All,
I uploaded the code for ridesf
(
https://code.google.com/p/sf-bike-planner/source/browse/branches/#branches/ridesf).
There are three main sub dirs, www, src, and data (they should be self
explanatory. Under "src" the main cgi app is named "route" it connects
to the DB and performs the search as well as massages the data
returned from the pgRouting functions. I also uploaded several utility
programs that attempt to do various things from segmenting shapefiles
into two point line segments, to reverse geocode points along a line
segment in order to identify unnamed streets. Overall the code is not
very well commented and very much a work in progress. The front end
is, of course, in "www".
For all you GIS guys the data is a shapefile that I painstakingly
conflated by hand from the MTC data and Tiger. I don't know how
accurate it is at this point and it still doesn't contain one way
streets or elevation data (hint, hint).
Data wise, if you look at the shapefile I uploaded, it contains 7
attributes (FID, FULLNAME, DIVROAD, DECKROAD, BIKELANE, ELEVATION,
LENGTH). They should mostly be self explanatory, DIVROAD is boolean
attribute for a road with a median (like Dolores St.) and DECKROAD is
a boolean for roads that are stacked (like the bay bridge), neither of
these is used anywhere by me, I just thought they might be useful
eventually. It would be great to have some form of one way data (not
sure how best to add that data, perhaps an attribute with a radian
bearing?), bike lane class type (wish I'd taken the time to transfer
this info) and elevation (the field is there and I think my elevation
program will fill it out properly?).
If anyone would like shell access to the server, send me your pub ssh
key and I'll add you to the authorized_keys.
Let me know if you have any questions,
John