At various places on the Web (including Mathematics StackExchange) are various methods of calculating the semimajor and semiminor parameters of a ellipse $(a, b)$ from the location of a focus on the $x$ axis at a specified distance $c$ from the origin and a perpendicular from a directrix along the $x$ axis. However, I have not found anywhere where these can be derived in closed form from only the information available in the focus-directrix definition of an ellipse, i.e., the locus of points, the ratio of whose distance from the focus point and a perpendicular from the directrix is a constant, $0<e<1$. The derivations I have found either explicitly state values for the location of the focus along the $x$ axis $(ae)$ and the equation for the directrix $(x=a/e)$ and/or presuppose the existence of a second focus equidistant from the origin. Neither of these conditions should be necessary for derivation from the definition alone.
To this end, consider the folowing diagram:

$F$ is the focus, at distance $c$ from the origin. $DD'$ is the directrix, at distance $d$ from the origin. Eccentricity is specified as $e$. This is all the information we have from the definition.
Let $A$ be a point on the ellipse which is also on the $x$ axis. Let $B$ be a point on the ellipse which is also on the $y$ axis. Find $a = f(c,d,e)$ and $b=f(c,d,e)$
$\frac{a-c}{d-a} = e $ by definition applied to the $x$ axis, so $de = (1+e)a-c$
$de = \sqrt{b^2+c^2}$ by definition applied to the $y$ axis.
So $(de)^2 = b^2+c^2= ((1+e)a-c)^2 = \\ =(1+e)^2a^2-2(1+e)ac+c^2 \\ b^2 = a^2+2ea^2+e^2a^2-2ac-2eac$
Also,
$c = (1+e)a-de \\ c^2=((1+e)a-de)^2 = (1+e)^2a^2-2(1+e)de + (de)^2 \\ = a^2+2ea^2+ e^2a^2-2de-2de^2+d^2e^2$
$(de)^2 = b^2+c^2 = a^2+2ea^2+e^2a^2-2ac-2eac + a^2+2ea^2+ e^2a^2-2de-2de^2+(de)^2 \\ 2de(1+e)= 2a^2 + 4ea^2 + 2e^2a^2 - 2ac - 2eac\\ de(1+e) = ((1+e)a)^2 -2(1+e)ac\\$
or
$de = (1+e)a^2 - 2ac\\$
Completing the square:
$de(1+e)+c^2 = ((1+e)a - c)^2$
If I'm correctly interpreting everything that has been known about ellipses for the past 2000 years or so, I should be able to reduce this to $a=de$, which obviously can't be done without some substitution for $c$. Any assistance is greatly appreciated.

$c$,$d$ and $e$ are not all independent variables; the value of 2 of them determines the third. The relationships are $c=de^2$, $d=\frac{c}{e^2}$ or $e=\sqrt{\frac{c}{d}}$. These all can be derived by substituting either of the accepted parameters $d=\frac{a}{e}$ or $c=ae$ into $\frac{a-c}{d-a}=e$. An expression for $a$ follows directly from this equation. An expression for $b$ comes from $\sqrt{b^2+c^2} = de =d\sqrt{\frac{c}{d}}$.
$e=\sqrt{1-\frac{b^2}{a^2}}$ can be derived directly from the latter equation by simple algebra. Proof of the general equation for a standard ellipse can then proceed as can be found in various sources.