I've always taken integration by substitution for granted but recently I've learned that differentials can't fully be treated as variables and that the process of integration by substitution is really just a shorthand. Now, writing something like $du=u'(x)dx$ during substitution feels weird and empty.
I understood the justification for $\int f(g(x))g'(x)\,dx = \int f(u)\,du$ (given $u = g(x)$) from this page, but how do I apply this process to prove $\int f(x)f'(x)\,dx = \int v\,dv$ (given $v=f(x)$)?
Sorry that I don't have any work to show for myself, it's because I don't even know how to approach this problem.
I'll flip the $f$ and $g$ around, i.e. lets say that we can assume $$ \int f(g(x))g'(x)\,dx = \int f(u)\,du$$ and we want to prove $$ \int g(x) g'(x)\, dx = \int u\, du $$ We can write this as $\int f(g(x))g'(x)\,dx $ using $f(x)=x$, so that the substitution $u=g(x)$ gives the result.
Alternatively, since $(g^2)' = 2gg'$,$$ \int g(x) g'(x)\, dx =\int \left(\frac{g^2}2\right)' \, dx = \frac{g^2}2 + C$$