From 8d084476522ff267a25b8faa58e48879b5d0cdb6 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@cs.cmu.edu>
Date: Sat, 10 Nov 2012 19:02:56 -0500
Subject: support multi-reference references

---
 mteval/ns_docscorer.cc | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

(limited to 'mteval/ns_docscorer.cc')

diff --git a/mteval/ns_docscorer.cc b/mteval/ns_docscorer.cc
index f72ad115..83bd1a29 100644
--- a/mteval/ns_docscorer.cc
+++ b/mteval/ns_docscorer.cc
@@ -11,11 +11,14 @@ using namespace std;
 
 DocumentScorer::~DocumentScorer() {}
 
+DocumentScorer::DocumentScorer() {}
+
 void DocumentScorer::Init(const EvaluationMetric* metric,
             const vector<string>& ref_files,
             const string& src_file,
             bool verbose) {
   scorers_.clear();
+  static const WordID kDIV = TD::Convert("|||");
   if (verbose) cerr << "Loading references (" << ref_files.size() << " files)\n";
   assert(src_file.empty());
   std::vector<ReadFile> ifs(ref_files.begin(),ref_files.end());
@@ -23,13 +26,14 @@ void DocumentScorer::Init(const EvaluationMetric* metric,
   char buf[64000];
   bool expect_eof = false;
   int line=0;
+  vector<WordID> tmp;
+  vector<vector<WordID> > refs;
   while (ifs[0].get()) {
-    vector<vector<WordID> > refs(ref_files.size());
+    refs.clear();
     for (int i=0; i < ref_files.size(); ++i) {
       istream &in=ifs[i].get();
       if (in.eof()) break;
       in.getline(buf, 64000);
-      refs[i].clear();
       if (strlen(buf) == 0) {
         if (in.eof()) {
           if (!expect_eof) {
@@ -38,9 +42,17 @@ void DocumentScorer::Init(const EvaluationMetric* metric,
           }
           break;
         }
-      } else {
-        TD::ConvertSentence(buf, &refs[i]);
-        assert(!refs[i].empty());
+      } else { // read a line from a reference file
+        tmp.clear();
+        TD::ConvertSentence(buf, &tmp);
+        unsigned last = 0;
+        for (unsigned j = 0; j < tmp.size(); ++j) {
+          if (tmp[j] == kDIV) {
+            refs.push_back(vector<WordID>(tmp.begin() + last, tmp.begin() + j));
+            last = j + 1;
+          }
+        }
+        refs.push_back(vector<WordID>(tmp.begin() + last, tmp.end()));
       }
       assert(!expect_eof);
     }
-- 
cgit v1.2.3