int scrabble_search_test(const char* filename)
{
long result = errors_counter();
trivial_scrabble_test();
exhaustive_scrabble_test(filename);
return errors_counter() - result;
}
int exhaustive_scrabble_test(const char* filename)
{
long result = errors_counter();
// note that dictsize*keys*3 tests are performed
size_t DICTSIZE = 10000, KEYS = 10;
typedef std::vector<std::string> Dictionary;
extern size_t fill_dictionary(const char*, Dictionary&, size_t, size_t = 0);
Dictionary dictionary;
size_t longest_in_file = fill_dictionary(filename, dictionary, DICTSIZE);
std::random_shuffle(dictionary.begin(), dictionary.end());
ScrabbleTst names;
Dictionary::iterator dit;
for (dit = dictionary.begin(); dit != dictionary.end(); dit += 4) {
const std::string& name(*dit);
names[name] = name.c_str();
}
int skip = dictionary.size() / KEYS;
for(dit = dictionary.begin() + skip; dit != dictionary.end(); dit += skip)
{
std::string name(*dit);
std::sort(name.begin(), name.end());
for (int jokercount = 0; jokercount < 3; ++jokercount) {
ScrabbleTst::search_results_list matchresults = names.create_search_results();
names.combinatorial_search(name, std::back_inserter(matchresults), jokercount);
//if (matchresults.size() == 0)
// std::cout << "couldn't find " << name << '\n';
BOOST_CHECK(matchresults.size() != 0);
ScrabbleTst matches;
ScrabbleTst matchcount;
size_t prevsize = 0;
for (size_t i = 0; i != matchresults.size(); ++i) {
std::string mcopy(matchresults[i].key());
matchcount[mcopy] = mcopy.c_str();
if (matchcount.size() > prevsize) {
BOOST_CHECK(check_scrabble_found(name, mcopy, jokercount));
++prevsize;
}
else {
std::cout << "scrabble found duplicate " << mcopy << "\n";
}
std::sort(mcopy.begin(), mcopy.end());
matches[mcopy] = mcopy.c_str();
}
// check for duplicates
scrabble_findcount += matchresults.size();
scrabble_duplicates += matchresults.size() - matchcount.size();
BOOST_CHECK(matchcount.size() == matchresults.size());
BOOST_CHECK(check_scrabble_missed(names, matches, jokercount));
}
}
//std::cout << "scrabble duplicate matches " << scrabble_duplicates << " of " << scrabble_findcount << "\n";
return errors_counter() - result;
}
int trivial_scrabble_test()
{
long result = errors_counter();
typedef containers::ternary_tree<std::string, const char*> Tst;
Tst names;
names["ABCD"] = "ABCD";
names["Abcd"] = "Abcd";
names["ABcd"] = "ABcd";
names["aBCd"] = "aBCd";
names["abc"] = "abc";
names["abcde"] = "abcde";
names["bcd"] = "bcd";
names["abCD"] = "abCD";
names["abCd"] = "abCd";
names["AbcD"] = "AbcD";
names["ABcD"] = "ABcD";
names["aBCD"] = "aBCD";
names["abCDE"] = "abCDE";
names["abCDd"] = "abCDd";
names["abCcd"] = "abCcd";
names["bcdc"] = "bcdc";
names["aab"] = "aab";
Tst::search_results_list matches = names.create_search_results();
typedef Tst::search_results_list::iterator ResultIter;
int i = 0;
std::string s("abcde");
names.combinatorial_search(s, std::back_inserter(matches));
BOOST_CHECK(matches.size() == 3);
matches.clear();
std::random_shuffle(s.begin(), s.end());
names.combinatorial_search(s, std::back_inserter(matches));
BOOST_CHECK(matches.size() == 3);
// with wildcards
matches.clear();
std::random_shuffle(s.begin(), s.end());
names.combinatorial_search(s, std::back_inserter(matches), 1);
BOOST_CHECK(matches.size() == 8);
//for (ResultIter rit = matches.begin(); rit != matches.end(); ++rit, ++i)
// std::cout << i << ": " << **rit << "\n";
matches.clear();
s.append(s);
std::random_shuffle(s.begin(), s.end());
names.combinatorial_search(s, std::back_inserter(matches));
BOOST_CHECK(matches.size() == 5);
//for (ResultIter rit = matches.begin(); rit != matches.end(); ++rit, ++i)
// std::cout << i << ": " << **rit << "\n";
return errors_counter() - result;
}
int check_empty(containers::ternary_tree<std::basic_string<CharT>, DataT>& tst)
{
typedef containers::ternary_tree<std::basic_string<CharT>, DataT> Tst;
long result = errors_counter();
BOOST_CHECK(tst.empty());
BOOST_CHECK(tst.node_count() == 0);
BOOST_CHECK(tst.item_count() == 0);
Tst::iterator i2(tst.begin());
BOOST_CHECK(i2 == tst.end());
return errors_counter() - result;
}
int iteration_quickcheck(Tst& tst, size_t expectcount,
CharT* unexpected = 0)
{
long result = errors_counter();
size_t count = 0;
Tst::iterator first = tst.begin();
Tst::iterator last = tst.end();
const char *last_value = 0;
while(first != last) {
if (first.key() != *first)
std::cout << first.key() << " != " << *first << '\n';
BOOST_CHECK(first.key() == *first);
last_value = first.value();
if (unexpected)
BOOST_CHECK(first.key() != unexpected);
++first;
count++;
}
BOOST_CHECK(first == tst.end());
if (count != expectcount)
std::cout << count << " < " << expectcount << '\n';
BOOST_CHECK(count == expectcount);
count = 0;
first = --tst.end();
Tst::const_reverse_iterator rit = tst.rbegin();
BOOST_CHECK(*first == last_value);
last = tst.end();
while(first != last) {
if (first.key() != *first)
std::cout << first.key() << " != " << *first << '\n';
if (*first != *rit)
std::cout << *first << " != " << *rit << '\n';
BOOST_CHECK(first.key() == *first);
BOOST_CHECK(*first == *rit);
if (unexpected)
BOOST_CHECK(first.key() != unexpected);
--first;
++rit;
count++;
}
BOOST_CHECK(rit == const_cast<const Tst&>(tst).rend());
if (count != expectcount)
std::cout << count << " < " << expectcount << '\n';
return errors_counter() - result;
}
inline void
report_error( const char* msg, const char* file, int line, const_string func_name, bool is_msg = false )
{
++errors_counter();
std::cerr << file << "(" << line << "): ";
if( is_msg )
std::cerr << msg;
else
std::cerr << "test " << msg << " failed";
if( func_name != "(unknown)" )
std::cerr << " in function: '" << func_name << "'";
std::cerr << std::endl;
}
int levenshtein_search_test()
{
long result = errors_counter();
// Create a tree with known levenshtein count = 100
std::string b0("abcdefjklmn");
int testcount = 3;
while ( testcount ) {
std::random_shuffle(b0.begin(), b0.end());
int baselen = (int)b0.size();
typedef containers::ternary_tree<std::string, std::string> Tst;
Tst names;
int L1count = 0, L2count = 0, L3count = 0, L4count = 0;
for (int i = 0; i < baselen; i++)
{
std::string s1(b0);
mutate(s1, i);
if (names.insert(std::make_pair(s1, s1)).second)
L1count++;
for (int j = 0; j < (int)s1.size(); j++)
{
if (j == i)
continue;
std::string s2(s1);
mutate(s2, j);
if (names.insert(std::make_pair(s2, s2)).second)
L2count++;
for (int k = 0; k < (int)s2.size(); k++)
{
if (abs(k - j) <= 1 || abs(k - i) <= 1)
continue;
std::string s3(s2);
mutate(s3, k);
//if (s3.size() < s2.size() && s3.size() == s1.size())
// continue;
//if (names.insert(std::make_pair(s3, s3)).second)
// L3count++;
for (int m = 2; m < (int)s3.size() - 2; m++)
{
if (abs(m - j) <= 2 || abs(m - i) <= 2 || abs(m - k) <= 2)
continue;
std::string s4(s3);
mutate(s4, m);
if (m + 1 >= (int)s4.size())
continue;
mutate(s4, m + 1);
if (m - 1 < 0)
continue;
mutate(s4, m - 1);
if (m + 2 >= (int)s4.size())
continue;
mutate(s4, m + 2);
if (m - 2 < 0)
continue;
mutate(s4, m - 2);
if (abs((int)s4.size() - (int)b0.size()) < 2) {
s4.insert(rand()%s4.size(), 1, 'Z');
s4.insert(rand()%s4.size(), 1, 'Z');
}
if (names.insert(std::make_pair(s4, s4)).second)
L4count++;
}
}
}
}
if (names.size() != L1count + L2count + L3count + L4count) {
std::cout << "levenshtein tree name collision, try again";
continue;
}
Tst::search_results_list results = names.create_search_results();
typedef Tst::search_results_list::iterator ResultsIter;
names.levenshtein_search(b0, std::back_inserter(results), 2);
if (results.size() != L1count + L2count + L3count)
{
Tst found;
for (ResultsIter rit = results.begin(); rit != results.end(); ++rit)
found[**rit] = **rit;
BOOST_CHECK(found.size() == L1count + L2count); // + L3count);
std::cout << "\nSearch " << b0 //<< ", L-count: " << names.levenshtein_search_count
<< ", found " << results.size() << " strings";
std::cout << " (" << found.size() << " in " << found.node_count() << " nodes)\n";
if (found.size() != L1count + L2count) { // + L3count) {
Tst::iterator fit = found.begin();
for (Tst::iterator nit = names.begin(); nit != names.end(); ++nit)
{
if (fit != found.end() && *nit == *fit) {
std::cout << *nit << "\n";
++fit;
}
else
std::cout << *nit << " ###\n";
}
// std::cout << "full-frontal:\n";
// for (ResultsIter rit = results.begin(); rit != results.end(); ++rit)
// std::cout << **rit << "\n";
}
}
testcount--;
}
return errors_counter() - result;
}
int hamming_search_test(const char* filename)
{
long result = errors_counter();
typedef containers::ternary_tree<std::string, const char*> Tst;
Tst names;
names["ABCD"] = "ABCD";
names["Abcd"] = "Abcd";
names["ABcd"] = "ABcd";
names["aBCd"] = "aBCd";
names["abc"] = "abc";
names["abcde"] = "abcde";
names["bcd"] = "bcd";
names["abCD"] = "abCD";
names["abCd"] = "abCd";
names["AbcD"] = "AbcD";
names["ABcD"] = "ABcD";
names["aBCD"] = "aBCD";
names["abCDE"] = "abCDE";
names["abCDd"] = "abCDd";
names["abCcd"] = "abCcd";
names["bcdc"] = "bcdc";
names["aab"] = "aab";
// hamming_search(2): ABcd, AbcD, Abcd, aBCd, abCD, abCd, abc, abcde
// levenshtein_search(2): hamming + abCDd, abCcd bcd, bcdc
Tst::search_results_list matches = names.create_search_results();
typedef Tst::search_results_list::iterator ResultIter;
//names.levenshtein_search_count = 0;
names.levenshtein_search("abcd", std::back_inserter(matches), 2);
//std::cout << "l-count: " << names.levenshtein_search_count << "\n";
//names.levenshtein_search_count = 0;
BOOST_CHECK(matches.size() == 12);
if (matches.size() != 12)
{
// compare with DDJ
{
Tptr root = 0;
for (Tst::iterator it = names.begin(); it != names.end(); ++it) {
insertstr = (char*)*it;
root = insert2(root, insertstr);
}
nearsearch(root, "abcd", 2);
std::cout << "DDJ nearsearch abcd:\n";
for (int i = 0; i != srchtop; i++)
std::cout << srcharr[i] << "\n";
std::cout << "\n";
cleanup2(root);
}
int i = 0;
for (ResultIter rit = matches.begin(); rit != matches.end(); ++rit, ++i) {
std::cout << **rit << " = " << (*rit).key();
std::cout /*<< " = " << matches[i].key()*/ << "\n";
}
}
typedef std::vector<std::string> Dictionary;
extern size_t fill_dictionary(const char*, Dictionary&, size_t, size_t = 0);
Dictionary dictionary;
size_t longest_in_file = fill_dictionary(filename, dictionary, 300);
std::random_shuffle(dictionary.begin(), dictionary.end());
names.clear();
Dictionary wildnames;
// Add names with mini-variations
long zeroes = 0;
Dictionary::iterator dit;
for (dit = dictionary.begin(); dit != dictionary.end(); ++dit) {
const std::string& name(*dit);
std::string namecopy(*dit);
names[namecopy] = name.c_str();
std::string searchstr(name);
for (int i = 0; i < 5; ++i) {
int where = rand() % name.size();
// make string with ? at place of changed char
if (searchstr[where] == '?')
continue;
searchstr[where] = '?';
wildnames.push_back(searchstr);
char c = (char)rand();
if (!c) zeroes++;
namecopy[where] = c;
names[namecopy] = name.c_str();
}
}
for(dit = wildnames.begin(); dit != wildnames.end(); ++dit) {
const std::string& name(*dit);
for (int diff = 1; diff != 3; ++diff) {
Tst::search_results_list matchresults = names.create_search_results();
names.hamming_search(name, std::back_inserter(matchresults), diff);
//if (matchresults.size() == 0)
// std::cout << "couldn't find " << name << '\n';
//BOOST_CHECK(found == matchresults.size());
for (size_t i = 0; i != matchresults.size(); ++i)
BOOST_CHECK(check_hamming_match(matchresults[i].key(), name, diff));
}
}
return errors_counter() - result;
}
int basic_insertion_test()
{
long result = errors_counter();
typedef containers::ternary_tree<std::basic_string<CharT>, int> Tst;
Tst tst;
const char* strings[] = { "\0", "aa", "aab", "aac", "add", "aee", "bab" };
widen<CharT> w;
// Construction postconditions
check_empty(tst);
// Standard insert()
tst.insert(std::make_pair(w(strings[1]), 1));
tst.insert(std::make_pair(w(strings[5]), 5));
std::pair<std::basic_string<CharT>, int> inval(w(strings[3]), 3);
std::pair<Tst::iterator, bool> vr = tst.insert(inval);
BOOST_CHECK(vr.first != tst.end());
BOOST_CHECK(vr.second == true);
vr = tst.insert(inval);
BOOST_CHECK(vr.first != tst.end());
BOOST_CHECK(vr.second == false);
BOOST_CHECK((*tst.find(w(strings[1]))) == 1);
BOOST_CHECK((*tst.find(w(strings[3]))) == 3);
BOOST_CHECK((*tst.find(w(strings[5]))) == 5);
BOOST_CHECK(tst.item_count() == 3);
BOOST_CHECK(tst.total_key_length() == 3*3-1);
// In present version, no reassignment
tst.insert(std::make_pair(w(strings[1]), 4));
BOOST_CHECK((*tst.find(w(strings[1]))) == 1);
BOOST_CHECK(tst.item_count() == 3);
BOOST_CHECK(tst.total_key_length() == 3*3-1);
// reference_proxy insert()
tst[w(strings[2])] = 2;
tst[w(strings[6])] = 6;
tst[w(strings[4])] = 4;
BOOST_CHECK(tst[w(strings[2])] == 2);
BOOST_CHECK(tst[w(strings[4])] == 4);
BOOST_CHECK(tst[w(strings[6])] == 6);
BOOST_CHECK(tst.item_count() == 6);
BOOST_CHECK(tst.total_key_length() == 6*3-1);
// Now do reassignment
tst[w(strings[1])] = 7;
BOOST_CHECK((*tst.find(w(strings[1]))) == 7); //1);
tst[w(strings[1])] = 1;
// iterator used to check sort order of strings
Tst::const_iterator it(tst.begin());
int expect_val = 1;
while (it != tst.end()) {
Tst::value_type val = *it;
// string == "a", "b" etc
BOOST_CHECK(expect_val == 1 || it.key().size() == 3);
BOOST_CHECK(it.key() == w(strings[expect_val]));
// value is 1, 2 etc
BOOST_CHECK(val == expect_val++);
++it;
}
BOOST_CHECK(expect_val == int(tst.item_count() + 1));
// Test clear() postconditions
tst.clear();
check_empty(tst);
return errors_counter() - result;
}