class TestClientExpreissions(unittest.TestCase):
def setUp(self):
self.client = FullClient(apiKey=conf.API_KEY, apiServer=conf.BASE_PATH, retinaName=conf.RETINA_NAME)
self.jsonBulkExpression = None
# path relative to current working dir
with open(bulkInput, "r") as f:
self.jsonBulkExpression = "".join(f.readlines())
def testExpressions(self):
fp = self.client.getFingerprintForExpression(oneTermInputJSON, sparsity=0.5)
self.assertGreater(len(fp.positions), 100)
def testContexts(self):
contexts = self.client.getContextsForExpression(simpleExpression, getFingerprint=True, startIndex=0, maxResults=3, sparsity=1.0)
self.assertTrue(contexts != None)
self.assertEqual(3, len(contexts))
c0 = contexts[0]
self.assertGreater(len(c0.fingerprint.positions), 100)
self.assertTrue(conf.isString(c0.context_label))
self.assertTrue(c0.context_id == 0)
def testSimilarTerms(self):
terms = self.client.getSimilarTermsForExpression(simpleExpression, contextId=None, posType="NOUN", getFingerprint=True, startIndex=0, maxResults=8, sparsity=1.0)
self.assertTrue(terms != None)
self.assertEqual(8, len(terms))
for term in terms:
self.assertGreater(len(term.fingerprint.positions), 100)
self.assertTrue(term != None)
self.assertTrue("NOUN" in term.pos_types)
def testExpressionBulk(self):
fps = self.client.getFingerprintsForExpressions(self.jsonBulkExpression)
self.assertEqual(6, len(fps))
for fp in fps:
self.assertGreater(len(fp.positions), 50)
def testExpressionContextsBulk(self):
contextsList = self.client.getContextsForExpressions(self.jsonBulkExpression, getFingerprint=True, startIndex=0, maxResults=3)
self.assertEqual(len(contextsList), 6)
for contextList in contextsList:
for i, context in enumerate(contextList):
self.assertGreater(len(context.fingerprint.positions), 50)
self.assertTrue(conf.isString(context.context_label))
self.assertTrue(context.context_id == i)
def testExpressionSimilarTermsBulk(self):
termsLists = self.client.getSimilarTermsForExpressions(self.jsonBulkExpression, getFingerprint=True, maxResults=7)
self.assertTrue(len(termsLists) == 6)
for termList in termsLists:
self.assertTrue(len(termList) == 7)
self.assertGreater(len(termList[0].fingerprint.positions), 100)
class LiteClient(object):
"""Minimalistic client for accessing core features of Cortical.io's Retina API in a simple way."""
def __init__(self, apiKey):
self._fullClient = FullClient(apiKey, apiServer="http://api.cortical.io/rest", retinaName="en_associative")
def _createDictionary(self, textOrFingerprint):
if type(textOrFingerprint) == str:
return {"text": textOrFingerprint}
elif type(textOrFingerprint) == list:
return {"positions": textOrFingerprint}
else:
raise CorticalioException("Invalid argument, cannot create input from: '%s'" % (str(textOrFingerprint)))
def getSimilarTerms(self, textOrFingerprint):
"""Get the similar terms for a given text or fingerprint
Args:
textOrFingerprint, str OR list of integers
Returns:
list of str: the 20 most similar terms
Raises:
CorticalioException: if the request was not successful
"""
expression = self._createDictionary(textOrFingerprint)
terms = self._fullClient.getSimilarTermsForExpression(json.dumps(expression), maxResults=20)
return [t.term for t in terms]
def getKeywords(self, text):
"""Get a list of keywords from the text
Args:
text, str: The input document
Returns:
list of str
Raises:
CorticalioException: if the request was not successful
"""
terms = self._fullClient.getKeywordsForText(text)
return terms
def getFingerprint(self, text):
"""Get the semantic fingerprint of the input text.
Args:
text, str: The text to be evaluated
Returns:
list of str: the positions of the semantic fingerprint
Raises:
CorticalioException: if the request was not successful
"""
fp = self._fullClient.getFingerprintForText(text)
return fp.positions
def compare(self, textOrFingerprint1, textOrFingerprint2):
"""Returns the semantic similarity of texts or fingerprints. Each argument can be eiter a text or a fingerprint.
Args:
textOrFingerprint1, str OR list of integers
textOrFingerprint2, str OR list of integers
Returns:
float: the semantic similarity in the range [0;1]
Raises:
CorticalioException: if the request was not successful
"""
compareList = [self._createDictionary(textOrFingerprint1), self._createDictionary(textOrFingerprint2)]
metric = self._fullClient.compare(json.dumps(compareList))
return metric.cosineSimilarity
def createCategoryFilter(self, positiveExamples):
"""Creates a filter fingerprint.
Args:
positiveExamples, list(str): The list of positive example texts.
Returns:
list of int: the positions representing the filter representing the texts
Raises:
CorticalioException: if the request was not successful
"""
categoryFilter = self._fullClient.createCategoryFilter("CategoryFilter", positiveExamples)
return categoryFilter.positions
class LiteClient(object):
"""Minimalistic client for accessing core features of Cortical.io's Retina API in a simple way."""
def __init__(self, apiKey):
self._fullClient = FullClient(apiKey,
apiServer="http://api.cortical.io/rest",
retinaName="en_associative")
def _createDictionary(self, textOrFingerprint):
if type(textOrFingerprint) == str:
return {"text": textOrFingerprint}
elif type(textOrFingerprint) == list:
return {"positions": textOrFingerprint}
else:
raise CorticalioException(
"Invalid argument, cannot create input from: '%s'" %
(str(textOrFingerprint)))
def getSimilarTerms(self, textOrFingerprint):
"""Get the similar terms for a given text or fingerprint
Args:
textOrFingerprint, str OR list of integers
Returns:
list of str: the 20 most similar terms
Raises:
CorticalioException: if the request was not successful
"""
expression = self._createDictionary(textOrFingerprint)
terms = self._fullClient.getSimilarTermsForExpression(
json.dumps(expression), maxResults=20)
return [t.term for t in terms]
def getKeywords(self, text):
"""Get a list of keywords from the text
Args:
text, str: The input document
Returns:
list of str
Raises:
CorticalioException: if the request was not successful
"""
terms = self._fullClient.getKeywordsForText(text)
return terms
def getFingerprint(self, text):
"""Get the semantic fingerprint of the input text.
Args:
text, str: The text to be evaluated
Returns:
list of str: the positions of the semantic fingerprint
Raises:
CorticalioException: if the request was not successful
"""
fp = self._fullClient.getFingerprintForText(text)
return fp.positions
def compare(self, textOrFingerprint1, textOrFingerprint2):
"""Returns the semantic similarity of texts or fingerprints. Each argument can be eiter a text or a fingerprint.
Args:
textOrFingerprint1, str OR list of integers
textOrFingerprint2, str OR list of integers
Returns:
float: the semantic similarity in the range [0;1]
Raises:
CorticalioException: if the request was not successful
"""
compareList = [
self._createDictionary(textOrFingerprint1),
self._createDictionary(textOrFingerprint2)
]
metric = self._fullClient.compare(json.dumps(compareList))
return metric.cosineSimilarity
def createCategoryFilter(self, positiveExamples):
"""Creates a filter fingerprint.
Args:
positiveExamples, list(str): The list of positive example texts.
Returns:
list of int: the positions representing the filter representing the texts
Raises:
CorticalioException: if the request was not successful
"""
categoryFilter = self._fullClient.createCategoryFilter(
"CategoryFilter", positiveExamples)
return categoryFilter.positions
class TestClientExpreissions(unittest.TestCase):
def setUp(self):
self.client = FullClient(apiKey=conf.API_KEY,
apiServer=conf.BASE_PATH,
retinaName=conf.RETINA_NAME)
self.jsonBulkExpression = None
# path relative to current working dir
with open(bulkInput, "r") as f:
self.jsonBulkExpression = "".join(f.readlines())
def testExpressions(self):
fp = self.client.getFingerprintForExpression(oneTermInputJSON,
sparsity=0.5)
self.assertGreater(len(fp.positions), 100)
def testContexts(self):
contexts = self.client.getContextsForExpression(simpleExpression,
getFingerprint=True,
startIndex=0,
maxResults=3,
sparsity=1.0)
self.assertTrue(contexts != None)
self.assertEqual(3, len(contexts))
c0 = contexts[0]
self.assertGreater(len(c0.fingerprint.positions), 100)
self.assertTrue(conf.isString(c0.context_label))
self.assertTrue(c0.context_id == 0)
def testSimilarTerms(self):
terms = self.client.getSimilarTermsForExpression(simpleExpression,
contextId=None,
posType="NOUN",
getFingerprint=True,
startIndex=0,
maxResults=8,
sparsity=1.0)
self.assertTrue(terms != None)
self.assertEqual(8, len(terms))
for term in terms:
self.assertGreater(len(term.fingerprint.positions), 100)
self.assertTrue(term != None)
self.assertTrue("NOUN" in term.pos_types)
def testExpressionBulk(self):
fps = self.client.getFingerprintsForExpressions(
self.jsonBulkExpression)
self.assertEqual(6, len(fps))
for fp in fps:
self.assertGreater(len(fp.positions), 50)
def testExpressionContextsBulk(self):
contextsList = self.client.getContextsForExpressions(
self.jsonBulkExpression,
getFingerprint=True,
startIndex=0,
maxResults=3)
self.assertEqual(len(contextsList), 6)
for contextList in contextsList:
for i, context in enumerate(contextList):
self.assertGreater(len(context.fingerprint.positions), 50)
self.assertTrue(conf.isString(context.context_label))
self.assertTrue(context.context_id == i)
def testExpressionSimilarTermsBulk(self):
termsLists = self.client.getSimilarTermsForExpressions(
self.jsonBulkExpression, getFingerprint=True, maxResults=7)
self.assertTrue(len(termsLists) == 6)
for termList in termsLists:
self.assertTrue(len(termList) == 7)
self.assertGreater(len(termList[0].fingerprint.positions), 100)
