diff options
Diffstat (limited to 'extractor/precomputation.h')
| -rw-r--r-- | extractor/precomputation.h | 15 |
1 files changed, 10 insertions, 5 deletions
diff --git a/extractor/precomputation.h b/extractor/precomputation.h index 0d1b269f..428505d8 100644 --- a/extractor/precomputation.h +++ b/extractor/precomputation.h @@ -16,8 +16,8 @@ using namespace tr1; class SuffixArray; -typedef boost::hash<vector<int> > vector_hash; -typedef unordered_map<vector<int>, vector<int>, vector_hash> Index; +typedef boost::hash<vector<int> > VectorHash; +typedef unordered_map<vector<int>, vector<int>, VectorHash> Index; class Precomputation { public: @@ -27,20 +27,25 @@ class Precomputation { int max_rule_symbols, int min_gap_size, int max_frequent_phrase_len, int min_frequency); + virtual ~Precomputation(); + void WriteBinary(const fs::path& filepath) const; - const Index& GetInvertedIndex() const; - const Index& GetCollocations() const; + virtual const Index& GetInvertedIndex() const; + virtual const Index& GetCollocations() const; static int NON_TERMINAL; + protected: + Precomputation(); + private: vector<vector<int> > FindMostFrequentPatterns( shared_ptr<SuffixArray> suffix_array, const vector<int>& data, int num_frequent_patterns, int max_frequent_phrase_len, int min_frequency); void AddCollocations( - const vector<tuple<int, int, int> >& matchings, const vector<int>& data, + const vector<std::tuple<int, int, int> >& matchings, const vector<int>& data, int max_rule_span, int min_gap_size, int max_rule_symbols); void AddStartPositions(vector<int>& positions, int pos1, int pos2); void AddStartPositions(vector<int>& positions, int pos1, int pos2, int pos3); |