# The faces — how each was derived (reproduce any of them yourself)

Source of truth: CapMetro's published GTFS feed, pinned byte-exact in this demo
(capmetro_gtfs_current.zip, sha256 3890b0b3…, from data.texas.gov; June 2018 edition
4210921a… for time travel). Every face below is derived 1:1 from tables inside that
zip — no values added, removed, or altered. All derivations are deterministic.

- stops.txt — NOT derived: the native table, extracted unmodified from the zip.
- network.sql — routes.txt, trips.txt, stops.txt loaded 1:1 as three TEXT-column
  tables (CREATE + INSERT, published row order, empty→NULL). Counts in its header.
- stops.xlsx — stops.txt rendered as a single-sheet workbook, header row + one row
  per stop, every cell as text, in published order.
- system.json — agency.txt, fare_attributes.txt, feed_info.txt re-encoded as JSON,
  field names and values unchanged; row counts appended under "_counts".
- route-801-timetable.pdf — all distinct northbound start times for route 801
  (trips.txt direction_id=0 joined to stop_times.txt stop_sequence=1), printed as a
  rider timetable; footer carries the total trip count both directions: 462.
- fares-and-agency.docx — agency.txt and fare_attributes.txt rendered as a document,
  values unchanged.

Check any number on this page against any face — and against the original zip.