#!/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 argparse
import io
import logging
# in normal state level should be debug.WARNING, debug.INFO and debug.DEBUG
# give additional information.
logging.basicConfig(format='%(levelname)s:%(message)s',
                    level=logging.INFO)
log = logging.getLogger('versification')

import re
import sys
try:
    import lxml.etree as ET
except ImportError:
    import xml.etree.ElementTree as ET

OSIS_NS = 'http://www.bibletechnologies.net/2003/OSIS/namespace'
VERSEID_RE = re.compile(r'^(.+\.\d+\.\d+).*$')

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

arg_parser = argparse.ArgumentParser(
    description='Compare OSIS file with available v11ns.')

arg_parser.add_argument('--verbose', '-v', action='count')
arg_parser.add_argument('filename', nargs=1)


args = arg_parser.parse_args()

if args.verbose:
    log.setLevel = logging.DEBUG

log.debug('args = %s', args)

# Open the file
log.debug('Opening %s' % args.filename[0])

tree = ET.parse(io.open(args.filename[0], encoding='utf8')).getroot()
# Get the list of versifications
log.debug('Fetching a list of v11ns')
vmgr = Sword.VersificationMgr.getSystemVersificationMgr()
av11ns = vmgr.getVersificationSystems()
log.debug('av11ns = %s', av11ns)

# Get the list of all osisIDs
log.debug('Fetching a list of OSIS IDs')
ids = set()
for item in tree.iter('{%s}verse' % OSIS_NS):
    if 'osisID' in item.attrib:
        ids.add(item.attrib['osisID'].split('!')[0])
log.debug('ids = len(%d)', len(ids))

# Iterate each versification scheme
for v11n in av11ns:
    v11n_name = v11n.c_str()
    print('\nChecking %s:\n%s' %
          (v11n_name, (len(v11n_name) + 10) * '-'))
    # 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 osisid in ids:
#        log.debug('Checking key %s', osisid)
        if osisid in otkeyList:
            otkeyList.remove(osisid)
        elif osisid in ntkeyList:
            ntkeyList.remove(osisid)
            inNT = True
        else:
            verse_match = VERSEID_RE.match(osisid)
            if verse_match and inNT:
                ntextraKeys.append(verse_match.group(1))
            elif verse_match and not inNT:
                otextraKeys.append(verse_match.group(1))
            # Ignore it if not VERSEID_RE.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:
        if len(keyList) < 100:
            log.info('\tThe following IDs don’t appear in your file:\n%s',
                         str(", ".join(keyList)))
        print ('\tThere are %d OT IDs and %d NT IDs ' +
               'in v11n which aren’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:
        # It doesn't make sense to print out lists longer than 100
        # they cannot be read anyway
        if len(keyList) < 100:
            log.info(
                '\tThe following IDs don’t appear in v11n:\n%s',
                str(", ".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'