Processing module should only register it's own basic providers.
Note that when reloading a plugin, subclasses of ProcessingProvider remains and new ones are added,
so in such cases QgsProcessingProvider.__subclasses__() return same provider class many times.
Instead of returning a single QgsFeature from processFeature, we now
return a list of features.
This allows feature based algorithms which return multiple features
per input feature, e.g. "explode" type algorithms which split a
single input feature into multiple output features.
in model child algorithms
Prior to this change if you edited an algorithm in a model and
tried to enter a preset string for certain parameter types, this
string would get silently discarded on closing the dialog. E.g.
with a dissolve algorithm it was not possible to have a fixed
field name within the model to dissolve by.
This was caused by WidgetWrapper.comboValue returning the customData
for these manually entered values in the parameters combo box,
yet manually entered values never have custom data.
To work around this we only return the custom data if its
set for the selected item - otherwise we return the text unchanged.
In order to handle the "[not set]" options, a new static custom
data value of WidgetWrapper.NOT_SET_OPTION is added to that
comboValue can detect this and return the appropriate None value.
always add a consistent amount of attributes regardless of the
code paths taken
Some algorithms were adding features to data providers with
different number of attributes vs the layer's fields - this is
not supported and depending on the data provider will have
different (bad) results.
be prepared
There's cases where this happens without reflecting an invalid
expression. So we can try to prepare, but not abort if the
preparation fails.
Fixes#18103
- Fix missing format call for string when calculated number
of points is 0
- Fix incorrect progress reports which were spamming the dialog
and causing UI hangs, and slow algorithm execution
if they are unique in the current project
Instead of always using the layer id, which is generally
gibberish and meaningless for users, instead prefer to use
the layer name as an input in the batch processing dialog. This
is done only if the name is unique within the current project's
loaded layers.
This change makes the dialog more user-friendly, but more importantly
it means that autofilling output values based on an input layer
parameter generates more meaningful automatic output file names.
Some filters are fast enough to return results that it's overkill
to run them in a background thread - add a flag to these filters
to allow them to run (blocking) in the main thread instead.
- add a clone() method to filters, and always search using the
clone instead of the original filter
- add a prepare() method to filters, which is always run in the
main thread and can be used to prepare the filter for safe
background execution (e.g. creating feature iterators in advance)
- don't use QtConcurrent to perform searches in background threads,
since it is not safe to use with QObjects
- instead manually create threads and ensure that cloned objects
are always moved to the thread that they will run in, to ensure
that they correctly have thread affinity with the thread in which
they are executed
