Struct gtk4::CellArea [−][src]
pub struct CellArea(_);
Expand description
An abstract class for laying out GtkCellRenderers
The CellArea
is an abstract class for CellLayout
widgets
(also referred to as “layouting widgets”) to interface with an
arbitrary number of CellRenderer
s and interact with the user
for a given TreeModel
row.
The cell area handles events, focus navigation, drawing and size requests and allocations for a given row of data.
Usually users dont have to interact with the CellArea
directly
unless they are implementing a cell-layouting widget themselves.
Requesting area sizes
As outlined in
[GtkWidget’s geometry management section][geometry-management],
GTK uses a height-for-width
geometry management system to compute the sizes of widgets and user
interfaces. CellArea
uses the same semantics to calculate the
size of an area for an arbitrary number of TreeModel
rows.
When requesting the size of a cell area one needs to calculate
the size for a handful of rows, and this will be done differently by
different layouting widgets. For instance a TreeViewColumn
always lines up the areas from top to bottom while a IconView
on the other hand might enforce that all areas received the same
width and wrap the areas around, requesting height for more cell
areas when allocated less width.
It’s also important for areas to maintain some cell
alignments with areas rendered for adjacent rows (cells can
appear “columnized” inside an area even when the size of
cells are different in each row). For this reason the CellArea
uses a CellArea
Context object to store the alignments
and sizes along the way (as well as the overall largest minimum
and natural size for all the rows which have been calculated
with the said context).
The CellArea
Context is an opaque object specific to the
CellArea
which created it (see CellAreaExt::create_context()
).
The owning cell-layouting widget can create as many contexts as
it wishes to calculate sizes of rows which should receive the
same size in at least one orientation (horizontally or vertically),
However, it’s important that the same CellArea
Context which
was used to request the sizes for a given TreeModel
row be
used when rendering or processing events for that row.
In order to request the width of all the rows at the root level
of a TreeModel
one would do the following:
⚠️ The following code is in C ⚠️
GtkTreeIter iter;
int minimum_width;
int natural_width;
valid = gtk_tree_model_get_iter_first (model, &iter);
while (valid)
{
gtk_cell_area_apply_attributes (area, model, &iter, FALSE, FALSE);
gtk_cell_area_get_preferred_width (area, context, widget, NULL, NULL);
valid = gtk_tree_model_iter_next (model, &iter);
}
gtk_cell_area_context_get_preferred_width (context, &minimum_width, &natural_width);
Note that in this example it’s not important to observe the
returned minimum and natural width of the area for each row
unless the cell-layouting object is actually interested in the
widths of individual rows. The overall width is however stored
in the accompanying CellArea
Context object and can be consulted
at any time.
This can be useful since CellLayout
widgets usually have to
support requesting and rendering rows in treemodels with an
exceedingly large amount of rows. The CellLayout
widget in
that case would calculate the required width of the rows in an
idle or timeout source (see g_timeout_add()
) and when the widget
is requested its actual width in vfunc::Gtk::Widget::measure
it can simply consult the width accumulated so far in the
CellArea
Context object.
A simple example where rows are rendered from top to bottom and take up the full width of the layouting widget would look like:
⚠️ The following code is in C ⚠️
static void
foo_get_preferred_width (GtkWidget *widget,
int *minimum_size,
int *natural_size)
{
Foo *foo = FOO (widget);
FooPrivate *priv = foo->priv;
foo_ensure_at_least_one_handfull_of_rows_have_been_requested (foo);
gtk_cell_area_context_get_preferred_width (priv->context, minimum_size, natural_size);
}
In the above example the Foo widget has to make sure that some
row sizes have been calculated (the amount of rows that Foo judged
was appropriate to request space for in a single timeout iteration)
before simply returning the amount of space required by the area via
the CellArea
Context.
Requesting the height for width (or width for height) of an area is
a similar task except in this case the CellArea
Context does not
store the data (actually, it does not know how much space the layouting
widget plans to allocate it for every row. It’s up to the layouting
widget to render each row of data with the appropriate height and
width which was requested by the CellArea
).
In order to request the height for width of all the rows at the
root level of a TreeModel
one would do the following:
⚠️ The following code is in C ⚠️
GtkTreeIter iter;
int minimum_height;
int natural_height;
int full_minimum_height = 0;
int full_natural_height = 0;
valid = gtk_tree_model_get_iter_first (model, &iter);
while (valid)
{
gtk_cell_area_apply_attributes (area, model, &iter, FALSE, FALSE);
gtk_cell_area_get_preferred_height_for_width (area, context, widget,
width, &minimum_height, &natural_height);
if (width_is_for_allocation)
cache_row_height (&iter, minimum_height, natural_height);
full_minimum_height += minimum_height;
full_natural_height += natural_height;
valid = gtk_tree_model_iter_next (model, &iter);
}
Note that in the above example we would need to cache the heights returned for each row so that we would know what sizes to render the areas for each row. However we would only want to really cache the heights if the request is intended for the layouting widgets real allocation.
In some cases the layouting widget is requested the height for an
arbitrary for_width, this is a special case for layouting widgets
who need to request size for tens of thousands of rows. For this
case it’s only important that the layouting widget calculate
one reasonably sized chunk of rows and return that height
synchronously. The reasoning here is that any layouting widget is
at least capable of synchronously calculating enough height to fill
the screen height (or scrolled window height) in response to a single
call to vfunc::Gtk::Widget::measure
. Returning
a perfect height for width that is larger than the screen area is
inconsequential since after the layouting receives an allocation
from a scrolled window it simply continues to drive the scrollbar
values while more and more height is required for the row heights
that are calculated in the background.
Rendering Areas
Once area sizes have been acquired at least for the rows in the
visible area of the layouting widget they can be rendered at
vfunc::Gtk::Widget::snapshot
time.
A crude example of how to render all the rows at the root level runs as follows:
⚠️ The following code is in C ⚠️
GtkAllocation allocation;
GdkRectangle cell_area = { 0, };
GtkTreeIter iter;
int minimum_width;
int natural_width;
gtk_widget_get_allocation (widget, &allocation);
cell_area.width = allocation.width;
valid = gtk_tree_model_get_iter_first (model, &iter);
while (valid)
{
cell_area.height = get_cached_height_for_row (&iter);
gtk_cell_area_apply_attributes (area, model, &iter, FALSE, FALSE);
gtk_cell_area_render (area, context, widget, cr,
&cell_area, &cell_area, state_flags, FALSE);
cell_area.y += cell_area.height;
valid = gtk_tree_model_iter_next (model, &iter);
}
Note that the cached height in this example really depends on how
the layouting widget works. The layouting widget might decide to
give every row its minimum or natural height or, if the model content
is expected to fit inside the layouting widget without scrolling, it
would make sense to calculate the allocation for each row at
the time the widget is allocated using gtk_distribute_natural_allocation()
.
Handling Events and Driving Keyboard Focus
Passing events to the area is as simple as handling events on any
normal widget and then passing them to the CellAreaExtManual::event()
API as they come in. Usually CellArea
is only interested in
button events, however some customized derived areas can be implemented
who are interested in handling other events. Handling an event can
trigger the CellArea
::focus-changed signal to fire; as well as
CellArea
::add-editable in the case that an editable cell was
clicked and needs to start editing. You can call
CellAreaExt::stop_editing()
at any time to cancel any cell editing
that is currently in progress.
The CellArea
drives keyboard focus from cell to cell in a way
similar to Widget
. For layouting widgets that support giving
focus to cells it’s important to remember to pass CellRendererState::FOCUSED
to the area functions for the row that has focus and to tell the
area to paint the focus at render time.
Layouting widgets that accept focus on cells should implement the
vfunc::Gtk::Widget::focus
virtual method. The layouting widget is always
responsible for knowing where TreeModel
rows are rendered inside
the widget, so at vfunc::Gtk::Widget::focus
time the layouting widget
should use the CellArea
methods to navigate focus inside the area
and then observe the GtkDirectionType to pass the focus to adjacent
rows and areas.
A basic example of how the vfunc::Gtk::Widget::focus
virtual method
should be implemented:
⚠️ The following code is in C ⚠️
static gboolean
foo_focus (GtkWidget *widget,
GtkDirectionType direction)
{
Foo *foo = FOO (widget);
FooPrivate *priv = foo->priv;
int focus_row;
gboolean have_focus = FALSE;
focus_row = priv->focus_row;
if (!gtk_widget_has_focus (widget))
gtk_widget_grab_focus (widget);
valid = gtk_tree_model_iter_nth_child (priv->model, &iter, NULL, priv->focus_row);
while (valid)
{
gtk_cell_area_apply_attributes (priv->area, priv->model, &iter, FALSE, FALSE);
if (gtk_cell_area_focus (priv->area, direction))
{
priv->focus_row = focus_row;
have_focus = TRUE;
break;
}
else
{
if (direction == GTK_DIR_RIGHT ||
direction == GTK_DIR_LEFT)
break;
else if (direction == GTK_DIR_UP ||
direction == GTK_DIR_TAB_BACKWARD)
{
if (focus_row == 0)
break;
else
{
focus_row--;
valid = gtk_tree_model_iter_nth_child (priv->model, &iter, NULL, focus_row);
}
}
else
{
if (focus_row == last_row)
break;
else
{
focus_row++;
valid = gtk_tree_model_iter_next (priv->model, &iter);
}
}
}
}
return have_focus;
}
Note that the layouting widget is responsible for matching the GtkDirectionType values to the way it lays out its cells.
Cell Properties
The CellArea
introduces cell properties for CellRenderer
s.
This provides some general interfaces for defining the relationship
cell areas have with their cells. For instance in a CellArea
Box
a cell might “expand” and receive extra space when the area is allocated
more than its full natural request, or a cell might be configured to “align”
with adjacent rows which were requested and rendered with the same
CellArea
Context.
Use gtk_cell_area_class_install_cell_property()
to install cell
properties for a cell area class and gtk_cell_area_class_find_cell_property()
or gtk_cell_area_class_list_cell_properties()
to get information about
existing cell properties.
To set the value of a cell property, use [CellAreaExtManual::cell_set_property()
][crate::prelude::CellAreaExtManual::cell_set_property()],
gtk_cell_area_cell_set()
or gtk_cell_area_cell_set_valist()
. To obtain
the value of a cell property, use [CellAreaExtManual::cell_get_property()
][crate::prelude::CellAreaExtManual::cell_get_property()],
gtk_cell_area_cell_get()
or gtk_cell_area_cell_get_valist()
.
This is an Abstract Base Class, you cannot instantiate it.
Implements
CellAreaExt
, glib::ObjectExt
, BuildableExt
, CellLayoutExt
, CellAreaExtManual
, CellLayoutExtManual
Trait Implementations
This method returns an ordering between self
and other
values if one exists. Read more
This method tests less than (for self
and other
) and is used by the <
operator. Read more
This method tests less than or equal to (for self
and other
) and is used by the <=
operator. Read more
This method tests greater than (for self
and other
) and is used by the >
operator. Read more
Returns the type identifier of Self
.
Auto Trait Implementations
impl RefUnwindSafe for CellArea
impl UnwindSafe for CellArea
Blanket Implementations
Mutably borrows from an owned value. Read more
Upcasts an object to a superclass or interface T
. Read more
Upcasts an object to a reference of its superclass or interface T
. Read more
Tries to downcast to a subclass or interface implementor T
. Read more
Tries to downcast to a reference of its subclass or interface implementor T
. Read more
Tries to cast to an object of type T
. This handles upcasting, downcasting
and casting between interface and interface implementors. All checks are performed at
runtime, while downcast
and upcast
will do many checks at compile-time already. Read more
Tries to cast to reference to an object of type T
. This handles upcasting, downcasting
and casting between interface and interface implementors. All checks are performed at
runtime, while downcast
and upcast
will do many checks at compile-time already. Read more
Casts to T
unconditionally. Read more
Casts to &T
unconditionally. Read more
Returns true
if the object is an instance of (can be cast to) T
.
pub fn set_property<'a, N, V>(
&self,
property_name: N,
value: V
) -> Result<(), BoolError> where
N: Into<&'a str>,
V: ToValue,
pub fn set_property_from_value<'a, N>(
&self,
property_name: N,
value: &Value
) -> Result<(), BoolError> where
N: Into<&'a str>,
pub fn set_properties_from_value(
&self,
property_values: &[(&str, Value)]
) -> Result<(), BoolError>
pub fn has_property<'a, N>(&self, property_name: N, type_: Option<Type>) -> bool where
N: Into<&'a str>,
pub fn find_property<'a, N>(&self, property_name: N) -> Option<ParamSpec> where
N: Into<&'a str>,
Safety Read more
Safety Read more
Safety Read more
Safety Read more
pub fn connect<'a, N, F>(
&self,
signal_name: N,
after: bool,
callback: F
) -> Result<SignalHandlerId, BoolError> where
N: Into<&'a str>,
F: 'static + Fn(&[Value]) -> Option<Value> + Send + Sync,
Same as connect
but takes a SignalId
instead of a signal name.
pub fn connect_local<'a, N, F>(
&self,
signal_name: N,
after: bool,
callback: F
) -> Result<SignalHandlerId, BoolError> where
N: Into<&'a str>,
F: 'static + Fn(&[Value]) -> Option<Value>,
Same as connect_local
but takes a SignalId
instead of a signal name.
pub unsafe fn connect_unsafe<'a, N, F>(
&self,
signal_name: N,
after: bool,
callback: F
) -> Result<SignalHandlerId, BoolError> where
N: Into<&'a str>,
F: Fn(&[Value]) -> Option<Value>,
Same as connect_unsafe
but takes a SignalId
instead of a signal name.
Emit signal by signal id.
Same as emit
but takes Value
for the arguments.
Emit signal by its name.
Same as emit_by_name
but takes Value
for the arguments.
Emit signal with details by signal id.
Same as emit_with_details
but takes Value
for the arguments.
pub fn connect_notify<F>(&self, name: Option<&str>, f: F) -> SignalHandlerId where
F: 'static + Fn(&T, &ParamSpec) + Send + Sync,
pub fn connect_notify_local<F>(
&self,
name: Option<&str>,
f: F
) -> SignalHandlerId where
F: 'static + Fn(&T, &ParamSpec),
pub unsafe fn connect_notify_unsafe<F>(
&self,
name: Option<&str>,
f: F
) -> SignalHandlerId where
F: Fn(&T, &ParamSpec),
pub fn bind_property<'a, O, N, M>(
&'a self,
source_property: N,
target: &'a O,
target_property: M
) -> BindingBuilder<'a> where
O: ObjectType,
N: Into<&'a str>,
M: Into<&'a str>,
Returns a SendValue
clone of self
.