WIP: Add Dockerfile and a script to do initial data bootstrap

[iandees]

See also tilezen/tileserver#93.

[nvkelso]

Can this be a variable so I could, say, load up San Francisco instead?

How might we support multiple metro extracts?

[nvkelso]

Would need to get the filename part of the earlier variable per comment https://github.com/tilezen/vector-datasource/pull/1254/files#r117051094.

[iandees]

The import script is currently failing with:

+ ./perform-sql-updates.sh -h postgres -p 5432 -U osm -d osm
Creating custom schema...
psql:schema.sql:6: ERROR:  relation "wof_neighbourhood_placetype" already exists

[iandees]

Also getting this:

Generating functions from YAML...
/usr/src/app/vectordatasource/meta /usr/src/app/data
Traceback (most recent call last):
  File "sql.py", line 1, in <module>
    from vectordatasource.meta.python import parse_layers, output_kind, \
ImportError: No module named vectordatasource.meta.python

I think part of the problem is that my .dockerignore was not ignoring the compiled stuff I had locally, so it was pulling .pyc and .egg, etc. stuff into the Docker image. When I added this stuff to .dockerignore and built the image again, it required several more changes. I'm still running into similar errors, though, so I'm not sure what's going on.

[iandees]

😭 I think the issue is that I'm building the docker image using the tilezen/vector-datasource tag and then trying to run the mapzen/vector-datasource image...

[iandees]

Part of the problem was that I was confusing the image names as I mentioned above, but I also was running into problems because the image I built pulled in compiled stuff from my Mac that it shouldn't have. I added those things to .dockerignore and got farther. But with the compiled stuff not showing up I had to actually build the vector-datasource stuff in the image, so I added that as well. The image seems to work now as a data loader. I'll continue tomorrow by adding the documentation to tilezen/tileserver#93.

[guyisra]

@iandees probably a bit old, but how did you end up fixing the relation "wof_neighbourhood_placetype" already exists
seems to be caused by the CREATE TABLE in schema.sql