wikimedia / parsoid / 577
85%

DEFAULT BRANCH: master
Ran
Jobs 4
Files 70
Run time 4min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
577

push

travis-ci

subbuss 
Run normalization after dom-diff to handle edited content

* Switched around passes

  This patch switches around the DOM-diff and normalization passes.
  In 004c6659, I had switched them around so as not have to deal
  with updating DOM-diff markers because of all the normalizations
  we were going to end up doing as part of the scrubWikitext support.
  As this patch shows, it is a fair amount of bookkeeping work.
  However, I bit the bullet since we were having these odd gaps in
  our normalization support for edited content which leads to baffling
  results for editors and Parsoid clients.

  The normalization pass now properly updates dom-diff markers
  during normalization. Updated and tweaked helpers in DOMUtils
  to aid this.

  In selser mode, normalizations are run only on modified portions
  of the DOM -- this lets us deal with edited content more broadly.

  In non-selser mode (which is only run in rt-testing, and in
  commandline mode), we run normalizations on the full DOM.
  This indirectly addresses the testing need in T108137.

* Fixed recursion to clarify the flow better. Instead of starting
  with children of <body>, we now start with <body>. This also
  enabled a cleaner DOM processing without the awkwardness that
  I had originally introduced back when this normalization pass
  was first written. We are now doing more normalizations and that
  can interact with each other.

* Updated tests

  Removed the scrubWikitext mode tests that were testing for
  absence of normalizations on "old" nodes since those tests
  were all in non-selser mode and are no longer relevant.

  The newly passing test is a real bug fix. The change to the
  blacklisted test is also an improvement (that segment now
  matches the WTS output better and eliminates a useless nowiki).

  Added several new tests that run in selser mode with manual edits
  that clarify expectations around normalization and edited content.

  The failing selser tests are because non-selser normalizes all
  content (not just edited content) and selser is more accurate.
  FIXME: We need an option to turn off auto-selser tests for
  scenarios like this.

  Uncommented tests added in 66df48e2 that had been originally
  rebased on top of the current patch.

Bug: T108137
Bug: T107774
Bug: T93116

Change-Id: Ia9f278f05

11734 of 13654 relevant lines covered (85.94%)

82752.33 hits per line

Jobs
ID Job ID Ran Files Coverage
1 577.1 0
85.94
 Travis Job 577.1
2 577.2 0
85.94
 Travis Job 577.2
3 577.3 0
85.93
 Travis Job 577.3
4 577.4 0
85.93
 Travis Job 577.4
Source Files on build 577
Detailed source file information is not available for this build.