few small fixes of alignment tools, add new orthographic similarity feature for word aligner, final naming of directories, libraries in cdec

1 files changed, 112 insertions, 0 deletions

diff --git a/decoder/lexcrf.cc b/decoder/lexcrf.cc
new file mode 100644
index 00000000..33455a3d
--- /dev/null
+++ b/decoder/lexcrf.cc
@@ -0,0 +1,112 @@
+#include "lexcrf.h"
+
+#include <iostream>
+
+#include "filelib.h"
+#include "hg.h"
+#include "tdict.h"
+#include "grammar.h"
+#include "sentence_metadata.h"
+
+using namespace std;
+
+struct LexicalCRFImpl {
+  LexicalCRFImpl(const boost::program_options::variables_map& conf) :
+      use_null(false),
+      kXCAT(TD::Convert("X")*-1),
+      kNULL(TD::Convert("<eps>")),
+      kBINARY(new TRule("[X] ||| [X,1] [X,2] ||| [1] [2]")),
+      kGOAL_RULE(new TRule("[Goal] ||| [X,1] ||| [1]")) {
+    vector<string> gfiles = conf["grammar"].as<vector<string> >();
+    assert(gfiles.size() == 1);
+    ReadFile rf(gfiles.front());
+    TextGrammar *tg = new TextGrammar;
+    grammar.reset(tg);
+    istream* in = rf.stream();
+    int lc = 0;
+    bool flag = false;
+    while(*in) {
+      string line;
+      getline(*in, line);
+      if (line.empty()) continue;
+      ++lc;
+      TRulePtr r(TRule::CreateRulePhrasetable(line));
+      tg->AddRule(r);
+      if (lc %   50000 == 0) { cerr << '.'; flag = true; }
+      if (lc % 2000000 == 0) { cerr << " [" << lc << "]\n"; flag = false; }
+    }
+    if (flag) cerr << endl;
+    cerr << "Loaded " << lc << " rules\n";
+  }
+
+  void BuildTrellis(const Lattice& lattice, const SentenceMetadata& smeta, Hypergraph* forest) {
+    const int e_len = smeta.GetTargetLength();
+    assert(e_len > 0);
+    const int f_len = lattice.size();
+    // hack to tell the feature function system how big the sentence pair is
+    const int f_start = (use_null ? -1 : 0);
+    int prev_node_id = -1;
+    for (int i = 0; i < e_len; ++i) {  // for each word in the *ref*
+      Hypergraph::Node* node = forest->AddNode(kXCAT);
+      const int new_node_id = node->id_;
+      for (int j = f_start; j < f_len; ++j) { // for each word in the source
+        const WordID src_sym = (j < 0 ? kNULL : lattice[j][0].label);
+        const GrammarIter* gi = grammar->GetRoot()->Extend(src_sym);
+        if (!gi) {
+          cerr << "No translations found for: " << TD::Convert(src_sym) << "\n";
+          abort();
+        }
+        const RuleBin* rb = gi->GetRules();
+        assert(rb);
+        for (int k = 0; k < rb->GetNumRules(); ++k) {
+          TRulePtr rule = rb->GetIthRule(k);
+          Hypergraph::Edge* edge = forest->AddEdge(rule, Hypergraph::TailNodeVector());
+          edge->i_ = j;
+          edge->j_ = j+1;
+          edge->prev_i_ = i;
+          edge->prev_j_ = i+1;
+          edge->feature_values_ += edge->rule_->GetFeatureValues();
+          forest->ConnectEdgeToHeadNode(edge->id_, new_node_id);
+        }
+      }
+      if (prev_node_id >= 0) {
+        const int comb_node_id = forest->AddNode(kXCAT)->id_;
+        Hypergraph::TailNodeVector tail(2, prev_node_id);
+        tail[1] = new_node_id;
+        const int edge_id = forest->AddEdge(kBINARY, tail)->id_;
+        forest->ConnectEdgeToHeadNode(edge_id, comb_node_id);
+        prev_node_id = comb_node_id;
+      } else {
+        prev_node_id = new_node_id;
+      }
+    }
+    Hypergraph::TailNodeVector tail(1, forest->nodes_.size() - 1);
+    Hypergraph::Node* goal = forest->AddNode(TD::Convert("[Goal]")*-1);
+    Hypergraph::Edge* hg_edge = forest->AddEdge(kGOAL_RULE, tail);
+    forest->ConnectEdgeToHeadNode(hg_edge, goal);
+  }
+
+ private:
+  const bool use_null;
+  const WordID kXCAT;
+  const WordID kNULL;
+  const TRulePtr kBINARY;
+  const TRulePtr kGOAL_RULE;
+  GrammarPtr grammar;
+};
+
+LexicalCRF::LexicalCRF(const boost::program_options::variables_map& conf) :
+  pimpl_(new LexicalCRFImpl(conf)) {}
+
+bool LexicalCRF::Translate(const string& input,
+                      SentenceMetadata* smeta,
+                      const vector<double>& weights,
+                      Hypergraph* forest) {
+  Lattice lattice;
+  LatticeTools::ConvertTextToLattice(input, &lattice);
+  smeta->SetSourceLength(lattice.size());
+  pimpl_->BuildTrellis(lattice, *smeta, forest);
+  forest->Reweight(weights);
+  return true;
+}
+