YouCompleteMe/python/ycm/completers/all/identifier_completer.py
2013-05-19 19:44:42 -07:00

153 lines
4.5 KiB
Python

#!/usr/bin/env python
#
# 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/>.
import vim
import ycm_core
from ycm.completers.general_completer import GeneralCompleter
from ycm import vimsupport
from ycm import utils
MAX_IDENTIFIER_COMPLETIONS_RETURNED = 10
MIN_NUM_CHARS = int( vimsupport.GetVariableValue(
"g:ycm_min_num_of_chars_for_completion" ) )
class IdentifierCompleter( GeneralCompleter ):
def __init__( self ):
super( IdentifierCompleter, self ).__init__()
self.completer = ycm_core.IdentifierCompleter()
self.completer.EnableThreading()
def ShouldUseNow( self, start_column ):
return self.QueryLengthAboveMinThreshold( start_column )
def CandidatesForQueryAsync( self, query, unused_start_column ):
filetype = vim.eval( "&filetype" )
self.completions_future = self.completer.CandidatesForQueryAndTypeAsync(
utils.SanitizeQuery( query ),
filetype )
def AddIdentifier( self, identifier ):
filetype = vim.eval( "&filetype" )
filepath = vim.eval( "expand('%:p')" )
if not filetype or not filepath or not identifier:
return
vector = ycm_core.StringVec()
vector.append( identifier )
self.completer.AddCandidatesToDatabase( vector,
filetype,
filepath )
def AddPreviousIdentifier( self ):
self.AddIdentifier( PreviousIdentifier() )
def AddIdentifierUnderCursor( self ):
cursor_identifier = vim.eval( 'expand("<cword>")' )
if not cursor_identifier:
return
stripped_cursor_identifier = ''.join( ( x for x in
cursor_identifier if
utils.IsIdentifierChar( x ) ) )
if not stripped_cursor_identifier:
return
self.AddIdentifier( stripped_cursor_identifier )
def AddBufferIdentifiers( self ):
filetype = vim.eval( "&filetype" )
filepath = vim.eval( "expand('%:p')" )
collect_from_comments_and_strings = vimsupport.GetBoolValue(
"g:ycm_collect_identifiers_from_comments_and_strings" )
if not filetype or not filepath:
return
text = "\n".join( vim.current.buffer )
self.completer.AddCandidatesToDatabaseFromBufferAsync(
text,
filetype,
filepath,
collect_from_comments_and_strings )
def OnFileReadyToParse( self ):
self.AddBufferIdentifiers()
def OnInsertLeave( self ):
self.AddIdentifierUnderCursor()
def OnCurrentIdentifierFinished( self ):
self.AddPreviousIdentifier()
def CandidatesFromStoredRequest( self ):
if not self.completions_future:
return []
completions = self.completions_future.GetResults()[
: MAX_IDENTIFIER_COMPLETIONS_RETURNED ]
# We will never have duplicates in completions so with 'dup':1 we tell Vim
# to add this candidate even if it's a duplicate of an existing one (which
# will never happen). This saves us some expensive string matching
# operations in Vim.
return [ { 'word': x, 'dup': 1 } for x in completions ]
def PreviousIdentifier():
line_num, column_num = vimsupport.CurrentLineAndColumn()
buffer = vim.current.buffer
line = buffer[ line_num ]
end_column = column_num
while end_column > 0 and not utils.IsIdentifierChar( line[ end_column - 1 ] ):
end_column -= 1
# Look at the previous line if we reached the end of the current one
if end_column == 0:
try:
line = buffer[ line_num - 1]
except:
return ""
end_column = len( line )
while end_column > 0 and not utils.IsIdentifierChar(
line[ end_column - 1 ] ):
end_column -= 1
start_column = end_column
while start_column > 0 and utils.IsIdentifierChar( line[ start_column - 1 ] ):
start_column -= 1
if end_column - start_column < MIN_NUM_CHARS:
return ""
return line[ start_column : end_column ]