Deployment on other Wikimedia wikis

deployment/celery.yaml

@@ -21,7 +21,7 @@ spec:
     spec:
       containers:
         - name: celery
-          image: docker-registry.tools.wmflabs.org/toollabs-python37-base:latest
+          image: docker-registry.tools.wmflabs.org/toolforge-python37-sssd-base:latest
           command: [ "/data/project/editgroups/www/python/src/tasks.sh" ]
           workingDir: /data/project/editgroups/www/python/src
           env:

deployment/listener.yaml

@@ -21,7 +21,7 @@ spec:
     spec:
       containers:
         - name: listener
-          image: docker-registry.tools.wmflabs.org/toollabs-python37-base:latest
+          image: docker-registry.tools.wmflabs.org/toolforge-python37-sssd-base:latest
           command: [ "/data/project/editgroups/www/python/src/listener.sh" ]
           workingDir: /data/project/editgroups/www/python/src
           env:

deployment/migrator.yaml

@@ -21,7 +21,7 @@ spec:
     spec:
       containers:
         - name: migrator
-          image: docker-registry.tools.wmflabs.org/toollabs-python37-base:latest
+          image: docker-registry.tools.wmflabs.org/toolforge-python37-sssd-base:latest
           command: [ "/data/project/editgroups/www/python/src/migrator.sh" ]
           workingDir: /data/project/editgroups/www/python/src
           env:

docs/install.rst

@@ -69,55 +69,69 @@ to fill in the following fields:
 Once you validate this form, the edits listener will pick up the new tool (after a few minutes) and start ingesting its edits. If you 
 need to ingest past edits again, you can use the listener script with a date in the past to retrieve the previous edits.
 
-Deploying on WMF Toollabs
--------------------------
+Deploying on WMF Toolforge
+--------------------------
 
 In what follows we assume that the tool is deployed as the ``editgroups`` project.
 
 -  ``become editgroups``
 -  ``mkdir -p www/python/src``
 
+Put the following contents in ``manifest.template`` in the home directory of the tool::
+
+  backend: kubernetes
+  type: python3.7
+
 Install the dependencies in the virtualenv::
 
+  webservice shell
   cd www/python
-  virtualenv venv --python /usr/bin/python3
+  python3 -m venv venv
   source venv/bin/activate
+  pip install --upgrade pip
   git clone https://github.com/Wikidata/editgroups.git src
   pip install -r src/requirements.txt
 
 Configure static files::
 
   mkdir -p src/static
-  ln -s src/static
+  ln -s src/static .
 
-Put the following content in ``~/www/uwsgi.ini``::
-
-  [uwsgi]
-  check-static = /data/project/editgroups/www/python
-
-and run ``./manage.py collectstatic``.
-
-Create the SQL database:
+Create the SQL database (outside of the ``webservice shell``):
 
 - ``sql tools`` 
-- ``CREATE DATABASE s1234__editgroups;`` where ``s1234`` is the SQL username of the tool
+- ``CREATE DATABASE s1234__editgroups;`` where ``s1234`` is the SQL username of the tool (can be found in ``~/replica.my.cnf``)
 - ``\q``
 
 Configure database access and other settings::
 
   cd ~/www/python/src/editgroups/settings/
   echo "from .prod import *" > __init__.py
-  cp secret_wmflabs.py secret.py
+  cp secret_toolforge.py secret.py
 
 Edit ``secret.py`` with the user
 and password of the table (they can be found in ``~/replica.my.cnf``).
 The name of the table is the one you used at creation above
 (``s1234__editgroups`` where ``s1234`` is replaced by the username of
-the tool).
+the tool). Also, pick a secret key to store in ``SECRET_KEY``.
+
+In the ``editgroups/settings/__init__.py`` you can also copy over
+settings line from ``editgroups/settings/common.py`` and adapt them to
+the wiki that you are running EditGroups for (for instance ``MEDIAWIKI_API_ENDPOINT`` and the following lines).
+You should also adapt the allowed hostname (taken from ``editgroups/settings/prod.py``). It's easier
+to add those to the ``__init__.py`` file to avoid editing files tracked by Git.
+
+Put the following content in ``~/www/python/uwsgi.ini``::
+
+  [uwsgi]
+  static-map = /static=/data/project/editgroups/www/python/src/static
+
+and run ``./manage.py collectstatic`` in the ``~/www/python/src`` directory.
+
 
 Configure OAuth login:
 
-- Request an OAuth client id at https://meta.wikimedia.org/wiki/Special:OAuthConsumerRegistration/propose. Beyond the normal editing scopes, you will also need to perform administrative actions (delete, restore) on behalf of users, so make sure you request these scopes too.
+- Request an OAuth client id at https://meta.wikimedia.org/wiki/Special:OAuthConsumerRegistration/propose. As OAuth protocol version, use "OAuth 1.0a". As callback URL, use the domain of the tool and tick the box to treat it as a prefix. Beyond the normal editing scopes, you will also need to perform administrative actions (delete, restore) on behalf of users, so make sure you request these scopes too.
 - Put the tokens in ``~/www/python/src/editgroups/settings/secret.py``
 
 Migrate the database:
@@ -126,9 +140,20 @@ Migrate the database:
 
 Run the webserver:
 
-- ``webservice --backend kubernetes python start``
+- ``webservice start``
+
+Go to the webservice, login with OAuth to the application. This will create a ``User`` object that you can then mark as staff in the Django shell, as follows::
+
+   $ webservice shell
+   source ~/www/python/venv/bin/activate
+   cd www/python/src
+   ./manage.py shell
+   from django.contrib.auth.models import User
+   user = User.objects.get()
+   user.is_staff = True
+   user.save()
 
-Launch the listener and Celery in Kubernetes:
+Launch the listener and Celery in Kubernetes. These deployment files may need to be adapted if you are not deploying the tool as the ``editgroups`` toolforge tool but another tool id:
 
 - ``kubectl create -f deployment/listener.yaml``
 - ``kubectl create -f deployment/celery.yaml``

dump_events.py

@@ -2,10 +2,10 @@
 import sys
 from sseclient import SSEClient as EventSource
 from dateutil import parser
-from store.stream import WikidataEditStream
+from store.stream import WikiEditStream
 
 if __name__ == '__main__':
-    s = WikidataEditStream()
+    s = WikiEditStream()
     offset = None
     if len(sys.argv) > 1:
         offset = parser.parse(sys.argv[1])

editgroups/context_processors.py

@@ -0,0 +1,20 @@
+from django.conf import settings
+
+def mediawiki_site_settings(request):
+    return {
+        'MEDIAWIKI_API_ENDPOINT': settings.MEDIAWIKI_API_ENDPOINT,
+        'MEDIAWIKI_BASE_URL': settings.MEDIAWIKI_BASE_URL,
+        'MEDIAWIKI_INDEX_ENDPOINT': settings.MEDIAWIKI_INDEX_ENDPOINT,
+        'PROPERTY_BASE_URL': settings.PROPERTY_BASE_URL,
+        'USER_BASE_URL': settings.USER_BASE_URL,
+        'USER_TALK_BASE_URL': settings.USER_TALK_BASE_URL,
+        'CONTRIBUTIONS_BASE_URL': settings.CONTRIBUTIONS_BASE_URL,
+        'WIKI_CODENAME': settings.WIKI_CODENAME,
+        'USER_DOCS_HOMEPAGE': settings.USER_DOCS_HOMEPAGE,
+        'MEDIAWIKI_NAME': settings.MEDIAWIKI_NAME,
+        'DISCUSS_PAGE_PREFIX': settings.DISCUSS_PAGE_PREFIX,
+        'DISCUSS_PAGE_PRELOAD': settings.DISCUSS_PAGE_PRELOAD,
+        'REVERT_PAGE': settings.REVERT_PAGE,
+        'REVERT_PRELOAD': settings.REVERT_PRELOAD,
+        'WIKILINK_BATCH_PREFIX': settings.WIKILINK_BATCH_PREFIX
+    }

editgroups/settings/common.py

@@ -95,6 +95,7 @@
                     "social_django.context_processors.backends",
                     "social_django.context_processors.login_redirect",
                     "tagging.filters.context_processor",
+                    "editgroups.context_processors.mediawiki_site_settings",
                 ),
                 'debug': True
             }
@@ -183,6 +184,23 @@
 SOCIAL_AUTH_EMAIL_LENGTH = 190
 
 MEDIAWIKI_API_ENDPOINT = 'https://www.wikidata.org/w/api.php'
+MEDIAWIKI_BASE_URL = 'https://www.wikidata.org/wiki/'
+MEDIAWIKI_INDEX_ENDPOINT = 'https://www.wikidata.org/w/index.php'
+PROPERTY_BASE_URL = MEDIAWIKI_BASE_URL + 'Property:'
+USER_BASE_URL = MEDIAWIKI_BASE_URL + 'User:'
+USER_TALK_BASE_URL = MEDIAWIKI_BASE_URL + 'User_talk:'
+CONTRIBUTIONS_BASE_URL = MEDIAWIKI_BASE_URL + 'Special:Contributions/'
+WIKI_CODENAME = 'wikidatawiki'
+USER_DOCS_HOMEPAGE = 'https://www.wikidata.org/wiki/Wikidata:Edit_groups'
+MEDIAWIKI_NAME = 'Wikidata'
+DISCUSS_PAGE_PREFIX = 'Wikidata:Edit_groups/'
+DISCUSS_PAGE_PRELOAD = 'Wikidata:Edit_groups/Preload'
+REVERT_PAGE = 'Wikidata:Requests_for_deletions'
+REVERT_PRELOAD = 'Wikidata:Edit_groups/Revert'
+WATCHED_NAMESPACES = [0, 120]
+
+WIKILINK_BATCH_PREFIX = ':toollabs:editgroups/b/'
+REVERT_COMMENT_STAMP = ' ([[:toollabs:editgroups/b/EG/{}|details]])'
 
 ### Celery config ###
 # Celery runs asynchronous tasks such as metadata harvesting or

editgroups/settings/prod.py

@@ -6,7 +6,7 @@
 # Static files (CSS, JavaScript, Images)
 # https://docs.djangoproject.com/en/1.7/howto/static-files/
 
-STATIC_URL = '/editgroups/static/'
+STATIC_URL = '/static/'
 STATICFILES_DIRS = [
     os.path.join(BASE_DIR, "static"),
 ]

editgroups/settings/secret_toolforge.py

@@ -0,0 +1,33 @@
+import os
+from .common import BASE_DIR
+
+# SECURITY WARNING: keep the secret key used in production secret!
+SECRET_KEY = 'insert_a_random_hash_here'
+
+# Database
+# https://docs.djangoproject.com/en/1.7/ref/settings/#databases
+DATABASES = {
+    'default': {
+        'ENGINE': 'django.db.backends.mysql',
+        'NAME': 's1234__editgroups', # adapt to the database you created
+        'HOST': 'tools.db.svc.eqiad.wmflabs',
+        'OPTIONS': {
+           'init_command': "SET sql_mode='STRICT_TRANS_TABLES'",
+           'charset': 'utf8mb4',
+           'read_default_file': os.path.expanduser("~/replica.my.cnf")
+        },
+    }
+}
+
+# Adapt those to the credentials you got
+SOCIAL_AUTH_MEDIAWIKI_KEY = ''
+SOCIAL_AUTH_MEDIAWIKI_SECRET = ''
+SOCIAL_AUTH_MEDIAWIKI_URL = 'https://www.wikidata.org/w/index.php'
+SOCIAL_AUTH_MEDIAWIKI_CALLBACK = 'https://editgroups.toolforge.org/oauth/complete/mediawiki/'
+
+# Redis (if you use it)
+REDIS_HOST = 'tools-redis'
+REDIS_PORT = 6379
+REDIS_DB = 3
+REDIS_PASSWORD = ''
+

editgroups/templates/editgroups/common.html

@@ -37,7 +37,7 @@
                             <button type="submit" class="btn btn-default">Submit</button>
                         </form> -->
                         <ul class="nav navbar-nav navbar-right">
-                            <li><a href="https://www.wikidata.org/wiki/Wikidata:Edit_groups">About</a></li>
+                            <li><a href="{{ USER_DOCS_HOMEPAGE }}">About</a></li>
                             {% if not user.is_authenticated %}
                             <li><a href="{% url "social:begin" 'mediawiki' %}">Login</a></li>
                             {% else %}

listener.py

@@ -16,12 +16,12 @@
     import django
     django.setup()
 
-    from store.stream import WikidataEditStream
+    from store.stream import WikiEditStream
     from store.utils import grouper
     from store.models import Edit
 
-    print('Listening to Wikidata edits...')
-    s = WikidataEditStream()
+    print('Listening to edits...')
+    s = WikiEditStream()
     utcnow = datetime.utcnow()
     try:
         latest_edit_seen = Edit.objects.order_by('-timestamp')[0].timestamp

revert/models.py

@@ -32,7 +32,7 @@ def __str__(self):
 
     def comment_with_stamp(self):
         return (self.comment +
-                ' ([[:toollabs:editgroups/b/EG/{}|details]])'.format(self.uid))
+                settings.REVERT_COMMENT_STAMP.format(self.uid))
 
     def undo_summary(self, edit):
         prefix = '/* undo:0||{}|{} */ '.format(edit.newrevid, edit.user)
@@ -63,13 +63,11 @@ def revert_edit(self, edit):
             self.oauth_tokens['oauth_token_secret'])
 
         # Get token
-        r = requests.get('https://www.wikidata.org/w/api.php', params={
+        r = requests.get(settings.MEDIAWIKI_API_ENDPOINT, params={
             'action':'query',
             'meta':'tokens',
             'format': 'json',
         }, auth=auth)
-        print('#### GET TOKEN')
-        print(r.text)
         r.raise_for_status()
         token = r.json()['query']['tokens']['csrftoken']
 
@@ -105,11 +103,9 @@ def revert_edit(self, edit):
                 'watchlist': 'nochange',
             }
 
-        r = requests.post('https://www.wikidata.org/w/api.php',
+        r = requests.post(settings.MEDIAWIKI_API_ENDPOINT,
                 data=data, auth=auth)
 
-        print('#### UNDO EDIT')
-        print(r.text)
         #r.raise_for_status()
 
 

revert/templates/revert/initiate.html

@@ -8,7 +8,7 @@
 
 {% block mainBody %}
 <div class="page-header">
-<h3>Undoing edit group by <a href="https://www.wikidata.org/wiki/User:{{ batch.user }}">{{ batch.user }}</a>: {{ batch.summary }} ({{ batch.uid }})</h3>
+<h3>Undoing edit group by <a href="{{ USER_BASE_URL }}{{ batch.user }}">{{ batch.user }}</a>: {{ batch.summary }} ({{ batch.uid }})</h3>
 </div>
 <div class="revert-dialog">
     <p>You are about to undo {{ batch.nb_revertable_edits }} edits{% if batch.nb_undeleted_new_pages %}, which will delete or restore {{ batch.nb_undeleted_new_pages }} items{% endif %}.</p>

store/models.py

@@ -274,7 +274,7 @@ def archive_old_batches(cls, batch_inspector):
 
 class Edit(models.Model):
     """
-    A wikidata edit as returned by the Event Stream API
+    A MediaWiki edit as returned by the Event Stream API
     """
     id = models.IntegerField(unique=True, primary_key=True)
     oldrevid = models.IntegerField(null=True)
@@ -306,16 +306,16 @@ class Meta:
 
     @property
     def url(self):
-        return 'https://www.wikidata.org/wiki/index.php?diff={}&oldid={}'.format(self.newrevid,self.oldrevid)
+        return '{}?diff={}&oldid={}'.format(settings.MEDIAWIKI_INDEX_ENDPOINT, self.newrevid, self.oldrevid)
 
     @property
     def revert_url(self):
         if self.oldrevid:
-            return 'https://www.wikidata.org/w/index.php?title={}&action=edit&undoafter={}&undo={}'.format(self.title, self.oldrevid, self.newrevid)
+            return '{}?title={}&action=edit&undoafter={}&undo={}'.format(settings.MEDIAWIKI_INDEX_ENDPOINT, self.title, self.oldrevid, self.newrevid)
         elif self.changetype == 'delete':
-            return 'https://www.wikidata.org/wiki/Special:Undelete/{}'.format(self.title)
+            return '{}Special:Undelete/{}'.format(settings.MEDIAWIKI_BASE_URL, self.title)
         else:
-            return 'https://www.wikidata.org/w/index.php?title={}&action=delete'.format(self.title)
+            return '{}?title={}&action=delete'.format(settings.MEDIAWIKI_INDEX_ENDPOINT, self.title)
 
     def __str__(self):
         return '<Edit {} >'.format(self.url)
@@ -366,7 +366,7 @@ def ingest_edits(cls, json_batch):
         tools = Tool.objects.all()
 
         for edit_json in json_batch:
-            if not edit_json or edit_json.get('namespace') not in [0,120]:
+            if not edit_json or edit_json.get('namespace') not in settings.WATCHED_NAMESPACES:
                 continue
             timestamp = datetime.fromtimestamp(edit_json['timestamp'], tz=UTC)
 

store/stream.py

@@ -1,10 +1,11 @@
 import json
 from sseclient import SSEClient as EventSource
+from django.conf import settings
 
-class WikidataEditStream(object):
+class WikiEditStream(object):
     def __init__(self):
         self.url = 'https://stream.wikimedia.org/v2/stream/recentchange'
-        self.wiki = 'wikidatawiki'
+        self.wiki = settings.WIKI_CODENAME
 
     def stream(self, from_time=None):
         url = self.url