May be, we could use
$$\log(1+x)\,\log(1-x)=-\sum_{n=1}^\infty \frac{a_n}{b_n}\,x^{2n}$$ where the $a_n$ and $b_n$ correspond to sequences $A049281$ and $A069685$ in $OEIS$ and use
$$\int x^{2 n-\frac{1}{2}} \log (x)\,dx=\frac{2 x^{2 n+\frac{1}{2}} ((4 n+1)\log (x)-2)}{(4 n+1)^2}$$
$$\int_0^1 x^{2 n-\frac{1}{2}} \log (x)\,dx=-\frac{4}{(4 n+1)^2}$$ which would lead to
$$\int_0^1\frac{\log (1-x) \log (x) \log
(1+x)}{\sqrt{x}}\,dx=\sum_{n=1}^\infty \frac{a_n}{b_n}\,\frac{4}{(4 n+1)^2}$$
If $c_n$ is the summand
$$\frac {c_{n+1}}{c_n} \sim 1-\frac 3n$$
Computing the partial sums
$$S_p=\sum_{n=1}^p \frac{a_n}{b_n}\,\frac{4}{(4 n+1)^2}$$ and converting to decimals, the converegnce is quite slow
$$\left(
\begin{array}{cc}
p & S_p\\
10 & 0.1927571 \\
20 & 0.1933295 \\
30 & 0.1934418 \\
40 & 0.1934820 \\
50 & 0.1935008 \\
60 & 0.1935110 \\
70 & 0.1935173 \\
80 & 0.1935213 \\
90 & 0.1935241 \\
100 & 0.1935261 \\
200 & 0.1935325 \\
300 & 0.1935337 \\
400 & 0.1935341 \\
500 & 0.1935343 \\
\end{array}
\right)$$
The various inverse symbolic calculators I tried do not seem to enjoy
$$0.193534656289778253633161982923\cdots$$
Edit
We can also write (using the Lerch transcendent function)
$$\log(1+x)\,\log(1-x)=-\sum_{n=1}^\infty \frac{\Phi (-1,1,2 k)+\log (2)}{k}\,x^{2n}$$ which leads to
$$I=\log (2) \left(32-8 C-2 \pi -\pi ^2-12 \log (2)\right)+4\sum_{k=1}^\infty \frac{\Phi (-1,1,2 k)}{k (4 k+1)^2}$$
If $c_n$ is the summand
$$\frac {c_{n+1}}{c_n} \sim 1-\frac 4n$$
Summing the first $100$ terms leads to $0.1935346$ which is much better than the previous and next solutions.
Comment
Using Feynman's trick for
$$I(a)=\int_0^1\frac{\log (1-x) \log (x) \log
(1+ax)}{\sqrt{x}}\,dx$$
$$I'(a)=\int_0^1\frac{\sqrt{x} \log (1-x) \log (x)}{1+a x}\,dx$$ leads to a monster involving the Lerch transcendent function and a few (even multiple and cross) derivatives of the generalized Gaussian hypergeometric function with respect to its second and third arguments.
It is a pity since
$$\,_2\tilde{F}_1\left(1,\frac{5}{2};\frac{7}{2};-a\right)=\frac{8 \left(\sqrt{a} (a-3)+3 \tan
^{-1}\left(\sqrt{a}\right)\right)}{9 \sqrt{\pi} a^{5/2}}$$
Edit
Using your idea of expanding only $\log(1+x)$ we should have
$$I=\sum_{n=1}^\infty \frac{(-1)^{n+1}}n \int_0^1 x^{n-\frac{1}{2}} \log (1-x) \log (x)\,dx$$ and
$$J_n=\int_0^1 x^{n-\frac{1}{2}} \log (1-x) \log (x)\,dx$$
$$J_n=\frac{8}{(2 n+1)^3}+\frac{4 \left(\psi
^{(0)}\left(n+\frac{1}{2}\right)+\gamma
\right)}{(2 n+1)^2}-\frac{2 \psi
^{(1)}\left(n+\frac{3}{2}\right)}{2 n+1}$$ which would give
$$I=K+2\sum_{n=1}^\infty\frac{ (-1)^{n+1}}{n(2n+1) }\Bigg(\frac{2 \psi ^{(0)}\left(\frac{2n+1}{2}\right)}{2
n+1}-\psi ^{(1)}\left(\frac{2n+3}{2}\right) \Bigg)$$ with
$$K=8 (2+\gamma ) C+\frac{\pi ^3}{2}-4 (12-\pi -2\log
(2))-2 \gamma (8-\pi -2\log (2))$$
The convergence of the summation will not be very fast since, if $a_n$ is the summand
$$\left|\frac{a_{n+1}}{a_n}\right|=1-\frac{3 \log (n)-4}{n (\log (n)-1)}+O\left(\frac{1}{n^2 \log ^2(n)}\right)$$
For $p=100$ the result would be $0.1935329$