"navtree" handling buggy for "level > 0"

[d-maurer]

Consider the following setup:

>>> from Products.ExtendedPathIndex.ExtendedPathIndex import ExtendedPathIndex
>>> 
>>> class _O(object):
...   def __init__(self, pi): self.pi = pi
... 
>>> pi = ExtendedPathIndex("pi")
>>> 
>>> paths = {}
>>> for leaf in ("1/1 1/12/121 1/12/122 2/21/1".split()):
...   comps = leaf.split("/")
...   pd = paths
...   for c in comps:
...     if c not in pd: pd[c] = {}
...     pd = pd[c]
... 
>>> def gen(prefix, pd):
...   for c in sorted(pd):
...     np = prefix + "/" + c
...     yield np
...     yield from gen(np, pd[c])
... 
>>> 
>>> for i, p in enumerate(gen("", paths)):
...   print (i, p)
...   indexed = pi.index_object(i, _O(p))
... 
0 /1
1 /1/1
2 /1/12
3 /1/12/121
4 /1/12/122
5 /2
6 /2/21
7 /2/21/1

This means that /1/1 is the only path with 1 at level 1. As a consequence, search("/1/1", default_level=0, ...) and search("1", default_level=1, ...) must give the same results but:

>>> pi.search("/1/1", default_level=0, depth=0, navtree=True)
IISet([0, 1])
>>> pi.search("1", default_level=1, depth=0, navtree=True)
IISet([1])
>>> 
>>> pi.search("/1/1", default_level=0, depth=1, navtree=True)
IISet([0, 1, 2, 5])
>>> pi.search("1", default_level=1, depth=1, navtree=True)
IISet([0, 1, 2, 5, 6])

The results for default_level=1 are obviously wrong.