Use quoted identifiers in generated SQL

[robert-werner]

I've tried to deploy a T-Rex server from OSM data.
I've generated a config with genconfig command with PostgreSQL.
I started to get PostgreSQL's syntax errors, so I've investigated through the code, and I discovered that replacing the braces with nothing is creating these errors, so I've decided to delete that code.

Tests are passed.

[pka]

Thanks for your PR!
I really like the feature of having readable SQL in the generated configuration. A better solution would probably be checking that no problematic characters are included before removing the quotes.

[robert-werner]

'problematic characters' are the reserved words of Postgre, so the 'better solution' is not the solution at all,

Just try to generate .toml from OSM2PGSQL imported database.

[pka]

Which problematic table or column names does osm2pgsql create?

I'm still in favor of letting basic names unquoted. Rule proposal (subset of https://www.postgresql.org/docs/current/sql-syntax-lexical.html#SQL-SYNTAX-IDENTIFIERS): "SQL identifiers must begin with a letter a-z or an underscore (_). Subsequent characters in an identifier can be a-z, underscores or digits (0-9)"

[joto]

Which problematic table or column names does osm2pgsql create?

Of course it depends on the configuration, but typical osm2pgsql configs generate column names such as addr:housenumber from the tags of the same name. Not a great idea, but we are pretty much stuck with that.

[robert-werner]

Which problematic table or column names does osm2pgsql create?

Of course it depends on the configuration, but typical osm2pgsql configs generate column names such as addr:housenumber from the tags of the same name. Not a great idea, but we are pretty much stuck with that.

That's why T-Rex doesn't need the readability instead of functionality.
If anyone wants to get readable SQL's, he may be pleased with dedicated option.