# HG changeset patch # User cavaliet # Date 1377860909 -7200 # Node ID c4e7d66b7dc219ed4785be50d2be45e0b9a5ee66 # Parent 3bc55f57b2b1ba0435e25fffc5fe504dd8cd5407 add sparql request for languages and countries labels diff -r 3bc55f57b2b1 -r c4e7d66b7dc2 src/p4l/management/commands/get_country_label.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/src/p4l/management/commands/get_country_label.py Fri Aug 30 13:08:29 2013 +0200 @@ -0,0 +1,69 @@ +# -*- coding: utf-8 -*- + +from django.conf import settings +from django.core.management import BaseCommand +from p4l.models import Country +from p4l.utils import show_progress +import logging +import requests + +logger = logging.getLogger(__name__) + +class Command(BaseCommand): + + query = """ +PREFIX skos: +PREFIX rdf: +PREFIX owl: +PREFIX rdfs: +SELECT ?uri ?label +WHERE { + ?uri skos:inScheme . + ?uri skos:prefLabel|skos:label ?label . + FILTER (?uri = $root) +} +""" + + def fill_label(self): + # Loads Countries label from sparkl query + countries = Country.objects.filter(label=None) + total_countries = len(countries) + writer = None + i = 0 + found = 0 + for c in countries: + i += 1 + res = requests.get( + settings.SPARQL_QUERY_ENDPOINT, + params={'query':self.query, 'timeout':10, '$root' : "<"+c.country+">"}, + headers={'accept':'application/sparql-results+json'}, + ) + if not res.ok: + continue + elif res.text: + json_res = res.json() + if 'results' in json_res and 'bindings' in json_res['results'] and len(json_res['results']['bindings'])>0: + # json_res['results']['bindings'] has several languages. If we find french, we save the french label. + # If not, we save the first one. + tmp_dict = {} + first_label = None + # We create a temporary dict with the lang code and the label + for b in json_res['results']['bindings']: + if 'label' in b and 'value' in b['label'] and 'xml:lang' in b['label']: + tmp_dict[b['label']['xml:lang']] = b['label']['value'] + if not first_label: + first_label = b['label']['value'] + if 'fr' in tmp_dict or first_label: + if 'fr' in tmp_dict: + c.label = tmp_dict['fr'] + else: + c.label = first_label + l = c.label + c.save() + found += 1 + writer = show_progress(i, total_countries, l, 50, writer=writer) + print("Processing Subjects Sparql Done. %d found on %d" % (found, total_countries)) + + def handle(self, *args, **options): + self.fill_label() + diff -r 3bc55f57b2b1 -r c4e7d66b7dc2 src/p4l/management/commands/get_language_label.py --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/src/p4l/management/commands/get_language_label.py Fri Aug 30 13:08:29 2013 +0200 @@ -0,0 +1,73 @@ +# -*- coding: utf-8 -*- + +from django.conf import settings +from django.core.management import BaseCommand +from p4l.models import Language +from p4l.utils import show_progress +import logging +import requests + +logger = logging.getLogger(__name__) + +class Command(BaseCommand): + + query = """ +PREFIX skos: +PREFIX rdf: +PREFIX owl: +PREFIX rdfs: +SELECT ?uri ?label +WHERE { + ?uri skos:inScheme . + ?uri skos:prefLabel|skos:label ?label . + FILTER (?uri = $root) +} +""" + + def fill_label(self): + # Loads Subjects label from sparkl query + langs = Language.objects.filter(label=None) + total_langs = len(langs) + writer = None + i = 0 + found = 0 + for l in langs: + i += 1 + logger.debug("1") + logger.debug(l) + logger.debug("2") + logger.debug(l.language) + res = requests.get( + settings.SPARQL_QUERY_ENDPOINT, + params={'query':self.query, 'timeout':10, '$root' : "<"+l.language+">"}, + headers={'accept':'application/sparql-results+json'}, + ) + if not res.ok: + continue + elif res.text: + json_res = res.json() + if 'results' in json_res and 'bindings' in json_res['results'] and len(json_res['results']['bindings'])>0: + # json_res['results']['bindings'] has several languages. If we find french, we save the french label. + # If not, we save the first one. + tmp_dict = {} + first_label = None + # We create a temporary dict with the lang code and the label + for b in json_res['results']['bindings']: + if 'label' in b and 'value' in b['label'] and 'xml:lang' in b['label']: + tmp_dict[b['label']['xml:lang']] = b['label']['value'] + if not first_label: + first_label = b['label']['value'] + if 'fr' in tmp_dict or first_label: + if 'fr' in tmp_dict: + l.label = tmp_dict['fr'] + else: + l.label = first_label + lab = l.label + l.save() + found += 1 + writer = show_progress(i, total_langs, lab, 50, writer=writer) + print("Processing Subjects Sparql Done. %d found on %d" % (found, total_langs)) + + def handle(self, *args, **options): + self.fill_label() + diff -r 3bc55f57b2b1 -r c4e7d66b7dc2 src/p4l/migrations/0001_initial.py --- a/src/p4l/migrations/0001_initial.py Fri Aug 30 12:40:01 2013 +0200 +++ b/src/p4l/migrations/0001_initial.py Fri Aug 30 13:08:29 2013 +0200 @@ -94,6 +94,7 @@ db.create_table(u'p4l_country', ( (u'id', self.gf('django.db.models.fields.AutoField')(primary_key=True)), ('country', self.gf('django.db.models.fields.URLField')(unique=True, max_length=2048, db_index=True)), + ('label', self.gf('django.db.models.fields.CharField')(db_index=True, max_length=2048, null=True, blank=True)), )) db.send_create_signal('p4l', ['Country']) @@ -128,6 +129,7 @@ db.create_table(u'p4l_language', ( (u'id', self.gf('django.db.models.fields.AutoField')(primary_key=True)), ('language', self.gf('django.db.models.fields.URLField')(unique=True, max_length=2048, db_index=True)), + ('label', self.gf('django.db.models.fields.CharField')(db_index=True, max_length=2048, null=True, blank=True)), )) db.send_create_signal('p4l', ['Language']) diff -r 3bc55f57b2b1 -r c4e7d66b7dc2 src/p4l/models/data.py --- a/src/p4l/models/data.py Fri Aug 30 12:40:01 2013 +0200 +++ b/src/p4l/models/data.py Fri Aug 30 13:08:29 2013 +0200 @@ -57,6 +57,7 @@ class Country(P4lModel): country = models.URLField(max_length=2048, unique=True, db_index=True) + label = models.CharField(max_length=2048, blank=True, null=True, db_index=True) # filled from sparql request after import class Isbn(P4lModelLang): @@ -74,6 +75,7 @@ class Language(P4lModel): language = models.URLField(max_length=2048, unique=True, db_index=True) + label = models.CharField(max_length=2048, blank=True, null=True, db_index=True) # filled from sparql request after import diff -r 3bc55f57b2b1 -r c4e7d66b7dc2 src/p4l/templates/p4l/p4l_record_view.html --- a/src/p4l/templates/p4l/p4l_record_view.html Fri Aug 30 12:40:01 2013 +0200 +++ b/src/p4l/templates/p4l/p4l_record_view.html Fri Aug 30 13:08:29 2013 +0200 @@ -25,7 +25,7 @@ {% trans 'countries' %} -
    {% for i in record.countries.all %}
  • {{ i.country|default:'' }}
  • {% endfor %}
+
    {% for i in record.countries.all %}
  • {{ i.label|default:'' }} ({{ i.country|default:'' }})
  • {% endfor %}
{% trans 'identifier' %} @@ -49,7 +49,7 @@ {% trans 'language' %} - {{ record.language.language|default:'' }} + {{ record.language.label|default:'' }} ({{ record.language.language|default:'' }}) {% trans 'other languages' %}