path: root/versification/av11n.py

#!/usr/bin/env python
# coding: utf-8
#
# This does a very roughshod attempt to compare the osisIDs found in an
# XML file with each of the versifications that SWORD knows about to help
# a user find the one which is most akin to the one they are using. It is
# limited in its need for your file to be at least segregated into OT/NT
# in the proper order, although within each testament, it requires nothing
# special as for ordering.
#
# Invoke simply by calling the program and the file name.
import logging
# in normal state level should be debug.WARNING, debug.INFO and debug.DEBUG
# give additional information.
logging.basicConfig(format='%(levelname)s:%(funcName)s:%(message)s',
                    level=logging.WARNING)
import re
import sys
verseid = re.compile(r'^.+\..+\..+$')

# Inform the user that we need the SWORD extension
try:
    import Sword
except ImportError:
    logging.exception(
        "You do not have the SWORD library installed. Please install it.")
    sys.exit(1)

# Inform the user that we need pyquery, as it makes parsing XML files
# that much easier
try:
    from pyquery import PyQuery as pq  # noqa
except ImportError:
    logging.exception(
        "You do not appear to have PyQuery installed. Please install it.")
    sys.exit(2)

# Without the name of a file, we cannot proceed any further
if len(sys.argv) < 2 or sys.argv[1] == '--help':
    print >>sys.stderr, "Usage: %s <OSISfile>" % sys.argv[0]
    sys.exit(1)

# Open the file
logging.debug('Opening %s' % (sys.argv[1],))
d = pq(filename=sys.argv[1])
# Get the list of versifications
logging.debug('Fetching a list of v11ns')
vmgr = Sword.VersificationMgr.getSystemVersificationMgr()
av11ns = vmgr.getVersificationSystems()

# Get the list of all osisIDs
logging.debug('Fetching a list of OSIS IDs')
ids = d("*[osisID]")
# Iterate each versification scheme
for v11n in av11ns:
    print('Checking %s' % v11n.c_str())
    # Construct a list of the IDs in this versification
    key = Sword.VerseKey()
    key.setVersificationSystem(v11n.c_str())
    # Anything left in this afterwards is missing from the OSIS ot
    otkeyList = []
    # Anything left in this afterwards is missing from the OSIS nt
    ntkeyList = []
    # Anything that gets placed in here is extraneous OT material (we think)
    otextraKeys = []
    # Anything that gets placed in here is extraneous NT material (we think)
    ntextraKeys = []

    inNT = False
    while key.popError() == '\x00':
        skey = key.getOSISRef()
        # Assume we enter the NT when we hit Matthew
        if not inNT and skey.startswith('Matt'):
            inNT = True
        if inNT:
            ntkeyList.append(skey)
        else:
            otkeyList.append(skey)
        key.increment()
    ntkeyList = set(ntkeyList)  # The 'in' operator only works on a set
    otkeyList = set(otkeyList)

    inNT = False
    # Now iterate the ones we have in this file
    for e in ids:
        logging.debug('e = %s', e)
        osisid = e.attrib.get('osisID')
        #print 'Checking key %s' % (osisid,)
        if osisid in otkeyList:
            otkeyList.remove(osisid)
        elif osisid in ntkeyList:
            ntkeyList.remove(osisid)
            inNT = True
        elif verseid.match(osisid) and inNT:
            ntextraKeys.append(osisid)
        elif verseid.match(osisid) and not inNT:
            otextraKeys.append(osisid)
        # Ignore it if not verseid.match()

    # Now let's see what is left over
    # Sets in Python cannot be ordered
    keyList = list(otkeyList.union(ntkeyList))
    keyList.sort()
    if len(keyList) > 0:
        logging.info('\tThe following IDs don’t appear in your file:\n%s',
                     str("\n".join(keyList)))
        print ('\tThere are %d OT IDs and %d NT IDs ' +
               'in v11n which arn’t in your file.') \
            % (len(otkeyList), len(ntkeyList))
    else:
        print '\tYour file has all the references in this v11n'

    # Now let's see if you had extra
    if len(otextraKeys + ntextraKeys) > 0:
        logging.info(
            '\tThe following IDs don’t appear in v11n:\n%s',
            str("\n".join(keyList)))
        print ('\tThere are %d OT IDs and %d NT IDs ' +
               'in your file which don’t appear in v11n.') \
            % (len(otextraKeys), len(ntextraKeys))
    else:
        print '\tYour file has no extra references'