I didnt check, but I think that CAnalyzer needs to access the source and xml files to compute lines of code. How did you import your model? Where is it on disk? Did you generate new xml files or use old ones?
No, you do not need the original source code for computing the lines of code. There is a bijection between srcML XML files and original source code. This is a great strenght of srcML.
Cyrille, did you solve the problem?
Cheers, Alexandre