Restore the WKT parser, example & tests

[eriktim]

This is a first attempt to restore the WKT parser.
It is probably preferred to rewrite the parser to the ol.format namespace as well,
but I would first like to have some confirmation on this before doing so.

[tschaub]

Thanks for taking this on @GingerIK! We briefly discussed starting this from scratch and writing a tokenizer based parser. But I think your approach is probably the shortest path to success. Assuming you intend to continue adapting this so it extends from ol.format.TextFeature, I think this sounds good.

I'll let others weigh in, but I'd say go for it (as long as it conforms to the same interface, we can rewrite later if someone is motivated).

[eriktim]

You're welcome, I am happy I can finally do something in return for this great library.

I am almost done extending from ol.format.TextFeature, but I do have a question about reading/writing multiple features: should the MULTI* geometries be split into mulitple features, or should this only be the case for GEOMETRYCOLLECTIONs?

P.S. I am willing to write a tokenizer based parser as well later on, if someone can provide me with an example to start from.

[tschaub]

I think by default all WKT should be read as as a single geometry. For readFeatures the result would be an array of one feature. The GEOMETRYCOLLECTION would be read as ol.geom.GeometryCollection.

We could have an option to split GEOMETRYCOLLECTION into an array of features. This could look like the following:

var format = new ol.format.WKT({splitCollection: true});
var features = format.readFeatures(wktWithGeometryCollection);
// one feature for each geometry in the collection

I don't think this option should also split/explode MULTI* geometries. This could be handled by yet another option (splitMulti maybe).

[eriktim]

Your solution makes sense, so I added the splitCollection option as well.
I'll push the new commits now, so let me know what you think.
By the way, is it required that readProjection is exported? For now I did not add the @todo api tag as it cannot return the projection anyway.

[eriktim]

Any news on this?

[tschaub]

Only the canvas renderer supports vector rendering. Can you remove this renderer option altogether ("canvas" is the default).

[tschaub]

Thanks for the continued work on this @GingerIK. I made a few additional comments that should be straightforward to address.

At some point I imagine we could rewrite this with a simple lexer (these regexes are starting to show their age), but I think this will make a useful addition.

[eriktim]

Thanks for your feedback! I just incorporated your comments and pushed a new commit.

When I find myself some extra time, I'll start trying to write a lexer if that's OK.

[tschaub]

Now that ol.View2D is gone, you'll want to change this to ol.View.

[tschaub]

@GingerIK you'll want to rebase this on master.

git pull --rebase openlayers master
git push --force origin restore-wkt-parser

(This assumes you've named the remote pointing to https://github.com/openlayers/ol3/ openlayers and your own remote is origin).

If you're comfortable with git rebase, you could also squash some of those commits. I'm happy to help with this in a separate branch (using the commits you authored) if you could use a hand.

[eriktim]

I'm comfortable with git rebase. Not with rewriting history on github, however ;-) But I think it worked.

[tschaub]

Very nice work on this @GingerIK. Thanks for all the effort.

[eriktim]

My pleasure.