algo: String Matching: Boyer-Moore-Horspool

Conflicts:
	astringology2/src/astringology.cpp

alib2algo/src/stringology/exact/BoyerMooreHorspool.cpp

+/*
+ * BoyerMooreHorspool.cpp
+ *
+ *  Created on: 5. 11. 2014
+ *      Author: Radomir Polach, Tomas Pecka
+ */
+
+#include "BoyerMooreHorspool.h"
+
+#include <exception/AlibException.h>
+#include <string/LinearString.h>
+#include <alphabet/Symbol.h>
+
+#include <map>
+
+namespace stringology {
+
+namespace exact {
+
+std::set<unsigned> BoyerMooreHorspool::match(const string::String& subject, const string::String& pattern) {
+	std::set<unsigned> data;
+	Accept((void*) &data, subject.getData(), pattern.getData(), BoyerMooreHorspool::BOYER_MOORE_HORSPOOL);
+	return data;
+}
+
+std::set<unsigned> BoyerMooreHorspool::match(const string::LinearString& string, const string::LinearString& pattern)
+{
+	std::set<unsigned> occ;
+	std::map<alphabet::Symbol, size_t> bcs;
+
+	/* Initialization of BCS to the length of the needle. */
+	for(const auto& symbol : string.getAlphabet())
+		bcs.insert(std::make_pair(symbol, pattern.getContent().size()));
+
+	/* Filling out BCS, ignoring last character. */
+	for(size_t i = 0; i < pattern.getContent().size() - 1; i++)
+		bcs[pattern.getContent().at(i)] = pattern.getContent().size() - i - 1;
+
+	/*
+	for(const auto& kv: bcs)
+		std::cout << std::string(kv.first) << " " << kv.second << std::endl;
+	for(const auto& s: string.getContent())
+		std::cout << std::string(s);std::cout << std::endl;
+	*/
+
+	size_t haystack_offset = 0;
+	while(haystack_offset + pattern.getContent().size() <= string.getContent().size())
+	{
+		size_t i = pattern.getContent().size();
+		while(string.getContent().at(haystack_offset + i - 1) == pattern.getContent().at(i - 1))
+		{
+			i--;
+
+			if(i == 0) // Yay, there is match!!!
+			{
+				occ.insert(haystack_offset);
+				haystack_offset ++;
+				break;
+			}
+		}
+
+		if(i != 0)
+		{
+			haystack_offset += bcs[string.getContent().at(haystack_offset + i - 1)];
+		}
+		//std::cout << haystack_offset << std::endl;
+	}
+	return occ;
+}
+
+void BoyerMooreHorspool::Visit(void*, const string::Epsilon&, const string::Epsilon&) const {
+	throw exception::AlibException("Unsupported string type Epsilon");
+}
+
+void BoyerMooreHorspool::Visit(void* data, const string::LinearString& subject, const string::LinearString& pattern) const {
+	std::set<unsigned> & res = *((std::set<unsigned>*) data);
+	res = this->match(subject, pattern);
+}
+
+void BoyerMooreHorspool::Visit(void*, const string::CyclicString&, const string::CyclicString&) const {
+	throw exception::AlibException("Unsupported string type CyclicString");
+}
+
+const BoyerMooreHorspool BoyerMooreHorspool::BOYER_MOORE_HORSPOOL;
+
+} /* namespace exact */
+
+} /* namespace stringology */

alib2algo/src/stringology/exact/BoyerMooreHorspool.h

+/*
+ * BoyerMooreHorspool.h
+ *
+ *  Created on: 5. 11. 2014
+ *      Author: Radomir Polach, Tomas Pecka
+ */
+
+#ifndef _BOYER_MOORE_HORSPOOL_H
+#define _BOYER_MOORE_HORSPOOL_H
+
+#include <string/String.h>
+
+namespace stringology {
+
+namespace exact {
+
+/**
+ * Implementation of BMH for MI(E+\eps)-EVY course 2014
+ * To get rid of zeros in BCS table we ignore last haystack character
+ */
+class BoyerMooreHorspool : public string::VisitableStringBase::const_same_visitor_type {
+public:
+	/**
+	 * Search for pattern in linear string.
+	 * @return set set of occurences
+	 */
+	static std::set<unsigned> match(const string::String& subject, const string::String& pattern);
+
+	static std::set<unsigned> match(const string::LinearString& subject, const string::LinearString& pattern);
+private:
+	void Visit(void*, const string::Epsilon& subject, const string::Epsilon& pattern) const;
+	void Visit(void*, const string::LinearString& subject, const string::LinearString& pattern) const;
+	void Visit(void*, const string::CyclicString& subject, const string::CyclicString& pattern) const;
+
+	static const BoyerMooreHorspool BOYER_MOORE_HORSPOOL;
+};
+
+} /* namespace exact */
+
+} /* namespace stringology */
+
+#endif /* _BOYER_MOORE_HORSPOOL_H */

astringology2/src/astringology.cpp

@@ -15,6 +15,7 @@
 #include <container/Container.h>
 #include <string/naive/ExactMatch.h>
+#include <stringology/exact/BoyerMooreHorspool.h>
 #include <stringology/exact/ExactMatchingAutomaton.h>
 #include <stringology/exact/BorderArray.h>
@@ -25,6 +26,7 @@ int main(int argc, char* argv[]) {
 		std::vector<std::string> allowed;
 		allowed.push_back("exactMatchingAutomaton");
 		allowed.push_back("exactMatch");
+		allowed.push_back("boyerMooreHorspool");
 		allowed.push_back("borderArray");
 		TCLAP::ValuesConstraint<std::string> allowedVals( allowed );
@@ -67,6 +69,12 @@ int main(int argc, char* argv[]) {
 			std::set<unsigned> res = string::naive::ExactMatch::match(subject, pattern);
 			alib::XmlDataFactory::toStdout(res);
 			return 0;
+		} else if( algorithm.getValue() == "boyerMooreHorspool") {
+			string::String subject = alib::XmlDataFactory::fromTokens<string::String>(subjectTokens);
+			string::String pattern = alib::XmlDataFactory::fromTokens<string::String>(patternTokens);
+			std::set<unsigned> res = stringology::exact::BoyerMooreHorspool::match(subject, pattern);
+			alib::XmlDataFactory::toStdout(res);
+			return 0;
 		} else if( algorithm.getValue() == "exactMatchingAutomaton") {
 			string::String pattern = alib::XmlDataFactory::fromTokens<string::String>(patternTokens);
 			automaton::Automaton automaton = stringology::exact::ExactMatchingAutomaton::construct(pattern);