After running a VASP simulation with Li, La, Zr, and O, I found a bug in vasprun.xml in which Zr is written as r.
$ grep 'PAW_PBE' vasprun.xml
<rc><c> 28</c><c>Li</c><c> 7.01000000</c><c> 1.00000000</c><c> PAW_PBE Li 17Jan2003 </c></rc>
<rc><c> 12</c><c>La</c><c> 138.90000000</c><c> 9.00000000</c><c> PAW_PBE La_s 06Sep2000 </c></rc>
<rc><c> 8</c><c>r </c><c> 91.22400000</c><c> 12.00000000</c><c> PAW_PBE Zr_sv 04Jan2005 </c></rc>
<rc><c> 48</c><c>O </c><c> 16.00000000</c><c> 6.00000000</c><c> PAW_PBE O_s 07Sep2000 </c></rc>
So reading the vasprun.xml file with using ASE module or something like that does not work.
My current solution is just to correct using sed command.
$ sed -i "" -e 's|<c>r </c>|<c>Zr</c>|' vasprun.xml
But this is kind of an annoying bug when one does some automatic parsing of vasprun.xml…