def minimize_context(context_str, from_lang_code, word_str):
_query = TranslationQuery.for_word_occurrence(word_str, context_str, 1, 7)
processor = RemoveUnnecessarySentences(from_lang_code)
query = processor.process_query(_query)
minimal_context = (query.before_context + ' ' + query.query +
query.after_context)
return minimal_context, query
class TestRemoveUnnecessarySentences(TestCase):
def setUp(self):
self.query_processor = RemoveUnnecessarySentences(language_code='nl')
def test_single_sentence(self):
q = TranslationQuery(
before_context='De directeur',
query='treedt',
after_context=' af.'
)
result = self.query_processor.process_query(q)
self.assertEqual(q, result)
def test_multiple_sentences(self):
query = TranslationQuery(
before_context='Justitieminister Koen Geens (CD&V) werkt aan een wetsontwerp dat burgerinfiltranten '
'mogelijk maakt in de strijd tegen',
query='terrorisme',
after_context='en georganiseerde misdaad. Dat zegt hij woensdag in Knack, nadat Brussels '
'procureur-generaal Johan Delmulle daar vorig najaar een lans voor had gebroken. '
'Daarnaast werkt Geens ook aan een regeling rond spijtoptanten. Een akkoord binnen de '
'meerderheid is er nog niet..'
)
result = self.query_processor.process_query(query)
self.assertEqual(query.before_context, result.before_context) # before context should remain the same
self.assertEqual(query.query, result.query) # query should be untouched
self.assertEqual(result.after_context, 'en georganiseerde misdaad.')
def minimize_context(context_str, from_lang_code, word_str):
_query = TranslationQuery.for_word_occurrence(word_str, context_str, 1, 7)
processor = RemoveUnnecessarySentences(from_lang_code)
query = processor.process_query(_query)
minimal_context = (
query.before_context + ' ' + query.query + query.after_context)
return minimal_context, query
def build_with_context(source_language: str,
target_language: str,
key=None) -> MicrosoftTranslator:
"""
Builds a Microsoft translator with suitable context processors for the given source and target language.
:param source_language:
:param target_language:
:param key:
:return:
"""
translator = MicrosoftTranslatorFactory.build_clean(
source_language, target_language, key)
# Right now only apply the processor to Dutch, English, German, French, Spanish
if source_language in ['nl', 'en', 'de', 'fr', 'es']:
translator.add_query_processor(
RemoveUnnecessarySentences(source_language))
if source_language in list(conjunctions.keys()):
translator.add_query_processor(
RemoveUnnecessaryConjunctions(conjunctions[source_language]))
translator.translator_name = 'Microsoft - with context'
translator.service_name = 'Microsoft - with context'
return translator
def setUp(self):
self.query_processor = RemoveUnnecessarySentences(language_code='nl')