$\sum_{p\leq x} \frac{1}{p} \sim \log\log x$ follows from the PNT, and there's also a short proof not relying on the PNT. However that proof still requires some nontrivial steps. I was wondering if there is a simpler proof which shows the weaker $$C_1\log\log x < \sum_{p\leq x} \frac{1}{p} < C_2\log\log x$$ for some constants $C_1, C_2$.
What made me wonder is that if I start from a proof that $\sum_p \frac{1}{p}$ diverges (which I take from Bateman & Diamond's ANT book): $$\sum_{n\leq x} \frac{1}{n} < \prod_{p\leq x} \left(1-\frac{1}{p}\right)^{-1}<\prod_{p\leq x} e^{2/p} = \exp(2\sum_{p\leq x}\frac{1}{p})$$ and take the $\log$ of each side I get $$\sum_{p\leq x} \frac{1}{p} > \frac{1}{2}\log\left(\sum_{n\leq x}\frac{1}{n}\right) > \frac{1}{2}\log\log x$$ which gives one side of the required inequality. Is there a way to get a similarly easy proof of the other side?
Relatedly, is there a proof of $\sum_{p\leq x}\frac{1}{p} \sim \log\log x$ which is simpler than the one in Merten's theorems (and which does not rely on the PNT)? Mertens gives more, $$\sum_{p\leq x}\frac{1}{p}=\log \log x + M + O\left(\frac{1}{\log n}\right)$$ which is why I ask if there is a simpler proof which does not give the asymptotic term.