autogenerate Riemann solver doctrings from the Fortan source

[cbm755]

Background

For the _py Riemann solvers that are written in python, I can get live help:

from clawpack import riemann

riemann.euler_1D_py?

which then shows me the docstring including lots of nice info.

But riemann.traffic_1D? gives much less helpful info:

Type:        module
String form: <module 'clawpack.riemann.traffic_1D' from '/home/cbm/.local/lib/python3.13/site-packages/clawpack/riemann/traffic_1D.cpython-313-x86_64-linux-gnu.so'>
File:        ~/.local/lib/python3.13/site-packages/clawpack/riemann/traffic_1D.cpython-313-x86_64-linux-gnu.so
Docstring:  
This module 'traffic_1D' is auto-generated with f2py (version:2.3.5).
Functions:
    wave,s,amdq,apdq = rp1(maxm,num_waves,num_ghost,num_cells,ql,qr,auxl,auxr,num_eqn=shape(ql, 0),num_aux=shape(auxl, 0))
COMMON blocks:
  /cparam/ umax

(and similarly for the other f2py modules).

Fortran source has docs

However, in, for example, rp1_traffic.f90, I see useful stuff:

! Riemann solver for the LWR traffic model:

! q_t + (u_max q(1-q))_x = 0

! waves: 1
! equations: 1

! Conserved quantities:
!       1 q

! The speed limit umax should be in the common block cparam.

! See http://www.clawpack.org/riemann.html for a detailed explanation
! of the Riemann solver API.

Proposal

I imagine its possible, and perhaps even easy, to pull the fortran comments into the docstring dynamically.

[mandli]

I would be curious if there's something that would do this, but I thought that maybe f2py did something like this. Doxygen definitely did do something like this, and some of the AMRClaw comments in the headers are parsed that way. Anything you can find that does not add a significant dependency would be great though.