importing_data.txt

Within data dir:

1. Make GTFS into GTFSDB.
   $ gs_gtfsdb_build trimet-2010-02-28.gtfs.zip trimet-2010-02-28.gtfsdb

2. Find extent of GTFSDB stations in sqlite.
   $ sqlite3 ~/data/pdx/trimet-2010-02-28.gtfsdb \
     "select min(stop_lon),min(stop_lat),max(stop_lon),max(stop_lat) from stops"

3. Extract Portland TriMet service area.
   Keeps nodes outside the box if they are part of ways inside the box.
   On approx 1GB cloudmade Oregon set takes around 3h30.
   $ ~/devel/osmosis-0.34/bin/osmosis \
     --read-xml  file=~/data/pdx/cloudmade_oregon.osm \
     --bounding-box completeWays=yes \
       top=45.68688202    left=-123.18748474 \
       bottom=45.23430252 right=-122.20288086 \
     --write-xml file=~/data/pdx/trimet_all.osm

4. Keep only ways that can be walked on and their nodes.
   On trimet area, takes around 13 seconds.
   $ ~/devel/osmosis-0.34/bin/osmosis \
     --read-xml file=~/data/pdx/trimet_all.osm \
     --tf accept-ways highway=* --tf reject-ways highway=motorway,motorway_link \
     --tf reject-relations \
     --used-node --write-xml file=~/data/pdx/trimet_walk.osm

5. Make osm (xml) into osmdb (sqlite database) also cleaning up and splitting ways.
   gs_compile_osm makes edges right away, but it should be done after running filters.
   This script hack takes care of the problem temporarily.
   $ python osmdb_clean_split_build.py \
     ~/data/pdx/trimet_walk.osm \
     ~/data/pdx/trimet_walk_clean.osmdb \
     ~/data/pdx/trimet-2010-02-28.gtfsdb

6. Import GTFS database into graph database
   $ gs_import_gtfs trimet.gsdb trimet-fjdkqsm-.gtfsdb
   (this step is slow, maybe back up the result)
   $ cp trimet.gsdb trimet_transit_only.gsdb

7. Import OSM database into graph database
   $ gs_import_osm trimet.gsdb trimet_walk_clean.osmdb

8. Link GTFS stations to OSM nodes in graph database
   (Stations should also be linked to one another within a short range, in case there is no road near them.)
   $ gs_link_osm_gtfs trimet.gsdb trimet_walk_clean.osmdb trimet-2010-02-28.gtfsdb

9. Add WKT strings to graph database, if desired
   $ gdb_add_wkt.py trimet.gsdb trimet_walk_clean.osmdb trimet-2010-02-28.gtfsdb