Problem in PyArg_ParseTuple

[varunc]

Hi,

I am writing an extension module in C++. I am facing a problem with PyArg_ParseTuple. Here is the relevant portion of my code:
TInt type = -1;
char *s_key, *s_value;//to address, data
s_key=NULL;
s_value=NULL;
int l_key, l_value; //lengths of strings
long int l_valuel;//from serviceid, to serviceid
int i_value; //protocol
double d_value1;
double d_value2;
float f_value;
Py_complex c;

/* Possible types are:
EIntType,
ELongType,
EStringType,
EFloatType,
EComplexType

Determine type:
*/
if(PyArg_ParseTuple(args, "u#s#", &s_key, &l_key, &s_value, &l_value)) /* type-checked */
type=CAppConfig::EStringType;
else if(PyArg_ParseTuple(args, "u#f", &s_key, &l_key, &f_value))
{
d_value1 = f_value;
type=CAppConfig::EFloatType;
}
else if(PyArg_ParseTuple(args, "u#d", &s_key, &l_key, &d_value1))
type=CAppConfig::EFloatType;
else if(PyArg_ParseTuple(args, "u#i", &s_key, &l_key, &i_value))
type=CAppConfig::EIntType;
else if(PyArg_ParseTuple(args, "u#l", &s_key, &l_key, &l_valuel))
type=CAppConfig::ELongType;
else if(PyArg_ParseTuple(args, "u#D", &s_key, &l_key, &c))
{
type=CAppConfig::EComplexType;
d_value1 = c.real;
d_value2 = c.imag;
}
else
{
PyErr_BadArgument();
return NULL;
}

This code always works with strings and ints with no problems. However, when i pass in a float like 12.22, it detects it as int. For complex type is always branches to PyErr_BadArgument(); same applies for long int's

Any info/help in this regard will be much appreciated!!!!

Regards,
Varun

[jplauril]

First of all, note that PyArg_ParseTuple will raise an exception if the conversion fails. Your code will thus raise an exception for all argument types except string.

Also, the Python "long" type cannot be stored in a C long int variable nor can it be parsed with the "l" format specifier.

What is the Python script you use for testing that code?

[varunc]

Hi,
The code I am using to test this is something like this:
>>>> import ghostConfig
>>>> ghostConfig.set(u'mykey', 'myvalue')
>>>> ghostConfig.get(u'mykey')
'myvalue'
>>>> ghostConfig.set(u'mykey', 12)
>>>> ghostConfig.get(u'mykey')
12
>>>> ghostConfig.set(u'mykey', 14.56)
>>>> ghostConfig.get(u'mykey')
14 <---- Problem why treated as int???
>>>> ghostConfig.set(u'mykey', 3+4j)
Error type mismatch....

-------------

This is really a toy project that I am using to understand value passing between python and c++. It is a simple key value registry...

"PyArg_ParseTuple will raise an exception if the conversion fails"
>> I thought PyArg_ParseTuple would return 0 if the conversion fails??
Hence this kind of nested ifs should be ok???
if(PyArg_ParseTuple(args, "u#s#", &s_key, &l_key, &s_value, &l_value)) /* type-checked */
type=CAppConfig::EStringType;
else if(PyArg_ParseTuple(args, "u#f", &s_key, &l_key, &f_value))
{
d_value1 = f_value;
type=CAppConfig::EFloatType;
}

Whats going on here?