void
Value::Read(ods::Ns &ns, ods::Attrs &attrs)
{
DeleteData();
auto *type_attr = attrs.Get(ns.office(), ods::ns::kValueType);
if (type_attr == nullptr)
{ // shouldn't happen
type_ = ods::Type::Fail;
return;
}
type_ = ods::TypeFromString(type_attr->value());
auto *value_attr = attrs.Get(ns.office(), ods::ns::kValue);
if (value_attr == nullptr)
{
type_ = ods::Type::Fail;
return;
}
if (IsDouble() || IsPercentage() || IsCurrency())
{
double num;
if (!value_attr->ToDouble(num))
{
mtl_warn("ToDouble()");
return;
}
set(new double(num), type_);
} else if (IsString()) {
set(new QString(value_attr->value()), ods::Type::String);
} else if (IsDate()) {
auto *custom_attr = attrs.Get(ns.office(), ods::ns::kDateValue);
if (custom_attr == nullptr)
{
mtl_warn("custom_attr == nullptr");
return;
}
auto dt = QDateTime::fromString(custom_attr->value(), Qt::ISODate);
set(new QDateTime(dt), type_);
} else if (IsDuration()) {
auto *custom_attr = attrs.Get(ns.office(), ods::ns::kTimeValue);
if (custom_attr == nullptr)
{
mtl_warn("custom_attr == nullptr");
return;
}
auto *t = new ods::Duration();
t->Decode(custom_attr->value());
set(t, type_);
} else {
type_ = ods::Type::NotSet;
}
}
Substyle::~Substyle()
{
if (IsNotSet())
return;
if (IsCurrency())
delete AsCurrency();
else if (IsDate())
delete AsDate();
else if (IsDuration())
delete AsDuration();
else if (IsPercent())
delete AsPercent();
}
QString
Value::toString() const
{
if (!Ok())
return "";
if (IsDouble() || IsPercentage())
return QString::number(*AsDouble());
if (IsString())
return *AsString();
if (IsDate())
return AsDate()->toString(Qt::ISODate);
if (IsCurrency())
mtl_qline("Currency not supported yet");
if (IsDuration())
return AsDuration()->ToString();
return "";
}
void
Value::CopyTo(ods::Value &v)
{
v.type_set(type_);
if (NoValue())
return;
if (IsDouble())
v.SetDouble(*AsDouble());
else if (IsString())
v.SetString(*AsString());
else if (IsCurrency())
v.SetCurrency(*AsCurrency());
else if (IsPercentage())
v.SetPercentage(*AsPercentage());
else if (IsDate())
v.SetDate(*AsDate());
else if (IsDuration())
v.SetDuration(*AsDuration());
else
mtl_warn("Not implemented");
}
void CScraperPreprocessor::PreprocessMonetaryString(CString *monetary_string)
{
// 1) Commas in numbers: "$3,45" should be raplaced by "$3.45".
// And "$3,456.78" should be replaced by "$3.456.78"
// 2) Dot as delimter-symbol in numbers: "$3.456.78".
// Dots followed by three (or more) digits should be removed: "$3456.78"
// 3) Spaces inside numbers: somebody reported blinds like "$40 000" in MTTs.
// They also should be removed.
// 4) Foreign-language money-signs like Euro € should be replaced by dollars.
int length = monetary_string->GetLength();
CString result;
// A small technical trick: append 3 spaces,
// so we don't have to check for index errors if we investigate the next 3 characters
*monetary_string += " ";
// Special handling required for the first character
char first_character = monetary_string->GetAt(0);
char second_character = monetary_string->GetAt(1);
if (!IsCurrency(first_character) || !isdigit(second_character))
{
result += first_character;
}
// Now caring about the rest, starting from 2nd character
int second_position = 1;
int last_position = length - 1;
char ith_character = ' ';
char previous_character = ' ';
char next_character = ' ';
for (int i=second_position; i<=last_position; i++)
{
ith_character = monetary_string->GetAt(i);
next_character = monetary_string->GetAt(i+1);
if (IsCurrency(ith_character))
{
// Keep them at the moment, but replace foreign currencys by dollars.
// The scraper will deal with them later
result += '$';
}
else if ((ith_character == '.') || (ith_character == ','))
{
// Accept commans and dots outside numbers
if (!isdigit(previous_character) || !isdigit(next_character))
{
result += ith_character;
}
else
{
// Comma or dot inside number
// If only the following 2 digits are numbers then we have a fractional-dot.
// But if the 3rd digit is a number too then it is only a delimiter-dot
// (for better readability) that has to be ignored.
char third_next_character = monetary_string->GetAt(i+3);
if (!isdigit(third_next_character))
{
// Append the fractional-dot (implicitly repalcing comma if necessary)
result += '.';
}
}
}
else if (ith_character == ' ')
{
// Spaces inside numbers have to be ignored
// Up to now we saw it only for very large blinds of SNGs / MTTs,
// for example "$40 000".
if (isdigit(previous_character)
&& next_character == '0'
&& monetary_string->GetAt(i+2) == '0'
&& monetary_string->GetAt(i+3) == '0')
{
// Ignore it
}
else
{
result += ' ';
}
}
else
{
// Everything else
// Keep it
result += ith_character;
}
previous_character = ith_character;
}
*monetary_string = result;
write_log(preferences.debug_scraper_preprocessor(), "Monetary string after preprocessing: %s\n", result);
}
