WordCram 0.6.1 is here, with improvements to the examples, and some bug fixes.
Two new examples show how to use the new ShapeBasedPlacer, and the Event callbacks introduced in 0.6.0. Also, some examples were still using the old WordCram.NO_SPACE constants – thanks to Pierre Haessig for a fix. Look under File > Examples > Contributed Libraries > WordCram to see them.
Happy word cramming!
WordCram 0.6.0 is here, and brings with it arbitrary shape-based layouts, SVG export, event callbacks so you can monitor your sketch’s progress, and as usual, a few bug fixes. Special thanks to simpsus for helping out so much!
Ever wonder what WordCram’s doing in there? Ever have a WordCram just hang, making your fan whir, and you have no idea why it’s taking so long? Callbacks can help you figure it out.
Add methods like this to your Processing sketch, the way you do for mouse or keyboard events:
void wordSkipped(Word word) {
println("Skipped: " + word);
}
void wordDrawn(Word word) {
println("Just drew: " + word);
}
As your sketch runs, you’ll see the print-outs. You can run any kind of code in there, too.
You can see the full list of callbacks in the javadoc for the Observer class.
WordCram has been able to export to PDF since the 0.5.0 release, but for a real scalable image, SVG is much better.
Now, it’s dirt-simple to export to SVG. Just add this to the WordCram setup:
toSvg("my-wordcram.svg", width, height)
…as in:
import wordcram.*;
size(800, 600);
try {
new WordCram(this)
.toSvg("nytimes.svg", width, height)
.fromWebPage("http://nytimes.com")
.drawAll();
}
catch (java.io.FileNotFoundException x) {
println(x);
}
(Unfortunately, you have to wrap it in the try-catch. I’ll work on smoothing that out for 0.6.1.)
Here’s what you get:
This is the big one. This is the last major feature WordCram lacked, one that I really wanted to add.
Start with an image like this:
Load the image, extract any color from it into a Shape (here, I’m using black), and make a ShapeBasedPlacer for it:
PImage image = loadImage("batman.gif");
Shape imageShape = new ImageShaper().shape(image, #000000);
ShapeBasedPlacer placer = new ShapeBasedPlacer(imageShape);
Use that ShapeBasedPlacer as both the WordPlacer and the WordNudger in your WordCram:
background(#110000);
new WordCram(this).
fromWebPage("http://en.wikipedia.org/wiki/Batman").
withPlacer(placer).
withNudger(placer).
sizedByWeight(12, 120).
withFont("Futura-CondensedExtraBold").
withColor(#DBC900).
drawAll();
And voilà:
It can be slow sometimes, though the event callbacks can help you see where it’s spending its time, and tune accordingly. And the API isn’t totally there yet – I’ll work on smoothing out both for the future. But the capability is there, so give it a try.
Download WordCram 0.6.0, and get cramming.
If you tweet, and you make some WordCrams and tweet them, let @wordcram know!
WordCram 0.5.7 is out, and it’s about making things just a little easier.
withWordPadding(0).This is a quick release, because simpsus is helping me smooth out making WordCrams from shapes, which will be the next release:
In the meantime, download the zip or the tar, and have fun cramming!
You installed WordCram, gave it a list of words, and ran it – and some of your words are missing. What’s up?
WordCram will try really hard to place all your words, but if it has a hard time placing one, it’ll eventually move on to the next word. (If it didn’t, you might still be waiting now for a WordCram you started to render a year-and-a-half ago.) Normally, this is ok, because WordCram isn’t rendering mission-critically important data – it’s for fun.
But if it’s not rendering some words that you really want to see, what can you do?
The first thing is to understand why it’s skipping your words. Paste this code into your sketch, right after yourWordCram renders:
for (Word word : yourWordCram.getSkippedWords()) {
switch (word.wasSkippedBecause()) {
case WordSkipReason.SHAPE_WAS_TOO_SMALL:
println(word.word + ": shape was too small");
break;
case WordSkipReason.WAS_OVER_MAX_NUMBER_OF_WORDS:
println(word.word + ": was over max # of words");
break;
case WordSkipReason.NO_SPACE:
println(word.word + ": no room to place it");
break;
}
}
2014-10-09: Updated that code sample to reflect the new WordSkipReason enum.
Let’s look at each of these cases:
sizedByWeight or sizedByRank.maxNumberOfWordsToDraw, telling WordCram to only render so many words, and this word was over the limit. You should only use this if your corpus has tons of distinct words, and you need to keep WordCram from chugging for ever. If this is you, try setting a higher limit. But this probably isn’t your problem – why would you be worried about a word that’s so low in weight?Here’s a bit more about that last case, not having enough room. Suppose your sketch only has 3 words, and you placed them all spread out, like below. WordCram should have no problem placing them where you want them to go.
But now suppose you put them all in the same corner. On the same spot, they’ll overlap:
To fix this, WordCram nudges them around, so they’re close to where you want:
With only 3 words, this is pretty easy. With 30 words, it’s still easy, if they’re small enough.
But with too many words, WordCram eventually will throw up its hands and move on to the next word.
It’s ok, no one should eat bubblegum ice cream anyway
This has been a feature request for a long time: making WordCrams in arbitrary shapes, like Tagxedo does.
It’s currently slow, and a bit limited – you have to provide a java.awt.Shape, not (say) an image mask – but it’s a first step. I hope to package it up nicely for the next WordCram release.
Below is the guts of the code, but here’s the output:
class ShapeBasedPlacer implements WordPlacer, WordNudger {
Area area;
float minX;
float minY;
float maxX;
float maxY;
public ShapeBasedPlacer(Shape shape) {
this.area = new Area(shape);
Rectangle2D areaBounds = area.getBounds2D();
this.minX = (float)areaBounds.getMinX();
this.minY = (float)areaBounds.getMinY();
this.maxX = (float)areaBounds.getMaxX();
this.maxY = (float)areaBounds.getMaxY();
}
public PVector place(Word w, int rank, int count, int ww, int wh, int fw, int fh) {
w.setProperty("width", ww);
w.setProperty("height", wh);
for (int i = 0; i < 1000; i++) {
float newX = randomBetween(minX, maxX);
float newY = randomBetween(minY, maxY);
if (area.contains(newX, newY, ww, wh)) {
return new PVector(newX, newY);
}
}
return new PVector(-1, -1);
}
public PVector nudgeFor(Word word, int attempt) {
PVector target = word.getTargetPlace();
float wx = target.x;
float wy = target.y;
float ww = (Integer)word.getProperty("width");
float wh = (Integer)word.getProperty("height");
for (int i = 0; i < 1000; i++) {
float newX = randomBetween(minX, maxX);
float newY = randomBetween(minY, maxY);
if (area.contains(newX, newY, ww, wh)) {
return new PVector(newX - wx, newY - wy);
}
}
return new PVector(-1, -1);
}
float randomBetween(float a, float b) {
return a + random(b - a);
}
}
void setup() {
Shape heart = makeHeart();
ShapeBasedPlacer shapeBasedPlacer = new ShapeBasedPlacer(heart);
new WordCram(this)
.withPlacer(shapeBasedPlacer)
.withNudger(shapeBasedPlacer)
.drawAll();
}
This release includes three bug fixes (#6, #7, and #8), and some small performance improvements, that should get us a tiny bit closer to faster wordcrams. Get the better bits, and happy hacking!
This release is mostly a fix for issue #4. dbasch couldn’t make a word cloud, because cue.language couldn’t decide what language his text was in. Next time cue gets confused, instead of stopping the whole show, WordCram will gracefully ignore it, and include the word.
I also made public some classes that weren’t before: BBTree, BBTreeBuilder, and WordShaper. If you’d like to make Processing sketches that do overlap-checking like WordCram does, but your shapes aren’t necessarily words, these might help you; you can see how they’re used by looking at the code in the WordCramEngine class. (Making them public also makes it easier for me to experiment with changing WordCram’s layout algorithms.)
As usual, you can download the new version at github. Happy word cramming!
Just a tiny bug fix for making word crams with the fromTextString method. Download only the freshest of WordCram bits!
Happy Fourth of July!
In the spirit of independence, here’s a WordCram made from the U.S. Declaration of Independence, with open-source software (Processing and WordCram), and open-source fonts (AveriaSerif-Regular and Jane Austen):
You can see it at OpenProcessing; here’s the source, in its entirety:
import wordcram.*;
size(800, 500);
background(#F5F1E3);
PFont averia = createFont("AveriaSerif-Regular.ttf", 1);
PFont janeAusten = createFont("JaneAust.ttf", 1);
new WordCram(this)
.fromTextFile("declaration.txt")
.withFonts(averia, janeAusten)
.withColors(#B20E0E, #0E12B2)
.angledAt(0)
.withWordPadding(1)
.drawAll();
WordCram 0.5.2 makes it easier for your users to click on words. You can download the .zip or the .tar.gz at the new github download page.
It’s common to let people interact with a word cloud by clicking on the words. But WordCram has always been very particular about where you click: you have to click directly on the part of the word that’s colored in. If you click in the middle of an O, or between the arms of an X, it won’t register. This is pretty frustrating, because people aren’t that precise, and no one expects it to work that way.
It does it this way because words can appear inside other words, or intertwined with them. So it can’t just count it as a click, if you click inside the word’s box. (That’s why it was so fussy, at first – being fussy side-stepped this problem.)
WordCram 0.5.2 solves this twist by giving you the smallest word whose box contains your click, on the theory that, if you’d wanted that bigger word, you probably wouldn’t have clicked so close to the smaller word.
There’s an example sketch on OpenProcessing. Try it out, and see whether it doesn’t feel natural. Or download it, and try it out in your own sketch!
