diff --git a/alib2algo/src/tree/properties/ExactSubtreeRepeats.h b/alib2algo/src/tree/properties/ExactSubtreeRepeats.h
index 8de0d9fd9267919461344bb5e9c9c61d460582ca..3f82debc7c2ad2eeae3a32ebcdb444cc2eb1d6e6 100644
--- a/alib2algo/src/tree/properties/ExactSubtreeRepeats.h
+++ b/alib2algo/src/tree/properties/ExactSubtreeRepeats.h
@@ -36,13 +36,40 @@ namespace properties {
*/
class ExactSubtreeRepeats : public std::SingleDispatch < ExactSubtreeRepeats, tree::Tree, const tree::TreeBase & > {
+ /**
+ * A nested class to hold and efficiently pass auxiliary arrays
+ */
class ExactSubtreeRepeatsAux {
+
+ /**
+ * Constructs array mu
+ * mu is a mapping from a ranked symbol to a number
+ * @param symbols The tree in postfix notation
+ */
template < class SymbolType, class RankType >
void buildMu ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols );
+
+ /**
+ * Constructs Parrent array P
+ * P[i] stores the index of the parent node for node i
+ * @param symbols The tree in postfix notation
+ */
template < class SymbolType, class RankType >
void buildP ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols );
+
+ /**
+ * Constructs array H
+ * H[i] stores height of the node i
+ * @param symbols The tree in postfix notation
+ */
template < class SymbolType, class RankType >
void buildH ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols );
+
+ /**
+ * Constructs array FC
+ * FC[i] is true if node i is the first child of its parent
+ * @param symbols The tree in postfix notation
+ */
template < class SymbolType, class RankType >
void buildFC ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols );
@@ -52,18 +79,39 @@ class ExactSubtreeRepeats : public std::SingleDispatch < ExactSubtreeRepeats, tr
std::vector < unsigned > P;
std::vector < unsigned > H;
std::vector < bool > FC;
- std::vector < unsigned > T;
- std::vector < unsigned > TL;
- std::vector < std::queue < std::tuple < std::deque < unsigned >, unsigned, int > > > LA;
- std::list < std::tuple < std::deque < unsigned >, unsigned, int > > found_repeats;
- unsigned alphabetSize;
- unsigned treeSize;
- unsigned sc;
+ std::vector < unsigned > T; /**< Stores ID of the last s-repeat found at this node. */
+ std::vector < unsigned > TL; /**< Complements array T. For every i in TL[i] stores length of s-repeat at T[i]. */
+ std::vector < std::queue < std::tuple < std::deque < unsigned >, unsigned, int > > > LA; /**< Level array. At index i stores repeats scheduled for processing at height i. */
+ std::list < std::tuple < std::deque < unsigned >, unsigned, int > > found_repeats; /**< Stores found repeats to be accessed later. */
+ unsigned alphabetSize; /**< Number of unique ranked symbols. */
+ unsigned treeSize; /**< Number of nodes in a tree. */
+ unsigned sc; /**< Tracks IDs of subtree repeats. */
};
+ /**
+ * Starting point of the algorithm
+ *
+ * @param symbols The tree in postfix notation
+ * @param res Array reference to store the result
+ */
template < class SymbolType, class RankType >
static void repeatsPostfixRanked ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols, std::vector < std::ranked_symbol < unsigned, RankType > > & res );
+
+ /**
+ * Assigns repeat triplet to the next height
+ * Checks whether some of the nodes in triplet->S can define subtrees on heigher levels of the tree.
+ * @param triplet Triplet to process
+ * @param aux Reference to auxiliary structures
+ */
static void assignLevel ( std::tuple < std::deque < unsigned >, unsigned, int > triplet, ExactSubtreeRepeats::ExactSubtreeRepeatsAux & aux );
+
+ /**
+ * Tries to expand triplet (S, l, ac) until it represents a subtree
+ *
+ * @param triplet Triplet to process
+ * @param symbols The tree in postfix notation
+ * @param aux Reference to auxiliary structures
+ */
template < class SymbolType, class RankType >
static void partition ( std::tuple < std::deque < unsigned >, unsigned, int > triplet, const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols, ExactSubtreeRepeats::ExactSubtreeRepeatsAux & aux );
@@ -104,7 +152,7 @@ ExactSubtreeRepeats::ExactSubtreeRepeatsAux::ExactSubtreeRepeatsAux ( const std:
template < class SymbolType, class RankType >
void ExactSubtreeRepeats::ExactSubtreeRepeatsAux::buildMu ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols ) {
- // Build mapping mu_map((Symb, Rank) -> Number)
+ // Build mapping mu_map((Symb, Rank) -> Number) and construct array mu from it
std::map < std::pair < SymbolType, RankType >, unsigned > mu_map;
this->alphabetSize = 0;
@@ -136,7 +184,7 @@ void ExactSubtreeRepeats::ExactSubtreeRepeatsAux::buildMu ( const std::vector <
template < class SymbolType, class RankType >
void ExactSubtreeRepeats::ExactSubtreeRepeatsAux::buildP ( const std::vector < std::ranked_symbol < SymbolType, RankType > > & symbols ) {
- // Build parrent array
+ // Build parent array
P = std::vector < unsigned > ( this->treeSize - 1 );
std::stack < unsigned > RP;
@@ -263,6 +311,10 @@ void ExactSubtreeRepeats::assignLevel ( std::tuple < std::deque < unsigned >, un
aux.LA[k].push ( std::make_tuple ( An[k], l, 0 ) );
Bn[k] = false;
+ /* This is line 15 (in paper) in the Assign-Level algorithm
+ * It has no effect here, as I made "An" local and the function ends right after the while loop.
+ * I leave it here to be consistent with the algorithm.
+ */
while ( !An[k].empty ( ) )
An[k].pop_front ( );
}
@@ -316,6 +368,11 @@ void ExactSubtreeRepeats::partition ( std::tuple < std::deque < unsigned >, unsi
unsigned k = Q1.front ( );
Q1.pop ( );
Q3.push ( En[k] );
+
+ /* The next two lines can be safely removed (Partition(), lines 23-24 in the paper)
+ * The variables are local and won't be used in the function again.
+ * Leaving them here to show every step of the algorithm.
+ */
En[k] = std::tuple < std::deque < unsigned >, unsigned, int > ( );
Bn[k] = false;
}
@@ -324,6 +381,11 @@ void ExactSubtreeRepeats::partition ( std::tuple < std::deque < unsigned >, unsi
unsigned k = Q2.front ( );
Q2.pop ( );
Q3.push ( Es[k] );
+
+ /* The next two lines can be safely removed (Partition(), lines 28-29 in the paper)
+ * The variables are local and won't be used in the function again.
+ * Leaving them here to show every step of the algorithm.
+ */
Es[k] = std::tuple < std::deque < unsigned >, unsigned, int > ( );
Bs[k] = false;
}
@@ -416,7 +478,12 @@ void ExactSubtreeRepeats::repeatsPostfixRanked ( const std::vector < std::ranked
aux.LA[i].pop ( );
}
- // prepare result
+ /* Prepare result :
+ * we have collected the triplets at this point and
+ * need to build a postfix representation of a tree from them.
+ * S-repeats IDs will be used as node-labels for the root (index_from_S + l)
+ * of each subtree.
+ */
std::vector < unsigned > post_repeats ( aux.treeSize );
unsigned curr_repeat = aux.found_repeats.size ( );