##Before Parsing :
##Change  RECORDX\xc3\x91_5\xc3\xaf\x03\x12  to RECORD
##Change & to and
##fix problems of <Tag_97> manualy.

import libxml2

record = []

def Record(i, node):
    sub_node = node.children
    while sub_node is not None:
        if sub_node.type == "element":
            record[i][sub_node.name] = sub_node.get_content()
            sub_node = sub_node.next
        else:
            sub_node = sub_node.next

# Memory debug specific
libxml2.debugMemory(1)
 
isis_data = libxml2.parseFile ('/root/fctvet.xml')
root = isis_data.getRootElement()
node = root.children
i = 0
while node is not None:
    if node.type != "element":
        node = node.next
        continue
    if node is None:
        break
    if node.name == "RECORD":
        ##print node.get_content()
        record.append({})
        Record(i, node)
        i = i + 1
        node = node.next
    else:
        print "unhandled node in <isis_xml>: " + node.name

isis_data.freeDoc()

# Memory debug specific
libxml2.cleanupParser()
if libxml2.debugMemory(1) == 0:
    print "OK"
else:
    print "Memory leak %d bytes" % (libxml2.debugMemory(1))
    libxml2.dumpMemory()

####################################################################

record_file = open('/root/fctveout.txt', 'w')
tag = ['Tag_4', 'Tag_7', 'Tag_11', 'Tag_12', 'Tag_13', 'Tag_25', 'Tag_60', 'Tag_70', 'Tag_80', 'Tag_18', 'Tag_19', 'Tag_20', 'Tag_21', 'Tag_24', 'Tag_38', 'Tag_10', 'Tag_114', 'Tag_115', 'Tag_110', 'Tag_5', 'Tag_122', 'Tag_97', 'Tag_151', 'Tag_153', 'Tag_123']

for j in record:
    if j.has_key('Tag_691'):
        record_file.write(j['Tag_691'])
##Check docs without 114 tag
##         if j.has_key('Tag_114'):
##             pass
##         else:
##             print j['Tag_691']
    else:
        record_file.write('_B_L_A_N_K_')
    for h in tag:
        if j.has_key(h):
            record_file.write('\t'+j[h])
        else:
            record_file.write('\t_B_L_A_N_K_')
    record_file.write('\n')