philo/contrib/sobol/models.py

   1 import datetime
   2 import itertools
   3
   4 from django.conf import settings
   5 from django.conf.urls.defaults import patterns, url
   6 from django.contrib import messages
   7 from django.core.exceptions import ValidationError
   8 from django.core.validators import URLValidator
   9 from django.db import models
  10 from django.http import HttpResponseRedirect, Http404, HttpResponse
  11 from django.utils import simplejson as json
  12 from django.utils.datastructures import SortedDict
  13
  14 from philo.contrib.sobol import registry, get_search_instance
  15 from philo.contrib.sobol.forms import SearchForm
  16 from philo.contrib.sobol.utils import HASH_REDIRECT_GET_KEY, URL_REDIRECT_GET_KEY, SEARCH_ARG_GET_KEY, check_redirect_hash, RegistryIterator
  17 from philo.exceptions import ViewCanNotProvideSubpath
  18 from philo.models import MultiView, Page
  19 from philo.models.fields import SlugMultipleChoiceField
  20
  21 eventlet = None
  22 if getattr(settings, 'SOBOL_USE_EVENTLET', False):
  23         try:
  24                 import eventlet
  25         except:
  26                 pass
  27
  28
  29 class Search(models.Model):
  30         """Represents all attempts to search for a unique string."""
  31         #: The string which was searched for.
  32         string = models.TextField()
  33
  34         def __unicode__(self):
  35                 return self.string
  36
  37         def get_weighted_results(self, threshhold=None):
  38                 """
  39                 Returns a list of :class:`ResultURL` instances related to the search and ordered by decreasing weight. This will be cached on the instance.
  40
  41                 :param threshhold: The earliest datetime that a :class:`Click` can have been made on a related :class:`ResultURL` in order to be included in the weighted results (or ``None`` to include all :class:`Click`\ s and :class:`ResultURL`\ s).
  42
  43                 """
  44                 if not hasattr(self, '_weighted_results'):
  45                         result_qs = self.result_urls.all()
  46
  47                         if threshhold is not None:
  48                                 result_qs = result_qs.filter(counts__datetime__gte=threshhold)
  49
  50                         results = [result for result in result_qs]
  51
  52                         results.sort(cmp=lambda x,y: cmp(y.weight, x.weight))
  53
  54                         self._weighted_results = results
  55
  56                 return self._weighted_results
  57
  58         def get_favored_results(self, error=5, threshhold=None):
  59                 """
  60                 Calculates the set of most-favored results based on their weight. Evenly-weighted results will be grouped together and either added or excluded as a group.
  61
  62                 :param error: An arbitrary number; higher values will cause this method to be more reticent about adding new items to the favored results.
  63                 :param threshhold: Will be passed directly into :meth:`get_weighted_results`
  64
  65                 """
  66                 if not hasattr(self, '_favored_results'):
  67                         results = self.get_weighted_results(threshhold)
  68
  69                         grouped_results = SortedDict()
  70
  71                         for result in results:
  72                                 grouped_results.setdefault(result.weight, []).append(result)
  73
  74                         self._favored_results = []
  75
  76                         for value, subresults in grouped_results.items():
  77                                 cost = error * sum([(value - result.weight)**2 for result in self._favored_results])
  78                                 if value > cost:
  79                                         self._favored_results += subresults
  80                                 else:
  81                                         break
  82                         if len(self._favored_results) == len(results):
  83                                 self._favored_results = []
  84                 return self._favored_results
  85
  86         class Meta:
  87                 ordering = ['string']
  88                 verbose_name_plural = 'searches'
  89
  90
  91 class ResultURL(models.Model):
  92         """Represents a URL which has been selected one or more times for a :class:`Search`."""
  93         #: A :class:`ForeignKey` to the :class:`Search` which the :class:`ResultURL` is related to.
  94         search = models.ForeignKey(Search, related_name='result_urls')
  95         #: The URL which was selected.
  96         url = models.TextField(validators=[URLValidator()])
  97
  98         def __unicode__(self):
  99                 return self.url
 100
 101         def get_weight(self, threshhold=None):
 102                 """
 103                 Calculates, caches, and returns the weight of the :class:`ResultURL`.
 104
 105                 :param threshhold: The datetime limit before which :class:`Click`\ s will not contribute to the weight of the :class:`ResultURL`.
 106
 107                 """
 108                 if not hasattr(self, '_weight'):
 109                         clicks = self.clicks.all()
 110
 111                         if threshhold is not None:
 112                                 clicks = clicks.filter(datetime__gte=threshhold)
 113
 114                         self._weight = sum([click.weight for click in clicks])
 115
 116                 return self._weight
 117         weight = property(get_weight)
 118
 119         class Meta:
 120                 ordering = ['url']
 121
 122
 123 class Click(models.Model):
 124         """Represents a click on a :class:`ResultURL`."""
 125         #: A :class:`ForeignKey` to the :class:`ResultURL` which the :class:`Click` is related to.
 126         result = models.ForeignKey(ResultURL, related_name='clicks')
 127         #: The datetime when the click was registered in the system.
 128         datetime = models.DateTimeField()
 129
 130         def __unicode__(self):
 131                 return self.datetime.strftime('%B %d, %Y %H:%M:%S')
 132
 133         def get_weight(self, default=1, weighted=lambda value, days: value/days**2):
 134                 """Calculates and returns the weight of the :class:`Click`."""
 135                 if not hasattr(self, '_weight'):
 136                         days = (datetime.datetime.now() - self.datetime).days
 137                         if days < 0:
 138                                 raise ValueError("Click dates must be in the past.")
 139                         default = float(default)
 140                         if days == 0:
 141                                 self._weight = float(default)
 142                         else:
 143                                 self._weight = weighted(default, days)
 144                 return self._weight
 145         weight = property(get_weight)
 146
 147         def clean(self):
 148                 if self.datetime > datetime.datetime.now():
 149                         raise ValidationError("Click dates must be in the past.")
 150
 151         class Meta:
 152                 ordering = ['datetime']
 153                 get_latest_by = 'datetime'
 154
 155
 156 class RegistryChoiceField(SlugMultipleChoiceField):
 157         def _get_choices(self):
 158                 if isinstance(self._choices, RegistryIterator):
 159                         return self._choices.copy()
 160                 elif hasattr(self._choices, 'next'):
 161                         choices, self._choices = itertools.tee(self._choices)
 162                         return choices
 163                 else:
 164                         return self._choices
 165         choices = property(_get_choices)
 166
 167
 168 try:
 169         from south.modelsinspector import add_introspection_rules
 170 except ImportError:
 171         pass
 172 else:
 173         add_introspection_rules([], ["^philo\.contrib\.sobol\.models\.RegistryChoiceField"])
 174
 175
 176 class SearchView(MultiView):
 177         """Handles a view for the results of a search, anonymously tracks the selections made by end users, and provides an AJAX API for asynchronous search result loading. This can be particularly useful if some searches are slow."""
 178         #: :class:`ForeignKey` to a :class:`.Page` which will be used to render the search results.
 179         results_page = models.ForeignKey(Page, related_name='search_results_related')
 180         #: A :class:`.SlugMultipleChoiceField` whose choices are the contents of the :class:`.SearchRegistry`
 181         searches = RegistryChoiceField(choices=registry.iterchoices())
 182         #: A :class:`BooleanField` which controls whether or not the AJAX API is enabled.
 183         #:
 184         #: .. note:: If the AJAX API is enabled, a ``ajax_api_url`` attribute will be added to each search instance containing the url and get parameters for an AJAX request to retrieve results for that search.
 185         #:
 186         #: .. note:: Be careful not to access :attr:`search_instance.results <.BaseSearch.results>` if the AJAX API is enabled - otherwise the search will be run immediately rather than on the AJAX request.
 187         enable_ajax_api = models.BooleanField("Enable AJAX API", default=True)
 188         #: A :class:`CharField` containing the placeholder text which is intended to be used for the search box for the :class:`SearchView`. It is the template author's responsibility to make use of this information.
 189         placeholder_text = models.CharField(max_length=75, default="Search")
 190
 191         #: The form which will be used to validate the input to the search box for this :class:`SearchView`.
 192         search_form = SearchForm
 193
 194         def __unicode__(self):
 195                 return u"%s (%s)" % (self.placeholder_text, u", ".join([display for slug, display in registry.iterchoices() if slug in self.searches]))
 196
 197         def get_reverse_params(self, obj):
 198                 raise ViewCanNotProvideSubpath
 199
 200         @property
 201         def urlpatterns(self):
 202                 urlpatterns = patterns('',
 203                         url(r'^$', self.results_view, name='results'),
 204                 )
 205                 if self.enable_ajax_api:
 206                         urlpatterns += patterns('',
 207                                 url(r'^(?P<slug>[\w-]+)$', self.ajax_api_view, name='ajax_api_view')
 208                         )
 209                 return urlpatterns
 210
 211         def results_view(self, request, extra_context=None):
 212                 """
 213                 Renders :attr:`results_page` with a context containing an instance of :attr:`search_form`. If the form was submitted and was valid, then one of two things has happened:
 214
 215                 * A search has been initiated. In this case, a list of search instances will be added to the context as ``searches``. If :attr:`enable_ajax_api` is enabled, each instance will have an ``ajax_api_url`` attribute containing the url needed to make an AJAX request for the search results.
 216                 * A link has been chosen. In this case, corresponding :class:`Search`, :class:`ResultURL`, and :class:`Click` instances will be created and the user will be redirected to the link's actual url.
 217
 218                 """
 219                 results = None
 220
 221                 context = self.get_context()
 222                 context.update(extra_context or {})
 223
 224                 if SEARCH_ARG_GET_KEY in request.GET:
 225                         form = self.search_form(request.GET)
 226
 227                         if form.is_valid():
 228                                 search_string = request.GET[SEARCH_ARG_GET_KEY].lower()
 229                                 url = request.GET.get(URL_REDIRECT_GET_KEY)
 230                                 hash = request.GET.get(HASH_REDIRECT_GET_KEY)
 231
 232                                 if url and hash:
 233                                         if check_redirect_hash(hash, search_string, url):
 234                                                 # Create the necessary models
 235                                                 search = Search.objects.get_or_create(string=search_string)[0]
 236                                                 result_url = search.result_urls.get_or_create(url=url)[0]
 237                                                 result_url.clicks.create(datetime=datetime.datetime.now())
 238                                                 return HttpResponseRedirect(url)
 239                                         else:
 240                                                 messages.add_message(request, messages.INFO, "The link you followed had been tampered with. Here are all the results for your search term instead!")
 241                                                 # TODO: Should search_string be escaped here?
 242                                                 return HttpResponseRedirect("%s?%s=%s" % (request.path, SEARCH_ARG_GET_KEY, search_string))
 243
 244                                 search_instances = []
 245                                 for slug in self.searches:
 246                                         if slug in registry:
 247                                                 search_instance = get_search_instance(slug, search_string)
 248                                                 search_instances.append(search_instance)
 249
 250                                                 if self.enable_ajax_api:
 251                                                         search_instance.ajax_api_url = "%s?%s=%s" % (self.reverse('ajax_api_view', kwargs={'slug': slug}, node=request.node), SEARCH_ARG_GET_KEY, search_string)
 252
 253                                 if eventlet and not self.enable_ajax_api:
 254                                         pool = eventlet.GreenPool()
 255                                         for instance in search_instances:
 256                                                 pool.spawn_n(lambda x: x.results, search_instance)
 257                                         pool.waitall()
 258
 259                                 context.update({
 260                                         'searches': search_instances,
 261                                         'favored_results': []
 262                                 })
 263
 264                                 try:
 265                                         search = Search.objects.get(string=search_string)
 266                                 except Search.DoesNotExist:
 267                                         pass
 268                                 else:
 269                                         context['favored_results'] = [r.url for r in search.get_favored_results()]
 270                 else:
 271                         form = SearchForm()
 272
 273                 context.update({
 274                         'form': form
 275                 })
 276                 return self.results_page.render_to_response(request, extra_context=context)
 277
 278         def ajax_api_view(self, request, slug, extra_context=None):
 279                 """
 280                 Returns a JSON object containing the following variables:
 281
 282                 search
 283                         Contains the slug for the search.
 284                 results
 285                         Contains the results of :meth:`.Result.get_context` for each result.
 286                 rendered
 287                         Contains the results of :meth:`.Result.render` for each result.
 288                 hasMoreResults
 289                         ``True`` or ``False`` whether the search has more results according to :meth:`BaseSearch.has_more_results`
 290                 moreResultsURL
 291                         Contains ``None`` or a querystring which, once accessed, will note the :class:`Click` and redirect the user to a page containing more results.
 292
 293                 """
 294                 search_string = request.GET.get(SEARCH_ARG_GET_KEY)
 295
 296                 if not request.is_ajax() or not self.enable_ajax_api or slug not in registry or slug not in self.searches or search_string is None:
 297                         raise Http404
 298
 299                 search_instance = get_search_instance(slug, search_string)
 300
 301                 return HttpResponse(json.dumps({
 302                         'search': search_instance.slug,
 303                         'results': [result.get_context() for result in search_instance.results],
 304                         'hasMoreResults': search_instance.has_more_results,
 305                         'moreResultsURL': search_instance.more_results_url,
 306                 }), mimetype="application/json")