def __init__(self, config):
# parse config file
parser = SafeConfigParser()
parser.read(config)
self.Aligner_object = Aligner_IBM1(parser)
self.Extractor_object = Extractor_Moses(parser)
self.Annotator_object = Annotator_onlinecache(parser)
self.logger = logging.getLogger('translation_log.updater')
def __init__(self, config):
# parse config file
parser = SafeConfigParser()
parser.read(config)
self.Aligner_object = Aligner_GIZA(parser)
self.Extractor_object = Extractor_Moses(parser)
self.Annotator_object = Annotator_onlinecache(parser)
self.logger = logging.getLogger('translation_log.updater')
class UpdaterProc(object):
def __init__(self, config):
# parse config file
parser = SafeConfigParser()
parser.read(config)
self.Aligner_object = Aligner_onlineGIZA(parser)
self.Extractor_object = Extractor_Moses(parser)
self.Annotator_object = Annotator_onlinecache(parser)
self.logger = logging.getLogger('translation_log.updater')
def update(self, source="", target=""):
# get alignment information for the (source,correction)
self.log("ALIGNER_INPUT source: " + str(source))
self.log("ALIGNER_INPUT correction: " + str(target))
aligner_output = self.Aligner_object.align(source=source,
correction=target)
self.log("ALIGNER_OUTPUT: " + str(aligner_output))
# get phrase pairs form the alignment information
bias, new, full = self.Extractor_object.extract_phrases(
source, target, aligner_output)
self.log("BIAS: " + str(bias))
self.log("NEW: " + str(new))
self.log("FULL: " + str(full))
self.Annotator_object.cbtm_update(new=new, bias=bias, full=full)
self.Annotator_object.cblm_update(target)
# read and annotate the next sentence
dummy_source = ""
annotated_source = self.Annotator_object.annotate(dummy_source)
return annotated_source
def reset(self):
annotated_source = ''
annotated_source = annotated_source + '<dlt cblm-command="clear"/>'
annotated_source = annotated_source + '<dlt cbtm-command="clear"/>'
return annotated_source
def log(self, message):
self.logger.info(message)
class UpdaterProc(object):
def __init__(self, config):
# parse config file
parser = SafeConfigParser()
parser.read(config)
self.Aligner_object = Aligner_GIZA(parser)
self.Extractor_object = Extractor_Moses(parser)
self.Annotator_object = Annotator_onlinecache(parser)
self.logger = logging.getLogger('translation_log.updater')
def update(self, source="", target=""):
# get alignment information for the (source,correction)
self.log("ALIGNER_INPUT source: "+str(source))
self.log("ALIGNER_INPUT correction: "+str(target))
aligner_output = self.Aligner_object.align(source=source,correction=target)
self.log("ALIGNER_OUTPUT: "+str(aligner_output))
# get phrase pairs form the alignment information
bias, new, full = self.Extractor_object.extract_phrases(source,target,aligner_output)
self.log("BIAS: "+str(bias))
self.log("NEW: "+str(new))
self.log("FULL: "+str(full))
self.Annotator_object.cbtm_update(new=new, bias=bias, full=full)
self.Annotator_object.cblm_update(target)
# read and annotate the next sentence
dummy_source = ""
annotated_source = self.Annotator_object.annotate(dummy_source)
return annotated_source
def reset(self):
annotated_source = ''
annotated_source = annotated_source + '<dlt cblm-command="clear"/>'
annotated_source = annotated_source + '<dlt cbtm-command="clear"/>'
return annotated_source
def log(self, message):
self.logger.info(message)
sys.exit(1)
if extractor_type == "Moses":
Extractor_object = Extractor_Moses(parser)
elif extractor_type == "Constrained_Search":
Extractor_object = Extractor_Constrained_Search(parser)
elif extractor_type == "Dummy":
Extractor_object = Extractor_Dummy(parser)
else:
logging.info("This extractor tool is UNKNOWN")
sys.exit(1)
if annotator_type == "onlinexml":
Annotator_object = Annotator_onlinexml(parser)
elif annotator_type == "onlinecache":
Annotator_object = Annotator_onlinecache(parser)
else:
logging.info("This annotation tool is UNKNOWN:")
sys.exit(1)
# main loop
# initialize: first sentence has no history
source = input.readline().strip()
annotated_source = source
s_id = 1
while source:
logging.info(str(s_id))
# talk to decoder
logging.info("DECODER_IN: " + annotated_source)
decoder_out, decoder_err = Decoder_object.communicate(annotated_source)
logging.info("DECODER_OUT: " + decoder_out)