def dothenlp():
if request.method == "POST":
try:
myData = request.json['myData']
recogniser = ternip.recogniser()
normaliser = ternip.normaliser()
doc = TernDocument(myData)
strdoc = str(doc)
ref_date = find_between(strdoc, "<DATE_TIME>", "</DATE_TIME>").replace("-","")
sents = recogniser.tag(doc.get_sents())
normaliser.annotate(sents, ref_date)
doc.reconcile(sents)
# Single Sentences
s = find_between(str(doc), "<TEXT>", "</TEXT>").replace("\n"," <br>")
s = s.replace("_QUOTE_",""")
s = s.replace("_APOSTROPHE_","'")
s = s.replace("_AND_","&")
sent_tokens = nltk.sent_tokenize(s)
t = str(doc).split("<TEXT>")[0]
t = t.replace("_QUOTE_",""")
t = t.replace("_APOSTROPHE_","'")
t = t.replace("_AND_","&")
output = t + "\n\n<SENTENCES>\n\n" + str(sent_tokens) + "\n\n</SENTENCES>"
return jsonify(result=output)
except:
return jsonify(result="something wrong")
def test_reconcile_DCT_sents_DATE(self):
d = TernDocument('<DOC><DOCNO>ABC123</DOCNO><DATE>20100801</DATE><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>')
s = d.get_dct_sents()
t = Timex()
t.value = 'ABCDEF'
s[0][0][2].add(t)
d.reconcile_dct(s)
self.assertEquals(str(d), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><DATE><TIMEX2 VAL="ABCDEF">20100801</TIMEX2></DATE><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>').toxml())
def test_create_from_sents(self):
s = TernDocument.create([[('This', 'POS', set()), ('is', 'POS', set()), ('some', 'POS', set()), ('annotated', 'POS', set()), ('text.', 'POS', set())],
[('This', 'POS', set()), ('is', 'POS', set()), ('a', 'POS', set()), ('second', 'POS', set()), ('sentence.', 'POS', set())]], 'ABC123')
self.assertEquals(str(s), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>').toxml())
def test_get_DCT_sents_None(self):
d = TernDocument('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>')
self.assertEquals([[]], d.get_dct_sents())
def test_get_DCT_sents_DATE(self):
d = TernDocument('<DOC><DOCNO>ABC123</DOCNO><DATE>20100801</DATE><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>')
self.assertEquals([[('20100801', 'CD', set())]], d.get_dct_sents())
def test_create_from_sents_with_offsets_tags(self):
sents = [[('This', 'POS', set()), ('is', 'POS', set()), ('some', 'POS', set()), ('annotated', 'POS', set()), ('text.', 'POS', set())],
[('This', 'POS', set()), ('is', 'POS', set()), ('a', 'POS', set()), ('second', 'POS', set()), ('sentence.', 'POS', set())]]
s = TernDocument.create(sents, 'ABC123', tok_offsets=[[2, 7, 11, 16, 28], [36, 41, 45, 46, 53]], add_S='s', add_LEX='lex', pos_attr='pos')
self.assertEquals(str(s), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT> <s><lex pos="POS">This</lex> <lex pos="POS">is</lex> <lex pos="POS">some</lex> <lex pos="POS">annotated</lex> <lex pos="POS">text.</lex></s> <s><lex pos="POS">This</lex> <lex pos="POS">is</lex> <lex pos="POS">a</lex><lex pos="POS">second</lex> <lex pos="POS">sentence.</lex></s></TEXT></BODY></DOC>').toxml())
self.assertEquals(sents, s.get_sents())
def test_create_from_sents_with_dct(self):
s = TernDocument.create([[('This', 'POS', set()), ('is', 'POS', set()), ('some', 'POS', set()), ('annotated', 'POS', set()), ('text.', 'POS', set())],
[('This', 'POS', set()), ('is', 'POS', set()), ('a', 'POS', set()), ('second', 'POS', set()), ('sentence.', 'POS', set())]], 'ABC123',
dct='20100802')
self.assertEquals(str(s), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><DATE_TIME>08/02/2010</DATE_TIME><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>').toxml())
ternip_extent_acts = 0
ternip_extent_cors = 0
ternip_norm_poss = 0
ternip_norm_acts = 0
ternip_norm_cors = 0
start = time.clock()
for i in range(len(unannotated)):
id = os.path.basename(unannotated[i])
# Open the document
try:
with open(unannotated[i]) as fd:
doc = TernDocument(fd.read())
except:
doc = None
print("Unable to load document", id)
continue
# Get DCT
dct_sents = doc.get_dct_sents()
dct_sents = recogniser.tag(dct_sents)
normaliser.annotate(dct_sents, 'XXXXXXXX')
doc.reconcile_dct(dct_sents)
if len(dct_sents) > 0 and len(dct_sents[0]) > 0 and len(
dct_sents[0][0][2]) > 0:
dct = dct_sents[0][0][2].pop().value
else:
dct = ''
import sys
import os
import os.path
import logging
sys.path.append('..')
from ternip.formats.tern import TernDocument
console = logging.StreamHandler()
console.setFormatter(logging.Formatter('[%(asctime)s] %(name)-12s %(levelname)-8s %(message)s'))
logging.getLogger().addHandler(console)
logging.getLogger('ternip').setLevel(logging.INFO)
LOGGER = logging.getLogger()
if not os.path.isdir('preprocessed'):
os.mkdir('preprocessed')
for fpath in glob(os.path.normpath('../sample_data/tern/data/english/ace_2004/*/*.sgm')):
with open(fpath) as fd:
try:
doc = TernDocument(fd.read())
LOGGER.info("Pre-processing " + os.path.basename(fpath))
doc.reconcile_dct(doc.get_dct_sents(), add_S='s', add_LEX='lex', pos_attr='pos')
doc.reconcile(doc.get_sents(), add_S='s', add_LEX='lex', pos_attr='pos')
with open(os.path.join('preprocessed', os.path.basename(fpath)), 'w') as ppfd:
ppfd.write(str(doc)[22:])
except Exception:
LOGGER.exception('Can not load document ' + os.path.basename(fpath))
ternip_extent_acts = 0
ternip_extent_cors = 0
ternip_norm_poss = 0
ternip_norm_acts = 0
ternip_norm_cors = 0
start = time.clock()
for i in range(len(unannotated)):
id = os.path.basename(unannotated[i])
# Open the document
try:
with open(unannotated[i]) as fd:
doc = TernDocument(fd.read())
except:
doc = None
print "Unable to load document", id
continue
# Get DCT
dct_sents = doc.get_dct_sents()
dct_sents = recogniser.tag(dct_sents)
normaliser.annotate(dct_sents, 'XXXXXXXX')
doc.reconcile_dct(dct_sents)
if len(dct_sents) > 0 and len(dct_sents[0]) > 0 and len(dct_sents[0][0][2]) > 0:
dct = dct_sents[0][0][2].pop().value
else:
dct = ''
sents = recogniser.tag(doc.get_sents())
def test_get_DCT_sents_None(self):
d = TernDocument('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>')
self.assertEqual([[]], d.get_dct_sents())
def test_get_DCT_sents_DATE(self):
d = TernDocument('<DOC><DOCNO>ABC123</DOCNO><DATE>20100801</DATE><BODY><TEXT>This is some annotated text. This is a second sentence.</TEXT></BODY></DOC>')
self.assertEqual([[('20100801', 'CD', set())]], d.get_dct_sents())
def test_create_from_sents_with_offsets_tags(self):
sents = [[('This', 'POS', set()), ('is', 'POS', set()), ('some', 'POS', set()), ('annotated', 'POS', set()), ('text.', 'POS', set())],
[('This', 'POS', set()), ('is', 'POS', set()), ('a', 'POS', set()), ('second', 'POS', set()), ('sentence.', 'POS', set())]]
s = TernDocument.create(sents, 'ABC123', tok_offsets=[[2, 7, 11, 16, 28], [36, 41, 45, 46, 53]], add_S='s', add_LEX='lex', pos_attr='pos')
self.assertEqual(str(s), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT> <s><lex pos="POS">This</lex> <lex pos="POS">is</lex> <lex pos="POS">some</lex> <lex pos="POS">annotated</lex> <lex pos="POS">text.</lex></s> <s><lex pos="POS">This</lex> <lex pos="POS">is</lex> <lex pos="POS">a</lex><lex pos="POS">second</lex> <lex pos="POS">sentence.</lex></s></TEXT></BODY></DOC>').toxml())
self.assertEqual(sents, s.get_sents())
def test_create_from_sents_with_offsets(self):
s = TernDocument.create([[('This', 'POS', set()), ('is', 'POS', set()), ('some', 'POS', set()), ('annotated', 'POS', set()), ('text.', 'POS', set())],
[('This', 'POS', set()), ('is', 'POS', set()), ('a', 'POS', set()), ('second', 'POS', set()), ('sentence.', 'POS', set())]],
'ABC123',
tok_offsets=[[2, 7, 11, 16, 28], [36, 41, 45, 46, 53]])
self.assertEqual(str(s), xml.dom.minidom.parseString('<DOC><DOCNO>ABC123</DOCNO><BODY><TEXT> This is some annotated text. This is asecond sentence.</TEXT></BODY></DOC>').toxml())
