From 7120911494711094c6dcd9dc0da741d686207aab Mon Sep 17 00:00:00 2001
From: Patrick Simianer <simianer@cl.uni-heidelberg.de>
Date: Fri, 25 Apr 2014 18:40:18 +0200
Subject: modifications

---
 src/functionalizer.py      | 31 +++++++++++++++++++++++++++++++
 src/moses.py               | 38 +++++++++++++++++++++++++++++++++++---
 src/smt_semparse_config.py |  4 ++--
 3 files changed, 68 insertions(+), 5 deletions(-)

(limited to 'src')

diff --git a/src/functionalizer.py b/src/functionalizer.py
index 66325a0..782b4e5 100644
--- a/src/functionalizer.py
+++ b/src/functionalizer.py
@@ -37,6 +37,37 @@ class Functionalizer:
           break
       counter += 1
 
+  def run_sentence(self, experiment_dir, temp_dir):
+    hyp_file = open('%s/nbest.tmp' % temp_dir, 'r')
+
+    hypsets = []
+    hypset = []
+    last_eid = 0
+    for line in hyp_file:
+      parts = line.split('|||')
+      eid = int(parts[0])
+      if eid != last_eid:
+        hypsets.append(hypset)
+        hypset = []
+        last_eid = eid
+      score = parts[2] + ' ||| ' + parts[3].strip()
+      hyp = parts[1].strip()
+      hypset.append((hyp,score))
+    hypsets.append(hypset)
+    hyp_file.close()
+
+    counter = 0
+    for hypset in hypsets:
+      hypset = list(reversed(hypset))
+      while hypset:
+        hyp, score = hypset.pop()
+        fun = self.functionalize(hyp)
+        if fun:
+          return fun
+          break
+      counter += 1
+    return ""
+
   #xc = 0
   def functionalize(self, mrl):
 
diff --git a/src/moses.py b/src/moses.py
index 857ddbf..9a159c3 100644
--- a/src/moses.py
+++ b/src/moses.py
@@ -3,6 +3,8 @@ import os
 import subprocess
 import gzip
 
+from subprocess import Popen, PIPE, STDOUT
+
 class Moses:
 
   def __init__(self, config):
@@ -17,7 +19,8 @@ class Moses:
             '--e', self.config.tgt,
             '--lm', '0:3:%s/%s.arpa' % (self.config.experiment_dir, self.config.tgt),
             #'-score-options', "'--OnlyDirect --NoPhraseCount'"
-            '--alignment', self.config.symm]
+            '--alignment', self.config.symm,
+			'-external-bin-dir', self.config.giza]
     if self.config.model == 'hier':
       args += ['-hierarchical', '-glue-grammar']
 
@@ -99,7 +102,7 @@ class Moses:
     else:
       args += [self.config.moses_decode_phrase]
     args += ['%s/model/moses.ini' % self.config.experiment_dir,
-             '--mertdir', '%s/dist/bin' % self.config.moses]
+             '--mertdir', '%s/bin' % self.config.moses]
     if self.config.model == 'hier':
       args += ['--filtercmd', 
                '%s/scripts/training/filter-model-given-input.pl --Hierarchical'\
@@ -119,7 +122,7 @@ class Moses:
     else:
       assert False
 
-    if self.config.run == 'test':
+    if self.config.run == 'test' or self.config.run == 'all':
       args += ['-f', '%s/mert-work/moses.ini' % self.config.experiment_dir]
     else:
       args += ['-f', '%s/model/moses.ini' % self.config.experiment_dir]
@@ -139,3 +142,32 @@ class Moses:
     infile.close()
     log.close()
     outfile.close()
+
+  def decode_sentence(self, experiment_dir, sentence, temp_dir):
+    if self.config.model == 'phrase':
+      args = [self.config.moses_decode_phrase]
+    elif self.config.model == 'hier':
+      args = [self.config.moses_decode_hier]
+    else:
+      assert False
+
+    if self.config.run == 'test' or self.config.run == 'all':
+      args += ['-f', '%s/mert-work/moses.ini' % experiment_dir]
+    else:
+      args += ['-f', '%s/model/moses.ini' % experiment_dir]
+
+    args += ['-drop-unknown',
+             '-n-best-list', '%s/nbest.tmp' % temp_dir,
+                             str(self.config.nbest), 'distinct',
+             '-threads', '1']
+
+    infile = open('%s/sent.tmp' % temp_dir, 'w')
+    print >>infile, sentence
+    infile.close
+    infile = open('%s/sent.tmp' % temp_dir, 'r')
+    nullfile = open(os.devnull, 'w')
+    p = subprocess.Popen(args, stdin=infile, stdout=nullfile, stderr=nullfile)
+    p.wait()
+    infile.close()
+    return
+
diff --git a/src/smt_semparse_config.py b/src/smt_semparse_config.py
index 71eaf24..6bf50d7 100644
--- a/src/smt_semparse_config.py
+++ b/src/smt_semparse_config.py
@@ -17,8 +17,8 @@ class SMTSemparseConfig(Config):
 
     self.put('moses_train', '%s/scripts/training/train-model.perl' % self.moses)
     self.put('moses_tune', '%s/scripts/training/mert-moses.pl' % self.moses)
-    self.put('moses_decode_phrase', '%s/dist/bin/moses' % self.moses)
-    self.put('moses_decode_hier', '%s/dist/bin/moses_chart' % self.moses)
+    self.put('moses_decode_phrase', '%s/bin/moses' % self.moses)
+    self.put('moses_decode_hier', '%s/bin/moses_chart' % self.moses)
     self.put('bleu_eval', '%s/scripts/generic/multi-bleu.perl' % self.moses)
 
     self.put('wasp_eval', '%s/data/geo-funql/eval/eval.pl' % self.wasp)
-- 
cgit v1.2.3