gen_osm_replacement.py

"""
Post processor for OSM tables imported by Imposm to rename suffixes and prefixes to shorter versions
Use like:
python gen_osm_replacment > gen_osm_replacment.sql
psql -d osm < gen_osm_replacmenet.sql

"""
replacements = [  
    ('Avenue', 'Ave'),    
    ('Boulevard', 'Blvd'),
    ('Causeway', 'Cswy'),
    ('Center', 'Ctr'),
    ('Circle', 'Cir'),
    ('Drive', 'Dr'),
    ('Expressway', 'Expy'),
    ('Extension', 'Ext'),
    ('Freeway', 'Fwy'),
    ('Highway', 'Hwy'),
    ('Junction', 'Jct'),
    ('Lane', 'Ln'),
    ('Parkway', 'Pkwy'),
    ('Road', 'Rd'),
    ('Street', 'St')    
]

directions = [    
    ('North', 'N'),
    ('Northwest', 'NW'),
    ('Northeast', 'NE'),
    ('South', 'S'),
    ('Southwest', 'SW'),
    ('Southeast', 'SE'),
    ('East', 'E'), 
    ('West', 'W')    
]

tables = [
     'osm_mainroads',
     'osm_motorways',
     'osm_mainroads_gen0',
     'osm_mainroads_gen1',
     'osm_motorways_gen0', 
     'osm_motorways_gen1',
     'osm_minorroads'
]



for i in directions:
    orig = i[0]
    replace = i[1]
    # Replace street names that end with a direction with the abbreviation.  "Main Street North" will become "Main Street N"
    for t in tables:        
        print "update %s set name = regexp_replace(name, ' %s$', ' %s');" % ( t, orig, replace)       
    print ""
    
    # Replace street names that begin with a direction with the abbreviation.  "North Main Street" will become "N Main Street"
    for t in tables:        
        print "update %s set name = regexp_replace(name, '^%s ', '%s ');" % ( t, orig, replace)       
    print ""
    
    
    

for i in replacements:
    orig = i[0]
    replace = i[1]
    # Replace street names that end with one of the words in the replacement list with their abbreviations.  "First Street" will become "First St".    
    for t in tables:
        print "update %s set name = regexp_replace(name, '%s$', '%s');" % ( t, orig, replace)               
    print ""   
    
    # Now replace any instance of the world that appear with spaces around them.  So part of the name but not the first word in the name
    # So "First Avenue N" would become "First Ave N"
    for t in tables:
        print "update %s set name = replace(name, ' %s ', ' %s ');" % ( t, orig, replace)
    print ""