Infrastructure for HPYPLM, config file management.

1 files changed, 30 insertions, 6 deletions

diff --git a/realtime/rt/rt.py b/realtime/rt/rt.py
index fc6e3929..b04b4ed5 100644
--- a/realtime/rt/rt.py
+++ b/realtime/rt/rt.py
@@ -10,11 +10,12 @@ import subprocess
 import tempfile
 import time
 
-import cdec.configobj
+from cdec.configobj import ConfigObj
 import cdec.sa
 
 import aligner
 import decoder
+import util
 
 class RealtimeDecoder:
 
@@ -32,21 +33,38 @@ class RealtimeDecoder:
         self.aligner = aligner.ForceAligner(fwd_params, fwd_err, rev_params, rev_err)
 
         # Grammar extractor
-        sa_config = os.path.join(configdir, 'sa.ini')
-        self.extractor = cdec.sa.GrammarExtractor(sa_config, online=True)
+        sa_config = ConfigObj(os.path.join(configdir, 'sa.ini'), unrepr=True)
+        sa_config.filename = os.path.join(self.tmp, 'sa.ini')
+        util.sa_ini_for_realtime(sa_config, os.path.abspath(configdir))
+        sa_config.write()
+        self.extractor = cdec.sa.GrammarExtractor(sa_config.filename, online=True)
         self.grammar_files = collections.deque()
         self.grammar_dict = {}
         self.cache_size = cache_size
 
+        # HPYPLM reference stream
+        ref_fifo_file = os.path.join(self.tmp, 'ref.fifo')
+        os.mkfifo(ref_fifo_file)
+        self.ref_fifo = open(ref_fifo_file, 'w+')
+        # Start with empty line (do not learn prior to first input)
+        self.ref_fifo.write('\n')
+        self.ref_fifo.flush()
+
         # Decoder
-        decoder_config = os.path.join(configdir, 'cdec.ini')
+        decoder_config = [[f.strip() for f in line.split('=')] for line in open(os.path.join(configdir, 'cdec.ini'))]
+        util.cdec_ini_for_realtime(decoder_config, os.path.abspath(configdir), ref_fifo_file)
+        decoder_config_file = os.path.join(self.tmp, 'cdec.ini')
+        with open(decoder_config_file, 'w') as output:
+            for (k, v) in decoder_config:
+                output.write('{}={}\n'.format(k, v))
         decoder_weights = os.path.join(configdir, 'weights.final')
-        self.decoder = decoder.MIRADecoder(decoder_config, decoder_weights)
+        self.decoder = decoder.MIRADecoder(decoder_config_file, decoder_weights)
 
     def close(self):
         logging.info('Closing processes')
         self.aligner.close()
         self.decoder.close()
+        self.ref_fifo.close()
         logging.info('Deleting {}'.format(self.tmp))
         shutil.rmtree(self.tmp)
 
@@ -75,6 +93,9 @@ class RealtimeDecoder:
         grammar_file = self.grammar(sentence)
         start_time = time.time()
         hyp = self.decoder.decode(sentence, grammar_file)
+        # Empty reference: HPYPLM does not learn prior to next translation
+        self.ref_fifo.write('\n')
+        self.ref_fifo.flush()
         stop_time = time.time()
         logging.info('Translation time: {} seconds'.format(stop_time - start_time))
         return hyp
@@ -91,4 +112,7 @@ class RealtimeDecoder:
         # Clear (old) cached grammar
         rm_grammar = self.grammar_dict.pop(source)
         os.remove(rm_grammar)
-        # TODO: Add to LM by writing to fifo
+        # Add to HPYPLM by writing to fifo (read on next translation)
+        logging.info('Adding to HPYPLM: {}'.format(target))
+        self.ref_fifo.write('{}\n'.format(target))
+        self.ref_fifo.flush()