class TranslatorInterface():
"""An interface to a single, possibly multilingual, model."""
def __init__(self, srclang, targetlang, service, model):
self.service = service
self.contentprocessor = ContentProcessor(
srclang,
targetlang,
sourcebpe=self.service.get('sourcebpe'),
targetbpe=self.service.get('targetbpe'),
sourcespm=self.service.get('sourcespm'),
targetspm=self.service.get('targetspm'))
self.worker = model
# becomes nonempty if there are multiple target languages
self.preamble = ""
def translate(self, text):
sentences = self.contentprocessor.preprocess(text)
translatedSentences = self.worker.translate(self.preamble +
'\n'.join(sentences))
translation = self.contentprocessor.postprocess(translatedSentences)
return ' '.join(translation)
def ready(self):
return self.worker != None and self.worker.ready()
def on_exit(self):
if self.worker != None:
self.worker.on_exit()
class TranslatorWorker():
def __init__(self, srclang, targetlang, service):
self.q = queues.Queue()
# Service definition
self.service = service
self.p = None
self.contentprocessor = ContentProcessor(
srclang,
targetlang,
sourcebpe=self.service.get('sourcebpe'),
targetbpe=self.service.get('targetbpe'),
sourcespm=self.service.get('sourcespm'),
targetspm=self.service.get('targetspm'))
self.ws_url = "ws://{}:{}/translate".format(self.service['host'],
self.service['port'])
if self.service['configuration']:
self.run()
@gen.coroutine
def run(self):
process.Subprocess.initialize()
self.p = process.Subprocess([
'marian-server',
'-c',
self.service['configuration'],
'-p',
self.service['port'],
'--allow-unk',
# enables translation with a mini-batch size of 64, i.e. translating 64 sentences at once, with a beam-size of 6.
'-b',
'6',
'--mini-batch',
'64',
# use a length-normalization weight of 0.6 (this usually increases BLEU a bit).
'--normalize',
'0.6',
'--maxi-batch-sort',
'src',
'--maxi-batch',
'100',
])
self.p.set_exit_callback(self.on_exit)
ret = yield self.p.wait_for_exit()
def on_exit(self):
print("Process exited")
def translate(self, srctxt):
ws = websocket.create_connection(self.ws_url)
sentences = self.contentprocessor.preprocess(srctxt)
ws.send('\n'.join(sentences))
translatedSentences = ws.recv().split('\n')
ws.close()
translation = self.contentprocessor.postprocess(translatedSentences)
return ' '.join(translation)
class TranslatorWorker():
def __init__(self, srclang, targetlang, service):
self.q = queues.Queue()
# Service definition
self.service = service
self.p = None
self.contentprocessor = ContentProcessor(
srclang,
targetlang,
sourcebpe=self.service.get('sourcebpe'),
targetbpe=self.service.get('targetbpe'),
sourcespm=self.service.get('sourcespm'),
targetspm=self.service.get('targetspm')
)
self.ws_url = "ws://{}:{}/translate".format(
self.service['host'], self.service['port'])
if self.service['configuration']:
self.run()
@gen.coroutine
def run(self):
process.Subprocess.initialize()
self.p = process.Subprocess(['marian-server', '-c',
self.service['configuration'],
'--quiet-translation',
'-p', self.service['port']])
self.p.set_exit_callback(self.on_exit)
ret = yield self.p.wait_for_exit()
def on_exit(self):
print("Process exited")
def translate(self, srctxt):
ws = websocket.create_connection(self.ws_url)
sentences = self.contentprocessor.preprocess(srctxt)
translatedSentences = []
for sentence in sentences:
ws.send(sentence)
translatedSentences.append(ws.recv())
ws.close()
translation = self.contentprocessor.postprocess(translatedSentences)
return ' '.join(translation)
