Sorting for reactor Flux - collections

Question

I am not familiar with java Collection sorting and reactor Flux.
Now, I got a requirement is that user need to prioritize the publishing for some of items from a Flux.
e.g.
p1 = Person{name="p1", country="US", gender="male", age=20}
p2 = Person{name="p2", country="US", gender="female", age=20}
p3 = Person{name="p3", country="Russia", gender="male", age=40}
p4 = Person{name="p4", country="China", gender="female" age=30}
p5 = Person{name="p5", country="Japan", gender="female" age=25}
p6 = Person{name="p6", country="Japan", gender="male" age=40}
A Flux is built like below:
Flux.fromIterable(Stream.of(p1, p2, p3, p4, p5, p6))
Now, we need to prioritize the females from Japan and China, the rest of person will keep orginal order.
e.g.
p5 = Person{name="p5", country="Japan", gender="female" age=25}
p6 = Person{name="p6", country="Japan", gender="male" age=40}
p4 = Person{name="p4", country="China", gender="female" age=30}
p1 = Person{name="p1", country="US", gender="male", age=20}
p2 = Person{name="p2", country="US", gender="female", age=20}
p3 = Person{name="p3", country="Russia", gender="male", age=40}
How to sort the Flux?

Answer 1

You can use sort method.
Flux<Person> flux = Flux.fromIterable(List.of(p1, p2, p3, p4, p5, p6));
Flux<Person> result = flux.sort(Comparator
.comparing((Person p) -> !p.getCountry().equals("Japan"))
.thenComparing((Person p) -> !p.getCountry().equals("China"))
);
result.subscribe(val -> System.out.println(val.toString()));
Output:
Person(name=p5, country=Japan, gender=female, age=25)
Person(name=p6, country=Japan, gender=male, age=40)
Person(name=p4, country=China, gender=female, age=30)
Person(name=p1, country=US, gender=male, age=20)
Person(name=p2, country=US, gender=female, age=20)
Person(name=p3, country=Russia, gender=male, age=40)

Answer 2

I assume you have a potentially infinite Flux, so you can't just collect everything, sort it and emit again. This would throw out Flux#sort, which would just do that.
Obviously you can't sort a sequence you do not (fully) have. What you can do is
Use Flux#groupBy and generate GroupedFluxes for every priority.
You get one Flux, which will emit a Flux for every group (prio). You can then work on each prioFlux as you like.
For example you could merge them again using Flux#mergeComparing, prioritizing one Flux over the other (if both have values, take the first - otherwise take what is there).
Use Flux#buffer and work on smaller buffers. Sort those as usual and emit again. This could be a reasonable and simple solution.
Regards,
Martin