#434

backend/app/models/history_access.rb

@@ -38,15 +38,38 @@ class HistoryAccess < ActiveSupport::CurrentAttributes
     table = model.table_name
     keys = KEYS.fetch(table)
     matches = keys.map { "boundary.#{ _1 } = #{ row }.#{ _1 }" }.join(' AND ')
-    clauses = ["NOT EXISTS (SELECT 1 FROM #{ table } boundary WHERE #{ matches } " +
-        "AND boundary.event_type = 'discard' AND boundary.version_no >= #{ row }.version_no)"]
+    boundaries = "SELECT #{ keys.join(', ') }, MAX(version_no) AS version_no FROM #{ table } " +
+        "WHERE event_type = 'discard' GROUP BY #{ keys.join(', ') }"
+    clauses = ["NOT EXISTS (SELECT 1 FROM (#{ boundaries }) boundary WHERE #{ matches } " +
+        "AND boundary.version_no >= #{ row }.version_no)"]
     clauses += reference_sql(table, row) if references
     clauses.map { "(#{ _1 })" }.join(' AND ')
   end
 
   def self.discard_reference(table, key, expression)
-    "NOT EXISTS (SELECT 1 FROM #{ table } blocked WHERE blocked.event_type = 'discard' " +
-        "AND #{ expression.call("blocked.#{ key }") })"
+    "NOT EXISTS (SELECT 1 FROM (#{ discarded_keys(table, key) }) blocked " +
+        "WHERE #{ expression.call("blocked.#{ key }") })"
+  end
+
+  def self.discarded_keys(table, key)
+    "SELECT #{ key } FROM #{ table } WHERE event_type = 'discard' GROUP BY #{ key }"
+  end
+
+  def self.json_reference(table, key, document)
+    "NOT EXISTS (SELECT 1 FROM JSON_TABLE(COALESCE(#{ document }, JSON_ARRAY()), " +
+        "'$[*]' COLUMNS (reference_id BIGINT PATH '$' NULL ON EMPTY)) reference_entry " +
+        "JOIN (#{ discarded_keys(table, key) }) blocked " +
+        "ON reference_entry.reference_id = blocked.#{ key })"
+  end
+
+  def self.list_reference(table, key, document)
+    # Preserve FIND_IN_SET's exact token semantics, including empty tokens.
+    array = "CONCAT('[', REPLACE(JSON_QUOTE(REPLACE(COALESCE(#{ document }, ''), " +
+        "' ', ',')), ',', '\",\"'), ']')"
+    "NOT EXISTS (SELECT 1 FROM JSON_TABLE(#{ array }, '$[*]' " +
+        "COLUMNS (reference_id VARCHAR(255) PATH '$')) reference_entry " +
+        "JOIN (#{ discarded_keys(table, key) }) blocked " +
+        "ON reference_entry.reference_id = CAST(blocked.#{ key } AS CHAR))"
   end
 
   def self.reference_sql(table, row)
@@ -54,16 +77,10 @@ class HistoryAccess < ActiveSupport::CurrentAttributes
     when 'post_versions'
       [['tag_versions', 'tag_id', 'tag_id'],
        ['nico_tag_versions', 'tag_id', 'external_tag_id']].map do |target, key, field|
-        discard_reference(target, key, ->(id) {
-          "JSON_CONTAINS(#{ row }.tags_json, JSON_OBJECT('#{ field }', #{ id }))"
-        })
-      end + [discard_reference('post_versions', 'post_id', ->(id) {
-        "FIND_IN_SET(#{ id }, REPLACE(#{ row }.parent_post_ids, ' ', ',')) > 0"
-      })]
+        json_reference(target, key, "JSON_EXTRACT(#{ row }.tags_json, '$[*].#{ field }')")
+      end + [list_reference('post_versions', 'post_id', "#{ row }.parent_post_ids")]
     when 'tag_versions'
-      [discard_reference('tag_versions', 'tag_id', ->(id) {
-        "FIND_IN_SET(#{ id }, REPLACE(#{ row }.parent_tag_ids, ' ', ',')) > 0"
-      })]
+      [list_reference('tag_versions', 'tag_id', "#{ row }.parent_tag_ids")]
     when 'tag_description_versions'
       [discard_reference('tag_versions', 'tag_id', ->(id) { "#{ row }.tag_id = #{ id }" })]
     when 'material_versions'
@@ -149,7 +166,7 @@ class HistoryAccess < ActiveSupport::CurrentAttributes
   # This number deliberately ignores time, identity and content filters to prevent count probes.
   def self.population_count(model) = model.unscoped.count
 
-  # Resource-wide totals provide a fast path; Counts computes filtered totals separately.
+  # Explicit whole-resource statistics; Counts pages do not run this aggregate.
   def self.population_counts(model)
     total, excluded = model.unscoped.optimizer_hints('MAX_EXECUTION_TIME(10000)').pick(
         Arel.sql('COUNT(*)'),