auf_savoirs_en_partage/savoirs/models.py

   1 # -*- encoding: utf-8 -*-
   2 import caldav
   3 import datetime
   4 import feedparser
   5 import operator
   6 import os
   7 import pytz
   8 import random
   9 import uuid
  10 import vobject
  11 from backend_config import RESOURCES
  12 from babel.dates import get_timezone_name
  13 from caldav.lib import error
  14 from babel.dates import get_timezone_name
  15 from datamaster_modeles.models import Region, Pays, Thematique
  16 from django.contrib.auth.models import User
  17 from django.db import models
  18 from django.db.models import Q, Max
  19 from django.db.models.signals import pre_delete
  20 from django.utils.encoding import smart_unicode
  21 from djangosphinx.models import SphinxQuerySet
  22 from savoirs.globals import META
  23 from settings import CALENDRIER_URL, SITE_ROOT_URL
  24
  25 # Fonctionnalités communes à tous les query sets
  26
  27 class RandomQuerySetMixin(object):
  28     """Mixin pour les modèles.
  29
  30     ORDER BY RAND() est très lent sous MySQL. On a besoin d'une autre
  31     méthode pour récupérer des objets au hasard.
  32     """
  33
  34     def random(self, n=1):
  35         """Récupère aléatoirement un nombre donné d'objets."""
  36         count = self.count()
  37         positions = random.sample(xrange(count), min(n, count))
  38         return [self[p] for p in positions]
  39
  40 class SEPQuerySet(models.query.QuerySet, RandomQuerySetMixin):
  41     pass
  42
  43 class SEPSphinxQuerySet(SphinxQuerySet, RandomQuerySetMixin):
  44     """Fonctionnalités communes aux query sets de Sphinx."""
  45
  46     def __init__(self, model=None, index=None, weights=None):
  47         SphinxQuerySet.__init__(self, model=model, index=index,
  48                                 mode='SPH_MATCH_EXTENDED2',
  49                                 rankmode='SPH_RANK_PROXIMITY_BM25',
  50                                 weights=weights)
  51
  52     def add_to_query(self, query):
  53         """Ajoute une partie à la requête texte."""
  54
  55         # Assurons-nous qu'il y a un nombre pair de guillemets
  56         if query.count('"') % 2 != 0:
  57             # Sinon, on enlève le dernier (faut choisir...)
  58             i = query.rindex('"')
  59             query = query[:i] + query[i+1:]
  60
  61         new_query = smart_unicode(self._query) + ' ' + query if self._query else query
  62         return self.query(new_query)
  63
  64     def search(self, text):
  65         """Recherche ``text`` dans tous les champs."""
  66         return self.add_to_query('@* ' + text)
  67
  68     def filter_discipline(self, discipline):
  69         """Par défaut, le filtre par discipline cherche le nom de la
  70            discipline dans tous les champs."""
  71         return self.search('"%s"' % discipline.nom)
  72
  73     def filter_region(self, region):
  74         """Par défaut, le filtre par région cherche le nom de la région dans
  75            tous les champs."""
  76         return self.search('"%s"' % region.nom)
  77
  78 class SEPManager(models.Manager):
  79     """Lorsque les méthodes ``search``, ``filter_region`` et
  80        ``filter_discipline`` sont appelées sur ce manager, le query set
  81        Sphinx est créé, sinon, c'est le query set Django qui est créé."""
  82
  83     def query(self, query):
  84         return self.get_sphinx_query_set().query(query)
  85
  86     def add_to_query(self, query):
  87         return self.get_sphinx_query_set().add_to_query(query)
  88
  89     def search(self, text):
  90         return self.get_sphinx_query_set().search(text)
  91
  92     def filter_region(self, region):
  93         return self.get_sphinx_query_set().filter_region(region)
  94
  95     def filter_discipline(self, discipline):
  96         return self.get_sphinx_query_set().filter_discipline(discipline)
  97
  98 # Disciplines
  99
 100 class Discipline(models.Model):
 101     id = models.IntegerField(primary_key=True, db_column='id_discipline')
 102     nom = models.CharField(max_length=765, db_column='nom_discipline')
 103
 104     def __unicode__ (self):
 105         return self.nom
 106
 107     class Meta:
 108         db_table = u'discipline'
 109         ordering = ["nom",]
 110
 111 # Actualités
 112
 113 class SourceActualite(models.Model):
 114     nom = models.CharField(max_length=255)
 115     url = models.CharField(max_length=255, verbose_name='URL')
 116
 117     class Meta:
 118         verbose_name = u'fil RSS syndiqué'
 119         verbose_name_plural = u'fils RSS syndiqués'
 120
 121     def __unicode__(self,):
 122         return u"%s" % self.nom
 123
 124     def update(self):
 125         """Mise à jour du fil RSS."""
 126         feed = feedparser.parse(self.url)
 127         for entry in feed.entries:
 128             if Actualite.all_objects.filter(url=entry.link).count() == 0:
 129                 ts = entry.updated_parsed
 130                 date = datetime.date(ts.tm_year, ts.tm_mon, ts.tm_mday)
 131                 a = self.actualites.create(titre=entry.title,
 132                                            texte=entry.summary_detail.value,
 133                                            url=entry.link, date=date)
 134
 135 class ActualiteQuerySet(SEPQuerySet):
 136
 137     def filter_date(self, min=None, max=None):
 138         qs = self
 139         if min:
 140             qs = qs.filter(date__gte=min)
 141         if max:
 142             qs = qs.filter(date__lte=max)
 143         return qs
 144
 145 class ActualiteSphinxQuerySet(SEPSphinxQuerySet):
 146
 147     def __init__(self, model=None):
 148         SEPSphinxQuerySet.__init__(self, model=model, index='savoirsenpartage_actualites',
 149                                    weights=dict(titre=3))
 150
 151 class ActualiteManager(SEPManager):
 152
 153     def get_query_set(self):
 154         return ActualiteQuerySet(self.model).filter(visible=True)
 155
 156     def get_sphinx_query_set(self):
 157         return ActualiteSphinxQuerySet(self.model).order_by('-date')
 158
 159     def filter_date(self, min=None, max=None):
 160         return self.get_query_set().filter_date(min=min, max=max)
 161
 162 class Actualite(models.Model):
 163     id = models.AutoField(primary_key=True, db_column='id_actualite')
 164     titre = models.CharField(max_length=765, db_column='titre_actualite')
 165     texte = models.TextField(db_column='texte_actualite')
 166     url = models.CharField(max_length=765, db_column='url_actualite')
 167     date = models.DateField(db_column='date_actualite')
 168     visible = models.BooleanField(db_column='visible_actualite', default=False)
 169     ancienid = models.IntegerField(db_column='ancienId_actualite', blank=True, null=True)
 170     source = models.ForeignKey(SourceActualite, blank=True, null=True, related_name='actualites')
 171     disciplines = models.ManyToManyField(Discipline, blank=True, related_name="actualites")
 172     regions = models.ManyToManyField(Region, blank=True, related_name="actualites", verbose_name='régions')
 173
 174     objects = ActualiteManager()
 175     all_objects = models.Manager()
 176
 177     class Meta:
 178         db_table = u'actualite'
 179         ordering = ["-date"]
 180
 181     def __unicode__ (self):
 182         return "%s" % (self.titre)
 183
 184     def assigner_disciplines(self, disciplines):
 185         self.disciplines.add(*disciplines)
 186
 187     def assigner_regions(self, regions):
 188         self.regions.add(*regions)
 189
 190 # Agenda
 191
 192 class EvenementQuerySet(SEPQuerySet):
 193
 194     def filter_type(self, type):
 195         return self.filter(type=type)
 196
 197     def filter_debut(self, min=None, max=None):
 198         qs = self
 199         if min:
 200             qs = qs.filter(debut__gte=min)
 201         if max:
 202             qs = qs.filter(debut__lt=max+datetime.timedelta(days=1))
 203         return qs
 204
 205 class EvenementSphinxQuerySet(SEPSphinxQuerySet):
 206
 207     def __init__(self, model=None):
 208         SEPSphinxQuerySet.__init__(self, model=model, index='savoirsenpartage_evenements',
 209                                    weights=dict(titre=3))
 210
 211     def filter_type(self, type):
 212         return self.add_to_query('@type "%s"' % type)
 213
 214     def filter_debut(self, min=None, max=None):
 215         qs = self
 216         if min:
 217             qs = qs.filter(debut__gte=min.toordinal()+365)
 218         if max:
 219             qs = qs.filter(debut__lte=max.toordinal()+365)
 220         return qs
 221
 222 class EvenementManager(SEPManager):
 223
 224     def get_query_set(self):
 225         return EvenementQuerySet(self.model).filter(approuve=True)
 226
 227     def get_sphinx_query_set(self):
 228         return EvenementSphinxQuerySet(self.model).order_by('-debut')
 229
 230     def filter_type(self, type):
 231         return self.get_query_set().filter_type(type)
 232
 233     def filter_debut(self, min=None, max=None):
 234         return self.get_query_set().filter_debut(min=min, max=max)
 235
 236 def build_time_zone_choices(pays=None):
 237     timezones = pytz.country_timezones[pays] if pays else pytz.common_timezones
 238     result = []
 239     now = datetime.datetime.now()
 240     for tzname in timezones:
 241         tz = pytz.timezone(tzname)
 242         fr_name = get_timezone_name(tz, locale='fr_FR')
 243         offset = tz.utcoffset(now)
 244         seconds = offset.seconds + offset.days * 86400
 245         (hours, minutes) = divmod(seconds // 60, 60)
 246         offset_str = 'UTC%+d:%d' % (hours, minutes) if minutes else 'UTC%+d' % hours
 247         result.append((seconds, tzname, '%s - %s' % (offset_str, fr_name)))
 248     result.sort()
 249     return [(x[1], x[2]) for x in result]
 250
 251 class Evenement(models.Model):
 252     TYPE_CHOICES = ((u'Colloque', u'Colloque'),
 253                     (u'Conférence', u'Conférence'),
 254                     (u'Appel à contribution', u'Appel à contribution'),
 255                     (u'Journée d\'étude', u'Journée d\'étude'),
 256                     (u'None', u'Autre'))
 257     TIME_ZONE_CHOICES = build_time_zone_choices()
 258
 259     uid = models.CharField(max_length=255, default=str(uuid.uuid1()))
 260     approuve = models.BooleanField(default=False, verbose_name=u'approuvé')
 261     titre = models.CharField(max_length=255)
 262     discipline = models.ForeignKey('Discipline', related_name = "discipline",
 263                                    blank = True, null = True)
 264     discipline_secondaire = models.ForeignKey('Discipline', related_name="discipline_secondaire",
 265                                               verbose_name=u"discipline secondaire",
 266                                               blank=True, null=True)
 267     mots_cles = models.TextField('Mots-Clés', blank=True, null=True)
 268     type = models.CharField(max_length=255, choices=TYPE_CHOICES)
 269     lieu = models.TextField()
 270     debut = models.DateTimeField(default=datetime.datetime.now)
 271     fin = models.DateTimeField(default=datetime.datetime.now)
 272     pays = models.ForeignKey(Pays, related_name='evenements', null=True, blank=True)
 273     fuseau = models.CharField(max_length=100, choices=TIME_ZONE_CHOICES, verbose_name='fuseau horaire')
 274     description = models.TextField(blank=True, null=True)
 275     contact = models.TextField(blank=True, null=True)
 276     url = models.CharField(max_length=255, blank=True, null=True)
 277     piece_jointe = models.FileField(upload_to='agenda/pj', blank=True, verbose_name='pièce jointe')
 278     regions = models.ManyToManyField(Region, blank=True, related_name="evenements", verbose_name='régions')
 279
 280     objects = EvenementManager()
 281     all_objects = models.Manager()
 282
 283     class Meta:
 284         ordering = ['-debut']
 285
 286     def __unicode__(self,):
 287         return "[%s] %s" % (self.uid, self.titre)
 288
 289     def duration_display(self):
 290         delta = self.fin - self.debut
 291         minutes, seconds = divmod(delta.seconds, 60)
 292         hours, minutes = divmod(minutes, 60)
 293         days = delta.days
 294         parts = []
 295         if days == 1:
 296             parts.append('1 jour')
 297         elif days > 1:
 298             parts.append('%d jours' % days)
 299         if hours == 1:
 300             parts.append('1 heure')
 301         elif hours > 1:
 302             parts.append('%d heures' % hours)
 303         if minutes == 1:
 304             parts.append('1 minute')
 305         elif minutes > 1:
 306             parts.append('%d minutes' % minutes)
 307         return ' '.join(parts)
 308
 309     def piece_jointe_display(self):
 310         return self.piece_jointe and os.path.basename(self.piece_jointe.name)
 311
 312     def clean(self):
 313         from django.core.exceptions import ValidationError
 314         if self.debut > self.fin:
 315             raise ValidationError('La date de fin ne doit pas être antérieure à la date de début')
 316
 317     def save(self, *args, **kwargs):
 318         """Sauvegarde l'objet dans django et le synchronise avec caldav s'il a été
 319         approuvé"""
 320         self.clean()
 321         super(Evenement, self).save(*args, **kwargs)
 322         self.update_vevent()
 323
 324     # methodes de commnunications avec CALDAV
 325     def as_ical(self,):
 326         """Retourne l'evenement django sous forme d'objet icalendar"""
 327         cal = vobject.iCalendar()
 328         cal.add('vevent')
 329
 330         # fournit son propre uid
 331         if self.uid in [None, ""]:
 332             self.uid = str(uuid.uuid1())
 333
 334         cal.vevent.add('uid').value = self.uid
 335
 336         cal.vevent.add('summary').value = self.titre
 337
 338         if self.mots_cles is None:
 339             kw = []
 340         else:
 341             kw = self.mots_cles.split(",")
 342
 343         try:
 344             kw.append(self.discipline.nom)
 345             kw.append(self.discipline_secondaire.nom)
 346             kw.append(self.type)
 347         except: pass
 348
 349         kw = [x.strip() for x in kw if len(x.strip()) > 0 and x is not None]
 350         for k in kw:
 351             cal.vevent.add('x-auf-keywords').value = k
 352
 353         description = self.description
 354         if len(kw) > 0:
 355             if len(self.description) > 0:
 356                 description += "\n"
 357             description += u"Mots-clés: " + ", ".join(kw)
 358
 359         cal.vevent.add('dtstart').value = combine(self.debut, pytz.timezone(self.fuseau))
 360         cal.vevent.add('dtend').value = combine(self.fin, pytz.timezone(self.fuseau))
 361         cal.vevent.add('created').value = combine(datetime.datetime.now(), "UTC")
 362         cal.vevent.add('dtstamp').value = combine(datetime.datetime.now(), "UTC")
 363         if len(description) > 0:
 364             cal.vevent.add('description').value = description
 365         if len(self.contact) > 0:
 366             cal.vevent.add('contact').value = self.contact
 367         if len(self.url) > 0:
 368             cal.vevent.add('url').value = self.url
 369         if len(self.lieu) > 0:
 370             cal.vevent.add('location').value = self.lieu
 371         if self.piece_jointe:
 372             url = self.piece_jointe.url
 373             if not url.startswith('http://'):
 374                 url = SITE_ROOT_URL + url
 375             cal.vevent.add('attach').value = url
 376         return cal
 377
 378     def update_vevent(self,):
 379         """Essaie de créer l'évènement sur le serveur ical.
 380         En cas de succès, l'évènement local devient donc inactif et approuvé"""
 381         try:
 382             if self.approuve:
 383                 event = self.as_ical()
 384                 client = caldav.DAVClient(CALENDRIER_URL)
 385                 cal = caldav.Calendar(client, url = CALENDRIER_URL)
 386                 e = caldav.Event(client, parent = cal, data = event.serialize(), id=self.uid)
 387                 e.save()
 388         except:
 389             self.approuve = False
 390
 391     def delete_vevent(self,):
 392         """Supprime l'evenement sur le serveur caldav"""
 393         try:
 394             if self.approuve:
 395                 event = self.as_ical()
 396                 client = caldav.DAVClient(CALENDRIER_URL)
 397                 cal = caldav.Calendar(client, url = CALENDRIER_URL)
 398                 e = cal.event(self.uid)
 399                 e.delete()
 400         except error.NotFoundError:
 401             pass
 402
 403     def assigner_regions(self, regions):
 404         self.regions.add(*regions)
 405
 406     def assigner_disciplines(self, disciplines):
 407         if len(disciplines) == 1:
 408             if self.discipline:
 409                 self.discipline_secondaire = disciplines[0]
 410             else:
 411                 self.discipline = disciplines[0]
 412         elif len(disciplines) >= 2:
 413             self.discipline = disciplines[0]
 414             self.discipline_secondaire = disciplines[1]
 415
 416 def delete_vevent(sender, instance, *args, **kwargs):
 417     # Surcharge du comportement de suppression
 418     # La méthode de connexion par signals est préférable à surcharger la méthode delete()
 419     # car dans le cas de la suppression par lots, cell-ci n'est pas invoquée
 420     instance.delete_vevent()
 421 pre_delete.connect(delete_vevent, sender=Evenement)
 422
 423 # Ressources
 424
 425 class ListSet(models.Model):
 426     spec = models.CharField(primary_key = True, max_length = 255)
 427     name = models.CharField(max_length = 255)
 428     server = models.CharField(max_length = 255)
 429     validated = models.BooleanField(default = True)
 430
 431     def __unicode__(self,):
 432         return self.name
 433
 434 class RecordSphinxQuerySet(SEPSphinxQuerySet):
 435
 436     def __init__(self, model=None):
 437         SEPSphinxQuerySet.__init__(self, model=model, index='savoirsenpartage_ressources',
 438                                    weights=dict(title=3))
 439
 440 class RecordManager(SEPManager):
 441
 442     def get_query_set(self):
 443         """Ne garder que les ressources validées et qui sont soit dans aucun
 444            listset ou au moins dans un listset validé."""
 445         qs = SEPQuerySet(self.model)
 446         qs = qs.filter(validated=True)
 447         qs = qs.filter(Q(listsets__isnull=True) | Q(listsets__validated=True))
 448         return qs.distinct()
 449
 450     def get_sphinx_query_set(self):
 451         return RecordSphinxQuerySet(self.model)
 452
 453 class Record(models.Model):
 454
 455     #fonctionnement interne
 456     id = models.AutoField(primary_key = True)
 457     server = models.CharField(max_length = 255, verbose_name=u'serveur')
 458     last_update = models.CharField(max_length = 255)
 459     last_checksum = models.CharField(max_length = 255)
 460     validated = models.BooleanField(default=True, verbose_name=u'validé')
 461
 462     #OAI
 463     title = models.TextField(null=True, blank=True, verbose_name=u'titre')
 464     creator = models.TextField(null=True, blank=True, verbose_name=u'auteur')
 465     description = models.TextField(null=True, blank=True)
 466     modified = models.CharField(max_length=255, null=True, blank=True)
 467     identifier = models.CharField(max_length = 255, null = True, blank = True, unique = True)
 468     uri = models.CharField(max_length = 255, null = True, blank = True, unique = True)
 469     source = models.TextField(null = True, blank = True)
 470     contributor = models.TextField(null = True, blank = True)
 471     subject = models.TextField(null=True, blank=True, verbose_name='sujet')
 472     publisher = models.TextField(null = True, blank = True)
 473     type = models.TextField(null = True, blank = True)
 474     format = models.TextField(null = True, blank = True)
 475     language = models.TextField(null = True, blank = True)
 476
 477     listsets = models.ManyToManyField(ListSet, null = True, blank = True)
 478
 479     #SEP 2 (aucune données récoltées)
 480     alt_title = models.TextField(null = True, blank = True)
 481     abstract = models.TextField(null = True, blank = True)
 482     creation = models.CharField(max_length = 255, null = True, blank = True)
 483     issued = models.CharField(max_length = 255, null = True, blank = True)
 484     isbn = models.TextField(null = True, blank = True)
 485     orig_lang = models.TextField(null = True, blank = True)
 486
 487     # Metadata AUF multivaluées
 488     disciplines = models.ManyToManyField(Discipline, blank=True)
 489     thematiques = models.ManyToManyField(Thematique, blank=True, verbose_name='thématiques')
 490     pays = models.ManyToManyField(Pays, blank=True)
 491     regions = models.ManyToManyField(Region, blank=True, verbose_name='régions')
 492
 493     # Managers
 494     objects = RecordManager()
 495     all_objects = models.Manager()
 496
 497     class Meta:
 498         verbose_name = 'ressource'
 499
 500     def __unicode__(self):
 501         return "[%s] %s" % (self.server, self.title)
 502
 503     def getServeurURL(self):
 504         """Retourne l'URL du serveur de provenance"""
 505         return RESOURCES[self.server]['url']
 506
 507     def est_complet(self):
 508         """teste si le record à toutes les données obligatoires"""
 509         return self.disciplines.count() > 0 and \
 510            self.thematiques.count() > 0 and \
 511            self.pays.count() > 0 and \
 512            self.regions.count() > 0
 513
 514     def assigner_regions(self, regions):
 515         self.regions.add(*regions)
 516
 517     def assigner_disciplines(self, disciplines):
 518         self.disciplines.add(*disciplines)
 519
 520 class Serveur(models.Model):
 521     """Identification d'un serveur d'ou proviennent les références"""
 522     nom = models.CharField(primary_key = True, max_length = 255)
 523
 524     def __unicode__(self,):
 525         return self.nom
 526
 527     def conf_2_db(self,):
 528         for k in RESOURCES.keys():
 529             s, created = Serveur.objects.get_or_create(nom=k)
 530             s.nom = k
 531             s.save()
 532
 533 class Profile(models.Model):
 534     user = models.ForeignKey(User, unique=True)
 535     serveurs = models.ManyToManyField(Serveur, null = True, blank = True)
 536
 537 class HarvestLog(models.Model):
 538     context = models.CharField(max_length = 255)
 539     name = models.CharField(max_length = 255)
 540     date = models.DateTimeField(auto_now = True)
 541     added = models.IntegerField(null = True, blank = True)
 542     updated = models.IntegerField(null = True, blank = True)
 543     processed = models.IntegerField(null = True, blank = True)
 544     record = models.ForeignKey(Record, null = True, blank = True)
 545
 546     @staticmethod
 547     def add(message):
 548         logger = HarvestLog()
 549         if message.has_key('record_id'):
 550             message['record'] = Record.all_objects.get(id=message['record_id'])
 551             del(message['record_id'])
 552
 553         for k,v in message.items():
 554             setattr(logger, k, v)
 555         logger.save()