Let say that you want to build a decision tree using a purity measure such as GINI.
If you want to split using some ranges of values for some continuous attribute such as age > 80 in a decision tree, a simple way is to calculate the GINI measures for different split: age > 80 / age <= 80 age > 75 / age <= 75 age > 70 / age <= 70 ...
Then, you choose the best split. For example if age > 70 / age <= 70 allows to obtain the best GINI value, then you choose this.