subtree repeats from subtree automaton algo

aarbology2/src/aarbology.cpp

@@ -42,6 +42,7 @@ int main ( int argc, char * argv[] ) {
 		allowed.push_back ( "nonlinearFullAndLinearIndex" );
 		allowed.push_back ( "exactSubtreeRepeatsNaive" );
+		allowed.push_back ( "exactSubtreeRepeatsFromSubtreeAutomaton" );
 		allowed.push_back ( "exactSubtreeRepeats" );
 		allowed.push_back ( "normalizeTreeLabels" );
 		allowed.push_back ( "badCharacterShiftTable" );
@@ -99,6 +100,7 @@ int main ( int argc, char * argv[] ) {
 		  || algorithm.getValue ( ) == "exactTreePatternAutomaton"
 		  || algorithm.getValue ( ) == "exactNonlinearTreePatternAutomaton"
 		  || algorithm.getValue ( ) == "exactSubtreeRepeatsNaive"
+		  || algorithm.getValue ( ) == "exactSubtreeRepeatsFromSubtreeAutomaton"
 		  || algorithm.getValue ( ) == "normalizeTreeLabels"
 		  || algorithm.getValue ( ) == "exactSubtreeRepeats"
 		  || algorithm.getValue ( ) == "compressedBitParallelIndex"
@@ -225,6 +227,8 @@ int main ( int argc, char * argv[] ) {
 				cliCommand = "execute arbology::exact::ExactNonlinearTreePatternAutomaton $subject $subtreeWildcard $nonlinearVariables > $output";
 		} else if ( algorithm.getValue ( ) == "exactSubtreeRepeatsNaive" ) {
 			cliCommand = "execute tree::properties::ExactSubtreeRepeatsNaive $subject > $output";
+		} else if ( algorithm.getValue ( ) == "exactSubtreeRepeatsFromSubtreeAutomaton" ) {
+			cliCommand = "execute arbology::properties::ExactSubtreeRepeatsFromSubtreeAutomaton $subject > $output";
 		} else if ( algorithm.getValue ( ) == "normalizeTreeLabels" ) {
 			cliCommand = "execute tree::NormalizeTreeLabels $subject > $output";
 		} else if ( algorithm.getValue ( ) == "exactSubtreeRepeats" ) {

alib2algo/src/arbology/properties/ExactSubtreeRepeatsFromSubtreeAutomaton.cpp

+/*
+ * ExactSubtreeRepeatsFromSubtreeAutomaton.cpp
+ *
+ *  Created on: 1. 4. 2016
+ *      Author: Jan Travnicek
+ */
+
+#include "ExactSubtreeRepeatsFromSubtreeAutomaton.h"
+#include <registration/AlgoRegistration.hpp>
+
+namespace {
+
+auto ExactRepeatsFromSubtreeAutomatonPrefixRankedTree = registration::AbstractRegister < arbology::properties::ExactSubtreeRepeatsFromSubtreeAutomaton, tree::PrefixRankedTree < unsigned, DefaultRankType >, const tree::PrefixRankedTree < > & > ( arbology::properties::ExactSubtreeRepeatsFromSubtreeAutomaton::repeats );
+
+} /* namespace tree */

alib2algo/src/arbology/properties/ExactSubtreeRepeatsFromSubtreeAutomaton.h

+/*
+ * ExactSubtreeRepeatsFromSubtreeAutomaton.h
+ *
+ *  Created on: 5. 11. 2014
+ *      Author: Jan Travnicek
+ */
+
+#ifndef _ARBOLOGY_SUBTREE_REPEATS_FROM_SUBTREE_AUTOMATON_H_
+#define _ARBOLOGY_SUBTREE_REPEATS_FROM_SUBTREE_AUTOMATON_H_
+
+#include <common/ranked_symbol.hpp>
+
+#include <tree/ranked/PrefixRankedTree.h>
+#include <arbology/exact/ExactSubtreeAutomaton.h>
+#include <automaton/determinize/Determinize.h>
+
+namespace arbology {
+
+namespace properties {
+
+/**
+ * Simple computation of subtree repeats
+ */
+class ExactSubtreeRepeatsFromSubtreeAutomaton {
+	template < class SymbolType, class RankType >
+	static void repeatsInternal ( const tree::PrefixRankedTree < SymbolType, RankType > & originalTree, const automaton::InputDrivenDPDA < common::ranked_symbol < SymbolType, RankType >, char, ext::set < unsigned > > & automaton, ext::vector < common::ranked_symbol < unsigned, RankType > > & repeats, const ext::set < unsigned > & state, unsigned size, unsigned ac );
+
+public:
+	/**
+	 * Compute a same shaped tree with nodes containing unique subtree ids.
+	 * @return Tree of repeats
+	 */
+	template < class SymbolType, class RankType >
+	static tree::PrefixRankedTree < unsigned, RankType > repeats ( const tree::PrefixRankedTree < SymbolType, RankType > & tree );
+
+};
+
+template < class SymbolType, class RankType >
+void ExactSubtreeRepeatsFromSubtreeAutomaton::repeatsInternal ( const tree::PrefixRankedTree < SymbolType, RankType > & originalTree, const automaton::InputDrivenDPDA < common::ranked_symbol < SymbolType, RankType >, char, ext::set < unsigned > > & automaton, ext::vector < common::ranked_symbol < unsigned, RankType > > & repeats, const ext::set < unsigned > & state, unsigned size, unsigned ac ) {
+	if ( state.size ( ) == 0 )
+		return;
+
+	if ( ac == 0 )
+		for ( unsigned label : state )
+			repeats [ label - size ] = common::ranked_symbol < unsigned, RankType > ( * state.begin ( ) - size, originalTree.getContent ( ) [ * state.begin ( ) - size ].getRank ( ) );
+	else
+		for ( const std::pair < ext::pair < ext::set < unsigned >, common::ranked_symbol < SymbolType, RankType > >, ext::set < unsigned > > transition : automaton.getTransitionsFromState ( state ) )
+			repeatsInternal ( originalTree, automaton, repeats, transition.second, size + 1, ac - 1 + ( unsigned ) transition.first.second.getRank ( ) );
+}
+
+template < class SymbolType, class RankType >
+tree::PrefixRankedTree < unsigned, RankType > ExactSubtreeRepeatsFromSubtreeAutomaton::repeats ( const tree::PrefixRankedTree < SymbolType, RankType > & tree ) {
+	automaton::InputDrivenNPDA < common::ranked_symbol < SymbolType, RankType >, char, unsigned > subtreePushdownAutomaton = arbology::exact::ExactSubtreeAutomaton::construct ( tree );
+	automaton::InputDrivenDPDA < common::ranked_symbol < SymbolType, RankType >, char, ext::set < unsigned > > deterministicSubtreePushdownAutomaton = automaton::determinize::Determinize::determinize ( subtreePushdownAutomaton );
+	ext::vector < common::ranked_symbol < unsigned, RankType > > data ( tree.getContent ( ).size ( ), common::ranked_symbol < unsigned, RankType > ( 0, RankType ( 0 ) ) );
+	repeatsInternal ( tree, deterministicSubtreePushdownAutomaton, data, deterministicSubtreePushdownAutomaton.getInitialState ( ), 0, 1 );
+	return tree::PrefixRankedTree < unsigned, RankType > ( data );
+}
+
+} /* namespace properties */
+
+} /* namespace arbology */
+
+#endif /* _ARBOLOGY_SUBTREE_REPEATS_FROM_SUBTREE_AUTOMATON_H_ */

tests.repeats.sh

@@ -157,9 +157,11 @@ function runRepeatsAlgorithmTest {
 	NAIVE_OUTPUT=$(mktemp)
 	ADVANCED_OUTPUT=$(mktemp)
+	ARBOLOGY_OUTPUT=$(mktemp)
 	./aarbology2 -s $1 -a exactSubtreeRepeatsNaive | ./aarbology2 -a normalizeTreeLabels > $NAIVE_OUTPUT
-	./acast2 -t PostfixRankedTree -i  $1 | ./aarbology2 -a exactSubtreeRepeats | ./acast2 -t RankedTree | ./aarbology2 -a normalizeTreeLabels > $ADVANCED_OUTPUT
+	./acast2 -t PostfixRankedTree -i $1 | ./aarbology2 -a exactSubtreeRepeats | ./acast2 -t RankedTree | ./aarbology2 -a normalizeTreeLabels > $ADVANCED_OUTPUT
+	./acast2 -t PrefixRankedTree -i $1 | ./aarbology2 -a exactSubtreeRepeatsFromSubtreeAutomaton | ./acast2 -t RankedTree | ./aarbology2 -a normalizeTreeLabels > $ARBOLOGY_OUTPUT
 	OUT=`timeout $TESTCASE_TIMEOUT bash -c "diff $NAIVE_OUTPUT $ADVANCED_OUTPUT"`
 	RET=$?
@@ -168,7 +170,14 @@ function runRepeatsAlgorithmTest {
 		log "$1" $RET "$OUT" "$1" "$NAIVE_OUTPUT" "$ADVANCED_OUTPUT"
 	fi
-	rm $1 $NAIVE_OUTPUT $ADVANCED_OUTPUT
+	OUT=`timeout $TESTCASE_TIMEOUT bash -c "diff $NAIVE_OUTPUT $ARBOLOGY_OUTPUT"`
+	RET=$?
+
+	if [ $RET != 0 ]; then # fail
+		log "$1" $RET "$OUT" "$1" "$NAIVE_OUTPUT" "$ARBOLOGY_OUTPUT"
+	fi
+
+	rm $1 $NAIVE_OUTPUT $ADVANCED_OUTPUT $ARBOLOGY_OUTPUT
 	if [ $RET == 124 ]; then # timeout
 		registerResult 2