Allomorphy.h

   1 // Driver and result type for alignment of similar signatures
   2 // Copyright © 2009 The University of Chicago
   3 #ifndef ALLOMORPHY_H
   4 #define ALLOMORPHY_H
   5
   6 #include <QTextStream>
   7 #include <QList>
   8 #include <QStringList>
   9 #include "Parse.h"
  10 #include "MiniLexicon.h"
  11
  12 enum eAlignmentStatus {IDENTICAL, DIFFERENT};
  13
  14 class AffixAlignment {
  15         QString m_OriginalAffix1;
  16         QString m_OriginalAffix2;
  17         QString m_Affix1;
  18         QString m_Affix2;
  19         QString m_Margin1;
  20         QString m_Margin2;
  21         QString m_Shift1;
  22         QString m_Shift2;
  23         enum eAlignmentStatus m_Status;
  24         /// size in bits of overlap of letters/phonemes
  25         double m_Agreement_unigram;
  26         /// size in bits of overlap of bigram-letters/phonemes
  27         double m_Agreement_bigram;
  28         /// size in bits of symmetric difference.
  29         double m_Disagreement_unigram;
  30         double m_Disagreement_bigram;
  31 public:
  32         // construction/destruction.
  33
  34         AffixAlignment(QString Affix1, QString Affix2);
  35         AffixAlignment(QString Margin1, QString Affix1,
  36                 QString Margin2, QString Affix2);
  37         // destructor implicitly defined.
  38
  39         // copy-construction and assignment implicitly defined.
  40 private:
  41         // disable default-construction
  42         AffixAlignment();
  43 public:
  44         // aligned affixes.
  45
  46         QString GetAffix1() const { return m_Affix1;}
  47         QString GetAffix2() const { return m_Affix2;}
  48         QString GetMargin1() const { return m_Margin1;}
  49         QString GetMargin2() const { return m_Margin2;}
  50         QString GetShift1() const { return m_Shift1;}
  51         QString GetShift2() const { return m_Shift2;}
  52 };
  53
  54 /// m_Sig1 is shorter than m_Sig2, or of same length, where we measure length as the total number of letters in the K longest affixes, where K is the min of the length of the two signatures.
  55 class CSignatureAlignment {
  56         class CSignature* m_SigPtr1;
  57         class CSignature* m_SigPtr2;
  58         QList<AffixAlignment*>  m_AffixAlignments;
  59         QStringList             m_Sig1AlignedAffixes;
  60         QStringList             m_Sig2AlignedAffixes;
  61         CSignature*             m_LongerSig;
  62         CSignature*             m_ShorterSig;
  63         QString                 m_difference;
  64 public:
  65         // construction/destruction.
  66
  67         CSignatureAlignment(class CSignature* Sig1, class CSignature* Sig2);
  68         CSignature*         LongerSig() {return m_LongerSig;};
  69         QString             difference() {return m_difference;};
  70         // destructor implicitly defined
  71
  72         // disable default-construction, copy
  73 private:
  74         CSignatureAlignment();
  75         CSignatureAlignment(const CSignatureAlignment& x);
  76         CSignatureAlignment& operator=(const CSignatureAlignment& x);
  77 public:
  78
  79         // Levenshtein’s string-edit-distance algorithm
  80
  81         void FindBestAlignment();
  82         /// success?
  83         bool AllSig1AffixesAligned();
  84         bool AllSig2AffixesAligned();
  85         /// result from string-edit-distance algorithm
  86         QList<AffixAlignment*>* GetAffixAlignments()
  87                 { return &m_AffixAlignments; }
  88
  89         /// Determines whether the two suffix signatures differ just by one having a prefix that the other does not (e.g., t, or at)
  90         bool FindWhetherOneIsSuffixOfTheOther();
  91
  92
  93         /// output to file
  94         void Display(QTextStream& LogStream);
  95         void Display(CMiniLexicon* p_mini);
  96 };
  97
  98 #endif // ALLOMORPHY_H