third_party/cld/encodings/compact_lang_det/tote.h

   1 // Copyright (c) 2006-2009 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef ENCODINGS_COMPACT_LANG_DET_TOTE_H_
   6 #define ENCODINGS_COMPACT_LANG_DET_TOTE_H_
   7
   8 #include <stdio.h>
   9 #include "encodings/compact_lang_det/win/cld_basictypes.h"
  10
  11 // Take a set of <key, value> pairs and tote them up.
  12 // After explicitly sorting, retrieve top key, value pairs
  13 class Tote {
  14  public:
  15   Tote();
  16   ~Tote();
  17   void Reinit();
  18   void AddGram();
  19   void Add(uint8 ikey, int idelta);
  20   void AddBytes(int ibytes) {byte_count_ += ibytes;}
  21   int CurrentTopKey();
  22   void Sort(int n);
  23   void Dump(FILE* f);
  24   uint16 GetGramCount() const {return gram_count_;}
  25   uint16 GetIncrCount() const {return incr_count_;}
  26   int GetByteCount() const {return byte_count_;}
  27   int MaxSize() const {return kMaxSize_;}
  28   uint8 Key(int i) const {return key_[i];}
  29   int Value(int i) const {return value_[i];}
  30   void SetGramCount(uint16 v) {gram_count_ = v;}
  31   void SetIncrCount(uint16 v) {incr_count_ = v;}
  32   void SetKey(int i, int v) {key_[i] = v;}
  33   void SetValue(int i, int v) {value_[i] = v;}
  34
  35  private:
  36   static const int kMaxSize_ = 24;
  37   uint16 gram_count_;       // Number of quadgrams/etc. scored
  38   uint16 incr_count_;       // Number of Add calls (1-3 per gram)
  39   int byte_count_;          // Bytes of text scored
  40   // Align at multiple of 8 bytes
  41   uint8 key_[kMaxSize_];    // Lang unassigned = 0, valid = 1..255
  42   int value_[kMaxSize_];    // Probability score sum
  43 };
  44
  45
  46 // Take a set of <key, value, reliability> triples and tote them up.
  47 // After explicitly sorting, retrieve top key, value, reliability triples
  48 class ToteWithReliability {
  49  public:
  50   ToteWithReliability();
  51   ~ToteWithReliability();
  52   void Reinit();
  53   void Add(uint8 ikey, int ibytes, int score, int ireliability);
  54   int Find(uint8 ikey);
  55   void AddClosePair(int subscr, int val) {closepair_[subscr] += val;}
  56   int CurrentTopKey();
  57   void Sort(int n);
  58   void Dump(FILE* f);
  59
  60   ////void AddSeq(uint8 ikey) {ss_.Add(ikey);}
  61   ////void ExtractSeq(int n, uint8* dst) {ss_.Extract(n, dst);}
  62
  63   int GetIncrCount() const {return incr_count_;}
  64   int GetClosePair(int subscr) const {return closepair_[subscr];}
  65   int MaxSize() const {return kMaxSize_;}
  66   uint8 Key(int i) const {return key_[i];}
  67   int Value(int i) const {return value_[i];}
  68   int Score(int i) const {return score_[i];}
  69   int Reliability(int i) const {return reliability_[i];}
  70   void SetKey(int i, int v) {key_[i] = v;}
  71   void SetValue(int i, int v) {value_[i] = v;}
  72   void SetScore(int i, int v) {score_[i] = v;}
  73   void SetReliability(int i, int v) {reliability_[i] = v;}
  74
  75  private:
  76   static const int kMaxSize_ = 24;
  77   static const int kMaxClosePairSize_ = 8;
  78   int incr_count_;         // Number of Add calls
  79   int sorted_;             // Contents have been sorted, cannot Add
  80   // Align at multiple of 8 bytes
  81   int closepair_[kMaxClosePairSize_];
  82   uint8 key_[kMaxSize_];    // Lang unassigned = 0, valid = 1..255
  83   int value_[kMaxSize_];    // Bytecount this lang
  84   int score_[kMaxSize_];    // Probability score sum
  85   int reliability_[kMaxSize_];  // Percentage 0..100
  86   ////SubsetSequence ss_;
  87 };
  88
  89 #endif  // ENCODINGS_COMPACT_LANG_DET_TOTE_H_