Strip junk from Pixiv URLs before indexing, #726

db/migrate/20130328092739_change_source_pattern_index_on_posts.rb

@@ -0,0 +1,24 @@
+class ChangeSourcePatternIndexOnPosts < ActiveRecord::Migration
+  def up
+    execute "set statement_timeout = 0"
+    execute "DROP INDEX index_posts_on_pixiv_suffix;"
+    execute "DROP INDEX index_posts_on_source_pattern;"
+    execute "CREATE FUNCTION SourcePattern(src text) RETURNS text AS $$
+               BEGIN
+                 RETURN regexp_replace(src, '^[^/]*(//)?[^/]*\.pixiv\.net/img.*(/[^/]*/[^/]*)$', 'pixiv\2');
+               END;
+             $$ LANGUAGE plpgsql IMMUTABLE RETURNS NULL ON NULL INPUT;"
+    execute "CREATE INDEX index_posts_on_source_pattern ON posts USING btree
+             ((SourcePattern(source)) text_pattern_ops);"
+  end
+
+  def down
+    execute "set statement_timeout = 0"
+    execute "DROP INDEX index_posts_on_source_pattern;"
+    execute "DROP FUNCTION SourcePattern(text);"
+    execute "CREATE INDEX index_posts_on_source_pattern ON posts USING btree
+             (source text_pattern_ops);"
+    execute "CREATE INDEX index_posts_on_pixiv_suffix ON posts USING btree
+             ((substring(source, 'pixiv.net/img.*/([^/]*/[^/]*)$')) text_pattern_ops);"
+  end
+end