GitLab

For clarity purposes we don't want the users to have to trigger the
mouseover event in order to display the jstree file selector. So using
the onload event is much better in that respect. Howver given the way
these DOM elements are loaded, that event is not triggered by the
browser. Therefore we need to trigger it ourselves.

For this we need to target the specific DOM node we want the event to be
triggered on since the 'load' event does not bubble, so triggering it on
for instance the div with id "db_content" does not work for us.

some ids became classes and the want to save space so these changes
should help with aesthetics and space economy

use the correct field identifiers with ids and such to decode the
filename correctly to ensure the filename is correct in the database and
the correct upload method is detected

serialises the form just like for sets. The success callback also
uploads all files, matching each one to the correct file_id by index in
the serialised version of the form.

During refactoring this feature was broken, rendering uploads
impossible, this reimplements the feature correctly and separates it
into it's own function

while refactoring the usage changed slightly, so we need to commit those
changes :)

same as for pre_process, we want to select it once and apply that rule
to all samples in the form.

no longer using a pre_process selection for each sample. One selection
affects each sample within the given form.

ensures that the hidden attribute doesn't get overridden by another css
display rule

This was problematic since file2 is not always used and when editing a
file we won't always be selecting a file to upload.

adapting to match what was done in the FormBuilder. This is was easier
given the way the FormBuilder was implemented.

self was undefined causing the serializeObject method to fail silently

the onchange callback used to set the value of the hidden field
"filename" needed altering to fit the multi upload paradigm.

split the actual form part into a partial as with sample_sets, this
allows us to have shorter files with less levels of indentation, making
a little easier to read.

also applies the multi-form naming convention so we can apply
serializeObject to the form.

refactors a couple elements from the controller into functions (such as
validation and pre_process listing) to shorten the controller itself.

working from the form submission was making the task of adapting the
tokeniser to an extendable form more complicated then simply allowing it
to be used on form input.

fixes uses of build_date in SetFormBuilder and subclasses

rename info to build_info and date to build_date

moves some methods up a level and renames some variables to be used in
future FormBuilders that don't inherit from SetFormBuilder

The build method intended for the GenericFormBuilder was assigned to the
SetFormBuilder prototype. Although it didn't cause any functionnal
issues, I feel it needed to be fixed.

make the inputs look a little bit nicer with a little bit of css from
bootstrap

reduces the size of fields in form to save more space.

Although we do have server-side validation (and still need it to prevent
tampering with forms or even the use of outdated browsers),  this will
serve two purposes:
1) alleviate the server slightly
2) have a faster response time for users

so win-win :).

makes better use of space for adding multiple patients/runs/sets in a
single form.

storing the result of the validation in the dict was a bad idea when no
errors are found since we were attempting to insert the error field in
into the database. Of course this column doesn't exist and was causing
an error.

The data attribute 'group-ids' had not been altered to the generic
attribute 'keys'  from the refactor to use with autocompletion on
samples

sets the 'type' argument for redirection after adding sets to the most
abundant type of set.

For ex. when adding mostly patients and some of the other types, the
user will be redirected to the list of patients once the form has been
successfully submitted.

when editing a set, the error key is not present in the patient/run/set,
so we needed to manage that case

When errors are detected in the form, the route was rendering a form
that no longer exists

The permission check in the form route was operating on the sample_set
id rather than the patient/run/set id

sample sets now share a single add/edit route.

moves the new patient/form and patient/submit routes to the sample_set
controller and removes any add/edit routes from all set controllers.
The same goes for the templates.

Also allows the inclusion of different types of sets in a single form.

For this, the sample_set/form.html template includes patials for each
set type.

See #2878

One form to rule them all, one form the find them, one form to bring
them all, and in the template bind them.