Unexpected behaviour of generate_subcatalogs #240
Description
Hello, first of all thanks so much for this fantastic library - it's really great! I observed an unexpected behaviour of the Catalog generate_subcatalogs method - I thought it might be worth reporting.
I am setting up a local catalog with some items organised in a sub-catalog structure. I would like to add items at different stages (e.g. whenever new assets become available). I would have expected that calling the generate_subcatalogs method every time I add new items to the root catalog would do the job. However, every time the method is called, the full sub-catalog structure is replicated. Here's a minimal working example:
import pystac
from datetime import datetime
# initialise my catalog
catalog = pystac.Catalog(id='my-catalog', description='My Awesome Catalog')
properties = dict(first=1, second=2, third=3)
# add first set of items
for id in ['A', 'B', 'C']:
item = pystac.Item(id, None, None, datetime.utcnow(), properties)
catalog.add_item(item)
# generate sub-catalog structure
catalog.generate_subcatalogs(template='${first}/${second}/${third}')
catalog.describe()
# * <Catalog id=my-catalog>
# * <Catalog id=1>
# * <Catalog id=2>
# * <Catalog id=3>
# * <Item id=A>
# * <Item id=B>
# * <Item id=C>
# add more items
for id in ['D', 'E', 'F']:
item = pystac.Item(id, None, None, datetime.utcnow(), properties)
catalog.add_item(item)
catalog.generate_subcatalogs(template='${first}/${second}/${third}')
catalog.describe()
# * <Catalog id=my-catalog>
# * <Catalog id=1>
# * <Catalog id=2>
# * <Catalog id=3>
# * <Catalog id=1>
# * <Catalog id=2>
# * <Catalog id=3>
# * <Item id=A>
# * <Item id=B>
# * <Item id=C>
# * <Catalog id=1>
# * <Catalog id=2>
# * <Catalog id=3>
# * <Item id=D>
# * <Item id=E>
# * <Item id=F>while I would like to obtain the following:
# * <Catalog id=my-catalog>
# * <Catalog id=1>
# * <Catalog id=2>
# * <Catalog id=3>
# * <Item id=A>
# * <Item id=B>
# * <Item id=C>
# * <Item id=D>
# * <Item id=E>
# * <Item id=F>If you agree this is unexpected, one could initialise the following sub-catalog dictionary:
Line 545 in 2a0c850
with all the sub-catalogs up to the current level:
subcat_id_to_cat = {}
curr_parent = self
while curr_parent is not None:
subcat_id_to_cat[curr_parent.id] = curr_parent
curr_parent = curr_parent.get_parent()In this way, sub-catalog levels that are already present are not replicated..
Thanks in advance!