Using thread score to help in thread relevance

Sergio Oliveira
1 parent d70cc0d4
Showing 3 changed files with 17 additions and 10 deletions Show diff stats
src/proxy/search_indexes.py
src/search/base_indexes.py
src/super_archives/search_indexes.py
@@ -143,11 +143,9 @@ class RevisionIndex(BaseIndex, indexes.Indexable):
     def get_updated_field(self):
         return 'created'
  
-    def get_boost(self, obj, data):
-        if obj.hits <= 10:
-            data['boost'] = 0.8
-        else:
-            data['boost'] = math.log(obj.hits) * 0.8
+    def get_boost(self, obj):
+        boost = super(RevisionIndex, self).get_boost(obj)
+        return boost * 0.8
  
     def prepare_icon_name(self, obj):
         return u'align-right'
@@ -22,15 +22,15 @@ class BaseIndex(indexes.SearchIndex):
     def get_updated_field(self):
         return 'modified'
  
-    def get_boost(self, obj, data):
+    def get_boost(self, obj):
         if obj.hits <= 10:
-            data['boost'] = 1
-        else:
-            data['boost'] = math.log(obj.hits)
+            return 1
+
+        return math.log(obj.hits)
  
     def prepare(self, obj):
         data = super(BaseIndex, self).prepare(obj)
-        self.get_boost(obj, data)
+        data['boost'] = self.get_boost(obj)
         return data
  
     def prepare_author(self, obj):
@@ -25,6 +25,7 @@ class ThreadIndex(BaseIndex, indexes.Indexable):
         model_attr='mailinglist__get_absolute_url',
         indexed=False,
     )
+    score = indexes.IntegerField(model_attr='score')
  
     def get_model(self):
         return Thread
@@ -64,3 +65,11 @@ class ThreadIndex(BaseIndex, indexes.Indexable):
         return self.get_model().objects.filter(
             spam=False
         ).exclude(subject_token='')
+
+    def get_boost(self, obj):
+        boost = super(ThreadIndex, self).get_boost(obj)
+
+        if obj.score >= 20:
+            boost = boost * math.log(obj.score, 20)
+
+        return boost
...	...	@@ -143,11 +143,9 @@ class RevisionIndex(BaseIndex, indexes.Indexable):
143	143	def get_updated_field(self):
144	144	return 'created'
145	145
146		- def get_boost(self, obj, data):
147		- if obj.hits <= 10:
148		- data['boost'] = 0.8
149		- else:
150		- data['boost'] = math.log(obj.hits) * 0.8
	146	+ def get_boost(self, obj):
	147	+ boost = super(RevisionIndex, self).get_boost(obj)
	148	+ return boost * 0.8
151	149
152	150	def prepare_icon_name(self, obj):
153	151	return u'align-right'
...	...
...	...	@@ -22,15 +22,15 @@ class BaseIndex(indexes.SearchIndex):
22	22	def get_updated_field(self):
23	23	return 'modified'
24	24
25		- def get_boost(self, obj, data):
	25	+ def get_boost(self, obj):
26	26	if obj.hits <= 10:
27		- data['boost'] = 1
28		- else:
29		- data['boost'] = math.log(obj.hits)
	27	+ return 1
	28	+
	29	+ return math.log(obj.hits)
30	30
31	31	def prepare(self, obj):
32	32	data = super(BaseIndex, self).prepare(obj)
33		- self.get_boost(obj, data)
	33	+ data['boost'] = self.get_boost(obj)
34	34	return data
35	35
36	36	def prepare_author(self, obj):
...	...
...	...	@@ -25,6 +25,7 @@ class ThreadIndex(BaseIndex, indexes.Indexable):
25	25	model_attr='mailinglist__get_absolute_url',
26	26	indexed=False,
27	27	)
	28	+ score = indexes.IntegerField(model_attr='score')
28	29
29	30	def get_model(self):
30	31	return Thread
...	...	@@ -64,3 +65,11 @@ class ThreadIndex(BaseIndex, indexes.Indexable):
64	65	return self.get_model().objects.filter(
65	66	spam=False
66	67	).exclude(subject_token='')
	68	+
	69	+ def get_boost(self, obj):
	70	+ boost = super(ThreadIndex, self).get_boost(obj)
	71	+
	72	+ if obj.score >= 20:
	73	+ boost = boost * math.log(obj.score, 20)
	74	+
	75	+ return boost
...	...