No, but this method on Connection provides DB-specific information that might include (some of) what you want: https://docs.oracle.com/en/java/javase/17/docs/api/java.sql/java/sql/Connection.html#getClientInfo()

For several databases, that returns the JDBC URL which you can pick apart to get dbtype/dbname etc.

That method gives me {"ApplicationName" "PostgreSQL JDBC Driver"}, but there is also getURL which returns the URL that can be parsed as you describe. I'm looking for the port, which is in there. I have my test fixtures set up similar to how next.jdbc sets it up, so I can get the connection set up in my tests. My code, however, just expects the db-spec (passed around in the config) and creates the connection on its own. Does parsing the URL seem like the logical approach here?

There's no getURL() method on Connection.

The test fixtures in next.jdbc have the db-spec and the DataSource available (not Connection) so I think you're a bit confused?

I'm taking the ds that the fixture makes available and creating a Connection in my tests.

(type (fix/ds))                                   
; => org.postgresql.ds.PGSimpleDataSource         
(type (jdbc/get-connection (fix/ds)))             
; => org.postgresql.jdbc.PgConnection             
(.getURL (jdbc/get-connection (fix/ds)))          
; => "jdbc:" 

;; this is just a dummy db-spec -- it's handled in with-test-db below

One other way:

(jdbc/execute-one! (fix/ds) ["select inet_server_port()"]) 
; => {:inet_server_port 33745}                             

If you load next.jdbc.datafy and then datafy the Connection, you get lots of information (`clojure.core.protocols/datafy`):

:DBVersionNumber - 14.5
:URL - jdbc:
:adaptiveFetch - false
:autoCommit - true
:autosave - #object[org.postgresql.jdbc.AutoSave 0x27002c5a NEVER]
:backendPID - 9973
:catalog - postgres
:class - org.postgresql.jdbc.PgConnection
:clientInfo - #object[java.util.Properties 0x19a920d1 {ApplicationName=PostgreSQL JDBC Driver}]
:closed - false
:columnSanitiserDisabled - false
:copyAPI - #object[org.postgresql.copy.CopyManager 0x6c861b25 org.postgresql.copy.CopyManager@6c861b25]
:cursorName - nil
...

So I think you were getting .getURL from the PgConnection type directly, since Connection does not have it. Overall, I would advise you to write code assuming a DataSource -- since in production you'd want to use a connection pooled datasource such as HikariCP produces, and that's what next.jdbc does for the tests now that I look at it:

(reset! test-datasource
                (.getPostgresDatabase ^EmbeddedPostgres @embedded-pg))

and for all other DBs:

(reset! test-datasource (jdbc/get-datasource db)))

all the next.jdbc functions accept a DataSource (and that's also going to be much more efficient than trying to work with db-spec hash maps!).

That clears up many of my questions. I just don't understand how that snippet shows that the next.jdbc tests are using a pooled datasource. Could you please explain that bit?

No, they're not using a pooled datasource -- just a regular datasource. But you can easily swap a pooled datasource for that. What I'm saying is: write your code to expect a datasource, not a db-spec hash map.

Ah, okay. I understand now. Thanks very much, @U04V70XH6.

Mostly, it's transparent: calling get-connection on a db-spec calls get-datasource under the hood and then calls get-connection on that.