Re: US Postal Service

["Aaron Swartz" <xx@aaronsw.com>]

> Corynne said she'd take us on, but is busy for a couple of weeks.

cool

> I can do the procurement and licenses piece.  Do you want to do the "how our
> collection of facts is not the same thing as they distribute" piece?  E.g.,
> how did we extract, filter out, and reassemble this so it is only factual
> information.  (I'm happy to do that piece as well if you'd prefer).  This
> doesn't have to be long or formal.

How's this:

USPS data is provided in a unique "fixed-width ASCII" format. We
developed new tools for parsing this format and adapted them to the
specific instances the Post Office provides. We convert encoded
numbers to actual numbers, split up and clarify text, decode
post-offfice-specific codes, and annotate each piece with what it
does. In short, we convert a line like:

D91101Z200132998AS  GRANITE                     DR    0000000902000000
0998E055862CA      Z22427

to:

{
  "_type": "Five-Digit ZIP Detail",
  "action_code": "ADD",
  "addr_primary_hi": "0000000900",
  "addr_primary_lo": "0000000899",
  "addr_primary_odd_even": "BOTH",
  "finance_no": "055862",
  "prefd_lastline_ctyst_key": "Z22427",
  "record_type": "STREET",
  "state_abbrev": "CA",
  "street_name": "GRANITE",
  "street_post_dir": "",
  "street_pre_dir": "",
  "street_suffix": "DR",
  "update_key": "Z200132997",
  "urbanization_ctyst_key": "",
  "zip": "91101"
}