rudimentary support for meteor via an external process. still needs configuration of path, but it should work

5 files changed, 85 insertions, 38 deletions

diff --git a/mteval/external_scorer.cc b/mteval/external_scorer.cc
index 4327ce9b..3757064b 100644
--- a/mteval/external_scorer.cc
+++ b/mteval/external_scorer.cc
@@ -2,20 +2,42 @@
 
 #include <cstdio> // popen
 #include <cstdlib>
+#include <cstring>
 #include <sstream>
 #include <iostream>
 #include <cassert>
 
+#include "stringlib.h"
 #include "tdict.h"
 
 using namespace std;
 
+map<string, boost::shared_ptr<ScoreServer> > ScoreServerManager::servers_;
+
+class METEORServer : public ScoreServer {
+ public:
+  METEORServer() : ScoreServer("java -Xmx1024m -jar /Users/cdyer/software/meteor/meteor-1.3.jar - - -mira -lower -t tune -l en") {}
+};
+
+ScoreServer* ScoreServerManager::Instance(const string& score_type) {
+  boost::shared_ptr<ScoreServer>& s = servers_[score_type];
+  if (!s) {
+    if (score_type == "meteor") {
+      s.reset(new METEORServer);
+    } else {
+      cerr << "Don't know how to create score server for type '" << score_type << "'\n";
+      abort();
+    }
+  }
+  return s.get();
+}
+
 ScoreServer::ScoreServer(const string& cmd) : pipe_() {
   cerr << "Invoking " << cmd << " ..." << endl;
   pipe_ = popen(cmd.c_str(), "r+");
-  assert(pipe_);
+  if (!pipe_) { perror("popen"); abort(); }
   string dummy;
-  RequestResponse("EVAL ||| Reference initialization string . ||| Testing initialization string .\n", &dummy);
+  RequestResponse("SCORE ||| Reference initialization string . ||| Testing initialization string .", &dummy);
   assert(dummy.size() > 0);
   cerr << "Connection established.\n";
 }
@@ -24,12 +46,11 @@ ScoreServer::~ScoreServer() {
   pclose(pipe_);
 }
 
-double ScoreServer::ComputeScore(const vector<float>& fields) {
+float ScoreServer::ComputeScore(const vector<float>& fields) {
   ostringstream os;
-  os << "EVAL";
+  os << "EVAL |||";
   for (unsigned i = 0; i < fields.size(); ++i)
     os << ' ' << fields[i];
-  os << endl;
   string sres;
   RequestResponse(os.str(), &sres);
   return strtod(sres.c_str(), NULL);
@@ -48,46 +69,42 @@ void ScoreServer::Evaluate(const vector<vector<WordID> >& refs, const vector<Wor
   for (unsigned i = 0; i < hyp.size(); ++i) {
     os << ' ' << TD::Convert(hyp[i]);
   }
-  os << endl;
   string sres;
   RequestResponse(os.str(), &sres);
   istringstream is(sres);
-  double val;
+  float val;
   fields->clear();
-  while(is >> val) {
+  while(is >> val)
     fields->push_back(val);
-  }
 }
 
 #define MAX_BUF 16000
 
 void ScoreServer::RequestResponse(const string& request, string* response) {
-  fprintf(pipe_, "%s", request.c_str());
+  //cerr << "@SERVER: " << request << endl;
+  fputs(request.c_str(), pipe_);
+  fputc('\n', pipe_);
   fflush(pipe_);
   char buf[MAX_BUF];
-  size_t cr = fread(buf, 1, MAX_BUF, pipe_);
-  if (cr == 0) {
+  if (NULL == fgets(buf, MAX_BUF, pipe_)) {
     cerr << "Read error. Request: " << request << endl;
     abort();
   }
-  while (buf[cr-1] != '\n') {
-    size_t n = fread(&buf[cr], 1, MAX_BUF-cr, pipe_);
-    assert(n > 0);
-    cr += n;
-    assert(cr < MAX_BUF);
+  size_t len = strlen(buf);
+  if (len < 2) {
+    cerr << "Malformed response: " << buf << endl;
   }
-  buf[cr - 1] = 0;
-  *response = buf;
+  *response = Trim(buf, " \t\n");
+  //cerr << "@RESPONSE: '" << *response << "'\n";
 }
 
 struct ExternalScore : public ScoreBase<ExternalScore> {
   ExternalScore() : score_server() {}
-  explicit ExternalScore(const ScoreServer* s) : score_server(s), fields() {}
-  ExternalScore(const ScoreServer* s, const vector<float>& f) : score_server(s), fields(f) {}
+  explicit ExternalScore(ScoreServer* s) : score_server(s), fields() {}
+  ExternalScore(ScoreServer* s, const vector<float>& f) : score_server(s), fields(f) {}
   float ComputePartialScore() const { return 0.0;}
   float ComputeScore() const {
-    // TODO make EVAL call
-    assert(!"not implemented");
+    return score_server->ComputeScore(fields);
   }
   void ScoreDetails(string* details) const {
     ostringstream os;
@@ -127,14 +144,17 @@ struct ExternalScore : public ScoreBase<ExternalScore> {
   }
   void Encode(string* out) const {
     ostringstream os;
+    for (unsigned i = 0; i < fields.size(); ++i)
+      os << (i == 0 ? "" : " ") << fields[i];
+    *out = os.str();
   }
   bool IsAdditiveIdentity() const {
-    for (int i = 0; i < fields.size(); ++i)
+    for (unsigned i = 0; i < fields.size(); ++i)
       if (fields[i]) return false;
     return true;
   }
 
-  const ScoreServer* score_server;
+  mutable ScoreServer* score_server;
   vector<float> fields;
 };
 
@@ -148,3 +168,12 @@ ScoreP ExternalSentenceScorer::ScoreCCandidate(const Sentence& hyp) const {
   assert(!"not implemented");
 }
 
+ScoreP ExternalSentenceScorer::ScoreFromString(ScoreServer* s, const string& data) {
+  istringstream is(data);
+  vector<float> fields;
+  float val;
+  while(is >> val)
+    fields.push_back(val);
+  return ScoreP(new ExternalScore(s, fields));
+}
+
diff --git a/mteval/external_scorer.h b/mteval/external_scorer.h
index a2c91960..59ece269 100644
--- a/mteval/external_scorer.h
+++ b/mteval/external_scorer.h
@@ -3,15 +3,20 @@
 
 #include <vector>
 #include <cstdio>
+#include <string>
+#include <map>
+#include <boost/shared_ptr.hpp>
 
 #include "scorer.h"
 
 class ScoreServer {
- public:
+  friend class ScoreServerManager;
+ protected:
   explicit ScoreServer(const std::string& cmd);
   virtual ~ScoreServer();
 
-  double ComputeScore(const std::vector<float>& fields);
+ public:
+  float ComputeScore(const std::vector<float>& fields);
   void Evaluate(const std::vector<std::vector<WordID> >& refs, const std::vector<WordID>& hyp, std::vector<float>* fields);
 
  private:
@@ -19,17 +24,22 @@ class ScoreServer {
   FILE* pipe_;
 };
 
+struct ScoreServerManager {
+  static ScoreServer* Instance(const std::string& score_type);
+ private:
+  static std::map<std::string, boost::shared_ptr<ScoreServer> > servers_;
+};
+
 class ExternalSentenceScorer : public SentenceScorer {
  public:
-  virtual ScoreP ScoreCandidate(const Sentence& hyp) const = 0;
-  virtual ScoreP ScoreCCandidate(const Sentence& hyp) const =0;
+  ExternalSentenceScorer(ScoreServer* server, const std::vector<std::vector<WordID> >& r) :
+    SentenceScorer("External", r), eval_server(server) {}
+  virtual ScoreP ScoreCandidate(const Sentence& hyp) const;
+  virtual ScoreP ScoreCCandidate(const Sentence& hyp) const;
+  static ScoreP ScoreFromString(ScoreServer* s, const std::string& data);
+
  protected:
   ScoreServer* eval_server;
 };
 
-class METEORServer : public ScoreServer {
- public:
-  METEORServer() : ScoreServer("java -Xmx1024m -jar meteor-1.3.jar - - -mira -lower") {}
-};
-
 #endif
diff --git a/mteval/scorer.cc b/mteval/scorer.cc
index 64ce63af..2daa0daa 100644
--- a/mteval/scorer.cc
+++ b/mteval/scorer.cc
@@ -17,11 +17,12 @@
 #include "comb_scorer.h"
 #include "tdict.h"
 #include "stringlib.h"
+#include "external_scorer.h"
 
 using boost::shared_ptr;
 using namespace std;
 
-void Score::TimesEquals(float scale) {
+void Score::TimesEquals(float /*scale*/) {
   cerr<<"UNIMPLEMENTED except for BLEU (for MIRA): Score::TimesEquals"<<endl;abort();
 }
 
@@ -43,12 +44,14 @@ ScoreType ScoreTypeFromString(const string& st) {
     return Koehn_BLEU;
   if (sl == "combi")
     return BLEU_minus_TER_over_2;
+  if (sl == "meteor")
+    return METEOR;
   cerr << "Don't understand score type '" << st << "', defaulting to ibm_bleu.\n";
   return IBM_BLEU;
 }
 
 static char const* score_names[]={
-  "IBM_BLEU", "NIST_BLEU", "Koehn_BLEU", "TER", "BLEU_minus_TER_over_2", "SER", "AER", "IBM_BLEU_3"
+  "IBM_BLEU", "NIST_BLEU", "Koehn_BLEU", "TER", "BLEU_minus_TER_over_2", "SER", "AER", "IBM_BLEU_3", "METEOR"
 };
 
 std::string StringFromScoreType(ScoreType st) {
@@ -356,6 +359,7 @@ ScorerP SentenceScorer::CreateSentenceScorer(const ScoreType type,
     case TER: r = new TERScorer(refs);break;
     case SER: r = new SERScorer(refs);break;
     case BLEU_minus_TER_over_2: r = new BLEUTERCombinationScorer(refs);break;
+    case METEOR: r = new ExternalSentenceScorer(ScoreServerManager::Instance("meteor"), refs); break;
     default:
       assert(!"Not implemented!");
   }
@@ -398,6 +402,8 @@ ScoreP SentenceScorer::CreateScoreFromString(const ScoreType type, const string&
       return SERScorer::ScoreFromString(in);
     case BLEU_minus_TER_over_2:
       return BLEUTERCombinationScorer::ScoreFromString(in);
+    case METEOR:
+      return ExternalSentenceScorer::ScoreFromString(ScoreServerManager::Instance("meteor"), in);
     default:
       assert(!"Not implemented!");
   }
diff --git a/mteval/scorer.h b/mteval/scorer.h
index f18c8c7f..8bd76982 100644
--- a/mteval/scorer.h
+++ b/mteval/scorer.h
@@ -17,7 +17,7 @@ class ErrorSurface;
 class Hypergraph;  // needed for alignment
 
 //TODO: BLEU N (N separate arg, not part of enum)?
-enum ScoreType { IBM_BLEU, NIST_BLEU, Koehn_BLEU, TER, BLEU_minus_TER_over_2, SER, AER, IBM_BLEU_3 };
+enum ScoreType { IBM_BLEU, NIST_BLEU, Koehn_BLEU, TER, BLEU_minus_TER_over_2, SER, AER, IBM_BLEU_3, METEOR };
 ScoreType ScoreTypeFromString(const std::string& st);
 std::string StringFromScoreType(ScoreType st);
 
@@ -66,7 +66,7 @@ class SentenceScorer {
   typedef std::vector<Sentence> Sentences;
   std::string desc;
   Sentences refs;
-  SentenceScorer(std::string desc="SentenceScorer_unknown", Sentences const& refs=Sentences()) : desc(desc),refs(refs) {  }
+  explicit SentenceScorer(std::string desc="SentenceScorer_unknown", Sentences const& refs=Sentences()) : desc(desc),refs(refs) {  }
   std::string verbose_desc() const;
   virtual float ComputeRefLength(const Sentence& hyp) const; // default: avg of refs.length
   virtual ~SentenceScorer();
diff --git a/vest/dist-vest.pl b/vest/dist-vest.pl
index 789b5b14..b7a862c4 100755
--- a/vest/dist-vest.pl
+++ b/vest/dist-vest.pl
@@ -118,6 +118,8 @@ if ($usefork) { $usefork = "--use-fork"; } else { $usefork = ''; }
 
 if ($metric =~ /^(combi|ter)$/i) {
   $lines_per_mapper = 40;
+} elsif ($metric =~ /^meteor$/i) {
+  $lines_per_mapper = 2000;   # start up time is really high
 }
 
 ($iniFile) = @ARGV;