It isn't trivial as I believe SPSS uses an overparameterized general linear model - for example see the reference here:
Green, S. B., Marquis, J. G., Thompson, M. S., Hershberger, S. L., & McCollam, K. M. (1999). The overparameterized analysis of variance model. Psychological Methods, 4(2), 214-233. 10.1037/1082-989X.4.2.214
It is further complicated because SPSS also runs a form of MANOVA on the difference scores as an additional model automatically and a few other bits and pieces that aren't strictly part of the ANOVA.
For a balanced ANOVA you can reproduce the equivalent model fairly easily as a classical ANOVA by partitioning sums of squares.