fix findtoken. it occur maxmempattern error.

autoload/zencoding/html.vim

@@ -1,23 +1,24 @@
+"let s:mx = '\([+>]\|<\+\)\{-}\((*\)\{-}\([@#.]\{-}[a-zA-Z\!][a-zA-Z0-9:_\!\-$]*\|'
 let s:mx = '\([+>]\|<\+\)\{-}\s*\((*\)\{-}\s*\([@#.]\{-}[a-zA-Z\!][a-zA-Z0-9:_\!\-$]*\|'
 \       .'{.\{-}}[ \t\r\n}]*\)\(\%(\%(#{[{}a-zA-Z0-9_\-\$]\+\|'
 \       .'#[a-zA-Z0-9_\-\$]\+\)\|\%(\[[^\]]\+\]\)\|'
 \       .'\%(\.{[{}a-zA-Z0-9_\-\$]\+\|'
-\       .'\.[a-zA-Z0-9_\-\$]\+\)\)*\)\%(\({[^}]\+}\+\)\)\{0,1}\%(\s*\*\s*\([0-9]\+\)\s*\)\{0,1}\(\%(\s*)\%(\s*\*\s*[0-9]\+\s*\)\{0,1}\)*\)'
+\       .'\.[a-zA-Z0-9_\-\$]\+\)\)*\)\%(\({[^}]\+}\+\)\)\{0,1}\%(\*\([0-9]\+\)\)\{0,1}\(\%()\%(\*[0-9]\+\)\{0,1}\)*\)'
 
 function! zencoding#html#findTokens(str)
   let str = a:str
+  let last_pos = 0
   while len(str) > 0
-    let token = matchstr(str, s:mx.'\s*$')
+    let token = matchstr(str, '\s'.s:mx)
     if token == ''
       break
-    elseif token =~ '^\s'
-      let token = substitute(token, '^\s*', '', '')
-      let str = str[0 : -len(token)-1]
-      break
     endif
-    let str = str[0 : -len(token)-1]
+	if token =~ '^\s'
+      let last_pos = stridx(str, matchstr(token, '^\s*\zs.*'))
+    endif
+    let str = str[stridx(str, token) + len(token):]
   endwhile
-  return a:str[len(str) :-1]
+  return a:str[last_pos :-1]
 endfunction
 
 function! zencoding#html#parseIntoTree(abbr, type)

unittest.vim

@@ -529,6 +529,12 @@ finish
       'type': "html",
       'result': "foo <p>bar baz</p>",
     },
+    {
+      'name': "f div.boxes>article.box2>header>(hgroup>h2{aaa}+h3{bbb})+p{ccc}$$$$",
+      'query': "f div.boxes>article.box2>header>(hgroup>h2{aaa}+h3{bbb})+p{ccc}$$$$",
+      'type': "html",
+      'result': "f <div class=\"boxes\">\n\t<article class=\"box2\">\n\t\t<header>\n\t\t\t<hgroup>\n\t\t\t\t<h2>aaa</h2>\n\t\t\t\t<h3>bbb</h3>\n\t\t\t</hgroup>\n\t\t\t<p>ccc</p>\n\t\t</header>\n\t</article>\n</div>",
+    },
   ],
 },
 {