Slice.h

   1 // fast “slice-and-count” calculation of string similarity
   2 // Copyright © 2009 The University of Chicago
   3 #ifndef SLICE_H
   4 #define SLICE_H
   5
   6 class snippet;
   7 class StringInventory;
   8 class StringPtrList;
   9 class SliceCollection;
  10
  11 // for snippet:
  12 #include <QString>
  13 // for StringInventory:
  14 #include <Q3Dict>
  15 #include <QString>
  16 namespace linguistica { namespace ui { class status_user_agent; } }
  17 // for StringPtrList:
  18 #include <QList>
  19 #include <QString>
  20 // for SliceCollection:
  21 #include <QList>
  22 namespace linguistica { namespace ui { class status_user_agent; } }
  23
  24 /**
  25  * A snippet is a representation of a string which also remembers the
  26  * number of times it is used.  The primary purpose of this object is
  27  * so that we can access a pointer to each one.
  28  */
  29 class snippet {
  30         QString m_string;
  31         int m_count;
  32 public:
  33         explicit snippet(QString);
  34         unsigned int getlength() const {return  m_string.length();}
  35         QString getkey() const {return m_string;}
  36 };
  37
  38 extern bool lessThan(const snippet* s1, const snippet* s2);
  39
  40 /**
  41  * This is simply a dictionary that allows us to go from a QString to a pointer
  42  * to a snippet with the same key that is in the inventory.
  43  */
  44 class StringInventory : public Q3Dict<snippet> {
  45 public:
  46         StringInventory();
  47
  48         void addwordcollection(class CWordCollection* words,
  49                 SliceCollection& slices,
  50                 linguistica::ui::status_user_agent& status_display);
  51         snippet* getsnippetpointer(QString key);
  52         snippet* operator<<(QString key);
  53         StringPtrList* addword(QString word);
  54 };
  55
  56 /**
  57  * A slice is a list of pointers to a snippet for every substring in the key of
  58  * the slice.  It is the principal class in this file.
  59  */
  60 class StringPtrList : public QList<snippet*> {  // a.k.a. Slice
  61 public:
  62         QString m_key;
  63         StringInventory* m_MyStringInventory;
  64 public:
  65         StringPtrList(QString key, StringInventory*);
  66
  67         void sort();
  68         int getlength() {return m_key.length();}
  69         QString getkey() {return m_key;}
  70 };
  71 typedef StringPtrList Slice;
  72
  73 class SliceCollection : public QList<Slice*> {
  74 public:
  75         void writetolog(class QTextStream& log);
  76         void FindWordPairs(linguistica::ui::status_user_agent& status_display,
  77                 class CLexicon* lex, unsigned int minimumlength = 0);
  78 };
  79
  80 extern int largestcommonsubstringlength(Slice* a, Slice* b,
  81         unsigned int minimum = 0);
  82 extern snippet* largestcommonsubstring(Slice* a, Slice* b,
  83         unsigned int minimum = 0);
  84
  85 #endif // SLICE_H