How to compress a string?

Question

I would like to have a reversible compression for a type of string so that i can include it in URLs without keeping track of what it refers to. The string i would like to compress is SVG path string, here is a short primer: http://apike.ca/prog_svg_paths.html

Basically, the string contains a character, followed by arbitrary number of integers, then another character followed by arbitrary number of integers and so on.

If anyone knows of a good resource for this, it would be much appreciated!

Jason

Answer 1

Many compression algorithms are well documented, a couple even have js implementations:

• GZip A common (reasonably) good compression algorithm, I know there's a JS impl, i'm just hunting the URL

• LZW Another question points to an LZW implementation in JS

• Arithmetic coding (i did this, but the model it uses is stupid so doesn't achieve the best compression rates it could)

Answer 2

Sounds like you might benefit from single and double RLE compression.

A primer on this can be seen here:

http://pp19dd.com/2011/10/query-string-limits-encoding-hundreds-of-checkboxes-with-rle/#demo

The library should be flexible enough to modify your compression pattern to something more preferable. The writeup explains how this works; might be a good start to optimize your SVG case.

Answer 3

You could try Huffman compression. Number of different chars is 20-30, and if the string is long, compression should be effective.