use language specific analyzer for body

shop/search/analyzers.py

@@ -0,0 +1,19 @@
+from elasticsearch_dsl.analysis import analyzer, token_filter
+
+html_strip = analyzer('html_strip',
+    tokenizer='standard',
+    filter=['lowercase', 'stop', 'snowball'],
+    char_filter=['html_strip'],
+)
+
+german_analyzer = analyzer('german_analyzer',
+    type='custom',
+    tokenizer='standard',
+    filter=[
+        'lowercase',
+        token_filter('asciifolding', type='asciifolding', preserve_original=False),
+        token_filter('german_stop', type='stop', language='german'),
+        token_filter('german_stemmer', type='snowball', language='german'),
+    ],
+    char_filter=['html_strip'],
+)

shop/search/documents.py

@@ -5,38 +5,26 @@
 from django.utils import translation
 
 from django_elasticsearch_dsl import fields, Document, Index
-from elasticsearch_dsl import analyzer
 
 from shop.models.product import ProductModel
 
 
-html_strip = analyzer(
-    'html_strip',
-    tokenizer='standard',
-    filter=['lowercase', 'stop', 'snowball'],
-    char_filter=['html_strip'],
-)
-
-
 class _ProductDocument(Document):
     product_code = fields.KeywordField(
         multi=True,
-        boost=5,
+        boost=3,
     )
 
     product_name = fields.TextField(
-        boost=3,
+        boost=2,
     )
 
     product_type = fields.TextField()
 
-    body = fields.TextField(
-        analyzer=html_strip,
-    )
-
     class Django:
         model = ProductModel
         fields = ['id']
+        ignore_signals = True  # performed by ProductModel.save()
 
     def __str__(self):
         return "{} {}: {}".format(self.product_type, self.id, self.product_name)
@@ -46,6 +34,9 @@ def get_queryset(self):
         return queryset.filter(active=True)
 
     def prepare_product_code(self, instance):
+        """
+        Create a list of textual representation for product codes.
+        """
         has_valid_product_code = lambda obj: isinstance(getattr(obj, 'product_code', None), str)
         variants = instance.get_product_variants()
         product_codes = [v.product_code for v in variants if has_valid_product_code(v)]
@@ -69,11 +60,17 @@ def prepare_body(self, instance):
         return body
 
     def update(self, thing, refresh=None, action='index', parallel=False, **kwargs):
-        with translation.override(self._language):
+        if self._language:
+            with translation.override(self._language):
+                super().update(thing, refresh=None, action='index', parallel=False, **kwargs)
+        else:
             super().update(thing, refresh=None, action='index', parallel=False, **kwargs)
 
 
 class ProductDocument:
+    """
+    Factory for building an elasticsearch-dsl Document class. This class
+    """
     def __new__(cls, language=None, settings=None, analyzer=None):
         if language:
             index_name = 'products-{}'.format(language.lower())
@@ -84,8 +81,8 @@ def __new__(cls, language=None, settings=None, analyzer=None):
         products_index = Index(index_name)
         if settings:
             products_index.settings(**settings)
-        if analyzer:
-            products_index.analyzer(analyzer)
-        doc_class = type(doc_name, (_ProductDocument,), {'_language': language})
+        kwargs = {'analyzer': analyzer} if analyzer else {}
+        attrs = {'_language': language, 'body': fields.TextField(**kwargs)}
+        doc_class = type(doc_name, (_ProductDocument,), attrs)
         products_index.document(doc_class)
         return doc_class

shop/search/mixins.py

@@ -16,7 +16,8 @@ def get_document(self, language):
 
 class ProductSearchViewMixin(SearchViewMixin):
     """
-    Mixin class to be added to the ProductListView so that autocompletion works.
+    Mixin class to be added to the ProductListView to restrict that list to entities matching
+    the query string.
     """
     search_fields = ['product_name', 'product_code']