How do I integrate a definite integral using the metropolis algorithm?

I don't quite understand how I would go about this? I understand that I have a random walk that samples the function, given by:

Fun = @x. ...;
x = 0;
delta =...;  

for
r=rand();
x_T=x+delta(2r-1);
P=f(x_T)/f(x)

if P>=1;
then x = x_T;

else 

if P>= r
then x = x_T;
else 
x=x;
end
end

To find the integral using this method, I understand I have to sample a PDF? What is the PDF/how do I find it? Moreover, how to I talk the sampling after the random walk and use it to approximate the integral of the function.