SEXP
convertItemToR(zorba::Item item, bool asXMLSubNode)
{
SEXP ans = R_NilValue;
int np = 0;
zorba::Item type;
type = item.getType();
zorba::String typeName;
typeName = type.getStringValue();
char *endp;
// return(mkString(typeName.c_str()));
#ifdef R_DEBUG_XQUERY
fprintf(stderr, "Type: %s\n", typeName.c_str());
#endif
try {
if(item.isNull()) {
ans = R_NilValue;
// else if(item.isNaN())
// ans = R_NilValue;
}else if(item.isNode() && (( compareString(typeName, "untyped") ) || compareString(typeName, "anyType"))) { // !item.isAtomic()
// fprintf(stderr, "node: %s %s\n", typeName.c_str(), item.getStringValue().c_str());
ans = makeXMLNode(item);
} else if(compareString(typeName, "integer") || compareString(typeName, "int") || compareString(typeName, "short") || compareString(typeName, "unsignedShort"))
ans = ScalarInteger(R_strtod(item.getStringValue().c_str(), &endp));
else if(compareString(typeName, "string") || compareString(typeName, "normalizedString") || compareString(typeName, "untypedAtomic"))
ans = mkString(item.getStringValue().c_str());
else if(compareString(typeName, "decimal") || compareString(typeName, "double") || compareString(typeName, "float"))
ans = ScalarReal(R_strtod(item.getStringValue().c_str(), &endp));
else if(compareString(typeName, "boolean"))
ans = ScalarLogical(item.getBooleanValue());
else if(compareString(typeName, "dateTime"))
ans = convertItemToPOSIXct(item);
else if(compareString(typeName, "date"))
ans = convertItemToPOSIXct(item, "date");
else if(compareString(typeName, "time"))
ans = convertItemToPOSIXct(item, "time");
else if(compareString(typeName, "anyURI"))
ans = mkString(item.getStringValue().c_str());
else {
if(asXMLSubNode) {
ans = makeXMLTextNode(item);
} else {
fprintf(stderr, "unhandled type: %s. Processed as strings for now!\n", typeName.c_str());
ans = mkString(item.getStringValue().c_str());
}
// ans = R_NilValue;
}
} catch(zorba::SystemException &e) {
ans = mkString(item.getStringValue().c_str());
}
if(np)
UNPROTECT(np);
return(ans);
}
double Column::get_numeric_value() {
numeric_value = R_strtod(char_value, &end_ptr);
if ((*end_ptr != '\0') || (isnan(numeric_value))) {
return numeric_limits<double>::quiet_NaN();
}
return numeric_value;
}
void CorrectedPvalueColumn::out(ofstream& output_stream) throw (ofstream::failure) {
numeric_value = R_strtod(pvalue_column->char_value, &end_ptr);
if ((*end_ptr != '\0') || (ISNAN(numeric_value))) {
output_stream << "NA";
} else {
if (lambda > 1.0) {
numeric_value = Rf_pchisq((Rf_qchisq(numeric_value, 1, 0, 0) / lambda), 1, 0, 0);
}
output_stream << numeric_value;
}
}
void MetaNumeric::put(char* value) throw (MetaException) {
if (numeric) {
if (strcmp(na_marker, value) == 0) {
na_value = true;
na += 1;
value_saved = false;
this->value = numeric_limits<double>::quiet_NaN();
return;
}
else {
na_value = false;
}
d_value = R_strtod(value, &end_ptr);
if (*end_ptr != '\0') {
numeric = false;
value_saved = false;
this->value = numeric_limits<double>::quiet_NaN();
free(data);
data = NULL;
return;
}
if (isnan(d_value)) {
na_value = true;
na += 1;
value_saved = false;
this->value = numeric_limits<double>::quiet_NaN();
return;
}
n += 1;
if (n > current_heap_size) {
current_heap_size += Meta::HEAP_INCREMENT;
new_data = (double*)realloc(data, current_heap_size * sizeof(double));
if (new_data == NULL) {
free(data);
data = NULL;
throw MetaException("MetaNumeric", "put( char* )", __LINE__, 3, current_heap_size * sizeof(double));
}
data = new_data;
}
value_saved = true;
this->value = d_value;
data[n - 1] = d_value;
}
}
double Formatter::calculate_lambda(int& n_total, int& n_filtered) throw (FormatterException) {
Descriptor* descriptor = NULL;
char* line = NULL;
int line_length = 0;
unsigned int line_number = 2;
char data_separator = '\0';
vector<double>* snp_hq = NULL;
double maf_filter_value = -numeric_limits<double>::infinity();
double oevar_imp_filter_value = -numeric_limits<double>::infinity();
bool maf_filter = false;
bool oevar_imp_filter = false;
char* token = NULL;
char* pvalue_token = NULL;
char* maf_token = NULL;
char* oevar_imp_token = NULL;
int column_position = 0;
char* end_ptr = NULL;
double d_value = 0.0;
int n = 0;
double* data = NULL;
double* new_data = NULL;
int current_heap_size = HEAP_SIZE;
double lambda = numeric_limits<double>::quiet_NaN();
n_total = 0;
n_filtered = 0;
if ((gwafile == NULL) || (pvalue_column_pos < 0)) {
return lambda;
}
try {
descriptor = gwafile->get_descriptor();
snp_hq = descriptor->get_threshold(Descriptor::SNP_HQ);
data_separator = gwafile->get_data_separator();
if (((maf_filter_value = snp_hq->at(0)) > 0) && (maf_column_pos >= 0)){
maf_filter = true;
}
if (((oevar_imp_filter_value = snp_hq->at(1)) > 0) && (oevar_imp_column_pos >= 0)) {
oevar_imp_filter = true;
}
data = (double*)malloc(HEAP_SIZE * sizeof(double));
if (data == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 2, HEAP_SIZE * sizeof(double));
}
if (maf_filter) {
if (oevar_imp_filter) {
// all filters
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
column_position = 0;
pvalue_token = NULL;
maf_token = NULL;
oevar_imp_token = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == pvalue_column_pos) {
auxiliary::trim(&token);
pvalue_token = token;
} else if (column_position == maf_column_pos) {
auxiliary::trim(&token);
maf_token = token;
} else if (column_position == oevar_imp_column_pos) {
auxiliary::trim(&token);
oevar_imp_token = token;
}
token = auxiliary::strtok(&line, data_separator);
column_position += 1;
}
if (pvalue_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, pvalue_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
if (maf_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, maf_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
if (oevar_imp_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, oevar_imp_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
d_value = R_strtod(maf_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
d_value = d_value > 0.5 ? 1.0 - d_value : d_value;
if (auxiliary::fcmp(d_value, maf_filter_value, EPSILON) == 1) {
d_value = R_strtod(oevar_imp_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
if (auxiliary::fcmp(d_value, oevar_imp_filter_value, EPSILON) == 1) {
d_value = R_strtod(pvalue_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
n += 1;
if (n > current_heap_size) {
current_heap_size += HEAP_INCREMENT;
new_data = (double*)realloc(data, current_heap_size * sizeof(double));
if (new_data == NULL) {
free(data);
data = NULL;
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 2, current_heap_size * sizeof(double));
}
data = new_data;
}
data[n - 1] = d_value;
}
}
line_number += 1;
}
} else {
// only maf filter
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
column_position = 0;
pvalue_token = NULL;
maf_token = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == pvalue_column_pos) {
auxiliary::trim(&token);
pvalue_token = token;
} else if (column_position == maf_column_pos) {
auxiliary::trim(&token);
maf_token = token;
}
token = auxiliary::strtok(&line, data_separator);
column_position += 1;
}
if (pvalue_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, pvalue_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
if (maf_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, maf_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
d_value = R_strtod(maf_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
d_value = d_value > 0.5 ? 1.0 - d_value : d_value;
if (auxiliary::fcmp(d_value, maf_filter_value, EPSILON) == 1) {
d_value = R_strtod(pvalue_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
n += 1;
if (n > current_heap_size) {
current_heap_size += HEAP_INCREMENT;
new_data = (double*)realloc(data, current_heap_size * sizeof(double));
if (new_data == NULL) {
free(data);
data = NULL;
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 2, current_heap_size * sizeof(double));
}
data = new_data;
}
data[n - 1] = d_value;
}
line_number += 1;
}
}
} else if (oevar_imp_filter) {
// only oevar_imp_filter
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
column_position = 0;
pvalue_token = NULL;
oevar_imp_token = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == pvalue_column_pos) {
auxiliary::trim(&token);
pvalue_token = token;
} else if (column_position == oevar_imp_column_pos) {
auxiliary::trim(&token);
oevar_imp_token = token;
}
token = auxiliary::strtok(&line, data_separator);
column_position += 1;
}
if (pvalue_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, pvalue_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
if (oevar_imp_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, oevar_imp_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
d_value = R_strtod(oevar_imp_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
if (auxiliary::fcmp(d_value, oevar_imp_filter_value, EPSILON) == 1) {
d_value = R_strtod(pvalue_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
n += 1;
if (n > current_heap_size) {
current_heap_size += HEAP_INCREMENT;
new_data = (double*)realloc(data, current_heap_size * sizeof(double));
if (new_data == NULL) {
free(data);
data = NULL;
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 2, current_heap_size * sizeof(double));
}
data = new_data;
}
data[n - 1] = d_value;
}
line_number += 1;
}
} else {
// no filters
while ((line_length = reader.read_line()) > 0) {
line = *reader.line;
column_position = 0;
pvalue_token = NULL;
token = auxiliary::strtok(&line, data_separator);
while (token != NULL) {
if (column_position == pvalue_column_pos) {
auxiliary::trim(&token);
pvalue_token = token;
break;
}
token = auxiliary::strtok(&line, data_separator);
column_position += 1;
}
if (pvalue_token == NULL) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 8, pvalue_column->get_header(), line_number, gwafile->get_descriptor()->get_name());
}
d_value = R_strtod(pvalue_token, &end_ptr);
if ((*end_ptr != '\0') || (isnan(d_value))) {
line_number += 1;
continue;
}
n += 1;
if (n > current_heap_size) {
current_heap_size += HEAP_INCREMENT;
new_data = (double*)realloc(data, current_heap_size * sizeof(double));
if (new_data == NULL) {
free(data);
data = NULL;
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 2, current_heap_size * sizeof(double));
}
data = new_data;
}
data[n - 1] = d_value;
line_number += 1;
}
}
n_filtered = n;
n_total = line_number - 2;
if (line_length == 0) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 7, line_number, gwafile->get_descriptor()->get_name());
}
reader.reset();
if (reader.read_line() <= 0) {
throw FormatterException("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 5, line_number, gwafile->get_descriptor()->get_name());
}
} catch (DescriptorException &e) {
FormatterException new_e(e);
new_e.add_message("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 9, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (ReaderException &e) {
FormatterException new_e(e);
new_e.add_message("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 9, gwafile->get_descriptor()->get_name());
throw new_e;
} catch (FormatterException &e) {
e.add_message("Formatter", "double calculate_lambda( int& , int& )", __LINE__, 9, gwafile->get_descriptor()->get_name());
throw;
}
if (n > 0) {
for (int i = 0; i < n; i++) {
data[i] = pow(Rf_qnorm5(0.5 * data[i], 0.0, 1.0, 0, 0), 2.0);
}
qsort(data, n, sizeof(double), auxiliary::dblcmp);
lambda = auxiliary::stats_median_from_sorted_data(data, n) / Rf_qchisq(0.5, 1.0, 0, 0);
}
free(data);
data = NULL;
return lambda;
}