// Count how many tests are missing first_error information when it should be available.
static void
countMissingErrors(const SqlDatabase::TransactionPtr &tx, const Settings &settings) {
std::vector<std::string> args;
SqlDatabase::StatementPtr q = tx->statement("select count(*)"
" from test_results test"
" join attachments att on test.id = att.test_id" +
sqlWhereClause(tx, settings, args) + " and"
" test.first_error is null and"
" test.status <> 'end' and"
" att.name = 'Final output'");
sqlBindArgs(q, args);
int n = q->execute_int();
std::cout <<n <<"\n";
}
static void
callLSH(const SqlDatabase::TransactionPtr &tx, const std::string databaseName, double similarity_threshold, const string& Exec,
int norm, size_t hash_function_size, size_t hash_table_count)
{
double distance = sqrt((1. - similarity_threshold) * 50.);
double false_negative_rate = ( similarity_threshold != 1.0) ? 0.0100 : 0;
vector<CloneRange> ranges = computeranges(distance, 50, 100000);
int maxNumElementsInGroup = -1;
int maxNumElementIdx = -1;
// FIXME: We can't pass parameters to the exec'd process this way because the parent's SQL statements are
// being executed in a transaction -- they won't be visible in the child. [Robb P. Matzke 2013-08-12]
tx->execute("delete from detection_parameters");
tx->statement("insert into detection_parameters (similarity_threshold, false_negative_rate) values (?, ?)")
->bind(0, similarity_threshold)
->bind(1, false_negative_rate)
->execute();
map<size_t, int> groupSizes;
std::cout << "Looking for the biggest group" << std::endl;
for (size_t i = 0; i < ranges.size(); ++i) {
std::string sql = std::string("select count(*) from vectors where sum_of_counts >= ?") +
(ranges[i].high != -1 ? " and sum_of_counts <= ?" : "");
SqlDatabase::StatementPtr cmd = tx->statement(sql);
cmd->bind(0, ranges[i].low);
if (ranges[i].high != -1)
cmd->bind(1, ranges[i].high);
int numElementsInGroup = cmd->execute_int();
groupSizes[i] = numElementsInGroup;
std::cerr << "The current group from " << ranges[i].low << " to " << ranges[i].high
<< " is of size " << numElementsInGroup << std::endl;
if (numElementsInGroup > maxNumElementsInGroup) {
maxNumElementsInGroup = numElementsInGroup;
maxNumElementIdx = i;
}
}
std::cout << "Biggest group found " << ranges[maxNumElementIdx].low << " " << ranges[maxNumElementIdx].high << std::endl;
char tempDirName[] = "/tmp/paramdirXXXXXX";
char* mkdtempResult = mkdtemp(tempDirName);
if (!mkdtempResult) {
perror("mkdtemp: ");
exit (1);
}
string paramFileName = string(tempDirName) + "/params";
paramFileName = "/tmp/lshparamdirE40hF1/params";
std::cout << "Number of groups :" << ranges.size() << std::endl;
for (int i = 0; i < (int)ranges.size(); ++i) {
size_t group = (i == 0) ? maxNumElementIdx : (i <= maxNumElementIdx) ? i - 1 : i;
if (groupSizes[group] > 1) {
std::cout << "Executing LSH code low " << ranges[group].low
<< " high " << ranges[group].high << " group " << group << " size " << groupSizes[group] << std::endl;
if(norm == 3) {
executeLSHCode(tx, databaseName, Exec, paramFileName, ranges[group]);
} else {
executeLSHCodeLLNL(tx, databaseName, Exec, paramFileName, ranges[group], norm, similarity_threshold,
false_negative_rate, groupSizes[group]);
}
}
}
unlink(paramFileName.c_str());
rmdir(tempDirName);
}