Merge branch 'feature-1738-add-before-index-extension'

ckan/lib/search/index.py

@@ -7,6 +7,8 @@
 
 from common import SearchIndexError, make_connection
 from ckan.model import PackageRelationship
+from ckan.plugins import (PluginImplementations,
+                          IPackageController)
 
 log = logging.getLogger(__name__)
 
@@ -138,6 +140,11 @@ def index_package(self, pkg_dict):
         import hashlib
         pkg_dict['index_id'] = hashlib.md5('%s%s' % (pkg_dict['id'],config.get('ckan.site_id'))).hexdigest()
 
+        for item in PluginImplementations(IPackageController):
+            pkg_dict = item.before_index(pkg_dict)
+
+        assert pkg_dict, 'Plugin must return non empty package dict on index'
+
         # send to solr:  
         try:
             conn.add_many([pkg_dict])

ckan/logic/action/create.py

@@ -66,6 +66,7 @@ def package_create(context, data_dict):
     model.setup_default_user_roles(pkg, admins)
     # Needed to let extensions know the package id
     model.Session.flush()
+
     for item in PluginImplementations(IPackageController):
         item.create(pkg)
 

ckan/plugins/interfaces.py

@@ -262,6 +262,16 @@ def after_search(self, search_results, search_params):
 
         return search_results
 
+    def before_index(self, pkg_dict):
+        '''
+             Extensions will recieve what will be given to the solr for indexing.
+             This is essentially a flattened dict (except for multlivlaued fields such as tags
+             of all the terms sent to the indexer.  The extension can modify this by returning
+             an altered version.
+        '''
+        return pkg_dict
+
+
 class IPluginObserver(Interface):
     """
     Plugin to the plugin loading mechanism

ckan/tests/functional/test_package.py

@@ -57,6 +57,10 @@ def after_search(self, search_results, search_params):
         self.calls['after_search'] += 1
         return search_results
 
+    def before_index(self, search_params):
+        self.calls['before_index'] += 1
+        return search_params
+
 
 existing_extra_html = ('<label class="field_opt" for="Package-%(package_id)s-extras-%(key)s">%(capitalized_key)s</label>', '<input id="Package-%(package_id)s-extras-%(key)s" name="Package-%(package_id)s-extras-%(key)s" size="20" type="text" value="%(value)s">')
 

ckan/tests/logic/test_action.py

@@ -1193,6 +1193,10 @@ def test_4_sort_by_metadata_modified(self):
 class MockPackageSearchPlugin(SingletonPlugin):
     implements(IPackageController, inherit=True)
 
+    def before_index(self, data_dict):
+        data_dict['extras_test'] = 'abcabcabc'
+        return data_dict
+
     def before_search(self, search_params):
         if 'extras' in search_params and 'ext_avoid' in search_params['extras']:
             assert 'q' in search_params
@@ -1268,3 +1272,28 @@ def test_search_plugin_interface_abort(self):
         assert res_dict['count'] == 0
         assert len(res_dict['results']) == 0
         plugins.unload(plugin)
+
+    def test_before_index(self):
+        plugin = MockPackageSearchPlugin()
+        plugins.load(plugin)
+        # no datasets get aaaaaaaa
+        search_params = '%s=1' % json.dumps({
+            'q': 'aaaaaaaa',
+        })
+
+        res = self.app.post('/api/action/package_search', params=search_params)
+
+        res_dict = json.loads(res.body)['result']
+        assert res_dict['count'] == 0 
+        assert len(res_dict['results']) == 0
+        plugins.unload(plugin)
+
+        # all datasets should get abcabcabc
+        search_params = '%s=1' % json.dumps({
+            'q': 'abcabcabc',
+        })
+        res = self.app.post('/api/action/package_search', params=search_params)
+
+        res_dict = json.loads(res.body)['result']
+        assert res_dict['count'] == 2
+        assert len(res_dict['results']) == 2