Merge pull request #59 from kho/mrmira

Mr.MIRA compatibility

3 files changed, 69 insertions, 1 deletions

diff --git a/utils/Makefile.am b/utils/Makefile.am
index 727fa8a5..64f6d433 100644
--- a/utils/Makefile.am
+++ b/utils/Makefile.am
@@ -22,6 +22,7 @@ libutils_a_SOURCES = \
   alias_sampler.h \
   alignment_io.h \
   array2d.h \
+  b64featvector.h \
   b64tools.h \
   batched_append.h \
   city.h \
@@ -70,6 +71,7 @@ libutils_a_SOURCES = \
   fast_lexical_cast.hpp \
   intrusive_refcount.hpp \
   alignment_io.cc \
+  b64featvector.cc \
   b64tools.cc \
   corpus_tools.cc \
   dict.cc \
@@ -117,4 +119,3 @@ stringlib_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_U
 # do NOT NOT NOT add any other -I includes NO NO NO NO NO ######
 AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -I. -I$(top_srcdir) -DTEST_DATA=\"$(top_srcdir)/utils/test_data\"
 ################################################################
-
diff --git a/utils/b64featvector.cc b/utils/b64featvector.cc
new file mode 100644
index 00000000..c7d08b29
--- /dev/null
+++ b/utils/b64featvector.cc
@@ -0,0 +1,55 @@
+#include "b64featvector.h"
+
+#include <sstream>
+#include <boost/scoped_array.hpp>
+#include "b64tools.h"
+#include "fdict.h"
+
+using namespace std;
+
+static inline void EncodeFeatureWeight(const string &featname, weight_t weight,
+                                       ostream *output) {
+  output->write(featname.data(), featname.size() + 1);
+  output->write(reinterpret_cast<char *>(&weight), sizeof(weight_t));
+}
+
+string EncodeFeatureVector(const SparseVector<weight_t> &vec) {
+  string b64;
+  {
+    ostringstream base64_strm;
+    {
+      ostringstream strm;
+      for (SparseVector<weight_t>::const_iterator it = vec.begin();
+           it != vec.end(); ++it)
+        if (it->second != 0)
+          EncodeFeatureWeight(FD::Convert(it->first), it->second, &strm);
+      string data(strm.str());
+      B64::b64encode(data.data(), data.size(), &base64_strm);
+    }
+    b64 = base64_strm.str();
+  }
+  return b64;
+}
+
+void DecodeFeatureVector(const string &data, SparseVector<weight_t> *vec) {
+  vec->clear();
+  if (data.empty()) return;
+  // Decode data
+  size_t b64_len = data.size(), len = b64_len / 4 * 3;
+  boost::scoped_array<char> buf(new char[len]);
+  bool res =
+      B64::b64decode(reinterpret_cast<const unsigned char *>(data.data()),
+                     b64_len, buf.get(), len);
+  assert(res);
+  // Apply updates
+  size_t cur = 0;
+  while (cur < len) {
+    string feat_name(buf.get() + cur);
+    if (feat_name.empty()) break;  // Encountered trailing \0
+    int feat_id = FD::Convert(feat_name);
+    weight_t feat_delta =
+        *reinterpret_cast<weight_t *>(buf.get() + cur + feat_name.size() + 1);
+    (*vec)[feat_id] = feat_delta;
+    cur += feat_name.size() + 1 + sizeof(weight_t);
+  }
+}
diff --git a/utils/b64featvector.h b/utils/b64featvector.h
new file mode 100644
index 00000000..6ac04d44
--- /dev/null
+++ b/utils/b64featvector.h
@@ -0,0 +1,12 @@
+#ifndef _B64FEATVECTOR_H_
+#define _B64FEATVECTOR_H_
+
+#include <string>
+
+#include "sparse_vector.h"
+#include "weights.h"
+
+std::string EncodeFeatureVector(const SparseVector<weight_t> &);
+void DecodeFeatureVector(const std::string &, SparseVector<weight_t> *);
+
+#endif  // _B64FEATVECTOR_H_