From 671c21451542e2dd20e45b4033d44d8e8735f87b Mon Sep 17 00:00:00 2001
From: Chris Dyer <redpony@gmail.com>
Date: Thu, 3 Dec 2009 16:33:55 -0500
Subject: initial check in

---
 src/ff_wordalign.h | 133 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 133 insertions(+)
 create mode 100644 src/ff_wordalign.h

(limited to 'src/ff_wordalign.h')

diff --git a/src/ff_wordalign.h b/src/ff_wordalign.h
new file mode 100644
index 00000000..1581641c
--- /dev/null
+++ b/src/ff_wordalign.h
@@ -0,0 +1,133 @@
+#ifndef _FF_WORD_ALIGN_H_
+#define _FF_WORD_ALIGN_H_
+
+#include "ff.h"
+#include "array2d.h"
+
+class RelativeSentencePosition : public FeatureFunction {
+ public:
+  RelativeSentencePosition(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* out_context) const;
+ private:
+  const int fid_;
+};
+
+class MarkovJump : public FeatureFunction {
+ public:
+  MarkovJump(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* out_context) const;
+ private:
+  const int fid_;
+  bool individual_params_per_jumpsize_;
+  bool condition_on_flen_;
+  std::string template_;
+};
+
+class AlignerResults : public FeatureFunction {
+ public:
+  AlignerResults(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* out_context) const;
+ private:
+  int fid_;
+  std::vector<boost::shared_ptr<Array2D<bool> > > is_aligned_;
+  mutable int cur_sent_;
+  const Array2D<bool> mutable* cur_grid_;
+};
+
+#include <tr1/unordered_map>
+#include <boost/functional/hash.hpp>
+#include <cassert>
+class BlunsomSynchronousParseHack : public FeatureFunction {
+ public:
+  BlunsomSynchronousParseHack(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* out_context) const;
+ private:
+  inline bool DoesNotBelong(const void* state) const {
+    for (int i = 0; i < NumBytesContext(); ++i) {
+      if (*(static_cast<const unsigned char*>(state) + i)) return false;
+    }
+    return true;
+  }
+
+  inline void AppendAntecedentString(const void* state, std::vector<WordID>* yield) const {
+    int i = 0;
+    int ind = 0;
+    while (i < NumBytesContext() && !(*(static_cast<const unsigned char*>(state) + i))) { ++i; ind += 8; }
+    // std::cerr << i << " " << NumBytesContext() << std::endl;
+    assert(i != NumBytesContext());
+    assert(ind < cur_ref_->size());
+    int cur = *(static_cast<const unsigned char*>(state) + i);
+    int comp = 1;
+    while (comp < 256 && (comp & cur) == 0) { comp <<= 1; ++ind; }
+    assert(ind < cur_ref_->size());
+    assert(comp < 256);
+    do {
+      assert(ind < cur_ref_->size());
+      yield->push_back((*cur_ref_)[ind]);
+      ++ind;
+      comp <<= 1;
+      if (comp == 256) {
+        comp = 1;
+        ++i;
+        cur = *(static_cast<const unsigned char*>(state) + i);
+      }
+    } while (comp & cur);
+  }
+
+  inline void SetStateMask(int start, int end, void* state) const {
+    assert((end / 8) < NumBytesContext());
+    int i = 0;
+    int comp = 1;
+    for (int j = 0; j < start; ++j) {
+      comp <<= 1;
+      if (comp == 256) {
+        ++i;
+        comp = 1;
+      }
+    }
+    //std::cerr << "SM: " << i << "\n";
+    for (int j = start; j < end; ++j) {
+      *(static_cast<unsigned char*>(state) + i) |= comp;
+      //std::cerr << "  " << comp << "\n";
+      comp <<= 1;
+      if (comp == 256) {
+        ++i;
+        comp = 1;
+      }
+    }
+    //std::cerr << "   MASK: " << ((int)*(static_cast<unsigned char*>(state))) << "\n";
+  }
+
+  const int fid_;
+  mutable int cur_sent_;
+  typedef std::tr1::unordered_map<std::vector<WordID>, int, boost::hash<std::vector<WordID> > > Vec2Int;
+  mutable Vec2Int cur_map_;
+  const std::vector<WordID> mutable * cur_ref_;
+  mutable std::vector<std::vector<WordID> > refs_;
+};
+
+#endif
-- 
cgit v1.2.3