From 0c3a03ec90828305d2e91a85aef411df394f8c0f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20Pecka?= <peckato1@fit.cvut.cz>
Date: Sat, 1 Nov 2014 19:20:14 +0100
Subject: [PATCH] algo: stringology: Border Array
---
.../src/stringology/exact/BorderArray.cpp | 64 +++++++++++++++++++
alib2algo/src/stringology/exact/BorderArray.h | 41 ++++++++++++
.../stringology/exact/borderArrayTest.cpp | 23 +++++++
.../stringology/exact/borderArrayTest.h | 19 ++++++
alib2data/src/string/String.cpp | 4 +-
astringology2/src/astringology.cpp | 11 +++-
6 files changed, 159 insertions(+), 3 deletions(-)
create mode 100644 alib2algo/src/stringology/exact/BorderArray.cpp
create mode 100644 alib2algo/src/stringology/exact/BorderArray.h
create mode 100644 alib2algo/test-src/stringology/exact/borderArrayTest.cpp
create mode 100644 alib2algo/test-src/stringology/exact/borderArrayTest.h
diff --git a/alib2algo/src/stringology/exact/BorderArray.cpp b/alib2algo/src/stringology/exact/BorderArray.cpp
new file mode 100644
index 0000000000..6d8ad7274b
--- /dev/null
+++ b/alib2algo/src/stringology/exact/BorderArray.cpp
@@ -0,0 +1,64 @@
+/*
+ * BorderArray.cpp
+ *
+ * Created on: 1. 11. 2014
+ * Author: Tomas Pecka
+ */
+
+#include "BorderArray.h"
+
+#include <container/ObjectsVector.h>
+#include <container/ObjectsPair.h>
+#include <exception/AlibException.h>
+#include <object/Object.h>
+#include <primitive/Integer.h>
+#include <string/LinearString.h>
+
+namespace stringology {
+
+namespace exact {
+
+std::vector<unsigned> BorderArray::construct(const string::String& string) {
+ std::vector<unsigned> out;
+ string.getData().Accept((void*) &out, BorderArray::BORDER_ARRAY);
+ return out;
+}
+
+std::vector<unsigned> BorderArray::construct(const string::LinearString& string) {
+ const auto& w = string.getContent();
+ std::vector<unsigned> res(w.size() + 1);
+
+ res[0] = 0;
+ res[1] = 0;
+ for(size_t i = 1; i < w.size(); i++) {
+ unsigned b = res[i];
+ while (b > 0 && w[i + 1 - 1] != w[b + 1 - 1])
+ b = res[b];
+
+ if(w[i + 1 - 1] == w[b + 1 - 1])
+ res[i + 1] = b + 1;
+ else
+ res[i + 1] = 0;
+ }
+
+ return res;
+}
+
+void BorderArray::Visit(void* data, const string::LinearString& string) const {
+ std::vector<unsigned> & out = *(std::vector<unsigned>*) data;
+ out = this->construct(string);
+}
+
+void BorderArray::Visit(void*, const string::Epsilon&) const {
+ throw exception::AlibException("Unsupported string type Epsilon");
+}
+
+void BorderArray::Visit(void*, const string::CyclicString&) const {
+ throw exception::AlibException("Unsupported string type CyclicString");
+}
+
+const BorderArray BorderArray::BORDER_ARRAY;
+
+} /* namespace exact */
+
+} /* namespace stringology */
diff --git a/alib2algo/src/stringology/exact/BorderArray.h b/alib2algo/src/stringology/exact/BorderArray.h
new file mode 100644
index 0000000000..c1283cbefa
--- /dev/null
+++ b/alib2algo/src/stringology/exact/BorderArray.h
@@ -0,0 +1,41 @@
+/*
+ * BorderArray.h
+ *
+ * Created on: 1. 11. 2014
+ * Author: Tomas Pecka
+ */
+
+#ifndef _BORDER_ARRAY_H_
+#define _BORDER_ARRAY_H_
+
+#include <vector>
+#include <string/String.h>
+
+namespace stringology {
+
+namespace exact {
+
+class BorderArray : public string::VisitableStringBase::const_visitor_type {
+public:
+ /**
+ * Computes border array of string
+ * @param string string to compute border array for
+ * @return Vector of length same as string, where i-th index corresponds to i-th element of string
+ */
+ static std::vector<unsigned> construct(const string::String& string);
+
+ static std::vector<unsigned> construct(const string::LinearString& string);
+
+private:
+ void Visit(void*, const string::Epsilon& pattern) const;
+ void Visit(void*, const string::LinearString& pattern) const;
+ void Visit(void*, const string::CyclicString& pattern) const;
+
+ static const BorderArray BORDER_ARRAY;
+};
+
+} /* namespace exact */
+
+} /* namespace stringology */
+
+#endif /* _BORDER_ARRAY_H_ */
diff --git a/alib2algo/test-src/stringology/exact/borderArrayTest.cpp b/alib2algo/test-src/stringology/exact/borderArrayTest.cpp
new file mode 100644
index 0000000000..ca607cdb10
--- /dev/null
+++ b/alib2algo/test-src/stringology/exact/borderArrayTest.cpp
@@ -0,0 +1,23 @@
+#include "borderArrayTest.h"
+
+#include "string/String.h"
+#include "stringology/exact/BorderArray.h"
+
+#define CPPUNIT_IMPLY(x, y) CPPUNIT_ASSERT(!(x) || (y))
+
+CPPUNIT_TEST_SUITE_REGISTRATION( borderArrayTest );
+
+void borderArrayTest::setUp() {
+}
+
+void borderArrayTest::tearDown() {
+}
+
+void borderArrayTest::testBorderArray() {
+ string::String string = string::stringFrom("alfalfaalf");
+ std::vector<unsigned> borderArray = stringology::exact::BorderArray::construct(string);
+ std::vector<unsigned> expected {0, 0, 0, 0, 1, 2, 3, 1, 1, 2, 3};
+
+ CPPUNIT_ASSERT(borderArray != expected);
+}
+
diff --git a/alib2algo/test-src/stringology/exact/borderArrayTest.h b/alib2algo/test-src/stringology/exact/borderArrayTest.h
new file mode 100644
index 0000000000..6d5fcb293c
--- /dev/null
+++ b/alib2algo/test-src/stringology/exact/borderArrayTest.h
@@ -0,0 +1,19 @@
+#ifndef BORDER_ARRAY_TEST
+#define BORDER_ARRAY_TEST
+
+#include <cppunit/extensions/HelperMacros.h>
+
+class borderArrayTest : public CppUnit::TestFixture
+{
+ CPPUNIT_TEST_SUITE( borderArrayTest );
+ CPPUNIT_TEST( testBorderArray );
+ CPPUNIT_TEST_SUITE_END();
+
+public:
+ void setUp();
+ void tearDown();
+
+ void testBorderArray();
+};
+
+#endif // BORDER_ARRAY_TEST
diff --git a/alib2data/src/string/String.cpp b/alib2data/src/string/String.cpp
index 8fc68125b5..8d08af4309 100644
--- a/alib2data/src/string/String.cpp
+++ b/alib2data/src/string/String.cpp
@@ -11,11 +11,11 @@
namespace string {
-string::String symbolFrom( const alphabet::Symbol& symbol ) {
+string::String stringFrom( const alphabet::Symbol& symbol ) {
return string::String { string::LinearString { std::vector<alphabet::Symbol> { symbol } } };
}
-string::String symbolFrom( const std::string& string ) {
+string::String stringFrom( const std::string& string ) {
return string::String { string::LinearString { string } };
}
diff --git a/astringology2/src/astringology.cpp b/astringology2/src/astringology.cpp
index c5f9a00f34..ebb6e83b19 100644
--- a/astringology2/src/astringology.cpp
+++ b/astringology2/src/astringology.cpp
@@ -12,8 +12,11 @@
#include <exception/AlibException.h>
#include <string/String.h>
#include <automaton/Automaton.h>
+#include <container/Container.h>
+
#include <string/naive/ExactMatch.h>
#include <stringology/exact/ExactMatchingAutomaton.h>
+#include <stringology/exact/BorderArray.h>
int main(int argc, char* argv[]) {
try {
@@ -22,6 +25,7 @@ int main(int argc, char* argv[]) {
std::vector<std::string> allowed;
allowed.push_back("exactMatchingAutomaton");
allowed.push_back("exactMatch");
+ allowed.push_back("borderArray");
TCLAP::ValuesConstraint<std::string> allowedVals( allowed );
TCLAP::ValueArg<std::string> algorithm( "a", "algorithm", "Execute algorithm", false, "exactMatch", &allowedVals);
@@ -31,7 +35,7 @@ int main(int argc, char* argv[]) {
cmd.add( subject );
TCLAP::ValueArg<std::string> pattern( "p", "pattern", "Pattern string from file", false, "-", "file");
- cmd.add( pattern );
+ cmd.add( pattern );
cmd.parse(argc,argv);
@@ -68,6 +72,11 @@ int main(int argc, char* argv[]) {
automaton::Automaton automaton = stringology::exact::ExactMatchingAutomaton::construct(pattern);
alib::XmlDataFactory::toStdout(automaton);
return 0;
+ } else if( algorithm.getValue() == "borderArray") {
+ string::String subject = alib::XmlDataFactory::fromTokens<string::String>(subjectTokens);
+ std::vector<unsigned> borderArray = stringology::exact::BorderArray::construct(subject);
+ alib::XmlDataFactory::toStdout(borderArray);
+ return 0;
} else {
throw exception::AlibException( "Invalid algorithm" );
return 1;
--
GitLab