diff -r 3a7ecd3ad34f -r c8627191f2d7 web/hdalab/management/commands/query_dbpedia.py --- a/web/hdalab/management/commands/query_dbpedia.py Thu Sep 06 12:16:27 2012 +0200 +++ b/web/hdalab/management/commands/query_dbpedia.py Sat Sep 08 03:49:10 2012 +0200 @@ -17,11 +17,11 @@ from hdabo.utils import show_progress from rdflib.graph import Graph from rdflib import URIRef -import re import traceback import sys from hdalab.models.dataviz import DbpediaFieldsTranslation from django import db +from SPARQLWrapper import SPARQLWrapper, RDF class Command(NoArgsCommand): ''' @@ -123,41 +123,54 @@ print "dbpedia query cancelled" return + endpoint = SPARQLWrapper("http://dbpedia.org/sparql", returnFormat=RDF) + writer = None for i,tag in enumerate(queryset): writer = show_progress(i+1, count, tag.label, 50, writer) db.reset_queries() - rdf_uri = re.sub('\/resource\/', "/data/", tag.dbpedia_uri) + ".n3" - g = Graph() + + #abstract query + #"select ?y + # where {<%s> ?y}" % (tag.dbpedia_uri) + + #rdf_uri = re.sub('\/resource\/', "/data/", tag.dbpedia_uri) + ".n3" + #g = Graph() try : - g.parse(rdf_uri, format="n3") - + abstracts = {} + labels = {} + thumbnail = None with transaction.commit_on_success(): + endpoint.setQuery("select distinct ?y where {<%s> ?y}" % (tag.dbpedia_uri)) + res_abstracts = endpoint.queryAndConvert() + for _,_,o in res_abstracts.triples((None, URIRef('http://www.w3.org/2005/sparql-results#value'), None)): + abstracts[o.language] = (unicode(o), True) + + endpoint.setQuery("select distinct ?y where {<%s> ?y}" % (tag.dbpedia_uri)) + res_labels = endpoint.queryAndConvert() + for _,_,o in res_labels.triples((None, URIRef('http://www.w3.org/2005/sparql-results#value'), None)): + labels[o.language] = (unicode(o), True) + + endpoint.setQuery("select distinct ?y where {<%s> ?y} limit 1" % (tag.dbpedia_uri)) + res_thumbnails = endpoint.queryAndConvert() + for _,_,o in res_thumbnails.triples((None, URIRef('http://www.w3.org/2005/sparql-results#value'), None)): + thumbnail = unicode(o) + + endpoint.setQuery('select distinct ?y where { <%s> ?p ?y . FILTER regex(?y, "^http://dbpedia.org/resource")}' % (tag.dbpedia_uri)) + res_links = endpoint.queryAndConvert() + for _,_,o in res_links.triples((None, URIRef('http://www.w3.org/2005/sparql-results#value'), None)): + tagqs = Tag.objects.filter(dbpedia_uri=unicode(o)) + if tagqs: + TagLinks.objects.get_or_create(subject=tag, object=tagqs[0]) - abstracts = {} - labels = {} - thumbnail = None - for t in g: - if t[1] == URIRef(u'http://dbpedia.org/ontology/abstract') and t[2] is not None \ - and hasattr(t[2], 'language'): - abstracts[t[2].language] = unicode(t[2]) - if t[1] == URIRef(u'http://www.w3.org/2000/01/rdf-schema#label') and t[2] is not None \ - and hasattr(t[2], 'language'): - labels[t[2].language] = unicode(t[2]) - if t[1] == URIRef(u'http://dbpedia.org/ontology/thumbnail') and t[2] is not None: - thumbnail = unicode(t[2]) - if u'http://dbpedia.org/resource' in t[2]: - tagqs = Tag.objects.filter(dbpedia_uri=unicode(t[2])) - if tagqs: - TagLinks.objects.get_or_create(subject=tag, object=tagqs[0]) - - ref_label = labels['fr'] if 'fr' in labels else labels['en'] if 'en' in labels else labels.values()[0] if len(labels) > 0 else tag.label - ref_abstract = abstracts['fr'] if 'fr' in abstracts else abstracts['en'] if 'en' in abstracts else abstracts.values()[0] if len(abstracts) > 0 else None + ref_label_lang, (ref_label, _) = ('fr',labels['fr']) if 'fr' in labels else ('en',labels['en']) if 'en' in labels else labels.items()[0] if len(labels) > 0 else ('fr',(tag.label, True)) + ref_abstract_lang, (ref_abstract, _) = ('fr',abstracts['fr']) if 'fr' in abstracts else ('en',abstracts['en']) if 'en' in abstracts else abstracts.items()[0] if len(abstracts) > 0 else ('fr',(None, 'True')) + for lang in settings.LANGUAGES: if lang[0] not in labels: - labels[lang[0]]= ref_label + labels[lang[0]]= (ref_label, False) if lang[0] not in abstracts: - abstracts[lang[0]] = ref_abstract + abstracts[lang[0]] = (ref_abstract, False) dbfield , created = DbpediaFields.objects.get_or_create(dbpedia_uri=tag.dbpedia_uri,tag=tag, defaults={'abstract':ref_abstract, 'thumbnail':thumbnail, 'label':ref_label}) #@UndefinedVariable if not created: @@ -169,16 +182,16 @@ consolidated_trans = {} for lang,label in labels.iteritems(): - consolidated_trans[lang] = [label,ref_abstract] + consolidated_trans[lang] = [label,(ref_abstract, lang==ref_abstract_lang)] for lang,abstract in abstracts.iteritems(): if lang in consolidated_trans: consolidated_trans[lang][1] = abstract else: - consolidated_trans[lang] = [ref_label, abstract] + consolidated_trans[lang] = [(ref_label, lang==ref_label_lang), abstract] for lang, trans in consolidated_trans.iteritems(): label, abstract = tuple(trans) - DbpediaFieldsTranslation.objects.create(master=dbfield, language_code=lang, label=label, abstract=abstract) + DbpediaFieldsTranslation.objects.create(master=dbfield, language_code=lang, label=label[0], is_label_translated=label[1], abstract=abstract[0], is_abstract_translated=abstract[1]) except Exception as e: