onchanged hooks

oyster/core.py

@@ -53,7 +53,8 @@ class Kernel(object):
             self.storage[name] = StorageCls(self)
 
         # set document classes
-        _doc_class_fields = ('update_mins', 'storage_engine')
+        _doc_class_fields = ('update_mins', 'storage_engine',
+                             'onchanged')
         self.doc_classes = doc_classes or {}
         for dc_name, dc_props in self.doc_classes.iteritems():
             for key in _doc_class_fields:
@@ -165,6 +166,9 @@ class Kernel(object):
                                     'storage_key': storage_id,
                                     'storage_type': storage.storage_type,
                                    })
+            # fire off onchanged functions
+            for onchanged in doc_class['onchanged']:
+                onchanged(doc)
 
         if error:
             # if there's been an error, increment the consecutive_errors count

oyster/tasks.py

@@ -14,7 +14,7 @@ class UpdateTask(Task):
     def run(self, doc_id):
         doc = kernel.db.tracked.find_one({'_id': doc_id})
         kernel.update(doc)
-        for hook in doc.get('post_update_hooks', []):
+        for task in doc.get('post_update_tasks', []):
             send_task(hook, (doc_id,))
         kernel.db.status.update({}, {'$inc': {'update_queue': -1}})
 

oyster/tests/test_kernel.py

@@ -7,14 +7,26 @@ import pymongo
 
 from oyster.core import Kernel
 
+def hook_fired(doc):
+    doc['hook_fired'] = doc.get('hook_fired', 0) + 1
+
+RANDOM_URL = 'http://www.random.org/integers/?num=1&min=-1000000000&max=1000000000&col=1&base=10&format=plain&rnd=new'
 
 class KernelTests(TestCase):
 
     def setUp(self):
         doc_classes = {'default':
-                        {'update_mins': 30, 'storage_engine': 'dummy'},
+                        {'update_mins': 30, 'storage_engine': 'dummy',
+                         'onchanged': []
+                        },
                        'fast-update':
-                        {'update_mins': 1/60., 'storage_engine': 'dummy'},
+                        {'update_mins': 1/60., 'storage_engine': 'dummy',
+                         'onchanged': []
+                        },
+                       'change-hook':
+                        {'update_mins': 30, 'storage_engine': 'dummy',
+                         'onchanged': [hook_fired]
+                        }
                       }
         self.kernel = Kernel(mongo_db='oyster_test', retry_wait_minutes=1/60.,
                              doc_classes=doc_classes)
@@ -136,6 +148,33 @@ class KernelTests(TestCase):
         assert obj['consecutive_errors'] == 2
 
 
+    def test_update_onchanged_fire_only_on_change(self):
+        self.kernel.track_url('http://example.com', 'change-hook')
+        obj = self.kernel.db.tracked.find_one()
+        self.kernel.update(obj)
+
+        doc = self.kernel.db.tracked.find_one()
+        assert doc['hook_fired'] == 1
+
+        # again, we rely on example.com not updating
+        self.kernel.update(obj)
+        doc = self.kernel.db.tracked.find_one()
+        assert doc['hook_fired'] == 1
+
+    def test_update_onchanged_fire_again_on_change(self):
+        self.kernel.track_url(RANDOM_URL, 'change-hook')
+        obj = self.kernel.db.tracked.find_one()
+        self.kernel.update(obj)
+
+        doc = self.kernel.db.tracked.find_one()
+        assert doc['hook_fired'] == 1
+
+        # we rely on this URL updating
+        self.kernel.update(obj)
+        doc = self.kernel.db.tracked.find_one()
+        assert doc['hook_fired'] == 2
+
+
     def test_get_update_queue(self):
         self.kernel.track_url('never-updates', 'fast-update')
         self.kernel.track_url('bad-uri', 'fast-update')