basic dogaru

alib2algo/src/stringology/exact/Dogaru.cpp

+/*
+ * Dogaru.cpp
+ *
+ *  Created on: 19. 3. 2020
+ *      Author: Jan Jirak
+ */
+
+#include "Dogaru.h"
+#include <registration/AlgoRegistration.hpp>
+
+namespace {
+
+    auto Dogaru = registration::AbstractRegister < stringology::exact::Dogaru, ext::set < unsigned >, const string::LinearString < > &, const string::LinearString < > & > ( stringology::exact::Dogaru::match );
+
+} /* namespace */

alib2algo/src/stringology/exact/Dogaru.h

+/*
+ * Dogaru.h
+ *
+ *  Created on: 19. 3. 2020
+ *      Author: Jan Jirak
+ */
+
+#ifndef _STRINGOLOGY_DOGARU_H_
+#define _STRINGOLOGY_DOGARU_H_
+
+#include <alib/measure>
+
+#include <alib/set>
+#include <alib/vector>
+
+#include <string/LinearString.h>
+
+
+namespace stringology {
+
+    namespace exact {
+
+
+/**
+ * Implementation of the Dogaru algorithm from article "On the All Occurrences of a Word in a Text" by O.C.Dogaru
+ */
+        class Dogaru{
+        public:
+            /**
+             * Search for pattern in linear string.
+             * @return set set of occurences
+             */
+            template < class SymbolType >
+            static ext::set < unsigned > match ( const string::LinearString < SymbolType > & subject, const string::LinearString < SymbolType > & pattern );
+
+        };
+
+        template < class SymbolType >
+        ext::set < unsigned > Dogaru::match ( const string::LinearString < SymbolType > & subject, const string::LinearString < SymbolType > & pattern ) {
+            ext::set<unsigned> occ;
+            const auto & text = subject.getContent();
+            const auto & pat = pattern.getContent();
+            long int n = text.size(), m = pat.size();
+
+            unsigned i = 0 , j = 0 , k = 0 ;
+
+            do {
+                j = 0 ;
+                while ( j < m && pat[j] == text[i] ) { ++ i ; ++ j ; }
+                if ( j == m ) { occ.insert( i - j ) ; continue ; }
+                one: ++ i ;
+                while ( i <= n - m + j && pat[j] != text[i] ) ++ i ;
+                if ( i > n - m + j ) return occ ;
+                k = 0 ;
+                while ( k <= m - 1 && pat[k] == text[ i - j + k ] ) ++ k ;
+                if ( k == m ) { occ.insert( i - j ) ; i = i - j + m ; }
+                else { goto one ;}
+
+            } while ( i < n - m + j ) ;
+
+            return occ;
+        }
+
+
+    } /* namespace exact */
+
+} /* namespace stringology */
+
+#endif /* _STRINGOLOGY_DOGARU_H_ */