouCompleteMe.
//
// YouCompleteMe is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// YouCompleteMe is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with YouCompleteMe. If not, see <http://www.gnu.org/licenses/>.
#include "Future.h"
#include "standard.h"
#include "Result.h"
namespace YouCompleteMe
{
Future::Future( boost::shared_future< AsyncResults > future )
: future_( boost::move( future ) )
{
}
bool Future::ResultsReady()
{
return future_.is_ready();
}
Pylist Future::GetResults()
{
Pylist candidates;
AsyncResults results;
try
{
results = future_.get();
}
catch ( boost::future_uninitialized & )
{
return candidates;
}
foreach ( const Result& result, *results )
{
candidates.append( *result.Text() );
}
return candidates;
}
} // namespace YouCompleteMe

View File

@ -19,8 +19,10 @@
#define FUTURE_H_NR1U6MZS
#include <boost/thread.hpp>
#include <boost/make_shared.hpp>
#include <boost/python.hpp>
#include <boost/shared_ptr.hpp>
#include <boost/function.hpp>
namespace YouCompleteMe
{
@ -29,22 +31,50 @@ class Result;
template< typename T > class ConcurrentLatestValue;
typedef boost::python::list Pylist;
typedef boost::shared_ptr< std::vector< Result > > AsyncResults;
typedef boost::shared_ptr< std::vector< std::string > > AsyncResults;
typedef ConcurrentLatestValue<
boost::shared_ptr<
boost::packaged_task< AsyncResults > > > LatestTask;
typedef boost::function< std::vector< std::string >() >
FunctionReturnsStringVector;
template< typename T >
boost::shared_ptr< T > ReturnValueAsShared(
boost::function< T() > func )
{
return boost::make_shared< T >( func() );
}
template< typename T >
class Future
{
public:
Future() {}
Future( boost::shared_future< AsyncResults > future );
bool ResultsReady();
Pylist GetResults();
Future() {};
Future( boost::shared_future< T > future )
: future_( boost::move( future ) ) {}
bool ResultsReady()
{
return future_.is_ready();
}
T GetResults()
{
try
{
return future_.get();
}
catch ( boost::future_uninitialized & )
{
// Do nothing and return a T()
}
return T();
}
private:
boost::shared_future< AsyncResults > future_;
boost::shared_future< T > future_;
};
} // namespace YouCompleteMe

View File

@ -36,12 +36,12 @@ using boost::thread;
namespace YouCompleteMe
{
extern const unsigned int MAX_ASYNC_THREADS = 4;
extern const unsigned int MIN_ASYNC_THREADS = 2;
namespace
{
const unsigned int MAX_ASYNC_THREADS = 4;
const unsigned int MIN_ASYNC_THREADS = 2;
void ThreadMain( LatestTask &latest_task )
{
while ( true )
@ -135,36 +135,29 @@ std::vector< std::string > IdentifierCompleter::CandidatesForQueryAndType(
}
Future IdentifierCompleter::CandidatesForQueryAndTypeAsync(
Future< AsyncResults > IdentifierCompleter::CandidatesForQueryAndTypeAsync(
const std::string &query,
const std::string &filetype ) const
{
// TODO: throw exception when threading is not enabled and this is called
if ( !threading_enabled_ )
return Future();
return Future< AsyncResults >();
// Try not to look at this too hard, it may burn your eyes.
// TODO: refactor this so it's more readable
shared_ptr< packaged_task< AsyncResults > > task =
make_shared< packaged_task< AsyncResults > >(
bind( &IdentifierCompleter::ResultsForQueryAndType,
bind( ReturnValueAsShared< std::vector< std::string > >,
static_cast< FunctionReturnsStringVector >(
bind( &IdentifierCompleter::CandidatesForQueryAndType,
boost::cref( *this ),
query,
filetype ) );
filetype ) ) ) );
unique_future< AsyncResults > future = task->get_future();
latest_task_.Set( task );
return Future( move( future ) );
}
AsyncResults IdentifierCompleter::ResultsForQueryAndType(
const std::string &query,
const std::string &filetype ) const
{
AsyncResults results = boost::make_shared< std::vector< Result > >();
ResultsForQueryAndType( query, filetype, *results );
return results;
return Future< AsyncResults >( move( future ) );
}

View File

@ -35,6 +35,7 @@ namespace YouCompleteMe
class Candidate;
class CandidateRepository;
// TODO: move to private
// filepath -> *( *candidate )
typedef boost::unordered_map< std::string,
boost::shared_ptr< std::list< const Candidate* > > >
@ -44,6 +45,9 @@ typedef boost::unordered_map< std::string,
typedef boost::unordered_map< std::string,
boost::shared_ptr< FilepathToCandidates > > FiletypeMap;
typedef ConcurrentLatestValue<
boost::shared_ptr<
boost::packaged_task< AsyncResults > > > LatestTask;
class IdentifierCompleter : boost::noncopyable
{
@ -70,14 +74,12 @@ public:
const std::string &query,
const std::string &filetype ) const;
Future CandidatesForQueryAndTypeAsync( const std::string &query,
Future< AsyncResults > CandidatesForQueryAndTypeAsync(
const std::string &query,
const std::string &filetype ) const;
private:
AsyncResults ResultsForQueryAndType( const std::string &query,
const std::string &filetype ) const;
void ResultsForQueryAndType( const std::string &query,
const std::string &filetype,
std::vector< Result > &results ) const;

View File

@ -28,12 +28,13 @@ BOOST_PYTHON_MODULE(indexer)
using namespace boost::python;
using namespace YouCompleteMe;
class_< std::vector< std::string > >( "StringVec" )
class_< std::vector< std::string >,
boost::shared_ptr< std::vector< std::string > > >( "StringVec" )
.def( vector_indexing_suite< std::vector< std::string > >() );
class_< Future >( "Future" )
.def( "ResultsReady", &Future::ResultsReady )
.def( "GetResults", &Future::GetResults );
class_< Future< AsyncResults > >( "Future" )
.def( "ResultsReady", &Future< AsyncResults >::ResultsReady )
.def( "GetResults", &Future< AsyncResults >::GetResults );
class_< IdentifierCompleter, boost::noncopyable >( "IdentifierCompleter" )
.def( "EnableThreading", &IdentifierCompleter::EnableThreading )
@ -61,9 +62,10 @@ BOOST_PYTHON_MODULE(indexer)
.def( vector_indexing_suite< std::vector< UnsavedFile > >() );
class_< ClangCompleter, boost::noncopyable >( "ClangCompleter" )
.def( "EnableThreading", &ClangCompleter::EnableThreading )
.def( "SetGlobalCompileFlags", &ClangCompleter::SetGlobalCompileFlags )
.def( "SetFileCompileFlags", &ClangCompleter::SetFileCompileFlags )
.def( "UpdateTranslationUnit", &ClangCompleter::UpdateTranslationUnit )
.def( "CandidatesForLocationInFile",
&ClangCompleter::CandidatesForLocationInFile );
.def( "CandidatesForQueryAndLocationInFileAsync",
&ClangCompleter::CandidatesForQueryAndLocationInFileAsync );
}

View File

@ -89,11 +89,15 @@ class IdentifierCompleter( Completer ):
filepath,
True )
class ClangCompleter( Completer ):
def __init__( self ):
self.completer = indexer.ClangCompleter()
self.completer.EnableThreading()
self.contents_holder = []
self.filename_holder = []
def CandidatesForQuery( self, query ):
def CandidatesForQueryAsync( self, query ):
# TODO: sanitize query
files = indexer.UnsavedFileVec()
@ -102,31 +106,32 @@ class ClangCompleter( Completer ):
# pointers to data members of python objects. We need to ensure that those
# objects outlive our UnsavedFile objects. This is why we need the
# contents_holder and filename_holder lists, to make sure the string objects
# are still around when we call CandidatesForLocationInFile. We do this to
# avoid an extra copy of the entire file contents.
# are still around when we call CandidatesForQueryAndLocationInFile. We do
# this to avoid an extra copy of the entire file contents.
contents_holder = []
filename_holder = []
if not query:
self.contents_holder = []
self.filename_holder = []
for buffer in GetUnsavedBuffers():
contents_holder.append( '\n'.join( buffer ) )
filename_holder.append( buffer.name )
self.contents_holder.append( '\n'.join( buffer ) )
self.filename_holder.append( buffer.name )
unsaved_file = indexer.UnsavedFile()
unsaved_file.contents_ = contents_holder[ -1 ]
unsaved_file.length_ = len( contents_holder[ -1 ] )
unsaved_file.filename_ = filename_holder[ -1 ]
unsaved_file.contents_ = self.contents_holder[ -1 ]
unsaved_file.length_ = len( self.contents_holder[ -1 ] )
unsaved_file.filename_ = self.filename_holder[ -1 ]
files.append( unsaved_file )
line, _ = vim.current.window.cursor
column = int( vim.eval( "s:completion_start_column" ) ) + 1
current_buffer = vim.current.buffer
results = self.completer.CandidatesForLocationInFile( query,
self.future = self.completer.CandidatesForQueryAndLocationInFileAsync(
query,
current_buffer.name,
line,
column,
files )
return list( results )
def GetUnsavedBuffers():
@ -137,6 +142,15 @@ def GetUnsavedBuffers():
return ( x for x in vim.buffers if BufferModified( x.number ) )
# TODO: just implement __str__/__repr__ on StringVec
def StringVectorToString( stringvec ):
result = [ "[" ]
for text in stringvec:
result.append( '"{0}",'.format( text ) )
result.append( "]" )
return ''.join( result )
def CurrentColumn():
"""Do NOT access the CurrentColumn in vim.current.line. It doesn't exist yet.
Only the chars before the current column exist in vim.current.line."""