evedataviewer.gui.model module
Model for the evedataviewer GUI application.
Note
A note for developers: Following the Model–View (MV) pattern, the model
does not care nor know about the view, but the view (here:
evedataviewer.gui.mainwindow.MainWindow) knows about the model
and connects to the appropriate signals and slots. The model is the
place to define the “business logic”, i.e. the (abstract) behaviour of
the application. Furthermore, the model is responsible to provide the
data to the connected view(s).
Core business logic
For the time being, a perhaps simplistic birds-eye view on what evedataviewer is supposed to do:
Allow the user to browse data files and select one or several files for display.
Allow for several different display modes of the data (graphical, tabular, …).
For the model, this boils down to:
Provide a list of datasets to be displayed.
Load data contained in the list of datasets to be selected.
Display data contained in the list of datasets to be displayed, with the display mode depending on internal settings.
Additional aspects that are handled from within the individual datasets:
Select channels and axes to plot (basically, x and y data)
Handle subscans within a dataset (and remember the current subscan number)
Things to be decided upon:
How to deal with displaying multiple datasets with incompatible axes?
Shall datasets remember the setting for channel/axis/subscan during one session of the GUI? Currently, they do. Do we need to have a “reset” button or similar?
Shall datasets remember plot-specific settings?
Implementation
Currently, the idea is very simplistic:
dictionary with loaded datasets
keys are the corresponding filenames
list of datasets to be displayed
list comes, e.g., from FileBrowser widget
if element in list has no corresponding entry in dict with loaded datasets, load dataset
if list changes, refresh plot
figure
handle of Matplotlib figure
if None, do not plot
Module documentation
- class evedataviewer.gui.model.Model
Bases:
QObjectModel for the evedataviewer GUI application.
The model in the model-view pattern contains the business logic. Furthermore, the model does not know nor care about views, but views are aware of the model. While views can modify/update the model, the model takes care of emitting signals whenever its internal state changes. Thus, views can subscribe to these signals and update themselves accordingly.
- datasets
Datasets loaded
The keys are the filenames, and the datasets are accessed accordingly by their filenames.
Note
Currently, every dataset once loaded is retained in this dict. If this ever gets a problem due to memory consumption, we will need to implement something taking care of a threshold of memory consumption of this object and starting to delete the oldest (i.e. first) entries. Note, however, that the first entry is not necessarily the longest not visited (hence least important) one. Therefore, it might be better to have a second structure containing keys and timestamps of last visit that allow for detecting the longest not-visited dataset.
For inspiration of a
sys.getsizeof()equivalent that works with containers such as adict, see https://code.activestate.com/recipes/577504/- Type:
- figure
Figure used to plot data
Note
In the long run, this should probably be replaced with a plotter object taking care of plotting the selected data. As a MultiPlotter can handle 1..*n* datasets, whereas a SinglePlotter can only handle a single dataset, using a MultiPlotter should be rather safe.
Perhaps both, a figure and a plotter are necessary, as this would allow to change/replace the plotter and take care from within the model that the plotter will always be (re)connected to the figure.
- Type:
- dataset_selection_changed
Signal emitted when the selection of datasets changed.
The signal contains the names of the selected datasets as
listparameter.- Returns:
datasets – Names of the selected datasets.
- Return type:
- dataset_changed
Signal that should be emitted whenever a dataset changes.
The signal contains the name of the dataset that has changed.
- Returns:
dataset – Name of the dataset that has been changed.
- Return type:
- current_dataset_changed
Signal that should be emitted whenever the current dataset changes.
The difference between current dataset and dataset selection: the latter is generally a list, the current dataset the one that is operated upon (i.e., the current selection from the list of datasets to be displayed).
The signal contains the name of the currently selected dataset.
- Returns:
dataset – Name of the dataset that is currently selected.
- Return type:
- plot_changed
Signal that should be emitted whenever the plot or its properties change.
- property datasets_to_display
Datasets that should be displayed.
- property current_dataset
Currently active dataset from the list of datasets to display.
While the list of datasets to be displayed contains the names of all datasets that are currently displayed in the GUI, the currently active dataset is the one element from this list currently acted upon. This is crucial for all those operations that can only work on one dataset at a time, be it displaying data or acting upon the dataset itself.
- display_data()
Display the datasets listed as to be displayed.
This method is a Qt slot events can be connected to. Hence, a GUI widget changing the state of the model or datasets contained therein usually connects to this slot.
Which kind of display will be used depends on the state of the internal property
_display_mode. Thus, you can easily set/add different display modes and need not care of how to display the data.Possibilities would be, besides the currently implemented graphical display (
plot), a tabular display or even a diff display for two or more datasets, showing the differences in the metadata.Important
Usually, you should call this method rather than those methods dealing with a concrete display mode, such as
plot_data(), as this method takes care of loading the data if they are not yet internally available.Note
For developers: For maximum flexibility and modularity, the actual display mode is used as first part of the method name. Hence, if you want to implement/support further display types, create a non-public method
<display_mode>_data()and implement all necessary functionality therein.
- load_data(filename='')
Load data from a given filename.
- Parameters:
filename (
str) – Name of the file the data should be loaded from
Note
For developers: For maximum flexibility and modularity, loading data is delegated to importers and datasets, and the importer obtained using an importer factory. These concepts are all implemented in the ASpecD framework and mimicked here, allowing for a later drop-in replacement with an ASpecD-derived package.