I'd like to have both the values and the data categories on a graph. This is a bar chart and I'd like to have the data values and a string printed in columns off to the left of the bar:
A 1 #
B 3 ###
I tried chaining a two add(pv.Label) calls onto my graph, but it seems to do nothing - the second label set is not added. Is this something that can even be done with protovis? any advice?
vis = new pv.Panel()
.def("j", -1)
.width(800)
.height(50)
.right(3);
vis.add(pv.Bar)
.data(wData)
.bottom(0)
.width(20)
.height(function(d) d[1] * 1.2)
.left(function() this.index * 27)
.fillStyle(function() vis.j() == this.index ? "orange" : "steelblue")
.add(pv.Label) **// does nothing!!**
.bottom(0)
.textAlign("center")
.textStyle("white")
.text(function(d) d[0] )
.event("mouseover", function() vis.j(this.index))
.event("mouseout", function() vis.j(-1))
.anchor("top").add(pv.Label)
.visible(function() vis.j() >= 0)
.textStyle("white")
.text(function(d) d[1]);
vis.render();
I actually did see both labels when I tried this out. But there are a couple of things that could be fixed here. The key point is that when you're chaining methods like this, when you add() a new mark, you change the context of the following method calls, e.g.:
vis.add(pv.Bar)
// this applies to the Bar
.width(10)
.add(pv.Label)
// this applies to the label
.top(5);
There are a couple issues with this in your code:
Your event() handlers are attached to the Label, not to the Bar - unfortunately, Labels can't receive events in Protovis.
Your second Label is attached to the first Label. While this actually seems to work somewhat, it's better to avoid it - you really want it attached to the Bar.
The easy way to deal with this is to only chain methods on a single mark. You can do this by assigning the parent mark to a variable, then using that variable several times for different child marks. You also have you first Label attached directly to the Bar, and not to an anchor - attaching it to an anchor will usually give you more predictable results.
Updated code:
// make a new variable to refer to the bars
var bars = vis.add(pv.Bar)
.data(wData)
.bottom(0)
.width(20)
.height(function(d) d[1] * 1.2)
.left(function() this.index * 27)
.fillStyle(function() vis.j() == this.index ? "orange" : "steelblue")
// you need to move the events up to apply
// to the bar - labels can't receive events,
// and the index will always be 0
.event("mouseover", function() vis.j(this.index))
.event("mouseout", function() vis.j(-1));
// now add each label to the bars
bars.anchor('bottom').add(pv.Label)
.bottom(0)
.textAlign("center")
.textStyle("white")
.text(function(d) d[0] );
// and again
bars.anchor("top").add(pv.Label)
.visible(function() vis.j() >= 0)
.textStyle("white")
.text(function(d) d[1]);
There's a working version here: http://jsfiddle.net/nrabinowitz/ABmuq/
Related
I want to hide the "top" part of all connector labels of a diagram. For this, I tried to set up a script, but it currently hides ALL labels (also the "bottom" labels which I want to preserve):
// Get a reference to the current diagram
var currentDiagram as EA.Diagram;
currentDiagram = Repository.GetCurrentDiagram();
if (currentDiagram != null)
{
for (var i = 0; i < currentDiagram.DiagramLinks.Count; i++)
{
var currentDiagramLink as EA.DiagramLink;
currentDiagramLink = currentDiagram.DiagramLinks.GetAt(i);
currentDiagramLink.Geometry = currentDiagramLink.Geometry
.replace(/HDN=0/g, "HDN=1")
.replace(/LLT=;/, "LLT=HDN=1;")
.replace(/LRT=;/, "LRT=HDN=1;");
if (!currentDiagramLink.Update())
{
Session.Output(currentDiagramLink.GetLastError());
}
}
}
When I hide only the top labels manually (context menu of a connector/Visibility/Set Label Visibility), the Geometry property of the DiagramLinks remains unchanged, so I guess the detailed label visibility information must be contained somewhere else in the model.
Does anyone know how to change my script?
Thanks in advance!
EDIT:
The dialog for editing the detailed label visibility looks as follows:
My goal is unchecking the "top label" checkboxes programmatically.
In the Geometry attribute you will find a partial string like
LLT=CX=36:CY=13:OX=0:OY=0:HDN=0:BLD=0:ITA=0:UND=0:CLR=-1:ALN=1:DIR=0:ROT=0;
So in between LLT and the next semi-colon you need to locate the HDN=0 and replace that with HDN=1. A simple global change like above wont work. You need a wild card like in the regex LLT=([^;]+); to work correctly.
Because I use Comboboxes that may contain text entries of very long size,
which leads to the combobox increasing its width far beyond reasonable size,
I am trying to give a maximum width to the combobox.
If I am doing this like this:
class MyCombo : public Gtk::ComboBox {
private:
CellRendererText render;
public:
MyCombo() {
render.property_width_chars() = 10;
render.property_ellipsize() = Pango::ELLIPSIZE_END;
pack_start(render, true);
}
};
The result will be an empty cell of the desired width, which seems logical since I did not specify which column to show. But how can I do this with that attempt? Using pack_start will just bypass the renderer...
Another approach is this one:
class MyCombo : public Gtk::ComboBox {
private:
CellRendererText render;
public:
MyCombo() {
pack_start(render, true);
set_cell_data_func(render, sigc::mem_fun(*this, &MyCombo::render_iter));
}
void render_iter(const TreeModel::const_iterator& iter) {
Glib::ustring data = get_string_from_iter(iter);
int desired_width_chars = 10; //for example
render.property_text() = ellipsize_string(data, desired_width_chars);
}
};
Using that approach, it works, but the text in the popup (what opens up when u click the combobox) is also shortened which is not what I want (obviously the user should be able to read the whole string and I dont care about the popup widht.)
Can you please help me with this? I would be happy for any advice/alternative solutions.
Regards tagelicht
NOTE: set_wrap_width is a function that wraps the total number of entries in the combo box over a number of columns specified; it does not answer the question.
Using set_wrap_width(1) | Using set_wrap_width(5)
Following Noup's answer as a guide I managed to get the below code; which directly answers the question and its requirements (C++/Gtkmm).
// Get the first cell renderer of the ComboBox.
auto v_cellRenderer = (Gtk::CellRendererText*)v_comboBox.get_first_cell();
// Probably obsolete; Sets character width to 1.
v_cellRenderer->property_width_chars() = 1;
// Sets the ellipses ("...") to be at the end, where text overflows.
// See Pango::ELLIPSIZE enum for other values.
v_cellRenderer->property_ellipsize() = Pango::ELLIPSIZE_END;
// Sets the size of the box, change this to suit your needs.
// -1 sets it to automatic scaling: (width, height).
v_cellRenderer->set_fixed_size(200, -1);
Result (image):
Result of code
BE AWARE: Depending on where you perform the above code; either all the cells will be the same size, or just the box itself (intended).
From experimenting, I've found:
In the parent object constructor: All cell sizes are the same.
In a separate function: Only the first cell (the box) is affected.
I'd recommend you put the code in a function that's connected to the comboBox's changed signal, such as:
v_comboBox.signal_changed().connect(sigc::mem_fun(*this, &YourClass::comboBox_changedFunction));
This may be what you are looking for:
cell_renderer_text.set_wrap_width(10)
This is for Python, but you get the idea :-)
Unfortunately, the documentation is scarce. I found this by poking around in Anjuta/Glade.
Edit:
the docs are here. They are not overly helpful, but they do exist.
As an alternative, the following works for me without having to set wrap_width nor to subclass ComboBox (in Gtk#):
ComboBoxText cb = new ComboBoxText();
cb.Hexpand = true; //If there's available space, we use it
CellRendererText renderer = (cb.Cells[0] as CellRendererText); //Get the ComboBoxText only renderer
renderer.WidthChars = 20; //Always show at least 20 chars
renderer.Ellipsize = Pango.EllipsizeMode.End;
Note: I'm using Expand to use space if it's available. If you just want to keep the combo box on a fixed width, just remove that bit.
I need to manipulate the text elements of the first and last tick of an axis to bring them more towards the center.
I am trying to select them, one at the time, with something like svg.select('.tick:last-child text') but it doesn't work. I'd then apply .transform('translate(4,0)')...
Am I doing something wrong? How can I achieve this?
One thing you could do is to create custom sub-selections by adding methods to d3.selection.prototype. You could create a selection.first() method that selects the first item in a selection, and a selection.last() method that selects the last item. For instance:
d3.selection.prototype.first = function() {
return d3.select(this[0][0]);
};
d3.selection.prototype.last = function() {
var last = this.size() - 1;
return d3.select(this[0][last]);
};
This would let you do the following:
var tickLabels = svg.selectAll('.tick text');
tickLabels.first()
.attr('transform','translate(4,0)');
tickLabels.last()
.attr('transform','translate(-4,0)');
Of course, you need to make sure that you only have one axis if you do it that way. Otherwise, specify the axis in your initial selection:
var tickLabels = svg.selectAll('.axis.x .tick text');
HERE is an example.
Here's the cleanest method I've found:
g.selectAll(".tick:first-of-type text").remove();
g.selectAll(".tick:last-of-type text").remove();
As google brought me here, I also want to add a cleaner method to what Adam Grey wrote.
Sometimes you just want to do it without taking a reference of selectAll .
svg.selectAll('.gridlines').filter(function(d, i,list) {
return i === list.length - 1;
}).attr('display', 'none');
the 3rd parameter of the filter function gives you the selected List of elements.
They don't exist in d3 specifically, but you can use the .firstChild and .lastChild methods on a node.
You can first select all of the parents of the node, and then operate within the scope of a .each() method, like so:
d3.selectAll('.myParentElements').each(function(d,i){
var firstChild = this.firstChild,
lastChild = this.lastChild;
//Do stuff with first and last child
});
Within the scope of .each(), this refers to the individual node, which is not wrapped by a d3 selection, so all of the standard methods on a node are available.
Using .filter() with a function also works selection.filter(filter) :
var gridlines;
gridlines = svg.selectAll('.gridlines');
gridlines.filter(function(d, i) {
return i === gridlines.size() - 1;
}).attr('display', 'none');
It's for D3.js v4
d3.selection.prototype.first = function() {
return d3.select(
this.nodes()[0]
);
};
d3.selection.prototype.last = function() {
return d3.select(
this.nodes()[this.size() - 1]
);
};
Example:
var lines = svg.selectAll('line');
lines.first()
.attr('transform','translate(4,0)');
lines.last()
.attr('transform','translate(-4,0)');
Here is another, even though I used Fered's solution for a problem I met.
d3.select(d3.selectAll('*').nodes().reverse()[0])
I have two line series and how do I make points on both series selected at the same time? Basically, my chart has 2 y values sharing the same x value and I'm representing them as two series. I want to display both points as selected for a given X Value.
Hi there,Thanks for the reply. I'm doing that in
- (void)sChart:(ShinobiChart *)chart toggledSelectionForPoint:(SChartDataPoint *)dataPoint inSeries:(SChartSeries *)series atPixelCoordinate:(CGPoint)pixelPoint
SChartDataPoint* point1Series1 = [chart.datasource sChart:chart dataPointAtIndex:dataPoint.index forSeriesAtIndex:0];
point1Series1.selected = YES;
SChartDataPoint* point1Series2 = [chart.datasource sChart:chart dataPointAtIndex:dataPoint.index forSeriesAtIndex:1];
point1Series2.selected = YES;
When I print the selected state of both points after this line of code, they return 1(selected) but they don't seem to appear as selected on the chart only the one I selected on the chart on device seem to appear as selected though I'm calling redrawChart after that. Any help would be appreciated
I think that it's likely (and I'm guessing because I can't see your code) that your chart data source isn't returning a reference to a datapoint which is part of the chart, but instead generating a new datapoint object each time you request one.
In order to cope with this you can request the data points from the chart itself, via the dataSeries property on SChartSeries objects.
The following delegate method should perform the selection you require.
- (void)sChart:(ShinobiChart *)chart toggledSelectionForPoint:(SChartDataPoint *)dataPoint inSeries:(SChartSeries *)series atPixelCoordinate:(CGPoint)pixelPoint
{
// Selection details
NSInteger dataPointIndex = dataPoint.index;
BOOL selected = dataPoint.selected;
for (SChartSeries *chartSeries in chart.series) {
// If only one data point in the series can be selected at once, then deselect the rest
if(!series.togglePointSelection && selected) {
for(SChartDataPoint *dp in chartSeries.dataSeries.dataPoints) {
dp.selected = NO;
}
}
// Find the data point and perform the selection
SChartDataPoint *dp = chartSeries.dataSeries.dataPoints[dataPointIndex];
dp.selected = selected;
}
}
Hope that helps.
You should be able to set .selected on the datapoints and customise the series.style.selectedPointStyle properties to display points how you wish :)
I have looked at the this link for a tutorial on dojo drag and drop feature. But one thing I have noticed is that in all cases of the examples, the items to be dragged around are always a simple item, just a string object...
I need to create something like an item group where you can drag an item into the item group to append into the group and to drag the item group around as a whole.
Hence my question, is it possible to drag and drop a dojo.dnd.Source item into another dojo.dnd.Source item?
Short answer: no. Many people tried to patch it, but found more and more non-working edge cases, so those patches never made the Dojo proper.
If you truly need to show and manipulate a hierarchical data, consider a Tree Dijit.
The problem is that when you start dragging and you drag over a Source of a child container, everything gets messed up. (Not exactly sure how). What you can do, is hide those child sources so that their overSource events never trigger:
1) Overrode the checkAcceptance function in Source.js. Just added the following for the if(!flag) return false;:
if(!flag){
/**
* Main Source
* - Group 1
* -- Child 1
* -- Child 2
* - Group 2
*/
var node = dojo.byId(this.node);
// If the node being moved is the source, skip, but don't hide from view.
if('#'+dojo.attr(source.node, 'id') != '#'+dojo.attr(node, 'id')){
// If the node being moved is an immediate child of the container, you can move it.
if(dojo.query('#'+dojo.attr(source.node, 'id') + '>#'+dojo.attr(node, 'id')).length) {
return true;
}
// If this source is not a parent of the element, hide it.
if(dojo.query('#'+dojo.attr(node, 'id') + ' #'+dojo.attr(source.node, 'id')).length == 0)
dojo.addClass(node, 'hiddenSource');
}
return false;
}
2) You need to also add the following as the first line under if(this.isDragging) in onMouseMove (important)
var node = dojo.byId(this.node);
// If this is immeditae child, drop it.
if(dojo.query('#'+dojo.attr(m.source.node, 'id') + '>#'+dojo.attr(node, 'id')).length){
m.canDrop(true);
return;
}
3) Extended onDndDrop to remove the added class to re display the hidden elements.
onDndDrop: function(source, nodes, copy, target)
{
this.inherited(arguments);
dojo.forEach(dojo.query('.hiddenSource'),
function(el){dojo.removeClass(el, 'hiddenSource');}
);
}
4) Extend onDndCancel to do the above
onDndCanel: function()
{
this.inherited(arguments);
dojo.forEach(dojo.query('.hiddenSource'),
function(el){dojo.removeClass(el, 'hiddenSource');}
);
}
This isn't the best solution since it hides the elements that can't be used with the current element that you are positioning, but it worked for me.