If there will only ever be one row in CSV1, you can read it in and output it to a cache sink.
You can include cached values in your CSV2 stream with a Derived Column transformation that uses a cached lookup expression, e.g. YourCacheSink#outputs()[1].f_ID