setup for hpyplm being optional, different metrics

2 files changed, 18 insertions, 8 deletions

diff --git a/realtime/mkconfig.py b/realtime/mkconfig.py
index f193c57a..71944080 100755
--- a/realtime/mkconfig.py
+++ b/realtime/mkconfig.py
@@ -12,11 +12,14 @@ from cdec.configobj import ConfigObj
 
 def main():
 
-    if len(sys.argv[1:]) != 12:
-        sys.stderr.write('usage: {} a.fwd_params a.fwd_err a.rev_params a.rev_err sa sa.ini mono.klm libcdec_ff_hpyplm.so corpus.hpyplm cdec.ini weights.final output.d\n'.format(sys.argv[0]))
+    if len(sys.argv[1:]) not in (10, 12):
+        sys.stderr.write('usage: {} a.fwd_params a.fwd_err a.rev_params a.rev_err sa sa.ini mono.klm cdec.ini weights.final output.d [libcdec_ff_hpyplm.so corpus.hpyplm]\n'.format(sys.argv[0]))
         sys.exit(2)
 
-    (a_fwd_params, a_fwd_err, a_rev_params, a_rev_err, sa, sa_ini, mono_klm, libcdec_ff_hpyplm_so, corpus_hpyplm, cdec_ini, weights_final, output_d) = sys.argv[1:]
+    (a_fwd_params, a_fwd_err, a_rev_params, a_rev_err, sa, sa_ini, mono_klm, cdec_ini, weights_final, output_d) = sys.argv[1:11]
+
+    # Optional
+    (libcdec_ff_hpyplm_so, corpus_hpyplm) = sys.argv[11:13] if len(sys.argv[1:]) == 12 else (None, None)
 
     if os.path.exists(output_d):
         sys.stderr.write('Directory {} exists, exiting.\n'.format(output_d))
@@ -40,8 +43,10 @@ def main():
 
     # language models
     shutil.copy(mono_klm, os.path.join(output_d, 'mono.klm'))
-    shutil.copy(libcdec_ff_hpyplm_so, os.path.join(output_d, 'libcdec_ff_hpyplm.so'))
-    shutil.copy(corpus_hpyplm, os.path.join(output_d, 'corpus.hpyplm'))
+    if libcdec_ff_hpyplm_so:
+        shutil.copy(libcdec_ff_hpyplm_so, os.path.join(output_d, 'libcdec_ff_hpyplm.so'))
+    if corpus_hpyplm:
+        shutil.copy(corpus_hpyplm, os.path.join(output_d, 'corpus.hpyplm'))
 
     # decoder config
     config = [[f.strip() for f in line.split('=')] for line in open(cdec_ini)]
@@ -52,6 +57,11 @@ def main():
 
     # weights
     shutil.copy(weights_final, os.path.join(output_d, 'weights.final'))
+
+    # other options
+    # TODO: automatically set some things here
+    with open(os.path.join(output_d, 'rt.ini'), 'w') as rt_ini:
+        pass
             
 if __name__ == '__main__':
     main()
diff --git a/realtime/rt/decoder.py b/realtime/rt/decoder.py
index ed45c248..5082911d 100644
--- a/realtime/rt/decoder.py
+++ b/realtime/rt/decoder.py
@@ -38,11 +38,11 @@ class CdecDecoder(Decoder):
 
 class MIRADecoder(Decoder):
 
-    def __init__(self, config, weights):
+    def __init__(self, config, weights, metric='bleu'):
         cdec_root = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
         mira = os.path.join(cdec_root, 'training', 'mira', 'kbest_cut_mira')
-        #                                              optimizer=2 step=0.001    best=500,    k=500,       uniq, stream
-        mira_cmd = [mira, '-c', config, '-w', weights, '-o', '2', '-C', '0.001', '-b', '500', '-k', '500', '-u', '-t']
+        #                                              optimizer=2 step=0.001    best=500,    k=500,       uniq, stream, metric
+        mira_cmd = [mira, '-c', config, '-w', weights, '-o', '2', '-C', '0.001', '-b', '500', '-k', '500', '-u', '-t', '-m', metric]
         logger.info('Executing: {}'.format(' '.join(mira_cmd)))
         self.decoder = util.popen_io(mira_cmd)
         self.lock = util.FIFOLock()