import sys, re
from lxml import etree as E

fn = r"D:/install/oracle/B19306_01/server.102/b14219/e0.htm"
#fn = r"D:/install/oracle/B19306_01/server.102/b14219/e12700.htm"


def parse(fn):
    parser = E.XMLParser(remove_blank_text=True)
    x = E.parse(open(fn, "rb"))
    root = x.getroot()

    reg1 = re.compile('''<span class="msg"><a name="ORA-\d\d\d\d\d" id="ORA-\d\d\d\d\d"></a>(.*)</span>''')
    reg2 = re.compile('''<span class="msgexplankw">(.*)</span>(.*)''', re.M)
    reg3 = re.compile('''<span class="msgactionkw">(.*)</span>(.*)''', re.M)

    regS = re.compile('''([^<>]*)<span class="italic">([a-zA-Z0-9]*)</span>(.*)$''', re.M)

    def clear_span(s):
        return s

    for el in x.getiterator():
        try:
            if el.attrib["class"] == "msgentry":

                msgerror = ""
                msgexpl = ""
                msgaction = ""

                for inner in el.getiterator():
                    try:
                        if inner.tag.find("span") and inner.attrib["class"] == "msg":
                            s = E.tostring(inner)
                            m = reg1.match(s)
                            if m:
                                msgerror = clear_span(m.group(1))
                        elif inner.tag.find("div") and inner.attrib["class"] == "msgexplankw":
                            s = E.tostring(inner)
                            m = reg2.match(s)
                            if m:
                                msgexpl = clear_span(m.group(2))
                        elif inner.tag.find("div") and inner.attrib["class"] == "msgactionkw":
                            s = E.tostring(inner)
                            m = reg3.match(s)
                            if m:
                                msgaction = clear_span(m.group(2))

                                print msgerror, '\t', msgexpl, '\t', msgaction
                                break
                    except Exception, e:
                        #print e
                        pass
        except Exception, e:
            pass

parse(fn)