If a snippet contains a .oe_structure element inside of it, it allows
the user to drop other snippets in that location. The web_editor
contains code which removes all empty parent of an element being
deleted. This is done so that, for example, when the last col of a
3-col snippet is removed, the whole snippet is removed.
However, there was an exception to this rule which was not considered,
if one of these empty parent is an .oe_structure, it should not be
removed.
Note: the only snippet which does this is the parallax snippet.
When pasting, the paste event is cancelled and done back with the
corrected data.
This works fine on firefox, but this can have unwanted results on other
browser like pasting two times or copying the end of the line at the
beginning.
This commit replace completely the summernote functionality with our
small implementation of pasting corrected text.
closes#14043
opw-690814
* web_editor, website_blog
Since recent refactoring, web editor dialogs automatically trigger
an event when closed (saved or cancel). The problem is that there
still was manual trigger of those events or variants ("save" instead
of "saved").
To prevent multi-saving and be consistent, all save events are now
automatically triggered on save/close (no more manual) and the event
name is "save".
Since saas-12, when deleting contents, parents are deleted too if they
become empty (while it was not possible to delete contents in that case
before). The problem here was that the history was recorded multiple
times if one deletion made multiple snippets be deleted and this
sometimes could crash.
Now the history is recorded only on first deletion when the user ask
for it.
There is a hackish feature in the web_editor which forces the
contenteditable to true on links which are clicked and removes it
anywhere else in the page so that only the link is editable after
click. This allows to add text at the end of a link without adding it
outside of the link by mistake.
The feature was badly implemented because, if the link was editable
without having any parent which is editable, the contenteditable was
forced to true, then immediatly forced to false when forcing to false
every page element. This was the case for lots of t-field. By example
the product categories in eCommerce are t-field links which are in non
editable <li/> elements...
The system was lost with the editor refactoring which came with saas-10
and commit bf39884a5c.
Indeed the show() method of the BuildingBlock class was removed as the
widget is now always showed but the show_blocks() method it called was
not removed (and thus was not called anymore).
As a fix, the show_blocks method is now called again, on snippets
initializations and each time a snippet is dropped or removed.
The .dropdown is in a .btn-group environment while a .btn-group can
only contain .btn or .btn-group elements.
This induced problems where the m2o options covers the full overlay
environment, preventing clicks on the customize dropdown.
The snippet options can define an on_remove method which was supposed
to be called when the snippet is removed. This was not the case since
the method was written so that the options were removed before the
on_remove method was called... (Un)fortunately, the on_remove callbacks
were not doing any critical job in current implementation. In saas-13
however, some elements will rely on this.
Commit 404ed0f788 tried to solve the
problem but only solved a part of it. Indeed, if the snippet child
has a bg color option but not a bg image option, the event propagation
stops in the parent handler, when it is too late.
Solution is checking that the event comes from the same element on
which the callback is attached to.
background and color snippet options
The background and color snippet options communicate so that if a
color is set while a background is already set, the backgrond is
removed. The problem is that the event propagation was not stop
so if a snippet had a background and one of its child set a color,
the background was removed.
Before this commit, the focus point the user can move on the image in
the background position configuration dialog was only following the
mouse as we want if the image original width was larger than the dialog
one. This is because the javascript relies on the fact the image takes
100% width of its container. To fix this, the container now does not
take 100% of modal width but the size of its internal image (with a
max-width of 100%).
In some snippets, the parent hierarchy has many elements. So, when
selecting a deep child of one of those, the customize dropdown had
many dropdown headers showing the parent hierarchy but without any
option for many of them.
Commit de0d9fd666 already fixed the case
where the dropdown only contains dropdown headers. This commit is an
adaptation to hide these headers one by one.
The call to the web_editor is made through an iframe with some get
parameters. While not really a problem in itself, this has the
disadvantage of generating long URIs. In some case, since all the
fields in the views are included as a GET parameter, URIs can be
generated that go up to several thousand characters. Most servers will
return a 414 for such URIs, preventing the edition of the mail
template in the view.
To avoid this problem, this commit extends the 'html_frame' widget and
blacklist 2 fields that can get too long once encoded.
Ideally, it would probably be better to stop sending all the fields and
simply add the *needed* fields one by one, but since it's difficult to
know exactly what fields might be needed and this is stable, I'm gonna
go with this 'workaround' but conservative fix.
opw-685646
This commit fixes two problems:
* Clicking on the save button of a web_editor dialog closed the dialog
two times because the button was created with the flag "close: true"
and binded to a handler which closes the dialog. This also prevented
to extend this handler by placing the this._super call in a deferred.
* Some web_editor dialog extended the save callback but forgot to use
the this._super, which prevents the "saved" event to be triggered
(currently not used anywhere but useful for custo and master).
Commit 89a4774c21 changed the kwargs
debug value of the FieldTextHtml controller to let it be equal to
"assets" instead of True or False. The problem is that an empty string
was considered True before the commit and False afterwards.
In fact, the original code was wrong and the debug value should equal
the already processed kwargs debug value in the HttpRequest object.
Indeed, the kwargs.debug value is equal to the URL value while the
request.debug value is already processed and is only one of the 3
values: None, True or "assets".
Commit 57c09e8a8e was supposed to fix the problem but only made it
worse in some cases. Since the new option parenting system, the
customize button visibility should not be determined on creation as
parent and own options are appended on snippet focus. Also, removing
the dropdown header if it was alone was a bad idea as this is it which
contains the information of options and all the options have not
necessarily a <li/> tag in the dropdown.
For example, the many2one option has no <li/> in the dropdown and is
the only option of the field snippet. So with the previous commit, the
dropdown header was removed as there is no <li/> in the dropdown other
than this header. So as the header was removed, on snippet focus, there
was no way to determine the many2one option exists, so its on_focus
method was not called.
Now, the dropdown-header is always kept. The visibility of customize
button is determine on snippet focus.
Before commit a8ee18e172, the paragraph alignment dropdown menu was
too large for its content, so it was ugly. After that commit, it became
too small for its content, which is even more ugly.
This fix sets the width of this specific dropdown so that its content
fills exactly in (as with original summernote).
qsm-odoo