rename aligner, add support for distinguishing translation / transliteration

8 files changed, 80 insertions, 33 deletions

diff --git a/decoder/aligner.cc b/decoder/aligner.cc
index 53e059fb..232e022a 100644
--- a/decoder/aligner.cc
+++ b/decoder/aligner.cc
@@ -11,7 +11,7 @@
 #include "sentence_metadata.h"
 #include "inside_outside.h"
 #include "viterbi.h"
-#include "alignment_pharaoh.h"
+#include "alignment_io.h"
 
 using namespace std;
 
@@ -300,7 +300,7 @@ void AlignerTools::WriteAlignment(const Lattice& src_lattice,
       cerr << grid << endl;
     }
     (*out) << TD::GetString(src_sent) << " ||| " << TD::GetString(trg_sent) << " ||| ";
-    AlignmentPharaoh::SerializePharaohFormat(grid, out);
+    AlignmentIO::SerializePharaohFormat(grid, out);
   }
 };
 
diff --git a/decoder/ff_wordalign.cc b/decoder/ff_wordalign.cc
index 9e7c618e..decdf9bc 100644
--- a/decoder/ff_wordalign.cc
+++ b/decoder/ff_wordalign.cc
@@ -15,7 +15,6 @@
 
 #include "factored_lexicon_helper.h"
 #include "verbose.h"
-#include "alignment_pharaoh.h"
 #include "stringlib.h"
 #include "sentence_metadata.h"
 #include "hg.h"
diff --git a/mteval/aer_scorer.cc b/mteval/aer_scorer.cc
index edd4390f..ae3192d4 100644
--- a/mteval/aer_scorer.cc
+++ b/mteval/aer_scorer.cc
@@ -5,7 +5,7 @@
 #include <sstream>
 
 #include "tdict.h"
-#include "alignment_pharaoh.h"
+#include "alignment_io.h"
 
 using namespace std;
 
@@ -85,7 +85,7 @@ AERScorer::AERScorer(const vector<vector<WordID> >& refs, const string& src) : s
     cerr << "AERScorer can only take a single reference!\n";
     abort();
   }
-  ref_ = AlignmentPharaoh::ReadPharaohAlignmentGrid(TD::GetString(refs.front()));
+  ref_ = AlignmentIO::ReadPharaohAlignmentGrid(TD::GetString(refs.front()));
 }
 
 static inline bool Safe(const Array2D<bool>& a, int i, int j) {
@@ -101,7 +101,7 @@ ScoreP AERScorer::ScoreCCandidate(const vector<WordID>& shyp) const {
 
 ScoreP AERScorer::ScoreCandidate(const vector<WordID>& shyp) const {
   boost::shared_ptr<Array2D<bool> > hyp =
-    AlignmentPharaoh::ReadPharaohAlignmentGrid(TD::GetString(shyp));
+    AlignmentIO::ReadPharaohAlignmentGrid(TD::GetString(shyp));
 
   int m = 0;
   int r = 0;
diff --git a/utils/Makefile.am b/utils/Makefile.am
index 3ea21835..2fc6ae21 100644
--- a/utils/Makefile.am
+++ b/utils/Makefile.am
@@ -23,7 +23,7 @@ atools_SOURCES = atools.cc
 noinst_LIBRARIES = libutils.a
 
 libutils_a_SOURCES = \
-  alignment_pharaoh.cc \
+  alignment_io.cc \
   b64tools.cc \
   corpus_tools.cc \
   dict.cc \
diff --git a/utils/alignment_pharaoh.cc b/utils/alignment_io.cc
index 890ff565..1d923f7f 100644
--- a/utils/alignment_pharaoh.cc
+++ b/utils/alignment_io.cc
@@ -1,12 +1,10 @@
-#include "utils/alignment_pharaoh.h"
-
-#include <set>
+#include "utils/alignment_io.h"
 
 using namespace std;
 
 static bool is_digit(char x) { return x >= '0' && x <= '9'; }
 
-boost::shared_ptr<Array2D<bool> > AlignmentPharaoh::ReadPharaohAlignmentGrid(const string& al) {
+boost::shared_ptr<Array2D<bool> > AlignmentIO::ReadPharaohAlignmentGrid(const string& al) {
   int max_x = 0;
   int max_y = 0;
   int i = 0;
@@ -64,14 +62,36 @@ boost::shared_ptr<Array2D<bool> > AlignmentPharaoh::ReadPharaohAlignmentGrid(con
   return grid;
 }
 
-void AlignmentPharaoh::SerializePharaohFormat(const Array2D<bool>& alignment, ostream* out) {
+void AlignmentIO::SerializePharaohFormat(const Array2D<bool>& alignment, ostream* o) {
+  ostream& out = *o;
   bool need_space = false;
   for (int i = 0; i < alignment.width(); ++i)
     for (int j = 0; j < alignment.height(); ++j)
       if (alignment(i,j)) {
-        if (need_space) (*out) << ' '; else need_space = true;
-        (*out) << i << '-' << j;
+        if (need_space) out << ' '; else need_space = true;
+        out << i << '-' << j;
       }
-  (*out) << endl;
+  out << endl;
+}
+
+void AlignmentIO::SerializeTypedAlignment(const Array2D<AlignmentType>& alignment, ostream* o) {
+  ostream& out = *o;
+  bool need_space = false;
+  for (int i = 0; i < alignment.width(); ++i)
+    for (int j = 0; j < alignment.height(); ++j) {
+      const AlignmentType& aij = alignment(i,j);
+      if (aij != kNONE) {
+        if (need_space) out << ' '; else need_space = true;
+        if (aij == kTRANSLATION) {}
+        else if (aij == kTRANSLITERATION) {
+          out << 'T' << ':';
+        } else {
+          cerr << "\nUnexpected alignment point type: " << static_cast<int>(aij) << endl;
+          abort();
+        }
+        out << i << '-' << j;
+      }
+    }
+  out << endl;
 }
 
diff --git a/utils/alignment_io.h b/utils/alignment_io.h
new file mode 100644
index 00000000..36bcecd7
--- /dev/null
+++ b/utils/alignment_io.h
@@ -0,0 +1,42 @@
+#ifndef _ALIGNMENT_IO_H_
+#define _ALIGNMENT_IO_H_
+
+#include <string>
+#include <iostream>
+#include <boost/shared_ptr.hpp>
+#include "array2d.h"
+
+struct AlignmentIO {
+  enum AlignmentType { kNONE = 0, kTRANSLATION = 1, kTRANSLITERATION = 2 };
+
+  static boost::shared_ptr<Array2D<bool> > ReadPharaohAlignmentGrid(const std::string& al);
+  static void SerializePharaohFormat(const Array2D<bool>& alignment, std::ostream* out);
+  static void SerializeTypedAlignment(const Array2D<AlignmentType>& alignment, std::ostream* out);
+};
+
+inline std::ostream& operator<<(std::ostream& os, const Array2D<AlignmentIO::AlignmentType>& m) {
+  os << ' ';
+  for (int j=0; j<m.height(); ++j)
+    os << (j%10);
+  os << "\n";
+  for (int i=0; i<m.width(); ++i) {
+    os << (i%10);
+    for (int j=0; j<m.height(); ++j) {
+      switch (m(i,j)) {
+        case AlignmentIO::kNONE:            os << '.'; break;
+        case AlignmentIO::kTRANSLATION:     os << '*'; break;
+        case AlignmentIO::kTRANSLITERATION: os << '#'; break;
+        default:                            os << '?'; break;
+      }
+    }
+    os << (i%10) << "\n";
+  }
+  os << ' ';
+  for (int j=0; j<m.height(); ++j)
+    os << (j%10);
+  os << "\n";
+  return os;
+}
+
+
+#endif
diff --git a/utils/alignment_pharaoh.h b/utils/alignment_pharaoh.h
deleted file mode 100644
index d111c8bf..00000000
--- a/utils/alignment_pharaoh.h
+++ /dev/null
@@ -1,14 +0,0 @@
-#ifndef _PHARAOH_ALIGNMENT_H_
-#define _PHARAOH_ALIGNMENT_H_
-
-#include <string>
-#include <iostream>
-#include <boost/shared_ptr.hpp>
-#include "array2d.h"
-
-struct AlignmentPharaoh {
-  static boost::shared_ptr<Array2D<bool> > ReadPharaohAlignmentGrid(const std::string& al);
-  static void SerializePharaohFormat(const Array2D<bool>& alignment, std::ostream* out);
-};
-
-#endif
diff --git a/utils/atools.cc b/utils/atools.cc
index ba56dd6c..bce7822e 100644
--- a/utils/atools.cc
+++ b/utils/atools.cc
@@ -8,7 +8,7 @@
 #include <boost/shared_ptr.hpp>
 
 #include "filelib.h"
-#include "alignment_pharaoh.h"
+#include "alignment_io.h"
 
 namespace po = boost::program_options;
 using namespace std;
@@ -348,9 +348,9 @@ int main(int argc, char **argv) {
     }
     if (line1.empty() && !*in1) break;
     boost::shared_ptr<Array2D<bool> > out(new Array2D<bool>);
-    boost::shared_ptr<Array2D<bool> > a1 = AlignmentPharaoh::ReadPharaohAlignmentGrid(line1);
+    boost::shared_ptr<Array2D<bool> > a1 = AlignmentIO::ReadPharaohAlignmentGrid(line1);
     if (in2) {
-      boost::shared_ptr<Array2D<bool> > a2 = AlignmentPharaoh::ReadPharaohAlignmentGrid(line2);
+      boost::shared_ptr<Array2D<bool> > a2 = AlignmentIO::ReadPharaohAlignmentGrid(line2);
       cmd.Apply(*a1, *a2, out.get());
     } else {
       Array2D<bool> dummy;
@@ -358,7 +358,7 @@ int main(int argc, char **argv) {
     }
     
     if (cmd.Result() == 1) {
-      AlignmentPharaoh::SerializePharaohFormat(*out, &cout);
+      AlignmentIO::SerializePharaohFormat(*out, &cout);
     }
   }
   if (cmd.Result() == 2)