Merge pull request #295 from stac-utils/mah/faster_generate_subcats

faster generate_subcatalogs

Author: matthewhanson

pystac/catalog.py

@@ -557,15 +557,19 @@ def generate_subcatalogs(self, template, defaults=None, parent_ids=None, **kwarg
 
         layout_template = LayoutTemplate(template, defaults=defaults)
 
-        items = list(self.get_items())
-        for item in items:
+        keep_item_links = []
+        item_links = [lk for lk in self.links if lk.rel == 'item']
+        for link in item_links:
+            link.resolve_stac_object(root=self.get_root())
+            item = link.target
             item_parts = layout_template.get_template_values(item)
             id_iter = reversed(parent_ids)
             if all(['{}'.format(id) == next(id_iter, None)
                     for id in reversed(item_parts.values())]):
                 # Skip items for which the sub-catalog structure already
                 # matches the template. The list of parent IDs can include more
                 # elements on the root side, so compare the reversed sequences.
+                keep_item_links.append(link)
                 continue
             curr_parent = self
             for k, v in item_parts.items():
@@ -578,13 +582,17 @@ def generate_subcatalogs(self, template, defaults=None, parent_ids=None, **kwarg
                     curr_parent.add_child(subcat)
                     result.append(subcat)
                 curr_parent = subcat
-            self.remove_item(item.id)
+
             # resolve collection link so when added back points to correct location
             link = item.get_single_link('collection')
             if link is not None:
                 link.resolve_stac_object()
+
             curr_parent.add_item(item)
 
+        # keep only non-item links and item links that have not been moved elsewhere
+        self.links = [lk for lk in self.links if lk.rel != 'item'] + keep_item_links
+
         return result
 
     def save(self, catalog_type=None):