My personal choice would be to do what Bill suggested in his comment. Pre-process the input file with a script task, which will add any missing commas to act as placeholders for your Flat File Source.
There are scenarios where this is not possible however. In such a case you could use the Script Component tool in your data flow task set up as a Source. A very good approach to what needs to be in that script can be found here:
This method is very flexible as it would deal with rows that have less number of columns coming in, as well as the cases where more columns might appear.