def index(self, persist=True):
"""(re)builds the interval tree index into the interval list you
have stored.
.. note: interval tree performance will not be obtained unless
the user *explicitly* calls this method. This is typically
called after a set of intervals has been added.
:param persist: permanently store the index on disk
:type persist: bool
"""
refcountdict = None
tbl = self.hdf_group._f_getChild(INTERVAL_TABLE)
tree_group = self.hdf_group._f_getChild(INTERVAL_INDEX_GROUP)
for rname, idx in self.indexes.iteritems():
# see if index needs to be generated
if (idx.tree is None) or idx.dirty:
# count number of items for each reference
if refcountdict is None:
refcountdict = self._get_ref_count_dict()
# iterate and insert all intervals
tree = IntervalTree(refcountdict[rname])
cur_id = 1
for row in tbl.where('ref == rname'):
start = row[START_COL_NAME]
end = row[END_COL_NAME]
tree.insert(cur_id, start, end, row.nrow)
cur_id += 1
# update index
idx.tree = tree
idx.dirty = False
# save for future use
if persist:
tree.tohdf(tree_group, rname)
def _init_index(self):
if not INTERVAL_INDEX_GROUP in self.hdf_group:
h5file = self._get_hdf_file()
h5file.createGroup(self.hdf_group, INTERVAL_INDEX_GROUP)
parentgroup = self.hdf_group._f_getChild(INTERVAL_INDEX_GROUP)
self.indexes = {}
for rname in self.get_rnames():
idx = self._Index()
# load index if it exists
if rname in parentgroup:
tree = IntervalTree.fromhdf(parentgroup._f_getChild(rname))
idx.tree = tree
self.indexes[rname] = idx
