tests/test_text.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 from __future__ import unicode_literals
   7
   8 import unittest
   9 from librarian import NoDublinCore
  10 from librarian.builders import builders
  11 from librarian.parser import WLDocument as LegacyWLDocument
  12 from librarian.document import WLDocument
  13 from nose.tools import *
  14 from .utils import get_fixture
  15
  16
  17 class TextTests(unittest.TestCase):
  18     maxDiff = None
  19
  20     def test_transform_legacy(self):
  21         expected_output_file_path = get_fixture('text', 'asnyk_miedzy_nami_expected.txt')
  22
  23         text = LegacyWLDocument.from_file(
  24             get_fixture('text', 'miedzy-nami-nic-nie-bylo.xml')
  25         ).as_text().get_bytes().decode('utf-8')
  26
  27         with open(expected_output_file_path, 'rb') as f:
  28             self.assertEqual(text, f.read().decode('utf-8'))
  29
  30     def test_transform(self):
  31         expected_output_file_path = get_fixture('text', 'asnyk_miedzy_nami_expected.txt')
  32
  33         text = WLDocument(
  34             filename=get_fixture('text', 'miedzy-nami-nic-nie-bylo.xml')
  35         ).build(builders['txt']).get_bytes().decode('utf-8')
  36
  37         with open(expected_output_file_path, 'rb') as f:
  38             self.assertEqual(text, f.read().decode('utf-8'))
  39
  40
  41     def test_transform_raw(self):
  42         expected_output_file_path = get_fixture('text', 'asnyk_miedzy_nami_expected_raw.txt')
  43
  44         text = LegacyWLDocument.from_file(
  45             get_fixture('text', 'miedzy-nami-nic-nie-bylo.xml')
  46         ).as_text(flags=['raw-text']).get_bytes().decode('utf-8')
  47
  48         with open(expected_output_file_path, 'rb') as f:
  49             self.assertEqual(text, f.read().decode('utf-8'))
  50
  51
  52 @raises(NoDublinCore)
  53 def test_no_dublincore():
  54     LegacyWLDocument.from_file(
  55             get_fixture('text', 'asnyk_miedzy_nami_nodc.xml')
  56         ).as_text()
  57
  58
  59 def test_passing_parse_dublincore_to_transform():
  60     """Passing parse_dublincore=False to the constructor omits DublinCore parsing."""
  61     LegacyWLDocument.from_file(
  62             get_fixture('text', 'asnyk_miedzy_nami_nodc.xml'),
  63             parse_dublincore=False,
  64         ).as_text()