Merge various file-based geodata and output as geojson

merge.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import optparse
import mapnik
import sys
import os

parser = optparse.OptionParser(usage="""%prog datasource files (like .geojson, .shp, .csv, .sqlite, .osm, .kml)

Merge various geodata and output as geojson

 $ %prog -h (or --help for possible options)

 $ %prog one.shp two.json | python -mjson.tool > merged.geojson
""", version='%prog ')

mapping = {
   'txt':'csv',
   'csv':'csv',
   'tsv':'csv',
   'osm':'osm',
   'shp':'shape',
   'db':'sqlite',
   'sqlite3':'sqlite',
   'sqlite':'sqlite',
   'kml':'ogr',
   'gml':'ogr',
   'gpx':'ogr',
   'geojson':'ogr',
   'json':'ogr'
}

def detect_mapnik_type(filename):
    ext = os.path.splitext(filename)
    if not ext[1]:
        sys.stderr.write('Could not detect file extension for %s\n' % filename)
        sys.exit(1)
    ext = ext[1][1:]
    if not mapping.has_key(ext):
        sys.stderr.write('No known mapping from this type of file "%s" to a known Mapnik datasource\n' % ext)
        sys.exit(1)
    ds_type = mapping[ext]
    if ds_type not in mapnik.DatasourceCache.plugin_names():
        sys.stderr.write('Whoops, looks like you a missing a great Mapnik datasource plugin (%s) to be able to read %s\n' % (ds_type,filename))
        sys.exit(1)
    params = {'type':ds_type}
    if ds_type == 'ogr':
        params['layer_by_index'] = 0
    params['file'] = filename
    return params

if __name__ == "__main__":

    (options, args) = parser.parse_args()
    if not len(args) > 0:
        parser.error("Please provide the path to a one or more datasource files")
    
    print '''{
  "type": "FeatureCollection",
  "features": ['''

    for idx, filename in enumerate(args):
        if idx > 0:
            print ','
        ds = mapnik.Datasource(**detect_mapnik_type(filename))
        e = ds.envelope();
        query = mapnik.Query(e)
        for fld in ds.fields():
            query.add_property_name(fld)
        fs = ds.features(query)

        feat = fs.next()
        while feat:
            try:
                print '    ' + feat.to_geojson(),
                feat = fs.next()
                print ','
            except StopIteration:
                feat = False
    print '''\n  ]
}'''

/cc @mousebird