Clarification needed.

In your sample data, if a node is type 1 then its parent is always also of type 1. (Equivalently, if a node is type 0, then its children - if any - are all of type 0.)

Is that also true in your real-life data?

 
drop table t_hierarchy;
create table t_hierarchy
(
	id		number	,
	id_sup	number	,
	id_type	number
);

insert into t_hierarchy values (1, null, 1);

insert into t_hierarchy values (2, 1, 0);
insert into t_hierarchy values (3, 2, 0);
insert into t_hierarchy values (30, 3, 0);
insert into t_hierarchy values (31, 3, 0);
insert into t_hierarchy values (32, 30, 0);

insert into t_hierarchy values (4, 2, 1);
insert into t_hierarchy values (5, 4, 0);

insert into t_hierarchy values (6, 2, 1);
insert into t_hierarchy values (7, 6, 0);

insert into t_hierarchy values (8, 2, 1);
insert into t_hierarchy values (9, 8, 0);

insert into t_hierarchy values (10, 2, 1);
insert into t_hierarchy values (11, 10, 0);

insert into t_hierarchy values (12, 2, 1);
insert into t_hierarchy values (13, 12, 0);

LIB_ID
---------------
-1 (1)
--2 (0)
---3 (0)
----30 (0)
-----32 (0)
----31 (0)
---4 (1)
----5 (0)
---6 (1)
----7 (0)
---8 (1)
----9 (0)
---10 (1)
----11 (0)
---12 (1)
----13 (0)

Or could you have a sub-hierarchy where 101 is parent of 102 who is parent of 103 who is parent of 104, and 101 and 103 are type 1 while 102 and 104 are type 0?

And, if such alternation of type between 1 and 0, as you descend from node to node (or as you ascend from node to node), are possible, WHAT do you actually need to find?

Suppose in my example you start with 104; do you stop immediately when you find that its parent, 103, has type 1? Or do you search further, and discover that 103's parent, which is 102, has type 0, and that must be considered too?

If it's the latter, the problem is a bit more interesting; if it's the former (stop as soon as a parent is type 1, don't look further) then the problem is trivial. Filter out all the nodes of type 1 first, and then apply a standard hierarchical query.