From 63b30ed9c8510da8c8e2f6a456576424fddacc0e Mon Sep 17 00:00:00 2001
From: Paul Baltescu <pauldb89@gmail.com>
Date: Thu, 14 Feb 2013 23:17:15 +0000
Subject: Working version of the grammar extractor.

---
 extractor/rule_extractor_helper.h | 82 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
 create mode 100644 extractor/rule_extractor_helper.h

(limited to 'extractor/rule_extractor_helper.h')

diff --git a/extractor/rule_extractor_helper.h b/extractor/rule_extractor_helper.h
new file mode 100644
index 00000000..3478bfc8
--- /dev/null
+++ b/extractor/rule_extractor_helper.h
@@ -0,0 +1,82 @@
+#ifndef _RULE_EXTRACTOR_HELPER_H_
+#define _RULE_EXTRACTOR_HELPER_H_
+
+#include <memory>
+#include <unordered_map>
+#include <vector>
+
+using namespace std;
+
+class Alignment;
+class DataArray;
+
+class RuleExtractorHelper {
+ public:
+  RuleExtractorHelper(shared_ptr<DataArray> source_data_array,
+                      shared_ptr<DataArray> target_data_array,
+                      shared_ptr<Alignment> alignment,
+                      int max_rule_span,
+                      int max_rule_symbols,
+                      bool require_aligned_terminal,
+                      bool require_aligned_chunks,
+                      bool require_tight_phrases);
+
+  virtual ~RuleExtractorHelper();
+
+  virtual void GetLinksSpans(vector<int>& source_low, vector<int>& source_high,
+                             vector<int>& target_low, vector<int>& target_high,
+                             int sentence_id) const;
+
+  virtual bool CheckAlignedTerminals(const vector<int>& matching,
+                                     const vector<int>& chunklen,
+                                     const vector<int>& source_low) const;
+
+  virtual bool CheckTightPhrases(const vector<int>& matching,
+                                 const vector<int>& chunklen,
+                                 const vector<int>& source_low) const;
+
+  virtual bool FindFixPoint(
+      int source_phrase_low, int source_phrase_high,
+      const vector<int>& source_low, const vector<int>& source_high,
+      int& target_phrase_low, int& target_phrase_high,
+      const vector<int>& target_low, const vector<int>& target_high,
+      int& source_back_low, int& source_back_high, int sentence_id,
+      int min_source_gap_size, int min_target_gap_size,
+      int max_new_x, bool allow_low_x, bool allow_high_x,
+      bool allow_arbitrary_expansion) const;
+
+  virtual bool GetGaps(
+      vector<pair<int, int> >& source_gaps, vector<pair<int, int> >& target_gaps,
+      const vector<int>& matching, const vector<int>& chunklen,
+      const vector<int>& source_low, const vector<int>& source_high,
+      const vector<int>& target_low, const vector<int>& target_high,
+      int source_phrase_low, int source_phrase_high, int source_back_low,
+      int source_back_high, int& num_symbols, bool& met_constraints) const;
+
+  virtual vector<int> GetGapOrder(const vector<pair<int, int> >& gaps) const;
+
+  // TODO(pauldb): Add unit tests.
+  virtual unordered_map<int, int> GetSourceIndexes(
+      const vector<int>& matching, const vector<int>& chunklen,
+      int starts_with_x) const;
+
+ protected:
+  RuleExtractorHelper();
+
+ private:
+  void FindProjection(
+      int source_phrase_low, int source_phrase_high,
+      const vector<int>& source_low, const vector<int>& source_high,
+      int& target_phrase_low, int& target_phrase_high) const;
+
+  shared_ptr<DataArray> source_data_array;
+  shared_ptr<DataArray> target_data_array;
+  shared_ptr<Alignment> alignment;
+  int max_rule_span;
+  int max_rule_symbols;
+  bool require_aligned_terminal;
+  bool require_aligned_chunks;
+  bool require_tight_phrases;
+};
+
+#endif
-- 
cgit v1.2.3