This integral is a generalization of another integral which arose during calculations related to a prior question of mine. Specifically, by using this integral representation for polylogarithms, $$\mathrm{Li}_n(z)=-\frac{1}{(n-2)!}\int_0^1\frac{\ln(1-tz)}{t}\ln^{n-2}(1/t)\,dt,$$ it can be shown that $$\sum_{k=1}^\infty\int_0^e(-1)^{k-1}\mathrm{Li}_{2k}(\ln{x})\,dx=\int_0^1 e^{1/t}\frac{\cos(\ln{t})\mathrm{Ei}(1-1/t)}{t}dt-e\int_0^1\frac{\cos(\ln{t})\ln(1-t)}{t}dt.$$ This equivalence holds numerically, so I believe that the above link's restriction of $\mathrm{Li}_s(z)$ to $|z|\lt 1$ should actually be extended to $z\lt1$. The first integral seems nigh impossible, even without the inclusion of $\cos(\ln{t})$.
Digressions aside, this brings me to my conjecture: $$-\int_0^1 x^a\cos(\ln{x})\ln(1-x)\,dx\stackrel{?}{=}\frac{a+1+i}{2a^2+4a+4}H_{a+1-i}+\frac{a+1-i}{2a^2+4a+4}H_{a+1+i},\tag{1}$$ where $H_n$ represents the $n$-th harmonic number and $a\geq-1$.
I don't know much about the analytical properties of the harmonic numbers, and I know even less about their extension into the complex plane. Therefore, I'm lost on how to proceed.
I'm not sure if it'll help much, but I have solved a similar integral by using integration by parts to obtain a rational zeta series. Maybe I'm missing something, but this approach doesn't seem to lead anywhere for my conjecture above. $$-\int_0^1\frac{\cos(\ln{x})\ln(1-x^b)}{x}dx=\frac{\pi}{2}\coth\left(\frac{\pi}{b}\right)-\frac{b}{2};\:b\gt0.\tag{2}$$It's easy to show that $\frac12(\pi\coth(\pi)-1)=\frac i2(H_{-i}-H_i)$.