User:Harej/sandbox
since 2005
{{Pageset definition
| namespaces =
| categories =
| category-depth =
| wdq1 =
| petscan1 =
| domain-links1 =
| sql1 =
| links-here1 =
| transclusions1 =
| links-on-page1 =
}}
- New thread on Talk:Easter Island
- Does easter bunny live in Easter Islands?
- 19:04, 27 April 2023 (UTC)
Notes
[edit]- Navigation header
- Metrics dashboard
- Alert list
- Blocks
- Standard icons
- Workspace intro
- Preview link
- Participant box
Missing articles
[edit]- Wikipedia:Find a Grave famous people/M/Mas
- Two separate requests under the same title. The title is a blue link, but the linked article is a living person and neither of the requested subjects
Citation watchlist script
[edit]https://en.wikipedia.org/w/api.php?action=compare&fromrev=1203018841&torev=1203024750&format=json
<a class="mw-changeslist-diff" href="/w/index.php?title=Zoology&curid=34413&diff=1203018841&oldid=1203024750">diff</a>
This diff adds a new sentence to the article and also adds a new link to a source.
In this one diff these two sources are cited:
- https://www.theguardian.com/world/2014/jan/17/dennis-mcguire-ohio-execution-untested-method-lawsuit
- https://www.cbsnews.com/news/ohio-delays-executions-until-2017-over-lack-of-lethal-drugs/
Given a watchlist:
- Isolate each revision id and previous id from each line in the watchlist
- Check every five seconds if there is a revision id / previous id pair that hasn't been checked yet.
Given a pair (or batch of them):
- Use the "action=compare" endpoint.
- Screen out URLs with a regular expression (joke about now having an additional problem to solve for)
- Isolate domain names from URLs
- Check those sources against internal representation of RSP (hardcoded in script for now)
- If there's a hit, add an indicator next to the diff. (Red Triangle "!" for warn-list, yellow circle "?" for caution-list)
The problems I have with this approach:
- Each user is doing the lookups and computations themselves, rather than going through a centralized service that does it for them
In the future when we have a centralized service doing this work, because we are doing something more complicated than screens against RSP,
The user script:
- Seeks consent to access the external service where data is coming from
- Scans each revision ID / prev ID on a watchlist
- Submits them to the service in batch
- Retrieves data
- Adds to HTML based on retrieved data
What about this "service"? If I set up WRDB as an ongoing, self-updated service, then all this service would need to do is check the revision ID in WRDB. At the moment, however, WRDB only supports a one-time build, and domain information is not directly stored in the database. However, this will help with support for non-URL references in the future.
Citation Watchlist testing
[edit]Diff, hist, prev, cur
[edit]Location | Revision(s) | Extracts URL from link label | "Type" | Old revision ID | New revision ID | Notes |
---|---|---|---|---|---|---|
Page history | First revision; no subsequent revisions | none! | new | Currently invisible to Test Wikipedia branch | ||
Page history | First revision | cur | new | none | (curid:) oldid= | It was the "curid" when it was new |
Page history | Subsequent revision | prev | diff | (diff:) extract previous revision ID from oldid= | (oldid:) oldid= | |
Watchlist and Recent Changes | First revision | hist | new | none | (curid:) curid= | |
Watchlist and Recent Changes | Subsequent revision | diff | diff | (diff:) extract previous revision ID from diff= | (oldid:) diff= |