summaryrefslogtreecommitdiff
path: root/decoder/scfg_translator.cc
diff options
context:
space:
mode:
authorAvneesh Saluja <asaluja@gmail.com>2013-03-28 18:28:16 -0700
committerAvneesh Saluja <asaluja@gmail.com>2013-03-28 18:28:16 -0700
commit3d8d656fa7911524e0e6885647173474524e0784 (patch)
tree81b1ee2fcb67980376d03f0aa48e42e53abff222 /decoder/scfg_translator.cc
parentbe7f57fdd484e063775d7abf083b9fa4c403b610 (diff)
parent96fedabebafe7a38a6d5928be8fff767e411d705 (diff)
fixed conflicts
Diffstat (limited to 'decoder/scfg_translator.cc')
-rw-r--r--decoder/scfg_translator.cc60
1 files changed, 60 insertions, 0 deletions
diff --git a/decoder/scfg_translator.cc b/decoder/scfg_translator.cc
index a978cfc2..3b43b586 100644
--- a/decoder/scfg_translator.cc
+++ b/decoder/scfg_translator.cc
@@ -15,13 +15,73 @@
#include "tdict.h"
#include "viterbi.h"
#include "verbose.h"
+#include <tr1/unordered_map>
#define foreach BOOST_FOREACH
#define reverse_foreach BOOST_REVERSE_FOREACH
using namespace std;
+using namespace std::tr1;
static bool printGrammarsUsed = false;
+struct GlueGrammar : public TextGrammar {
+ // read glue grammar from file
+ explicit GlueGrammar(const std::string& file);
+ GlueGrammar(const std::string& goal_nt, const std::string& default_nt, const unsigned int ctf_level=0); // "S", "X"
+ virtual bool HasRuleForSpan(int i, int j, int distance) const;
+};
+
+struct PassThroughGrammar : public TextGrammar {
+ PassThroughGrammar(const Lattice& input, const std::string& cat, const unsigned int ctf_level=0);
+ virtual bool HasRuleForSpan(int i, int j, int distance) const;
+};
+
+GlueGrammar::GlueGrammar(const string& file) : TextGrammar(file) {}
+
+static void RefineRule(TRulePtr pt, const unsigned int ctf_level){
+ for (unsigned int i=0; i<ctf_level; ++i){
+ TRulePtr r(new TRule(*pt));
+ pt->fine_rules_.reset(new vector<TRulePtr>);
+ pt->fine_rules_->push_back(r);
+ pt = r;
+ }
+}
+
+GlueGrammar::GlueGrammar(const string& goal_nt, const string& default_nt, const unsigned int ctf_level) {
+ TRulePtr stop_glue(new TRule("[" + goal_nt + "] ||| [" + default_nt + ",1] ||| [1]"));
+ AddRule(stop_glue);
+ RefineRule(stop_glue, ctf_level);
+ TRulePtr glue(new TRule("[" + goal_nt + "] ||| [" + goal_nt + ",1] ["+ default_nt + ",2] ||| [1] [2] ||| Glue=1"));
+ AddRule(glue);
+ RefineRule(glue, ctf_level);
+}
+
+bool GlueGrammar::HasRuleForSpan(int i, int /* j */, int /* distance */) const {
+ return (i == 0);
+}
+
+PassThroughGrammar::PassThroughGrammar(const Lattice& input, const string& cat, const unsigned int ctf_level) {
+ unordered_set<WordID> ss;
+ for (int i = 0; i < input.size(); ++i) {
+ const vector<LatticeArc>& alts = input[i];
+ for (int k = 0; k < alts.size(); ++k) {
+ const int j = alts[k].dist2next + i;
+ const string& src = TD::Convert(alts[k].label);
+ if (ss.count(alts[k].label) == 0) {
+ TRulePtr pt(new TRule("[" + cat + "] ||| " + src + " ||| " + src + " ||| PassThrough=1"));
+ pt->a_.push_back(AlignmentPoint(0,0));
+ AddRule(pt);
+ RefineRule(pt, ctf_level);
+ ss.insert(alts[k].label);
+ }
+ }
+ }
+}
+
+bool PassThroughGrammar::HasRuleForSpan(int, int, int distance) const {
+ return (distance < 2);
+}
+
struct SCFGTranslatorImpl {
SCFGTranslatorImpl(const boost::program_options::variables_map& conf) :
max_span_limit(conf["scfg_max_span_limit"].as<int>()),