test_bible_xml.py

from lxml import etree

def bib2html(xmlString):

    root = etree.fromstring(xmlString)

    out = etree.Element("out")

    for beg_verse in root.findall(".//verse[@sID]"):

        this_verse = etree.tostring(beg_verse)

        span = etree.Element("span")
        out.append(span)

        verse_text=this_verse[this_verse.find("/>")+2:]

        span.attrib["class"]="verse"
        span.attrib["id"]=beg_verse.attrib["sID"]

        span2=etree.Element("span")
        span2.attrib["class"]="nobreak"

        span.append(span2)

        vnum=etree.Element("span")
        vnum.attrib["class"]="vnum"
        vnum.text=beg_verse.attrib["sID"].split(".")[-1]

        span2.append(vnum)

        first,rest=verse_text.split(" ",1)
        vnum.tail=" "+first

        span2.tail=rest


    return etree.tostring(out)

def test_judges_3_2():

    xmlString = '<start><verse osisID="Judg.3.2" sID="Judg.3.2"/>Only that the generations of the children of Israel might know, to teach them war, at the least such as before knew nothing thereof;<verse eID="Judg.3.2"/></start>'

    out = bib2html(xmlString)


    assert '<out><span class="verse" id="Judg.3.2"><span class="nobreak"><span class="vnum">2</span> Only</span>that the generations of the children of Israel might know, to teach them war, at the least such as before knew nothing thereof;</span></out>'==out