mosesdecoder/util/tokenize_piece_test.cc
Kenneth Heafield 974a708ddd Updated kenlm 96ef3f2c11.
Invalidates old gcc and 32-bit formats, replacing these with one consistent format: 64-bit new gcc.  Backwards compatible with these files.
2011-11-17 19:12:19 +00:00

49 lines
1.1 KiB
C++

#include "util/tokenize_piece.hh"
#include "util/string_piece.hh"
#define BOOST_TEST_MODULE TokenIteratorTest
#include <boost/test/unit_test.hpp>
#include <iostream>
namespace util {
namespace {
BOOST_AUTO_TEST_CASE(pipe_pipe_none) {
const char str[] = "nodelimit at all";
TokenIter<MultiCharacter> it(str, MultiCharacter("|||"));
BOOST_REQUIRE(it);
BOOST_CHECK_EQUAL(StringPiece(str), *it);
++it;
BOOST_CHECK(!it);
}
BOOST_AUTO_TEST_CASE(pipe_pipe_two) {
const char str[] = "|||";
TokenIter<MultiCharacter> it(str, MultiCharacter("|||"));
BOOST_REQUIRE(it);
BOOST_CHECK_EQUAL(StringPiece(), *it);
++it;
BOOST_REQUIRE(it);
BOOST_CHECK_EQUAL(StringPiece(), *it);
++it;
BOOST_CHECK(!it);
}
BOOST_AUTO_TEST_CASE(remove_empty) {
const char str[] = "|||";
TokenIter<MultiCharacter, true> it(str, MultiCharacter("|||"));
BOOST_CHECK(!it);
}
BOOST_AUTO_TEST_CASE(remove_empty_keep) {
const char str[] = " |||";
TokenIter<MultiCharacter, true> it(str, MultiCharacter("|||"));
BOOST_REQUIRE(it);
BOOST_CHECK_EQUAL(StringPiece(" "), *it);
++it;
BOOST_CHECK(!it);
}
} // namespace
} // namespace util