search-engine-for-typesense 30.1.6.16 → 30.1.6.17

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e4f73039b28d08542d582f09279574bf2d982218434259e17e2ef341a98959cf
-  data.tar.gz: 1c4b811b429fd29d13b878a1cc9e292ca268e79f8746c1632ac29810bf82c0c6
+  metadata.gz: 1bd17994c46590322f8d32265bf5f1a5ae91be3b911a120c48730ba6339fb9e3
+  data.tar.gz: e7d9785f65c500ddffaef1898b8620c13142ade4f737d6b93f7f1b88a31ee582
 SHA512:
-  metadata.gz: afeca782edfa298e76b459e6ec7307d2b896a6a77987d70e3525602a2f052e2556df973d204f6be4ec38ef562e0b90ef0ce679c737c715ae278b783c014eab30
-  data.tar.gz: b7eb1fb839c37de2999bc1bdbcc78598aaa3b151961ae04fd8f3964d84207550f562b8d72840d03e2a97d0ca6ffc109ec9d34ea7608f8f6aea88c56c5e866c53
+  metadata.gz: 744c09d6abd7bce828543b0cd11bb354d94cdff2c17a89d690efd25948993901341c0177890e07256e0b1a9e376925f6ed790c1b5ad64f629e68a99dca9b6ad9
+  data.tar.gz: aee5a996ae1f5075533786e58347e0771737ecc8278cd25b380df7b8c21b6ea917e052a9ad34216eab7f6a2b8a80c90e6701241843eb9698572f1adc9fd201c1

data/app/search_engine/search_engine/index_partition_job.rb

@@ -33,6 +33,7 @@ module SearchEngine
     # @param metadata [Hash]
     # @return [void]
     def perform(collection_class_name, partition, into: nil, metadata: {})
+      payload = nil
       klass = constantize_collection!(collection_class_name)
       payload = base_payload(klass, partition: partition, into: into)
       instrument('search_engine.dispatcher.job_started',
@@ -54,7 +55,8 @@ module SearchEngine
       )
       nil
     rescue StandardError => error
-      instrument_error(error, payload: payload.merge(metadata: metadata))
+      safe_payload = payload || error_payload(error)
+      instrument_error(error, payload: safe_payload.merge(metadata: metadata || {}))
       raise
     end
 

data/lib/search_engine/cascade.rb

@@ -18,7 +18,7 @@ module SearchEngine
       # @param context [Symbol] :update or :full
       # @param client [SearchEngine::Client, nil]
       # @return [Hash]
-      # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/BlockNesting
+      # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/BlockNesting, Metrics/PerceivedComplexity, Metrics/BlockLength
       def cascade_reindex!(source:, ids:, context: :update, client: nil)
         raise ArgumentError, 'context must be :update or :full' unless %i[update full].include?(context.to_sym)
 
@@ -28,6 +28,7 @@ module SearchEngine
 
         graph = build_reverse_graph(client: ts_client)
         referencers = Array(graph[src_collection])
+        target_collections = referencers.map { |edge| edge[:referrer] }.uniq
 
         # Detect immediate cycles (A <-> B) and skip those pairs
         cycle_pairs = detect_immediate_cycles(graph)
@@ -43,7 +44,7 @@ module SearchEngine
           alias_cache: alias_cache
         )
 
-        outcomes = []
+        outcomes_by_collection = {}
         partial_count = 0
         full_count = 0
         skipped_unregistered = 0
@@ -57,18 +58,19 @@ module SearchEngine
           # Skip cycle pairs deterministically (avoid ping-pong)
           if cycle_pairs.include?([src_collection, referrer_coll])
             skipped_cycles << { pair: [src_collection, referrer_coll] }
-            outcomes << { collection: referrer_coll, mode: :skipped_cycle }
+            merge_cascade_outcome!(outcomes_by_collection, { collection: referrer_coll, mode: :skipped_cycle })
             next
           end
 
           ref_klass = safe_collection_class(referrer_coll)
           unless ref_klass
             skipped_unregistered += 1
-            outcomes << { collection: referrer_coll, mode: :skipped_unregistered }
+            merge_cascade_outcome!(outcomes_by_collection, { collection: referrer_coll, mode: :skipped_unregistered })
             next
           end
 
           mode = :full
+          partial_failure = nil
           if context.to_sym == :update && can_partial_reindex?(ref_klass)
             begin
               SearchEngine::Indexer.rebuild_partition!(ref_klass, partition: { local_key.to_sym => Array(ids) },
@@ -90,9 +92,8 @@ into: nil
                   mode = :skipped_no_partitions
                 end
               end
-              # Record diagnostic on the outcome for visibility upstream
-              outcomes << { collection: referrer_coll, mode: :partial_failed, error_class: error.class.name,
-                            message: error.message.to_s[0, 200] }
+              # Preserve the partial failure details on the final single outcome row.
+              partial_failure = { error_class: error.class.name, message: error.message.to_s[0, 200] }
             end
           elsif seen_full[referrer_coll]
             mode = :skipped_duplicate
@@ -107,14 +108,17 @@ into: nil
             end
           end
 
-          outcomes << { collection: referrer_coll, mode: mode }
+          outcome = { collection: referrer_coll, mode: mode }
+          outcome.merge!(partial_failure) if partial_failure
+          merge_cascade_outcome!(outcomes_by_collection, outcome)
         end
+        outcomes = outcomes_by_collection.values
 
         payload = {
           source_collection: src_collection,
           ids_count: Array(ids).size,
           context: context.to_sym,
-          targets_total: referencers.size,
+          targets_total: target_collections.size,
           partial_count: partial_count,
           full_count: full_count,
           skipped_unregistered: skipped_unregistered,
@@ -124,7 +128,7 @@ into: nil
 
         payload.merge(outcomes: outcomes)
       end
-      # rubocop:enable Metrics/AbcSize, Metrics/MethodLength, Metrics/BlockNesting
+      # rubocop:enable Metrics/AbcSize, Metrics/MethodLength, Metrics/BlockNesting, Metrics/PerceivedComplexity, Metrics/BlockLength
 
       # Build a reverse graph from Typesense live schemas when possible, falling
       # back to compiled local schemas for registered models.
@@ -469,6 +473,42 @@ foreign_key: fk }
         pairs.uniq
       end
 
+      def merge_cascade_outcome!(outcomes_by_collection, outcome)
+        key = outcome[:collection]
+        existing = outcomes_by_collection[key]
+        if existing.nil?
+          outcomes_by_collection[key] = outcome
+          return
+        end
+
+        winner = if cascade_outcome_priority(outcome[:mode]) > cascade_outcome_priority(existing[:mode])
+                   outcome.dup
+                 else
+                   existing.dup
+                 end
+
+        error_class = existing[:error_class] || outcome[:error_class]
+        message = existing[:message] || outcome[:message]
+        if error_class
+          winner[:error_class] = error_class
+          winner[:message] = message
+        end
+
+        outcomes_by_collection[key] = winner
+      end
+
+      def cascade_outcome_priority(mode)
+        case mode.to_sym
+        when :full then 60
+        when :partial then 50
+        when :skipped_no_partitions then 40
+        when :skipped_duplicate then 30
+        when :skipped_cycle then 20
+        when :skipped_unregistered then 10
+        else 0
+        end
+      end
+
       def safe_collection_class(name)
         SearchEngine::CollectionResolver.model_for_logical(name)
       end

data/lib/search_engine/cli/doctor.rb

@@ -166,7 +166,10 @@ module SearchEngine
           started = monotonic_ms
           client = client_with_overrides
           health = client.health
-          ok = !(health && (health[:ok] == true || health['ok'] == true)).nil?
+          ok_value = if health.is_a?(Hash)
+                       health.key?(:ok) ? health[:ok] : health['ok']
+                     end
+          ok = (ok_value == true)
           details = { response: redacted_value(health) }
           hint = ok ? nil : 'Verify host/port/protocol and network reachability to Typesense.'
 

data/lib/search_engine/dsl/parser.rb

@@ -66,7 +66,9 @@ module SearchEngine
       # @param joins [Array<Symbol>, nil]
       # @return [Array<SearchEngine::AST::Node>]
       def parse_list(list, klass:, joins: nil)
-        items = Array(list).flatten.compact
+        # Preserve input shape so template bind arrays stay intact:
+        # parse_list(['brand_id IN ?', [1, 2]]) should keep [1, 2] as one bind arg.
+        items = Array(list).compact
         return [] if items.empty?
 
         nodes = []
@@ -92,7 +94,7 @@ module SearchEngine
             nodes << parse_raw(entry.to_s)
             i += 1
           when Array
-            nodes << parse_array_entry(entry, klass: klass, joins: joins)
+            nodes.concat(Array(parse_array_entry(entry, klass: klass, joins: joins)))
             i += 1
           else
             raise ArgumentError, "Parser: unsupported where argument #{entry.class}"
@@ -104,7 +106,19 @@ module SearchEngine
       # --- Internals -------------------------------------------------------
 
       def parse_array_entry(entry, klass:, joins: nil)
-        return parse_raw(entry.to_s) unless entry.first.is_a?(String)
+        raise ArgumentError, 'Unsupported argument type: [] (Array)' if entry.empty?
+
+        first = entry.first
+        unless first.is_a?(String)
+          # Support wrapped nested where-lists while preserving template bind arrays:
+          # where([['a > ?', 1], ['b < ?', 2]]) should parse as two predicates.
+          if first.is_a?(Array) || first.is_a?(Hash) || first.is_a?(Symbol)
+            return parse_list(entry, klass: klass, joins: joins)
+          end
+
+          return parse_raw(entry.to_s)
+        end
+
         return parse_list(entry, klass: klass, joins: joins) unless placeholders?(entry.first)
 
         template = entry.first

data/lib/search_engine/indexer/retry_policy.rb

@@ -103,8 +103,12 @@ module SearchEngine
         rng = (Thread.current[:__se_retry_rng__] ||= Random.new)
         min = range.begin.to_f
         max = range.end.to_f
-        # Range may be exclusive; normalize to inclusive space
+        # Honor exclusive-end ranges by shifting the upper bound down one
+        # representable float step. Inclusive ranges preserve the original end.
+        max = max.prev_float if range.exclude_end?
         span = max - min
+        return min if span <= 0.0
+
         min + (rng.rand * span)
       end
     end

data/lib/search_engine/relation/dsl/filters.rb

@@ -17,6 +17,7 @@ module SearchEngine
           # @param args [Array<Object>]
           # @return [SearchEngine::Relation]
           def not(*args)
+            @relation.send(:validate_not_args!, args)
             nodes = Array(@relation.send(:build_ast_with_empty_array_rewrites, args, negated: true))
 
             # Invert non-hidden predicates (Eq, In) returned by the builder
@@ -44,6 +45,7 @@ module SearchEngine
         # @return [SearchEngine::Relation, WhereChain]
         def where(*args)
           return self if args.nil? || args.empty?
+          return self if blank_where_args?(args)
 
           ast_nodes = build_ast_with_empty_array_rewrites(args, negated: false)
           fragments = normalize_where(args)
@@ -58,6 +60,7 @@ module SearchEngine
         # @param args [Array<Object>]
         # @return [SearchEngine::Relation]
         def not(*args)
+          validate_not_args!(args)
           nodes = Array(build_ast_with_empty_array_rewrites(args, negated: true))
 
           negated = nodes.map do |node|
@@ -122,6 +125,30 @@ module SearchEngine
 
         private
 
+        def blank_where_args?(args)
+          return false unless args.length == 1
+
+          blank_where_arg?(args.first)
+        end
+
+        def blank_where_arg?(arg)
+          return true if arg.nil?
+          return true if arg.respond_to?(:empty?) && arg.empty?
+
+          arg.is_a?(String) && arg.strip.empty?
+        end
+
+        def validate_not_args!(args)
+          raise ArgumentError, 'wrong number of arguments (given 0, expected 1+)' if args.nil? || args.empty?
+
+          return unless args.length == 1
+
+          value = args.first
+          return unless value.nil? || (value.is_a?(Array) && value.empty?)
+
+          raise ArgumentError, 'Unsupported argument type:  (NilClass)'
+        end
+
         def merge_relation(other, assoc: nil)
           return self if other.nil?
 
@@ -227,7 +254,9 @@ module SearchEngine
         # - nil predicates to hidden *_blank flags when `optional` is enabled (new behavior)
         # Delegates other inputs to the DSL parser.
         def build_ast_with_empty_array_rewrites(args, negated: false)
-          items = Array(args).flatten.compact
+          # Preserve input shape so template bind arrays stay intact:
+          # where('field IN ?', [1, 2]) => ['field IN ?', [1, 2]]
+          items = Array(args).compact
           return [] if items.empty?
 
           out_nodes = []
@@ -560,7 +589,9 @@ module SearchEngine
 
         # Normalize where arguments into an array of string fragments safe for Typesense.
         def normalize_where(args)
-          list = Array(args).flatten.compact
+          # Preserve input shape so template bind arrays stay intact:
+          # where('field IN ?', [1, 2]) => ['field IN ?', [1, 2]]
+          list = Array(args).compact
           return [] if list.empty?
 
           fragments = []

data/lib/search_engine/schema.rb

@@ -436,8 +436,13 @@ module SearchEngine
         existing = list_physicals_starting_with(prefix, client: client)
         used_sequences = existing.map { |name| name.split('_').last.to_i }
 
-        seq = 1
-        seq += 1 while used_sequences.include?(seq) && seq < 999
+        seq = (1..999).find { |candidate| !used_sequences.include?(candidate) }
+        unless seq
+          raise SearchEngine::Errors::InvalidParams,
+                "Schema physical name sequence space exhausted for prefix '#{prefix}' (001..999 all occupied). " \
+                'Wait one second and retry.'
+        end
+
         format('%<prefix>s%<seq>03d', prefix: prefix, seq: seq)
       end
 

data/lib/search_engine/sources/lambda_source.rb

@@ -4,8 +4,13 @@ module SearchEngine
   module Sources
     # Adapter that delegates batch enumeration to a provided callable.
     #
-    # The callable is expected to implement `call(cursor:, partition:)` and return either
-    # an Enumerator or yield arrays of rows. Shapes are application-defined.
+    # The callable is expected to implement `call(cursor:, partition:)` and use one mode:
+    # - return mode: return an Enumerable of batches
+    # - yield mode: yield batches via the provided block argument
+    #
+    # Returning batch-like data (Array/Enumerator) while also yielding is treated as
+    # an ambiguous mixed mode and raises an error.
+    # Shapes are application-defined.
     #
     # @example
     #   src = SearchEngine::Sources::LambdaSource.new(->(cursor:, partition:) { [[row1, row2]] })
@@ -32,16 +37,32 @@ module SearchEngine
         return enum_for(:each_batch, partition: partition, cursor: cursor) unless block_given?
 
         started = monotonic_ms
+        consume_rows = lambda do |rows|
+          duration = monotonic_ms - started
+          instrument_batch_fetched(source: 'lambda', batch_index: nil, rows_count: Array(rows).size,
+                                   duration_ms: duration, partition: partition, cursor: cursor,
+                                   adapter_options: { callable: @callable.class.name }
+          )
+          yield(rows)
+          started = monotonic_ms
+          nil
+        end
+
         begin
-          enum = @callable.call(cursor: cursor, partition: partition)
-          Array(enum).each do |rows|
-            duration = monotonic_ms - started
-            instrument_batch_fetched(source: 'lambda', batch_index: nil, rows_count: Array(rows).size,
-                                     duration_ms: duration, partition: partition, cursor: cursor,
-                                     adapter_options: { callable: @callable.class.name }
-            )
-            yield(rows)
-            started = monotonic_ms
+          yielded = false
+          returned = @callable.call(cursor: cursor, partition: partition) do |rows|
+            yielded = true
+            consume_rows.call(rows)
+          end
+
+          if yielded
+            if mixed_mode_batch_return?(returned)
+              raise SearchEngine::Errors::InvalidParams,
+                    'lambda source callable must either yield batches or return an Enumerable of batches, not both'
+            end
+          else
+            to_iterate = returned.respond_to?(:each) ? returned : Array(returned)
+            to_iterate.each { |rows| consume_rows.call(rows) }
           end
         rescue StandardError => error
           instrument_error(source: 'lambda', error: error, partition: partition, cursor: cursor,
@@ -50,6 +71,12 @@ module SearchEngine
           raise
         end
       end
+
+      private
+
+      def mixed_mode_batch_return?(returned)
+        returned.is_a?(Array) || returned.is_a?(Enumerator)
+      end
     end
   end
 end

data/lib/search_engine/sources.rb

@@ -12,6 +12,11 @@ module SearchEngine
   #     Enumerator.new { |y| external_api.each_page(cursor) { |rows| y << rows } }
   #   end
   #
+  # Lambda source contract:
+  # - return mode: return an Enumerable of batches
+  # - yield mode: call the provided block with each batch
+  # Returning batch-like values while also yielding is treated as mixed mode and raises.
+  #
   # All adapters implement `each_batch(partition:, cursor:)` and return an Enumerator
   # when no block is provided.
   module Sources
@@ -19,7 +24,7 @@ module SearchEngine
     #
     # @param type [Symbol] :active_record, :sql, or :lambda
     # @param options [Hash] adapter-specific options
-    # @yield for :lambda sources, a block taking (cursor:, partition:) and returning an Enumerator
+    # @yield for :lambda sources, block taking (cursor:, partition:) and either yielding or returning batches
     # @return [Object] adapter responding to `each_batch(partition:, cursor:)`
     def self.build(type, **options, &block)
       case type.to_sym

data/lib/search_engine/version.rb

@@ -3,5 +3,5 @@
 module SearchEngine
   # Current gem version.
   # @return [String]
-  VERSION = '30.1.6.16'
+  VERSION = '30.1.6.17'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: search-engine-for-typesense
 version: !ruby/object:Gem::Version
-  version: 30.1.6.16
+  version: 30.1.6.17
 platform: ruby
 authors:
 - Nikita Shkoda