words_split.py

def txt_split(s):
    import re
    indx_srch = '<[a-z0-9 ]+>'
    txt_frags = re.split(indx_srch, s)
    indx_lst = re.findall(indx_srch,s)
    ret=[]
    ret.extend(re.split(' +', txt_frags[0]))
    for i,indx in enumerate( indx_lst):

        if ret[-1]=='':
            ret[-1]=[indx]
        else:
            if not isinstance( ret[-1], list):
                ret[-1]=[ret[-1]]
            ret[-1].append( indx)
        cur_lst = re.split(' +',  txt_frags[i+1])
        if cur_lst[0]!='':
            ret[-1].append(cur_lst.pop(0))
        else:
            cur_lst.pop(0)
        ret.extend(cur_lst)
    return ret