tei_import.py

# -*- coding: utf-8 -*-
from django.core.management.base import BaseCommand, CommandError
from eoa.models import *

import xmltodict, re, os, argparse, django
from collections import OrderedDict

from ._importer_class import *

############################
# Add arguments:
# Path to file and file_name
#
# Open file, substitute formating rules,
# and transform into dict
############################
default_path = 'eoa/teidoc/'

default_filename = 'minimalTEI.xml'

class Command(BaseCommand):
    help = 'Import TEI document to Django DB'

    def add_arguments(self, parser):
        parser.add_argument('-p', '--import_path', dest='p', type=str, default=default_path,
                            help='The path to the /tei/xml/file')

        parser.add_argument('-f','--file_name',dest='f', type=str, default=default_filename,
                            help='Name of file to import.')

    def handle(self,*args, **options):
        with open(os.path.join(options['p'],options['f']),'r') as infile:
            # skip this namespaces, i.e. a key in the dict "{namespace}key" will change to "key"
            namespaces = {
                'http://www.tei-c.org/ns/1.0': None,
                'http://www.w3.org/XML/1998/namespace': None
                }
            # REformat highlighting from <hi> ...  </hi> to {rend: italic, text}
            # otherwise it makes the dictionary structure unreadable
            #readfile = re.sub('<hi rend="(\w+)">(.+?)</hi>','{rend: \g<1>, \g<2>}',infile.read())
            readfile = infile.read()
            # parsing to dict
            tei_dict = xmltodict.parse(readfile, process_namespaces=True, namespaces=namespaces)

        ######################
        # Instantiate and add object via models, using _importer_class
        ######################

        text = TEIimporter(tei_dict)

        pub = createPublication(text)

        authorList = createAuthors(text,pub)

        chapters = createChapters(text,pub)

        sections = createSections(text,pub)

        #paragraphs = createParagraphs(text,pub)

        citations = createCitations(text,pub)

        self.stdout.write(
            'Successfully read TEI file, by {0}'.format(text.authors())
                )