Use three dots instead of elipsis

[suricactus]

pylupdate5 hardcodes the output to ascii symbols, therefore any unicode is XML encoded and cannot be reverted to the original state.

Enforcing codecForTr to be utf8 does not work:
https://github.com/baoboa/pyqt5/blob/31ef42563dbcaf3885da8803e2e5d84507141c9b/pylupdate/pylupdate_main.py#L231
As the output is hardcoded as ASCII in
https://github.com/baoboa/pyqt5/blob/31ef42563dbcaf3885da8803e2e5d84507141c9b/pylupdate/metatranslator.cpp#L435

[m-kuhn]

😢

[m-kuhn]

does this help us to recover utf-8 from those mangled poor characters? Will need a bit of preprocessing of the string though, but having proper unicode support in qgispluginci would be great.

[suricactus]

Unfortunately not, because the characters are encoded byte by byte, not char by char.

>>> import html
>>> html.unescape('опа') # cyrillic "опа" (latin: opa)
'опа'
>>> html.unescape('…') # elipsis is encoded for each individual byte
'…'

[m-kuhn]

what about creating a .pro?

[suricactus]

It is using the same codecForTr parameter as in the first post, it does not work.

Fake news up there, actually it works, but because of fetchtr_py function that is called only when .pro file is present. I will create such file.

[m-kuhn]

How far did you go with migration of pylupdate to qgis-plugin-ci?

[suricactus]

I am testing opengisch/qgis-plugin-ci#18 and if all good, then I will revert the changes here.

[suricactus]

Just to unblock the change to go further, temporary use the mask as obsolete method trUtf8 instead of tr. There is no other disadvantage of using the trUtf8 for now, except a warning and possible post-qt5 era need for porting.

https://doc.qt.io/qt-5/qobject-obsolete.html#trUtf8

When the .pro workaround is proved to work properly and merged at qgis-plugin-ci, we can get rid of trUtf8 calls.