fix: limit number of search indices added to Bugs (#4285)

[openSUSE-SU-2024:12948-1](https://osv.dev/vulnerability/openSUSE-SU-2024:12948-1)
has so many upstream vulns that it made >6500 search_indices, which
(combined with the other fields) was too many indexed properties for
datastore to handle.
Put a limit on the number of search indices populated from
alias/upstreams to avoid this problem.

Author: michaelkedar

osv/models.py

@@ -420,13 +420,15 @@ def _pre_put_hook(self):  # pylint: disable=arguments-differ
     for ecosystem in self.ecosystem:
       search_indices.update(_tokenize(ecosystem))
 
+    extra_search_indices = set()
+
     for alias in self.aliases:
-      search_indices.update(_tokenize(alias))
+      extra_search_indices.update(_tokenize(alias))
 
     # Please note this will not include exhaustive transitive upstream
     # so may not appear for all cases.
     for upstream in self.upstream_raw:
-      search_indices.update(_tokenize(upstream))
+      extra_search_indices.update(_tokenize(upstream))
 
     for affected_package in self.affected_packages:
       for affected_range in affected_package.ranges:
@@ -437,7 +439,12 @@ def _pre_put_hook(self):  # pylint: disable=arguments-differ
           repo_url_indices.append(url_no_https)  # add url without https://
           search_indices.update(repo_url_indices)
 
-    self.search_indices = list(set(search_indices))
+    # Excessive amounts of search indices cause the record to have too many
+    # indexed properties to store in the database. Arbitrarily limit the number
+    # of search indices from aliases/upstream to curb this.
+    # TODO(michaelkedar): handle this more gracefully.
+    search_indices.update(list(extra_search_indices)[:5000])
+    self.search_indices = list(search_indices)
     self.search_indices.sort()
 
     search_tags = set()