Better elisp support for identifier completer

Elisp is a lisp, so it supports all kinds of crazy identifiers. AFAIR
using the same regexes as Clojure should work more than well enough
because Clojure is too a Lisp descendent.

ycmd/identifier_utils.py

@@ -104,6 +104,10 @@
 FILETYPE_TO_IDENTIFIER_REGEX[ 'scss' ] = FILETYPE_TO_IDENTIFIER_REGEX[ 'css' ]
 FILETYPE_TO_IDENTIFIER_REGEX[ 'sass' ] = FILETYPE_TO_IDENTIFIER_REGEX[ 'css' ]
 FILETYPE_TO_IDENTIFIER_REGEX[ 'less' ] = FILETYPE_TO_IDENTIFIER_REGEX[ 'css' ]
+FILETYPE_TO_IDENTIFIER_REGEX[ 'elisp' ] = (
+  FILETYPE_TO_IDENTIFIER_REGEX[ 'clojure' ] )
+FILETYPE_TO_IDENTIFIER_REGEX[ 'lisp' ] = (
+  FILETYPE_TO_IDENTIFIER_REGEX[ 'clojure' ] )
 
 
 def IdentifierRegexForFiletype( filetype ):

ycmd/tests/identifier_utils_test.py

@@ -234,6 +234,20 @@ def IsIdentifier_Clojure_test():
   ok_( not iu.IsIdentifier( '' , 'clojure' ) )
 
 
+def IsIdentifier_Elisp_test():
+  # elisp is using the clojure regexes, so we're testing this more lightly
+  ok_( iu.IsIdentifier( 'foo'  , 'elisp' ) )
+  ok_( iu.IsIdentifier( 'f9'   , 'elisp' ) )
+  ok_( iu.IsIdentifier( 'a.b.c', 'elisp' ) )
+  ok_( iu.IsIdentifier( 'a/c'  , 'elisp' ) )
+
+  ok_( not iu.IsIdentifier( '9f'   , 'elisp' ) )
+  ok_( not iu.IsIdentifier( '9'    , 'elisp' ) )
+  ok_( not iu.IsIdentifier( 'a/b/c', 'elisp' ) )
+  ok_( not iu.IsIdentifier( '(a)'  , 'elisp' ) )
+  ok_( not iu.IsIdentifier( '' , 'elisp' ) )
+
+
 def IsIdentifier_Haskell_test():
   ok_( iu.IsIdentifier( 'foo' , 'haskell' ) )
   ok_( iu.IsIdentifier( "foo'", 'haskell' ) )