Adding an overflow option that will draw words even if they don't fit

[john-guerra]

When creating word clouds in svgs with small sizes the big words (arguably the most important ones) might not be drawn because they won't fit. This change adds an overflow option to the wordcloud that will draw words even if they don't fit on the available space, this will truncate words, but at least will guarantee that the most significant word is (at least partially) displayed. The default value is false.

[jasondavies]

Related #36.

[SebT]

@jasondavies Hi. First thanks for your work.

But why isn't this PR merged yet?

[ptandler]

I did some testing with this patch, and it works nicely in most cases. However, I noticed that it has a HUGE impact on performance when creating the word cloud. With overflow=true the time to create a 250-300 words cloud increased on my machine from something around 1-3 sec to 40-60 sec.

I'm not sure what exactly causes this dramatically increase, one thought is that it might be sufficient to only have the overflow option enabled for the first few biggest words, and not for every word in the cloud. But I didn't have the time to test this or to understand how the algorithm actually works ...

Does someone have any experiences with this option?

[bjuergens]

here is my guess why this takes so much longer: with this option instead of silently dropping the words, it tries to fit them in anyways. Which requires looking for collisions as well as actually placing the words, which are both things that will probably take a longer time, if more words are there to begin with. If you have many words, that would be dropped without overflow, than enabling overflow will slow down your code a lot.

My guess is that in your cloud with 250+ words, less than half are actually rendered. You can check the number of words that actually get rendered with something like this:

  .on("end", function(output) {
      if (word.length !== output.length) console.log(word.length - output.length)
   })