4.7 KiB
String conversions
Contents
operator << overload for std::ostream
Catch::StringMaker specialisation
Catch::is_range specialisation
Exceptions
Enums
Floating point precision
Catch needs to be able to convert types you use in assertions and logging expressions into strings (for logging and reporting purposes). Most built-in or std types are supported out of the box but there are two ways that you can tell Catch how to convert your own types (or other, third-party types) into strings.
operator << overload for std::ostream
This is the standard way of providing string conversions in C++ - and the chances are you may already provide this for your own purposes. If you're not familiar with this idiom it involves writing a free function of the form:
std::ostream& operator << ( std::ostream& os, T const& value ) {
os << convertMyTypeToString( value );
return os;
}
(where T
is your type and convertMyTypeToString
is where you'll write whatever code is necessary to make your type printable - it doesn't have to be in another function).
You should put this function in the same namespace as your type, or the global namespace, and have it declared before including Catch's header.
Catch::StringMaker specialisation
If you don't want to provide an operator <<
overload, or you want to convert your type differently for testing purposes, you can provide a specialization for Catch::StringMaker<T>
:
namespace Catch {
template<>
struct StringMaker<T> {
static std::string convert( T const& value ) {
return convertMyTypeToString( value );
}
};
}
Catch::is_range specialisation
As a fallback, Catch attempts to detect if the type can be iterated
(begin(T)
and end(T)
are valid) and if it can be, it is stringified
as a range. For certain types this can lead to infinite recursion, so
it can be disabled by specializing Catch::is_range
like so:
namespace Catch {
template<>
struct is_range<T> {
static const bool value = false;
};
}
Exceptions
By default all exceptions deriving from std::exception
will be translated to strings by calling the what()
method. For exception types that do not derive from std::exception
- or if what()
does not return a suitable string - use CATCH_TRANSLATE_EXCEPTION
. This defines a function that takes your exception type, by reference, and returns a string. It can appear anywhere in the code - it doesn't have to be in the same translation unit. For example:
CATCH_TRANSLATE_EXCEPTION( MyType const& ex ) {
return ex.message();
}
Enums
Introduced in Catch2 2.8.0.
Enums that already have a <<
overload for std::ostream
will convert to strings as expected.
If you only need to convert enums to strings for test reporting purposes you can provide a StringMaker
specialisations as any other type.
However, as a convenience, Catch provides the REGISTER_ENUM
helper macro that will generate the StringMaker
specialisation for you with minimal code.
Simply provide it the (qualified) enum name, followed by all the enum values, and you're done!
E.g.
enum class Fruits { Banana, Apple, Mango };
CATCH_REGISTER_ENUM( Fruits, Fruits::Banana, Fruits::Apple, Fruits::Mango )
TEST_CASE() {
REQUIRE( Fruits::Mango == Fruits::Apple );
}
... or if the enum is in a namespace:
namespace Bikeshed {
enum class Colours { Red, Green, Blue };
}
// Important!: This macro must appear at top level scope - not inside a namespace
// You can fully qualify the names, or use a using if you prefer
CATCH_REGISTER_ENUM( Bikeshed::Colours,
Bikeshed::Colours::Red,
Bikeshed::Colours::Green,
Bikeshed::Colours::Blue )
TEST_CASE() {
REQUIRE( Bikeshed::Colours::Red == Bikeshed::Colours::Blue );
}
Floating point precision
Introduced in Catch2 2.8.0.
Catch provides a built-in StringMaker
specialization for both float
and double
. By default, it uses what we think is a reasonable precision,
but you can customize it by modifying the precision
static variable
inside the StringMaker
specialization, like so:
Catch::StringMaker<float>::precision = 15;
const float testFloat1 = 1.12345678901234567899f;
const float testFloat2 = 1.12345678991234567899f;
REQUIRE(testFloat1 == testFloat2);
This assertion will fail and print out the testFloat1
and testFloat2
to 15 decimal places.