src/hdalab/management/commands/geojson_transform.py
author cavaliet
Thu, 26 Jun 2014 10:44:05 +0200
changeset 284 f01235a1dcc2
parent 281 bc0f26b1acc2
child 693 09e00f38d177
permissions -rw-r--r--
correct ontology requests
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
'''
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
     5
from django.conf import settings
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     6
from django.core.management.base import BaseCommand, CommandError
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     7
from django.utils.http import urlquote
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
     8
import json
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
     9
from SPARQLWrapper import SPARQLWrapper, JSON
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    10
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    11
#import pydevd #@UnresolvedImport
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    12
#pydevd.settrace(suspend=False)
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    13
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    14
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    15
class Command(BaseCommand):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    16
    '''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    17
    Command to export tags
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    18
    '''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    19
    args = '<path_to_geojson_file>'
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    20
    options = ''
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    21
    help = """Adds semantic data to countries.geo.json files"""
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    22
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    23
    def handle(self, *args, **options):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    24
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    25
        if len(args) == 0 or not args[0]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    26
            raise CommandError("Give a Geo Json File to process")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    27
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    28
        geojsonfile = args[0]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    29
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    30
        f = open(geojsonfile,"r")
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    31
        geojson = json.loads(f.read())
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    32
        f.close()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    33
        
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    34
        labels = [feature['properties']['label_fr'] for feature in geojson['features']]
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    35
        #uris = dict([(label, 'http://dbpedia.org/resource/' + urlquote(label.replace(' ','_'))) for label in labels])
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    36
        uris = dict([(label, settings.DBPEDIA_URI_TEMPLATE % ( 'resource', urlquote(label.replace(' ','_')) )) for label in labels])
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    37
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    38
        uri_list = [uris[label] for label in uris]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    39
        base_uris = dict([(uris[label],uris[label]) for label in uris])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    40
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    41
        grp = 10
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    42
        #endpoint = SPARQLWrapper("http://dbpedia.org/sparql")
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    43
        endpoint = SPARQLWrapper(settings.DBPEDIA_URI_TEMPLATE % ( 'sparql', '' ))
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    44
        endpoint.setReturnFormat(JSON)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    45
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    46
        start = 0
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    47
        while start < len(uri_list):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    48
            sparql = """
284
f01235a1dcc2 correct ontology requests
cavaliet
parents: 281
diff changeset
    49
                PREFIX dbpedia-owl: <http://dbpedia.org/ontology/>
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    50
                SELECT ?source ?target
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    51
                WHERE {
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    52
                    ?source dbpedia-owl:wikiPageRedirects ?target .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    53
                    Filter (?source in (<%s>)) .
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    54
                }
284
f01235a1dcc2 correct ontology requests
cavaliet
parents: 281
diff changeset
    55
            """ % '>, <'.join(uri_list[start:start + grp])
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    56
            endpoint.setQuery(sparql)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    57
            results = endpoint.query().convert()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    58
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    59
            for r in results["results"]["bindings"]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    60
                source = r["source"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    61
                target = r["target"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    62
                base_uris[source] = target
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    63
                #print "%s redirects to %s"%(source, target)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    64
            start += grp
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    65
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    66
        base_uri_list = [base_uris[uri] for uri in base_uris]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    67
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    68
        french_labels = {}
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    69
        labels = {}
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    70
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    71
        start = 0
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    72
        while start < len(base_uri_list):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    73
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    74
            sparql = """
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    75
                PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    76
                SELECT ?country ?label WHERE {
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    77
                  ?country rdfs:label ?label .
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    78
                  FILTER (?country in (<%s>))
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    79
                }
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    80
            """ % '>, <'.join(base_uri_list[start:start + grp])
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    81
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    82
            endpoint.setQuery(sparql)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    83
            results = endpoint.query().convert()
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    84
            
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
    85
            #print repr(results)
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    86
            
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    87
            for r in results["results"]["bindings"]:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    88
                country = r["country"]["value"]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    89
                label = r["label"]["value"]
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    90
                if country not in labels:
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    91
                    labels[country] = {}
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    92
                labels[country][r["label"]["xml:lang"]] = label 
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    93
                if r["label"]["xml:lang"] == 'fr':
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
    94
                    french_labels[country] = label
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    95
                print "%s label: %s"%(country, label)
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    96
            
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    97
            start += grp
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    98
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
    99
        for feature in geojson['features']:
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
   100
            base_uri = base_uris[uris[feature['properties']['label_fr']]]
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   101
            feature['properties']['dbpedia_uri'] = base_uri
135
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
   102
                        
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
   103
            if labels.has_key(base_uri):
dd6578e36a57 translate interface
ymh <ymh.work@gmail.com>
parents: 123
diff changeset
   104
                feature['properties']['labels'] = labels[base_uri]
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   105
            if french_labels.has_key(base_uri):
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   106
                feature['properties']['label_fr'] = french_labels[base_uri]
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   107
            else:
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   108
                print "No label for %s"%base_uri
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   109
        
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   110
        f = open(geojsonfile,"w")
281
bc0f26b1acc2 Hdalab : commands now work after update. Requests update with a dbpedia url from settings.
cavaliet
parents: 266
diff changeset
   111
        f.write(json.dumps(geojson, indent=2))
123
94fc5f5b5cfd Added Insee Codes
veltr
parents:
diff changeset
   112
        f.close()