web/hdalab/management/commands/geojson_transform.py
author veltr
Thu, 23 Feb 2012 19:45:00 +0100
changeset 123 94fc5f5b5cfd
child 135 dd6578e36a57
permissions -rw-r--r--
Added Insee Codes
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     2
'''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     3
@author: raphv
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     4
'''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     5
from django.core.management.base import BaseCommand, CommandError
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     6
from django.utils.http import urlquote
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     7
import django.utils.simplejson as json
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     8
from SPARQLWrapper import SPARQLWrapper, JSON
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     9
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    10
class Command(BaseCommand):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    11
    '''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    12
    Command to export tags
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    13
    '''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    14
    args = '<path_to_geojson_file>'
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    15
    options = ''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    16
    help = """Adds semantic data to countries.geo.json files"""
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    17
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    18
    def handle(self, *args, **options):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    19
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    20
        if len(args) == 0 or not args[0]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    21
            raise CommandError("Give a Geo Json File to process")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    22
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    23
        geojsonfile = args[0]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    24
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    25
        f = open(geojsonfile,"r")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    26
        geojson = json.loads(f.read())
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    27
        f.close()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    28
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    29
        labels = [feature['properties']['name'] for feature in geojson['features']]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    30
        uris = dict([(label, 'http://dbpedia.org/resource/' + urlquote(label.replace(' ','_'))) for label in labels])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    31
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    32
        uri_list = [uris[label] for label in uris]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    33
        base_uris = dict([(uris[label],uris[label]) for label in uris])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    34
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    35
        grp = 10
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    36
        endpoint = SPARQLWrapper("http://dbpedia.org/sparql")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    37
        endpoint.setReturnFormat(JSON)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    38
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    39
        start = 0
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    40
        while start < len(uri_list):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    41
            sparql = """
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    42
                PREFIX dbpedia-owl: <http://dbpedia.org/ontology/>
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    43
                SELECT ?source ?target
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    44
                WHERE {
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    45
                    ?source dbpedia-owl:wikiPageRedirects ?target .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    46
                    Filter (?source in (<%s>)) .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    47
                }
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    48
            """ % '>, <'.join(uri_list[start:start + grp])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    49
            endpoint.setQuery(sparql)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    50
            results = endpoint.query().convert()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    51
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    52
            for r in results["results"]["bindings"]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    53
                source = r["source"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    54
                target = r["target"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    55
                base_uris[source] = target
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    56
                #print "%s redirects to %s"%(source, target)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    57
            start += grp
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    58
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    59
        base_uri_list = [base_uris[uri] for uri in base_uris]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    60
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    61
        french_labels = {}
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    62
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    63
        start = 0
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    64
        while start < len(base_uri_list):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    65
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    66
            sparql = """
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    67
                PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    68
                SELECT ?country ?label WHERE {
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    69
                  ?country rdfs:label ?label .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    70
                  FILTER (?country in (<%s>)) .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    71
                  FILTER langMatches( lang(?label), "fr" )
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    72
                }
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    73
            """ % '>, <'.join(base_uri_list[start:start + grp])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    74
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    75
            endpoint.setQuery(sparql)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    76
            results = endpoint.query().convert()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    77
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    78
            for r in results["results"]["bindings"]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    79
                country = r["country"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    80
                label = r["label"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    81
                french_labels[country] = label
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    82
                print "%s label: %s"%(country, label)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    83
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    84
            start += grp
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    85
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    86
        for feature in geojson['features']:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    87
            base_uri = base_uris[uris[feature['properties']['name']]]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    88
            feature['properties']['dbpedia_uri'] = base_uri
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    89
            if french_labels.has_key(base_uri):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    90
                feature['properties']['label_fr'] = french_labels[base_uri]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    91
            else:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    92
                print "No label for %s"%base_uri
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    93
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    94
        f = open(geojsonfile,"w")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    95
        f.write(json.dumps(geojson))
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    96
        f.close()