Incorrect widget key retrieval : inconsistency is caused by a dot in the key string

[lionpeloux]

Version

PyPDFForm=1.4.37

Issue Description

In the attached pdf form, I am trying to rename the keys using update_widget_key (and a custom key-> key json map). During that process, I've remarked that a widget key was incorrectly retrieved :

The checkbox Gain de 2 classes on page 2 should be named "Gain de 2 classes.0" (according to Acrobat DC) but PyPDFForm is retrieving it as "0".

Moreover, when I try to rename this checkbox widget with :
update_widget_key ("0", "situation_finale_gain_2_classes")

what I get for the renamed widget is :
key="Gain de 2 classes. situation_finale_gain_2_classes"

looks like something is broken because of the dot . in the widget key ...

202402_AttestationTravauxMPRaccompagne-devis_WEBAI.pdf

[chinapandaman]

Hey thanks for posting.

It appears that other non-Acrobat PDF viewers like DocFly also sees that checkbox with a key of 0:

I think both PyPDFForm and DocFly retrieved the key based on the /T property of the widget. However, the way Acrobat retrieved the key is by combining the /TU and /T properties with a dot inbetween:

Which would explain why when you tried to modify the key by doing update_widget_key ("0", "situation_finale_gain_2_classes") it only modified the part after the dot (although I'm not sure where that extra space comes from).

In the meantime I suggest try to modify that key to remove the dot by using some PDF viewers instead of the library and see what happens. I need to do some researches around the /TU property and see whether it should be part of a widget's key.