class CoreNlpClient:
SERVER_URL = 'http://localhost:9000'
COMMAND_PATTERN = 'java -Xmx{} -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLP {}'
SERVER_COMMAND_PATTERN = 'java -Xmx{} -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -timeout {}'
def __init__(self, cwd, memory='4g', timeout=15000, verbose=False):
self._cwd = cwd
self._memory = memory
self._timeout = timeout
self._verbose = verbose
self._process = None
self._http_client = None
def start(self):
command = self.SERVER_COMMAND_PATTERN.format(self._memory,
self._timeout)
self._process = self._open_process(command, wait=False)
self._wait_for_server()
self._http_client = StanfordCoreNLP(self.SERVER_URL)
def stop(self):
self._process.kill()
def annotate(self, text, annotators, properties=None, http=False):
# Build properties
properties = properties or {}
properties = {
**properties,
**{
'outputFormat': 'json',
'annotators': ','.join(annotators)
}
}
# Run annotators via HTTP request or command line execution
if http:
if self._http_client is None:
raise Exception('CoreNLP client is not running!')
result = self._http_client.annotate(text, properties)
else:
result = self._annotate_cmd(text, properties)
# Raise exceptions thrown by CoreNLP
if not isinstance(result, dict):
raise Exception(result)
return result
def _annotate_cmd(self, text, properties):
with NamedTemporaryFile(mode='w') as text_file:
# Write text to temporary file so that is can be processed by
# CoreNLP process
text_file.write(text)
text_file.flush()
# Run CoreNLP as subprocess
self._run_cmd({**properties, **{'file': text_file.name}})
# Read results and delete result file afterwards
result_file_name = os.path.join(
self._cwd,
os.path.basename(text_file.name) + '.json')
with open(result_file_name) as result_file:
try:
results = json.load(result_file)
except json.decoder.JSONDecodeError:
results = result_file.read()
os.remove(result_file_name)
return results
def semgrex(self, text, pattern, filter=False):
if self._http_client is None:
raise Exception('CoreNLP client is not running!')
return self._http_client.semgrex(text, pattern=pattern, filter=filter)
def _run_cmd(self, properties):
arguments = ' '.join(
['-{} {}'.format(key, value) for key, value in properties.items()])
command = self.COMMAND_PATTERN.format(self._memory, arguments)
return self._open_process(command)
def _open_process(self, command, wait=True):
output = None if self._verbose else subprocess.DEVNULL
process = subprocess.Popen(command,
cwd=self._cwd,
shell=True,
stdout=output,
stderr=output)
if wait:
process.wait()
else:
return process
def _wait_for_server(self):
ready = False
try:
response = requests.get(self.SERVER_URL + '/ready')
ready = response.status_code == 200 and response.text.startswith(
'ready')
except ConnectionError:
pass
if not ready:
time.sleep(1)
self._wait_for_server()
corenlp_jars=["E:/software/stanford-corenlp-full-2017-06-09/to/stanford-corenlp-full-2015-04-20/*", "E:/software/to/stanford-srparser-2014-10-23-models.jar"])
'''
res = nlp.annotate("""I love you. I hate him. You are nice. He is dumb.
The best way to hope for any chance of enjoying
this film is by lowering your expectations.The show starts out as competent
but unremarkable and gradually grows in to a considerable power. will kill you. don't come. fool""",
properties={
'annotators': 'sentiment',
'outputFormat': 'json'
})
#print(res)
#The average sentiment of tweets is between Neutral (2) and Negative (1), the range is from
#VeryNegative (0) to VeryPositive (4) which appear to be quite rare.
for s in res["sentences"]:
print("%d: '%s': %s %s" % (s["index"], " ".join(
[t["word"]
for t in s["tokens"]]), s["sentimentValue"], s["sentiment"]))
text = ('Pusheen and Smitha walked along the beach. Pusheen wanted to surf,'
'but fell off the surfboard.')
output = nlp.annotate(text,
properties={
'annotators': 'tokenize,ssplit,pos,depparse,parse',
'outputFormat': 'json'
})
print(output['sentences'][0]['parse'])
output = nlp.tokensregex(text, pattern='/Pusheen|Sumitha/', filter=False)
print(output)
output = nlp.semgrex(text, pattern='{tag: VBD}', filter=False)
print(output)
from pycorenlp import StanfordCoreNLP
if __name__ == '__main__':
nlp = StanfordCoreNLP('http://localhost:9000')
text = (
'Pusheen and Smitha walked along the beach. Pusheen wanted to surf,'
'but fell off the surfboard.')
output = nlp.annotate(text, properties={
'annotators': 'tokenize,ssplit,pos,depparse,parse',
'outputFormat': 'json'
})
print(output['sentences'][0]['parse'])
output = nlp.tokensregex(text, pattern='/Pusheen|Smitha/', filter=False)
print(output)
output = nlp.semgrex(text, pattern='{tag: VBD}', filter=False)
print(output)
