[allura] 02/02: [#8272] make per-forum feeds not create a potentially gigantic $in query by default

[br...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

brondsem pushed a commit to branch db/8272
in repository https://gitbox.apache.org/repos/asf/allura.git

commit 02511fca3c484c178cdbe65f8ee371a226d01464
Author: Dave Brondsema <da...@brondsema.net>
AuthorDate: Wed Mar 13 14:42:27 2019 -0400

    [#8272] make per-forum feeds not create a potentially gigantic $in query by default
---
 Allura/allura/controllers/discuss.py | 14 ++++++++++++--
 Allura/allura/controllers/feed.py    |  9 +++++++++
 Allura/allura/model/artifact.py      |  2 ++
 3 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/Allura/allura/controllers/discuss.py b/Allura/allura/controllers/discuss.py
index 662c085..fdc89cc 100644
--- a/Allura/allura/controllers/discuss.py
+++ b/Allura/allura/controllers/discuss.py
@@ -19,6 +19,7 @@ from urllib import unquote
 from datetime import datetime
 import logging
 
+import pymongo
 from tg import expose, redirect, validate, request, flash, response
 from tg.decorators import with_trailing_slash, without_trailing_slash, before_render, before_validate
 from decorator import decorator
@@ -118,9 +119,18 @@ class DiscussionController(BaseController, FeedController):
         Overrides :meth:`allura.controllers.feed.FeedController.get_feed`.
 
         """
+        def query(since, until, page, limit, **kwargs):
+            if not since and not until and not page:
+                # simplest default case, so make the threads list shorter by grabbing only needed ones
+                discussion_threads = self.discussion.thread_class().query.find(dict(
+                    discussion_id=self.discussion._id,
+                    num_replies={'$gt': 0},  # exclude empty threads (spam/deleted) like ForumController does
+                )).sort([('last_post_date', pymongo.DESCENDING)]).limit(limit)
+            else:
+                discussion_threads = self.discussion.threads
+            return dict(ref_id={'$in': [t.index_id() for t in discussion_threads]})
         return FeedArgs(
-            dict(ref_id={'$in': [t.index_id()
-                 for t in self.discussion.threads]}),
+            query,
             'Recent posts to %s' % self.discussion.name,
             self.discussion.url())
 
diff --git a/Allura/allura/controllers/feed.py b/Allura/allura/controllers/feed.py
index 7aaa789..2fef9f3 100644
--- a/Allura/allura/controllers/feed.py
+++ b/Allura/allura/controllers/feed.py
@@ -35,6 +35,15 @@ class FeedArgs(object):
     """
 
     def __init__(self, query, title, url, description=None):
+        """
+        :param query: mongo criteria to query the Feed collection.
+                      Pagination & filter criteria will be added in automatically
+                      Can be a function to return criteria, which will be passed args (since, until, page, limit) for
+                        advanced optimization.
+        :param title: feed title
+        :param url: feed's own url
+        :param description: feed description
+        """
         self.query = query
         self.title = title
         self.url = url
diff --git a/Allura/allura/model/artifact.py b/Allura/allura/model/artifact.py
index 8f9c5b6..bf0716c 100644
--- a/Allura/allura/model/artifact.py
+++ b/Allura/allura/model/artifact.py
@@ -968,6 +968,8 @@ class Feed(MappedClass):
             feed = RssFeed(**d)
         limit, page = h.paging_sanitizer(limit or 10, page)
         query = defaultdict(dict)
+        if callable(q):
+            q = q(since, until, page, limit)
         query.update(q)
         if since is not None:
             query['pubdate']['$gte'] = since