Considering $g : \mathbb{R} \rightarrow \mathbb{R}$, $f>0$, $f \in C(\mathbb{R})$ and $\phi : \mathbb{R} \rightarrow \mathbb{R}$, As I see we only have to prove $d\!\left(\phi(x),\phi(a)\right)>d\!\left(g(x),g(a)\right) \;\forall_a$.
I’ve got $d\!\left(\phi(x),\phi(a)\right) = \left|\int_{0}^{g(x)}f(t)\;dt - \int_{0}^{g(a)}f(t)\;dt\right| = \left|\int_{g(a)}^{g(x)}f(t)\;dt\right| = \left|(g(a)-g(x))f(\zeta)\right|$
for some $\zeta$, if $1 \leq f(\zeta) \forall \zeta$ we got it, but I only have $f>0$, any idea about how it follows?, am I on the right way?
Suppose that $g$ is not continuous at $y$, there exists $c>0$, a sequence $x_n$ such that $|x_n-y|<{1\over n}$ and $|g(x_n)-g(y)|>c$.
We deduce that $|\phi(x_n)-\phi(y)|=|\int_{g(y)}^{g(x_n)}f(t)dt|=f(u_n)||g(y)-g(x_n)|\geq c|f(u_n)$ where $u_n\in [g(y),g(x_n)]$ or $[g(x_n),g(y)]$.
Since $\phi$ is continuous at $y$, $lim_{n\rightarrow +\infty}|\phi(x_n)-\phi(y)|=0$, we deduce that $lim_{n\rightarrow+\infty}c|f(u_n)|=0$.
Suppose that the sequence $u_n$ is bounded, there exists a converging subsequence $u_{n_k}$ towards $u$ and $lim_{n_k}f(u_{n_k})=0=f(u)$ contradiction since $f>0$.
Suppose that $u_n$ is not bounded. This implies that $g(x_n)$ is not bounded, Let $v_n$ a subsequence of $x_n$ such that $g(v_n)$ is increasing and $lim_ng(v_n)=+\infty,g(v_n)>g(y)$, or $g(v_n)$ is decreasing and $lim_ng(v_n)=-\infty, g(v_n)<g(y)$. Without restriction the generality, we suppose that $lim_ng(v_n)=+\infty$, the other case is similar. We have $|\phi(v_n)-\phi(y)|=\int_{g(y)}^{g(v_n)}f(t)dt>\int_{g(y)}^{g(v_1)}f(t)dt>0$ since $f>0$.
This implies that $lim_nv_n\phi(v_n)\neq \phi(y)$ which is a contradiction with the fact that $\phi$ is continuous and $lim_nv_n=y$.