From 5f1824742d5adc2968532f67b9084974e995708c Mon Sep 17 00:00:00 2001
From: Schuyler Erle <schuyler@entropyfree.com>
Date: Sat, 27 Aug 2011 18:26:00 -0700
Subject: [PATCH] Add .similar_features() method to Feature

---
 gazetteer/places/models.py | 30 +++++++++++++++++++++++++++++-
 1 file changed, 29 insertions(+), 1 deletion(-)

diff --git a/gazetteer/places/models.py b/gazetteer/places/models.py
index 1257507..e5057fc 100644
--- a/gazetteer/places/models.py
+++ b/gazetteer/places/models.py
@@ -22,7 +22,7 @@ class FeatureSearchManager(manager.GeoManager):
         if text:
             text = text.replace("'", "''") # escape the '
             # use the pg_trgm index
-            qset = qset.extra(where=["preferred_name %% '%s']" % text)
+            qset = qset.extra(where=["preferred_name %% '%s'" % text])
         return qset
 
 class Feature(models.Model):
@@ -65,6 +65,34 @@ class Feature(models.Model):
             return ''
     time_end.short_description = "End Date"
 
+    def similar_features(self, max_distance=15000, scale_factor=2000):
+        cursor = connection.cursor()
+        name = unicode(self).replace("'", "''") # escape '
+        cursor.execute("""
+            SELECT *, %f * similarity / distance AS score FROM (
+                SELECT url, preferred_name, feature_type,
+                       admin1, admin2, is_primary,
+                       similarity(preferred_name, '%s') AS similarity,
+                       st_distance_sphere(geometry, '%s') AS distance
+                       FROM places_feature
+                       WHERE geometry && st_buffer('%s', %f)
+                       AND preferred_name % '%s'
+                ) AS whatever
+                WHERE %f * similarity / distance >= 1.0
+                ORDER BY similarity / distance DESC"""
+            % (scale_factor, name, self.geometry, self.geometry,
+               max_distance*0.00001, name, scale_factor)
+        )
+        result_list = []
+        fields = ('url', 'preferred_name', 'feature_type', 'admin1', 'admin2', 'is_primary')
+        for row in cursor.fetchall():
+            vals = dict(zip(fields, row[:len(fields)]))
+            p = type(self)(**vals)
+            p.similarity = row[-3]
+            p.distance = row[-2]
+            p.score = row[-1]
+            result_list.append(p)
+        return result_list
 
 LANGUAGE_CHOICES = (
     ('en', 'English'),