/// <summary>
/// Try to evaluate the given expression in the context of the variable.
/// Imagine the variable is a class and the expression is in a method of that class.
/// Member variables can be accessed and take precedence over global variables of the same name.
/// Globals can be accessed as well. Returns a variable representing the result of the evaluation
/// or null if the expression can't be evaluated.
/// </summary>
Task <IVariableInformation> EvaluateExpressionInVariableScopeAsync(
IVariableInformation variable, VsExpression vsExpression, string displayName)
{
if (variable.IsPointer || variable.IsReference)
{
variable = variable.Dereference();
}
return(variable.EvaluateExpressionAsync(displayName, vsExpression));
}
/// <summary>
/// Declare a variable in using the given variable scope to execute the value expression.
/// Token replacement using scopedNames is done against both the variable name and the
/// value expression.
/// </summary>
/// <exception cref="ExpressionEvaluationFailed">
/// Expression to declare the variable failed to evaluate.
/// </exception>
public async Task DeclareVariableAsync(IVariableInformation variable, string variableName,
string valueExpression, NatvisScope natvisScope)
{
string scratchVar =
ReplaceScopedNames(variableName, natvisScope?.ScopedNames, out bool ignore);
VsExpression vsExpression = _vsExpressionCreator.Create(valueExpression, "")
.MapValue(e => ReplaceScopedNames(e, natvisScope?.ScopedNames, out ignore));
// Declare variable and return it. Pure declaration expressions will always return
// error because these expressions don't return a valid value.
VsExpression createExpression =
vsExpression.MapValue(e => $"auto {scratchVar}={e}; {scratchVar}");
if (variable.IsPointer || variable.IsReference)
{
variable = variable.Dereference();
if (variable == null)
{
string failMsg = $"Failed to dereference pointer: Name: {variableName}";
_logger.Error(failMsg);
throw new ExpressionEvaluationFailed(failMsg);
}
}
// TODO: Split the logic for LLDB and lldb-eval. Currently, LLDB is always
// used to create a scratch variable (even if lldb-eval is the selected engine).
IVariableInformation result =
await variable.EvaluateExpressionAsync(variableName, createExpression);
if (result != null && !result.Error && natvisScope != null)
{
// Result of 'auto {scratchVar}={e}; {scratchVar}' creates a copy of the scratch
// variable. Evaluating '{scratchVar}' returns the reference to the original
// variable. By using the original variable we make sure that the we always use its
// up-to-date value.
// TODO: Use RemoteFrame.FindValue to get the scratch variable.
// EvaluateExpression method already is optimised for the case of fetching scratch
// variables, but it isn't a convenient one.
result = await variable.EvaluateExpressionAsync(
variableName, _vsExpressionCreator.Create($"{scratchVar}", ""));
if (result != null && !result.Error)
{
natvisScope.AddContextVariable(scratchVar, result.GetRemoteValue());
return;
}
}
string msg = $"Failed to declare variable: Name: {variableName}, " +
$"Expression: {valueExpression}";
string resultMessage = result?.ErrorMessage;
if (!string.IsNullOrEmpty(resultMessage))
{
msg += $", Info: {{{resultMessage}}}";
}
_logger.Error(msg);
throw new ExpressionEvaluationFailed(msg);
}