From 9a026ba2db8fa7723374109e6a4a8dcaff8733cd 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.h | 92 +++++++++++++++++-----------------------------
 1 file changed, 33 insertions(+), 59 deletions(-)

(limited to 'extractor/rule_extractor.h')

diff --git a/extractor/rule_extractor.h b/extractor/rule_extractor.h
index f668de24..a087dc6d 100644
--- a/extractor/rule_extractor.h
+++ b/extractor/rule_extractor.h
@@ -2,6 +2,7 @@
 #define _RULE_EXTRACTOR_H_
 
 #include <memory>
+#include <unordered_map>
 #include <vector>
 
 #include "phrase.h"
@@ -13,8 +14,9 @@ class DataArray;
 class PhraseBuilder;
 class PhraseLocation;
 class Rule;
+class RuleExtractorHelper;
 class Scorer;
-class Vocabulary;
+class TargetPhraseExtractor;
 
 typedef vector<pair<int, int> > PhraseAlignment;
 
@@ -46,84 +48,56 @@ class RuleExtractor {
                 bool require_aligned_chunks,
                 bool require_tight_phrases);
 
-  vector<Rule> ExtractRules(const Phrase& phrase,
-                            const PhraseLocation& location) const;
+  // For testing only.
+  RuleExtractor(shared_ptr<DataArray> source_data_array,
+                shared_ptr<PhraseBuilder> phrase_builder,
+                shared_ptr<Scorer> scorer,
+                shared_ptr<TargetPhraseExtractor> target_phrase_extractor,
+                shared_ptr<RuleExtractorHelper> helper,
+                int max_rule_span,
+                int min_gap_size,
+                int max_nonterminals,
+                int max_rule_symbols,
+                bool require_tight_phrases);
+
+  virtual ~RuleExtractor();
+
+  virtual vector<Rule> ExtractRules(const Phrase& phrase,
+                                    const PhraseLocation& location) const;
+
+ protected:
+  RuleExtractor();
 
  private:
   vector<Extract> ExtractAlignments(const Phrase& phrase,
                                     const vector<int>& matching) const;
 
-  void GetLinksSpans(vector<int>& source_low, vector<int>& source_high,
-                     vector<int>& target_low, vector<int>& target_high,
-                     int sentence_id) const;
-
-  bool CheckAlignedTerminals(const vector<int>& matching,
-                             const vector<int>& chunklen,
-                             const vector<int>& source_low) const;
-
-  bool CheckTightPhrases(const vector<int>& matching,
-                         const vector<int>& chunklen,
-                         const vector<int>& source_low) const;
-
-  bool FindFixPoint(
-      int source_phrase_start, int source_phrase_end,
-      const vector<int>& source_low, const vector<int>& source_high,
-      int& target_phrase_start, int& target_phrase_end,
-      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, int max_low_x, int max_high_x,
-      bool allow_arbitrary_expansion) const;
-
-  void FindProjection(
-      int source_phrase_start, int source_phrase_end,
-      const vector<int>& source_low, const vector<int>& source_high,
-      int& target_phrase_low, int& target_phrase_end) const;
-
-  bool CheckGaps(
-     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;
-
   void AddExtracts(
       vector<Extract>& extracts, const Phrase& source_phrase,
+      const unordered_map<int, int>& source_indexes,
       const vector<pair<int, int> >& target_gaps, const vector<int>& target_low,
       int target_phrase_low, int target_phrase_high, int sentence_id) const;
 
-  vector<pair<Phrase, PhraseAlignment> > ExtractTargetPhrases(
-      const vector<pair<int, int> >& target_gaps, const vector<int>& target_low,
-      int target_phrase_low, int target_phrase_high, int sentence_id) const;
-
-  void GeneratePhrases(
-      vector<pair<Phrase, PhraseAlignment> >& target_phrases,
-      const vector<pair<int, int> >& ranges, int index,
-      vector<int>& subpatterns, const vector<int>& target_gap_order,
-      int target_phrase_low, int target_phrase_high, int sentence_id) const;
-
   void AddNonterminalExtremities(
-      vector<Extract>& extracts, const Phrase& source_phrase,
-      int source_phrase_low, int source_phrase_high, int source_back_low,
-      int source_back_high, const vector<int>& source_low,
+      vector<Extract>& extracts, const vector<int>& matching,
+      const vector<int>& chunklen, const Phrase& source_phrase,
+      int source_back_low, int source_back_high, const vector<int>& source_low,
       const vector<int>& source_high, const vector<int>& target_low,
-      const vector<int>& target_high,
-      const vector<pair<int, int> >& target_gaps, int sentence_id,
-      int extend_left, int extend_right) const;
+      const vector<int>& target_high, vector<pair<int, int> > target_gaps,
+      int sentence_id, int starts_with_x, int ends_with_x, int extend_left,
+      int extend_right) const;
 
-  shared_ptr<DataArray> source_data_array;
+ private:
   shared_ptr<DataArray> target_data_array;
-  shared_ptr<Alignment> alignment;
+  shared_ptr<DataArray> source_data_array;
   shared_ptr<PhraseBuilder> phrase_builder;
   shared_ptr<Scorer> scorer;
-  shared_ptr<Vocabulary> vocabulary;
+  shared_ptr<TargetPhraseExtractor> target_phrase_extractor;
+  shared_ptr<RuleExtractorHelper> helper;
   int max_rule_span;
   int min_gap_size;
   int max_nonterminals;
   int max_rule_symbols;
-  bool require_aligned_terminal;
-  bool require_aligned_chunks;
   bool require_tight_phrases;
 };
 
-- 
cgit v1.2.3