def test_shouldExtractSummary(self):
summarizer = mock()
document_text = "This is one summary sentence. This is what you are trying to summarize. " \
+ "In this test it does not matter what is the kind of text you supply, this is another summary sentence. " \
+ "That is because it uses a mock and is trying to test the behavior of api. The API sanity check for " \
+ "summary mechanism itself is showcased in summary test"
document = Document(doc_id="123",
text=document_text,
summarizer=summarizer)
sentence_map = {
0:
'This is one summary sentence.',
1:
'This is what you are trying to summarize.',
2:
'In this test it does not matter what is the kind of text you supply, this is another summary sentence.',
3:
'That is because it uses a mock and is trying to test the behavior of api.',
4:
'The API sanity check for summary mechanism itself is showcased in summary test'
}
tokenised_sentence_map = {
0: ['summary', 'sentence'],
1: ['summarize'],
2: [
'test', 'matter', 'kind', 'text', 'supply', 'summary',
'sentence'
],
3: ['mock', 'test', 'behavior', 'api'],
4: [
'API', 'sanity', 'check', 'summary', 'mechanism', 'showcased',
'summary', 'test'
]
}
tokens = [
'summary', 'sentence', 'summarize', 'test', 'matter', 'kind',
'text', 'supply', 'summary', 'sentence', 'mock', 'test',
'behavior', 'api', 'API', 'sanity', 'check', 'summary',
'mechanism', 'showcased', 'summary', 'test'
]
processed_document = ProcessedDocument(
sentence_map=sentence_map,
tokenised_sentence_map=tokenised_sentence_map,
tokens=tokens)
summary_sentences = [
"This is one summary sentence", "this is another summary sentence"
]
expected_summary = DocumentSummary("123", summary_sentences)
when(summarizer).summarize_using_weighing_measures(
processed_document).thenReturn(summary_sentences)
document_summary = document.summary()
verify(summarizer).summarize_using_weighing_measures(
processed_document)
self.assertEquals(expected_summary, document_summary)
def put(self):
request_body = self.request.body
logger.debug("Request received for summarisation; Request body: %s" %
request_body[0:20])
parameters = json.loads(request_body)
callback_url = parameters["callback"]
document_id = parameters["documentId"]
extracted_text = parameters["extractedText"]
compression_ratio = int(parameters["compressionRatio"])
summarizer = Summarizer(compression_ratio=compression_ratio)
document = Document(doc_id=document_id,
text=extracted_text,
summarizer=summarizer)
try:
logger.info("Generating Summary for document %s" % document_id)
document_summary = document.summary()
except Exception as e:
logger.error(
"Error while generating summary for document %s" % document_id,
e)
requests.post(callback_url,
data=json.dumps({
"status": "failure",
"message": e.message,
"documentId": document_id
}),
headers={'Content-Type': 'application/json'})
else:
logger.info(
"Summarisation completed for document %s. Updating to callback %s and sample summary %s"
%
(document_id, callback_url, document_summary.to_json()[0:20]))
requests.post(callback_url,
data=(document_summary.to_json()),
headers={'Content-Type': 'application/json'})