1. The URL table function should support file, s3, and other schemas available for similar engines.

Examples:

file://data.csv - file path relative to the user_files directory (which is the current directory for clickhouse-local).
file:///home/alexey/data.csv - absolute path.
s3://clickhouse-public-datasets/hits_compatible/hits.csv - an URL with s3 schema; currently works for environment credentials.

Implementation: consider the URL engine and table function as a wrapper on top of other engines.

2. The URL table function should support relative URLs.

Examples:

data.csv - a path-relative URL.
/test/data.csv - a host-relative URL.
//example.com/test/data.csv - a schema-relative URL.

All of them should be controlled by the setting url_base.
For example, if url_base is https://abc.xyz/def/, the path-relative URL will resolve to https://abc.xyz/def/data.csv, the host-relative URL will resolve to https://abc.xyz/test/data.csv, and the schema-relative URL will resolve to https://example.com/test/data.csv.

The base URL can have a foreign schema like file:// or s3://.

3. S3-like table functions should support relative URLs.

Similarly to the previous item, it can be controlled by the setting s3_base.

For example, if s3_base is s3://clickhouse-public-datasets/, you can write SELECT * FROM s3('hits_compatible/hits.csv').

4. Support URL as a database engine (to use inside the Overlay engine).

If it has file://path/to/current/dir/ as its base URL, it will allow handling file, url, and s3 uniformly.

The default database engine in clickhouse-local should be switched to the Overlay with URL (currently it is Overlay with File).