lex-apollo 0.4.24 → 0.4.25

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2dceee29e5b2af12bab5833beac8b837d2366eaf9509c7e4ce04e88b44a5c28f
-  data.tar.gz: 97d62a6385f375b19a6f48b1881fef8525e5949c4348bfe0cd07581d9ff710f2
+  metadata.gz: 96a0b3d72ad8895d264315b2da3710c323d31280cf08fe615f8db74c9c762e82
+  data.tar.gz: a6d938a7409cfd0f1b5f3a0dc6c7f1f3c0399dc93a8b7ba8ffcfff0c82536bc3
 SHA512:
-  metadata.gz: 90065729231f3322eeaaeaa0e44e03b9b79350157674bec30f174fa6447d74bc53b4794e3385eee9d617c13c62cb22198de39fc21ebece7d2d512e7b1dd09ba6
-  data.tar.gz: f1c1b607fd9d9d244c0937b654ea95ca361980f5732c6e513719a6163004595ce0aaa9baab4181db56c8e6e8d1c511c1d6695c9110905e2094a252ba06e5a2c6
+  metadata.gz: efd4519deb1935da01026f3149460fafd110be233f7cdaf522c1980c4cab70a2debcb4ca4ba8a7e3b5b8be0b409d94fdf566fdba6cc8bcdcc1a3ebef879ef6fe
+  data.tar.gz: b0b446a0af1010dde65247cff125c17d086ebba9cf9bd0269575fc7d94fed6eab06a2d836ee6419d742b4fe150aa7b055deefb11e6160b30bde83f79b5c63d71

data/CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## [0.4.25] - 2026-05-08
+
+### Fixed
+- Entity watchdog now remembers processed task-log text during the process lifetime so unchanged logs do not trigger repeated structured LLM extraction on every watchdog interval.
+
 ## [0.4.24] - 2026-05-07
 
 ### Fixed

data/lib/legion/extensions/apollo/actors/entity_watchdog.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+require 'digest'
 require 'legion/extensions/actors/every'
 require_relative '../runners/knowledge'
 require_relative '../runners/entity_extractor'
@@ -40,6 +41,8 @@ module Legion
 
             ingested = 0
             texts.each do |text|
+              next if task_log_text_processed?(text)
+
               result = extract_entities(
                 text:           text,
                 entity_types:   entity_types,
@@ -47,6 +50,7 @@ module Legion
               )
               next unless result[:success]
 
+              mark_task_log_text_processed(text) unless result[:source] == :unavailable
               result[:entities].each do |entity|
                 next if entity_exists_in_apollo?(entity)
 
@@ -124,6 +128,28 @@ module Legion
           def dedup_similarity_threshold
             settings[:entity_watchdog][:dedup_threshold].to_f
           end
+
+          def task_log_text_processed?(text)
+            processed_task_log_hashes.key?(task_log_text_hash(text))
+          end
+
+          def mark_task_log_text_processed(text)
+            hashes = processed_task_log_hashes
+            hashes[task_log_text_hash(text)] = true
+            hashes.shift while hashes.size > processed_task_log_hash_limit
+          end
+
+          def processed_task_log_hashes
+            @processed_task_log_hashes ||= {}
+          end
+
+          def processed_task_log_hash_limit
+            [settings[:entity_watchdog][:log_limit].to_i * 4, 100].max
+          end
+
+          def task_log_text_hash(text)
+            Digest::SHA256.hexdigest(text.to_s)
+          end
         end
       end
     end

data/lib/legion/extensions/apollo/version.rb

@@ -3,7 +3,7 @@
 module Legion
   module Extensions
     module Apollo
-      VERSION = '0.4.24'
+      VERSION = '0.4.25'
     end
   end
 end

data/spec/legion/extensions/apollo/actors/entity_watchdog_spec.rb

@@ -67,6 +67,13 @@ RSpec.describe Legion::Extensions::Apollo::Actor::EntityWatchdog do
       expect(actor).to have_received(:publish_entity_ingest).once
     end
 
+    it 'does not extract entities from the same task log text twice' do
+      actor.scan_and_ingest
+      actor.scan_and_ingest
+
+      expect(actor).to have_received(:extract_entities).once
+    end
+
     context 'when entity already exists in Apollo (high similarity)' do
       let(:existing_match) do
         { success: true, entries: [{ id: 42, content: 'lex-synapse', distance: 0.02 }], count: 1 }

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: lex-apollo
 version: !ruby/object:Gem::Version
-  version: 0.4.24
+  version: 0.4.25
 platform: ruby
 authors:
 - Esity