Vim Tips Wiki
Tip 1639 Printable Monobook Previous Next

created 2009 · complexity basic · author EUEU · version 7.0

After running a cscope command, you may have duplicate entries in your quickfix list, for instance when a header file is shared by multiple projects. This tip shows how to sort the quickfix list by file name and line number, and remove duplicates.


Create file ~/.vim/plugin/quickfix.vim (Unix) or $HOME/vimfiles/plugin/quickfix.vim (Windows) containing the script below, then restart Vim. After a quickfix command such as :grep is run, the script will automatically be invoked by the QuickfixCmdPost event, and the quickfix list will be sorted by file name and line number, and duplicates will be removed.

function! s:CompareQuickfixEntries(i1, i2)
  if bufname(a:i1.bufnr) == bufname(a:i2.bufnr)
    return a:i1.lnum == a:i2.lnum ? 0 : (a:i1.lnum < a:i2.lnum ? -1 : 1)
    return bufname(a:i1.bufnr) < bufname(a:i2.bufnr) ? -1 : 1

function! s:SortUniqQFList()
  let sortedList = sort(getqflist(), 's:CompareQuickfixEntries')
  let uniqedList = []
  let last = ''
  for item in sortedList
    let this = bufname(item.bufnr) . "\t" . item.lnum
    if this !=# last
      call add(uniqedList, item)
      let last = this
  call setqflist(uniqedList)
autocmd! QuickfixCmdPost * call s:SortUniqQFList()

Original script[]


  • Following is the original code. I am fairly confident that the version shown above is equivalent, but I'm keeping the original code for a while in the hope it will get more checking.
  • I do not undertand how two buffers could have the same name while having different buffer numbers (buffer names seem to be unique). I would therefore think that 'bufname()' could be omitted so the tests simply check buffer numbers, rather than buffer names. JohnBeckett 12:07, May 13, 2010 (UTC)
As far as I can remember I had problems with duplicated entries in my quickfix list because on Ms Windows some file would appear with both 8.3 short names and long name in wy quickfix list as a result of a CScope command. This is why I did that but I am not sure whether this is really required.EUEU 16:06, May 18, 2010 (UTC)
function! <SID>CompareQuickfixEntries(i1, i2)
  if bufname((a:i1).bufnr) == bufname((a:i2).bufnr)
    return (a:i1).lnum == (a:i2).lnum ? 0 : ( (a:i1).lnum < (a:i2).lnum ? -1 : 1)
  elseif bufname((a:i1).bufnr) < bufname((a:i2).bufnr)
    return -1
    return 1

function! SortUniqQFList()
  let s:sortedList = sort(getqflist(), "<SID>CompareQuickfixEntries")
  let s:uniqedList = []
  let s:olditem = {}
  for s:item in s:sortedList
    if s:olditem == {}
      let s:uniqedList += [s:item]
    elseif bufname((s:item).bufnr) != bufname((s:olditem).bufnr)
      let s:uniqedList += [s:item]
    elseif (s:item).lnum != (s:olditem).lnum
      let s:uniqedList += [s:item]
    let s:olditem = s:item
  call setqflist(s:uniqedList)
au QuickfixCmdPost * call SortUniqQFList()


Possibly bufname() could be omitted from each expression like bufname(a:i1.bufnr) because buffer names are unique, except that more than one buffer can have an empty name. Also should contemplate Windows systems where buffer names (like file names) are not case sensitive. Using == to compare two names is either case sensitive or not, depending on the setting of 'ignorecase'. JohnBeckett 05:01, May 15, 2010 (UTC)