Test structured-contents parsing with the new WMF update in T335512
Closed, ResolvedPublic3 Estimated Story Points
Actions

Assigned To

Authored By

	ROdonnell-WMF
	Dec 12 2023, 6:42 PM

Description

User Story: “As a developer, I want to run tests on the structured-contents parsing and check that the T335512 WMF changes to the Parsoid version don't break our API features so that I can upgrade the parser logic to Parsoid V2.8.”

Acceptance criteria
Use the WMF branch in T335512 and validate the structured content unit tests

ToDo

[ x] Change WMF submodule to use the branch that Prabhat created for Parsoid 2.8
[x ] Run the snapshot tests and check from breaking changes
[ x] Update the go-query HTML queries and extract them into a configuration file (Ask Ricardo what his preference is here for abstracting HTML dependencies)

Test Strategy

Use a normal parser.go snapshot unit tests

Checklist for testing

[x ] Check for 2.8 breaking changes in the parser.go HTML queries
[x ] Abstract HTML selection out of code and into config file?
[ x] Re-run tests until the parser works correctly using Parsoid v2.8 HTML

Related Objects
Search...

Status	Subtype	Assigned	Task
Stalled		None	T324931 Clean up open RESTBase related tickets
Open		None	T262315 <CORE TECHNOLOGY> API Migration & RESTBase Sunset
Resolved		Jgiannelos	T344944 Move Parsoid endpoints out of RESTBase
In Progress		None	T374135 Migrate RESTbase page content endpoints
Open		MSantos	T334238 Create deprecation plan for public parsoid endpoints
Resolved		MSantos	T335512 Transition Enterprise MediaWiki from RESTBase to core page HTML endpoints
Resolved		ROdonnell-WMF	T353286 Test structured-contents parsing with the new WMF update in T335512
Resolved	BUG REPORT	MSantos	T353689 REST HTML endpoint returning 500 for JSON page
Resolved	BUG REPORT	daniel	T353688 REST: Trying to load the content "magic" image description pages leads to redirect loop
Resolved	BUG REPORT	daniel	T353687 REST: page content endpoints fail on page "0"
Open	BUG REPORT	None	T362187 REST HTML endpoint returning 500 for the first 3 - 5 calls for several pages

Event Timeline

ROdonnell-WMF created this task.Dec 12 2023, 6:42 PM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptDec 12 2023, 6:42 PM

ROdonnell-WMF added parent tasks: T335512: Transition Enterprise MediaWiki from RESTBase to core page HTML endpoints, T334238: Create deprecation plan for public parsoid endpoints.Dec 12 2023, 6:43 PM

ROdonnell-WMF moved this task from Incoming to Sprint 52 on the Wikimedia Enterprise board.

ROdonnell-WMF edited projects, added Wikimedia Enterprise (Sprint 52); removed Wikimedia Enterprise.

@JArguello-WMF I'm adding this to this Sprint, because it's a breaking change and will eventually affect my code in structured-contents. I'd like to preempt any breakingchanges before the Xmas holidays

ROdonnell-WMF moved this task from Next Up to In Progress on the Wikimedia Enterprise (Sprint 52) board.Dec 12 2023, 6:47 PM

I ran an experiment with a sample of two sets of 500 random articles, a small differences were found just in Infobox and Sections. The good news is Templates, abstracts, and categories are identical.

Differences in samples

-------------------------------------------------------
| Sample | Infobox different |   Sections different |
-------------------------------------------------------
| 1st 500 |             17    |            19       |
-------------------------------------------------------
| 2st 500 |             2     |             14      |
-------------------------------------------------------

So, a difference rate of 0.36% for Sections and 0.16% for Infoboxes. On Monday, I'll investigate why they are different.

ROdonnell-WMF moved this task from In Progress to QA on the Wikimedia Enterprise (Sprint 52) board.Dec 17 2023, 8:54 PM

After comparing the old and new differences they are identical JSON. The screenshots show some examples highlighting RED and GREEN differences. The only difference is a prefix I put in the JSON to identify if it's OLD or NEW JSON.

The only minor difference in one category list output is that the order of items differs, but this is expected.

In conclusion, after running experiments on 4,000 articles the difference is near zero. The test code reported a 0.2% error rate, but on further inspection, all these were false-positive tests.

ROdonnell-WMF moved this task from QA to Sign Off on the Wikimedia Enterprise (Sprint 52) board.Dec 18 2023, 12:22 PM

JArguello-WMF updated the task description. (Show Details)Dec 18 2023, 2:06 PM

ROdonnell-WMF set the point value for this task to 3.Dec 20 2023, 10:17 AM

ROdonnell-WMF moved this task from Sign Off to Done on the Wikimedia Enterprise (Sprint 52) board.

JArguello-WMF closed this task as Resolved.Dec 22 2023, 8:10 PM

	F41610922: image.png
	Dec 18 2023, 12:21 PM

	F41610920: image.png
	Dec 18 2023, 12:21 PM

	F41610918: image.png
	Dec 18 2023, 12:21 PM

	F41610915: image.png
	Dec 18 2023, 12:21 PM

Test structured-contents parsing with the new WMF update in T335512Closed, ResolvedPublic3 Estimated Story PointsActions