panikweb/search.py

   1 import haystack.backends.solr_backend
   2 from haystack.query import SearchQuerySet, RelatedSearchQuerySet
   3 from haystack.constants import ID, DJANGO_CT, DJANGO_ID
   4 from haystack.utils import get_identifier
   5 from haystack.backends import EmptyResults
   6
   7 from pysolr import SolrError
   8
   9 from django import forms
  10 from django.conf import settings
  11 from django.utils.translation import ugettext_lazy as _
  12
  13 from emissions.models import Emission, Episode, NewsItem, SoundFile
  14
  15
  16 class MoreLikeThisSearchQuerySet(SearchQuerySet):
  17     def more_like_this(self, model_instance, **kwargs):
  18         clone = self._clone()
  19         clone.query.more_like_this(model_instance, **kwargs)
  20         return clone
  21
  22
  23 class CustomSolrSearchQuery(haystack.backends.solr_backend.SolrSearchQuery):
  24     def more_like_this(self, model_instance, **kwargs):
  25         self._more_like_this = True
  26         self._mlt_instance = model_instance
  27
  28     def run_mlt(self, **kwargs):
  29         """Builds and executes the query. Returns a list of search results."""
  30         if self._more_like_this is False or self._mlt_instance is None:
  31             raise MoreLikeThisError("No instance was provided to determine 'More Like This' results.")
  32
  33         additional_query_string = self.build_query()
  34         search_kwargs = {
  35             'start_offset': self.start_offset,
  36             'result_class': self.result_class,
  37             'models': self.models,
  38             'mlt.fl': 'tags',
  39         }
  40
  41         if self.end_offset is not None:
  42             search_kwargs['end_offset'] = self.end_offset - self.start_offset
  43
  44         results = self.backend.more_like_this(self._mlt_instance, additional_query_string, **search_kwargs)
  45         self._results = results.get('results', [])
  46         self._hit_count = results.get('hits', 0)
  47
  48
  49 class CustomSolrSearchBackend(haystack.backends.solr_backend.SolrSearchBackend):
  50     def more_like_this(self, model_instance, additional_query_string=None,
  51                        start_offset=0, end_offset=None, models=None,
  52                        limit_to_registered_models=None, result_class=None, **kwargs):
  53         from haystack import connections
  54
  55         # Deferred models will have a different class ("RealClass_Deferred_fieldname")
  56         # which won't be in our registry:
  57         model_klass = model_instance._meta.concrete_model
  58
  59         index = connections[self.connection_alias].get_unified_index().get_index(model_klass)
  60         field_name = index.get_content_field()
  61         params = {
  62             'fl': '*,score',
  63             'mlt.fl': 'text,title',
  64             'mlt.qf': 'text^0.3 tags^2.0 title^1.0',
  65         }
  66
  67         if start_offset is not None:
  68             params['start'] = start_offset
  69
  70         if end_offset is not None:
  71             params['rows'] = end_offset
  72
  73         narrow_queries = set()
  74
  75         if limit_to_registered_models is None:
  76             limit_to_registered_models = getattr(settings, 'HAYSTACK_LIMIT_TO_REGISTERED_MODELS', True)
  77
  78         if models and len(models):
  79             model_choices = sorted(['%s.%s' % (model._meta.app_label, model._meta.module_name) for model in models])
  80         elif limit_to_registered_models:
  81             # Using narrow queries, limit the results to only models handled
  82             # with the current routers.
  83             model_choices = self.build_models_list()
  84         else:
  85             model_choices = []
  86
  87         if len(model_choices) > 0:
  88             if narrow_queries is None:
  89                 narrow_queries = set()
  90
  91             narrow_queries.add('%s:(%s)' % (DJANGO_CT, ' OR '.join(model_choices)))
  92
  93         if additional_query_string:
  94             narrow_queries.add(additional_query_string)
  95
  96         if narrow_queries:
  97             params['fq'] = list(narrow_queries)
  98
  99         query = "%s:%s" % (ID, get_identifier(model_instance))
 100
 101         try:
 102             raw_results = self.conn.more_like_this(query, field_name, **params)
 103         except (IOError, SolrError) as e:
 104             if not self.silently_fail:
 105                 raise
 106
 107             self.log.error("Failed to fetch More Like This from Solr for document '%s': %s", query, e)
 108             raw_results = EmptyResults()
 109
 110         return self._process_results(raw_results, result_class=result_class)
 111
 112
 113 haystack.backends.solr_backend.SolrEngine.query = CustomSolrSearchQuery
 114 haystack.backends.solr_backend.SolrEngine.backend = CustomSolrSearchBackend
 115
 116 import haystack.views
 117 from haystack.views import search_view_factory, FacetedSearchView
 118 from haystack.forms import FacetedSearchForm, SearchForm
 119
 120
 121 class GlobalSearchForm(FacetedSearchForm):
 122     def no_query_found(self):
 123         sqs = super(GlobalSearchForm, self).no_query_found()
 124         if self.selected_facets:
 125             sqs = self.searchqueryset.all()
 126             for facet in self.selected_facets:
 127                 if ":" not in facet:
 128                     continue
 129                 field, value = facet.split(":", 1)
 130                 if value:
 131                     sqs = sqs.narrow(u'%s:"%s"' % (field, sqs.query.clean(value)))
 132         return sqs
 133
 134
 135 class SearchView(FacetedSearchView):
 136     def extra_context(self):
 137         context = super(SearchView, self).extra_context()
 138         context['sectionName'] = 'Search'
 139         if self.request.GET.getlist('selected_facets'):
 140             context['facets_qs'] = '&selected_facets=' + '&'.join(self.request.GET.getlist('selected_facets'))
 141         context['selected_categories'] = [
 142                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 143                 if x.startswith('categories_exact')]
 144         context['selected_tags'] = [
 145                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 146                 if x.startswith('tags_exact')]
 147         if 'categories' in context['facets'].get('fields', []):
 148             context['facets']['fields']['categories'] = [x for x in
 149                     context['facets']['fields']['categories'] if x[1] > 0]
 150             context['facets']['fields']['categories'].sort()
 151         if 'tags' in context['facets'].get('fields', []):
 152             context['facets']['fields']['tags'] = [x for x in
 153                     context['facets']['fields']['tags'] if x[1] > 0]
 154         return context
 155
 156 sqs = SearchQuerySet().models(Emission, Episode, NewsItem).facet('categories').facet('tags')
 157
 158 view = search_view_factory(SearchView,
 159         form_class=GlobalSearchForm,
 160         searchqueryset=sqs)
 161
 162
 163
 164 class ListenArchivesForm(FacetedSearchForm):
 165     q = forms.CharField(required=False, label='')
 166
 167     def no_query_found(self):
 168         return self.searchqueryset.all()
 169
 170     def search(self):
 171         sqs = super(ListenArchivesForm, self).search()
 172         return sqs.load_all()
 173
 174
 175 class ListenArchivesView(FacetedSearchView):
 176     template = 'listen/archives.html'
 177
 178     def __init__(self):
 179         sqs = RelatedSearchQuerySet().models(SoundFile).facet('format').facet('tags').order_by('-date')
 180         super(ListenArchivesView, self).__init__(searchqueryset=sqs,
 181                 form_class=ListenArchivesForm, results_per_page=20)
 182
 183     def extra_context(self):
 184         context = super(ListenArchivesView, self).extra_context()
 185         context['sectionName'] = "Listen"
 186         if self.request.GET.getlist('selected_facets'):
 187             context['facets_qs'] = '&selected_facets=' + '&'.join(self.request.GET.getlist('selected_facets'))
 188         context['selected_format'] = [
 189                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 190                 if x.startswith('format_exact')]
 191         context['selected_tags'] = [
 192                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 193                 if x.startswith('tags_exact')]
 194         if 'format' in context['facets'].get('fields', []):
 195             context['facets']['fields']['format'] = [x for x in
 196                     context['facets']['fields']['format'] if x[1] > 0]
 197             context['facets']['fields']['format'].sort()
 198         if 'tags' in context['facets'].get('fields', []):
 199             context['facets']['fields']['tags'] = [x for x in
 200                     context['facets']['fields']['tags'] if x[1] > 0]
 201         return context
 202
 203 listenArchives = search_view_factory(ListenArchivesView)
 204
 205
 206 class NewsArchivesForm(FacetedSearchForm):
 207     q = forms.CharField(required=False, label='')
 208
 209     def no_query_found(self):
 210         return self.searchqueryset.all()
 211
 212     def search(self):
 213         sqs = super(NewsArchivesForm, self).search()
 214         return sqs.load_all()
 215
 216
 217 class NewsArchivesView(FacetedSearchView):
 218     template = 'news/archives.html'
 219
 220     def __init__(self):
 221         sqs = RelatedSearchQuerySet().models(NewsItem).facet('news_categories').facet('tags').order_by('-date')
 222         super(NewsArchivesView, self).__init__(searchqueryset=sqs,
 223                 form_class=NewsArchivesForm, results_per_page=20)
 224
 225     def extra_context(self):
 226         context = super(NewsArchivesView, self).extra_context()
 227         if self.request.GET.getlist('selected_facets'):
 228             context['facets_qs'] = '&selected_facets=' + '&'.join(self.request.GET.getlist('selected_facets'))
 229         context['sectionName'] = 'News'
 230         context['selected_news_categories'] = [
 231                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 232                 if x.startswith('news_categories_exact')]
 233         context['selected_tags'] = [
 234                 x.split(':', 1)[1] for x in self.request.GET.getlist('selected_facets')
 235                 if x.startswith('tags_exact')]
 236         if 'news_categories' in context['facets'].get('fields', []):
 237             context['facets']['fields']['news_categories'] = [x for x in
 238                     context['facets']['fields']['news_categories'] if x[1] > 0]
 239             context['facets']['fields']['news_categories'].sort()
 240         if 'tags' in context['facets'].get('fields', []):
 241             context['facets']['fields']['tags'] = [x for x in
 242                     context['facets']['fields']['tags'] if x[1] > 0]
 243         return context
 244
 245 newsArchives = search_view_factory(NewsArchivesView)