Option to configure ID generation

[lordi]

If a CityObject does not have a gml:id, an ID will be generated for this object when reading. These IDs will also be used when transforming CityGML to CityJSON.

While it is nice to have deterministic IDs, the following does not work:

• Convert a.gml to a.json => 1 object gets ID_1
• Convert b.gml to b.json => 1 object gets ID_1
• Merge a.json + b.json => Non-unique ID ID_1

Can we make the IDs globally unique (i.e. random/non-deterministic) or have something like --id-prefix that will be used in the ID generation process?

[clausnagel]

Hi @lordi, it's not supposed to be like this. Missing IDs should be replaced with "ID_" + a random UUID. Thus, the resulting IDs should already be globally unique. I did a small test and the result looks ok. I wonder why you always get ID_1.

Are you using citygml-tools v2 or v1? Could you share an excerpt of your data for testing?

[lordi]

Thanks for the quick response. The ID_1 was just a placeholder for a non-random UUID. I'm using v2.1.

I did some more research. Most IDs are indeed random, but certain objects get a non-random UUID. have a dataset where some objects get assigned the same ID every run (specifically, these are Buildings in a CityObjectGroup). This ID does not appear in the CityGML. Does it ring a bell? Otherwise I can send you a dataset by mail.

[clausnagel]

Ok, seems to be reproducible then, that's good. Yes, please send a dataset by email because I actually don't know why buildings inside a group would make a difference.

[clausnagel]

Thanks @lordi, the test dataset was really helpful. I fixed the bug with bd361ad4e7da911c82ed1ded78d878a6869a9960.

The bugfix will be included in the next release. In the meantime, you can either use the edge Docker image from Docker Hub or build citygml-tools from the master branch.

[lordi]

Awesome, thanks @clausnagel . Will test the change.

[lordi]

I can confirm that the patch works as expected. Thanks again!