factbase 0.18.0 → 0.19.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 90b1ab3d82ed371b4de11b7eceb98137b2a7687bc2625e61d4d54a6538849b34
-  data.tar.gz: f9fd82d8254d7cfa646a742db6989790791ce157ce53d9e5a310f8f8f8c0db11
+  metadata.gz: c20d4add4664149fc5c070ec64321a79c8ac07ef81c69ae2b64bac1e9fd0c4e9
+  data.tar.gz: ff9a8f121d9cfafa6ad97a634c79fc2ec834a07ee0d17042eb3fc5fa8c529501
 SHA512:
-  metadata.gz: ae0d064e075f304c9fae6f4903c961293d07e186ee6bdf4e180d5f407f5b166f2b3cc345cec612d8184675394a2ee504197d4c13d9ce0f480cffd850d7d10169
-  data.tar.gz: b53da29f44f9ad5520bed3602c9c7dadf882115856ddb344dfba95a32017cd2e89c4582a7a160baa2ed611ff37acd0ba39e8fb9a8aa8ed7bec62f0acf9f0ee70
+  metadata.gz: 3f93283308d6f15766fca09f253da85145efc28975e51259ee88f0954cc6a7f584700de2ff604478bd5a48f1c86153243949a6280732a466412a739f38e9a1e2
+  data.tar.gz: e21602f7612821dd2389e831056389900f62a3ef3b1bf998f6ebbe722cfc9684357383c754ad0556a346534cf7a6bc76489706e0f0f58ea23336a918be71b442

data/Gemfile.lock

@@ -126,6 +126,7 @@ PLATFORMS
   arm64-darwin-22
   arm64-darwin-23
   arm64-darwin-24
+  arm64-darwin-25
   x64-mingw-ucrt
   x86_64-darwin-20
   x86_64-darwin-21

data/README.md

@@ -210,36 +210,54 @@ This is the result of the benchmark:
 
 <!-- benchmark_begin -->
 ```text
-                                                                   user
-insert 20000 facts                                             0.637574
-export 20000 facts                                             0.019855
-import 410750 bytes (20000 facts)                              0.035065
-insert 10 facts                                                0.044615
-query 10 times w/txn                                           2.455739
-query 10 times w/o txn                                         0.050423
-modify 10 attrs w/txn                                          1.894240
-delete 10 facts w/txn                                          1.043079
-(and (eq what 'issue-was-closed') (exists... -> 200            1.279827
-(and (eq what 'issue-was-closed') (exists... -> 200/txn        1.263302
-(and (eq what 'issue-was-closed') (exists... -> zero           1.253773
-(and (eq what 'issue-was-closed') (exists... -> zero/txn       1.292489
-(gt time '2024-03-23T03:21:43Z')                               0.398074
-(gt cost 50)                                                   0.254020
-(eq title 'Object Thinking 5000')                              0.037238
-(and (eq foo 42.998) (or (gt bar 200) (absent z...             0.047599
-(and (exists foo) (not (exists blue)))                         1.115156
-(eq id (agg (always) (max id)))                                0.711832
-(join "c<=cost,b<=bar" (eq id (agg (always) (ma...             1.393622
-(and (eq what "foo") (join "w<=what" (and (eq i...             7.428505
-delete!                                                        0.272962
-Taped.append() x50000                                          0.020619
-Taped.each() x125                                              1.721876
-Taped.delete_if() x375                                         0.844673
+                                                                       
+query all facts from an empty factbase                             0.00
+insert 20000 facts                                                 0.67
+export 20000 facts                                                 0.02
+import 411032 bytes (20000 facts)                                  0.03
+insert 10 facts                                                    0.04
+query 10 times w/txn                                               2.52
+query 10 times w/o txn                                             0.05
+modify 10 attrs w/txn                                              1.84
+delete 10 facts w/txn                                              2.98
+build index on 5000 facts                                          0.05
+export 5000 facts with index                                       0.03
+import 5000 facts with persisted index                             0.05
+query 5000 facts using persisted index                             0.11
+export 5000 facts without index                                    0.00
+import 5000 facts without index                                    0.04
+query 5000 facts building index on-the-fly                         0.10
+(and (eq what 'issue-was-closed') (exists... -> 200                1.21
+(and (eq what 'issue-was-closed') (exists... -> 200/txn            1.20
+(and (eq what 'issue-was-closed') (exists... -> zero               1.21
+(and (eq what 'issue-was-closed') (exists... -> zero/txn           1.23
+transaction rollback on factbase with 100000 facts                 0.25
+(gt time '2024-03-23T03:21:43Z')                                   0.33
+(gt cost 50)                                                       0.21
+(eq title 'Object Thinking 5000')                                  0.05
+(and (eq foo 42.998) (or (gt bar 200) (absent z...                 0.05
+(and (exists foo) (not (exists blue)))                             1.68
+(eq id (agg (always) (max id)))                                    2.76
+(join "c<=cost,b<=bar" (eq id (agg (always) (ma...                 4.02
+(and (eq what "foo") (join "w<=what" (and (eq i...                 7.68
+delete!                                                            0.56
+(and (eq issue *) (eq repository *) (eq what '*') (eq where '*'))  2.75
+Taped.append() x50000                                              0.16
+Taped.each() x125                                                  1.63
+Taped.delete_if() x375                                             0.84
+50000 facts: read-only txn (no copy needed)                        7.60
+50000 facts: rollback txn (no copy needed)                         7.47
+50000 facts: insert in txn (copy triggered)                        3.64
+50000 facts: modify in txn (copy triggered)                       39.42
+100000 facts: read-only txn (no copy needed)                      15.23
+100000 facts: rollback txn (no copy needed)                       15.17
+100000 facts: insert in txn (copy triggered)                       7.45
+100000 facts: modify in txn (copy triggered)                      73.97
 ```
 
 The results were calculated in [this GHA job][benchmark-gha]
-on 2025-10-15 at 14:53,
+on 2025-12-11 at 13:29,
 on Linux with 4 CPUs.
 <!-- benchmark_end -->
 
-[benchmark-gha]: https://github.com/yegor256/factbase/actions/runs/18533080805
+[benchmark-gha]: https://github.com/yegor256/factbase/actions/runs/20134646978

data/lib/factbase/cached/cached_fact.rb

@@ -6,7 +6,7 @@
 require 'others'
 require_relative '../../factbase'
 
-# A single fact in a factbase, which is sentitive to changes.
+# A single fact in a factbase, which is sensitive to changes.
 #
 # Author:: Yegor Bugayenko (yegor256@gmail.com)
 # Copyright:: Copyright (c) 2024-2025 Yegor Bugayenko
@@ -15,9 +15,11 @@ class Factbase::CachedFact
   # Ctor.
   # @param [Factbase::Fact] origin The original fact
   # @param [Hash] cache Cache of queries (to clean it on attribute addition)
-  def initialize(origin, cache)
+  # @param [Boolean] fresh True if this is a newly inserted fact (not yet in cache)
+  def initialize(origin, cache, fresh: false)
     @origin = origin
     @cache = cache
+    @fresh = fresh
   end
 
   def to_s
@@ -26,7 +28,9 @@ class Factbase::CachedFact
 
   # When a method is missing, this method is called.
   others do |*args|
-    @cache.clear if args[0].to_s.end_with?('=')
+    # Only clear cache when modifying properties on existing (non-fresh) facts
+    # Fresh facts are not in the cache yet, so modifications don't affect it
+    @cache.clear if args[0].to_s.end_with?('=') && !@fresh
     @origin.send(*args)
   end
 end

data/lib/factbase/cached/cached_factbase.rb

@@ -31,8 +31,7 @@ class Factbase::CachedFactbase
   # Insert a new fact and return it.
   # @return [Factbase::Fact] The fact just inserted
   def insert
-    @cache.clear
-    Factbase::CachedFact.new(@origin.insert, @cache)
+    Factbase::CachedFact.new(@origin.insert, @cache, fresh: true)
   end
 
   # Convert a query to a term.

data/lib/factbase/indexed/indexed_absent.rb

@@ -13,13 +13,19 @@ class Factbase::IndexedAbsent
   def predict(maps, _fb, _params)
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
-    if @idx[key].nil?
-      @idx[key] = []
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { facts: [], indexed_count: 0 }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       prop = @term.operands.first.to_s
-      maps.to_a.each do |m|
-        @idx[key].append(m) if m[prop].nil?
+      maps_array[entry[:indexed_count]..].each do |m|
+        entry[:facts] << m if m[prop].nil?
       end
+      entry[:indexed_count] = maps_array.size
     end
-    (maps & []) | @idx[key]
+    (maps & []) | entry[:facts]
   end
 end

data/lib/factbase/indexed/indexed_and.rb

@@ -18,14 +18,20 @@ class Factbase::IndexedAnd
       && @term.operands.all? { |o| o.operands.first.is_a?(Symbol) && _scalar?(o.operands[1]) }
       props = @term.operands.map { |o| o.operands.first }.sort
       key = [maps.object_id, props, :multi_and_eq]
-      if @idx[key].nil?
-        @idx[key] = {}
-        maps.to_a.each do |m|
+      entry = @idx[key]
+      maps_array = maps.to_a
+      if entry.nil?
+        entry = { index: {}, indexed_count: 0 }
+        @idx[key] = entry
+      end
+      if entry[:indexed_count] < maps_array.size
+        maps_array[entry[:indexed_count]..].each do |m|
           _all_tuples(m, props).each do |t|
-            @idx[key][t] = [] if @idx[key][t].nil?
-            @idx[key][t].append(m)
+            entry[:index][t] ||= []
+            entry[:index][t] << m
           end
         end
+        entry[:indexed_count] = maps_array.size
       end
       tuples = Enumerator.product(
         *@term.operands.sort_by { |o| o.operands.first }.map do |o|
@@ -36,7 +42,7 @@ class Factbase::IndexedAnd
           end
         end
       )
-      j = tuples.map { |t| @idx[key][t] || [] }.reduce(&:|)
+      j = tuples.map { |t| entry[:index][t] || [] }.reduce(&:|)
       r = (maps & []) | j
     else
       @term.operands.each do |o|

data/lib/factbase/indexed/indexed_eq.rb

@@ -14,15 +14,21 @@ class Factbase::IndexedEq
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
     return unless @term.operands.first.is_a?(Symbol) && _scalar?(@term.operands[1])
-    if @idx[key].nil?
-      @idx[key] = {}
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { index: {}, indexed_count: 0 }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       prop = @term.operands.first.to_s
-      maps.to_a.each do |m|
+      maps_array[entry[:indexed_count]..].each do |m|
         m[prop]&.each do |v|
-          @idx[key][v] = [] if @idx[key][v].nil?
-          @idx[key][v].append(m)
+          entry[:index][v] ||= []
+          entry[:index][v] << m
         end
       end
+      entry[:indexed_count] = maps_array.size
     end
     vv =
       if @term.operands[1].is_a?(Symbol)
@@ -33,7 +39,7 @@ class Factbase::IndexedEq
     if vv.empty?
       (maps & [])
     else
-      j = vv.map { |v| @idx[key][v] || [] }.reduce(&:|)
+      j = vv.map { |v| entry[:index][v] || [] }.reduce(&:|)
       (maps & []) | j
     end
   end

data/lib/factbase/indexed/indexed_exists.rb

@@ -13,13 +13,19 @@ class Factbase::IndexedExists
   def predict(maps, _fb, _params)
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
-    if @idx[key].nil?
-      @idx[key] = []
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { facts: [], indexed_count: 0 }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       prop = @term.operands.first.to_s
-      maps.to_a.each do |m|
-        @idx[key].append(m) unless m[prop].nil?
+      maps_array[entry[:indexed_count]..].each do |m|
+        entry[:facts] << m unless m[prop].nil?
       end
+      entry[:indexed_count] = maps_array.size
     end
-    (maps & []) | @idx[key]
+    (maps & []) | entry[:facts]
   end
 end

data/lib/factbase/indexed/indexed_fact.rb

@@ -6,7 +6,7 @@
 require 'others'
 require_relative '../../factbase'
 
-# A single fact in a factbase, which is sentitive to changes.
+# A single fact in a factbase, which is sensitive to changes.
 #
 # Author:: Yegor Bugayenko (yegor256@gmail.com)
 # Copyright:: Copyright (c) 2024-2025 Yegor Bugayenko
@@ -15,9 +15,11 @@ class Factbase::IndexedFact
   # Ctor.
   # @param [Factbase::Fact] origin The original fact
   # @param [Hash] idx The index
-  def initialize(origin, idx)
+  # @param [Boolean] fresh True if this is a newly inserted fact (not yet in index)
+  def initialize(origin, idx, fresh: false)
     @origin = origin
     @idx = idx
+    @fresh = fresh
   end
 
   def to_s
@@ -26,7 +28,9 @@ class Factbase::IndexedFact
 
   # When a method is missing, this method is called.
   others do |*args|
-    @idx.clear if args[0].to_s.end_with?('=')
+    # Only clear index when modifying properties on existing (non-fresh) facts
+    # Fresh facts are not in the index yet, so modifications don't affect it
+    @idx.clear if args[0].to_s.end_with?('=') && !@fresh
     @origin.send(*args)
   end
 end

data/lib/factbase/indexed/indexed_factbase.rb

@@ -31,8 +31,7 @@ class Factbase::IndexedFactbase
   # Insert a new fact and return it.
   # @return [Factbase::Fact] The fact just inserted
   def insert
-    @idx.clear
-    Factbase::IndexedFact.new(@origin.insert, @idx)
+    Factbase::IndexedFact.new(@origin.insert, @idx, fresh: true)
   end
 
   # Convert a query to a term.

data/lib/factbase/indexed/indexed_gt.rb

@@ -15,21 +15,27 @@ class Factbase::IndexedGt
     return unless @term.operands.first.is_a?(Symbol) && _scalar?(@term.operands[1])
     prop = @term.operands.first.to_s
     cache_key = [maps.object_id, @term.operands.first, :sorted]
-    if @idx[cache_key].nil?
-      @idx[cache_key] = []
-      maps.to_a.each do |m|
+    entry = @idx[cache_key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { sorted: [], indexed_count: 0 }
+      @idx[cache_key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
+      maps_array[entry[:indexed_count]..].each do |m|
         values = m[prop]
         next if values.nil?
         values.each do |v|
-          @idx[cache_key] << [v, m]
+          entry[:sorted] << [v, m]
         end
       end
-      @idx[cache_key].sort_by! { |pair| pair[0] }
+      entry[:sorted].sort_by! { |pair| pair[0] }
+      entry[:indexed_count] = maps_array.size
     end
     threshold = @term.operands[1].is_a?(Symbol) ? params[@term.operands[1].to_s]&.first : @term.operands[1]
     return nil if threshold.nil?
-    i = @idx[cache_key].bsearch_index { |pair| pair[0] > threshold } || @idx[cache_key].size
-    result = @idx[cache_key][i..].map { |pair| pair[1] }.uniq
+    i = entry[:sorted].bsearch_index { |pair| pair[0] > threshold } || entry[:sorted].size
+    result = entry[:sorted][i..].map { |pair| pair[1] }.uniq
     (maps & []) | result
   end
 

data/lib/factbase/indexed/indexed_lt.rb

@@ -15,21 +15,32 @@ class Factbase::IndexedLt
     return unless @term.operands.first.is_a?(Symbol) && _scalar?(@term.operands[1])
     prop = @term.operands.first.to_s
     cache_key = [maps.object_id, @term.operands.first, :sorted]
-    if @idx[cache_key].nil?
-      @idx[cache_key] = []
-      maps.to_a.each do |m|
+    entry = @idx[cache_key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { sorted: [], indexed_count: 0 }
+      @idx[cache_key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
+      new_pairs = []
+      maps_array[entry[:indexed_count]..].each do |m|
         values = m[prop]
         next if values.nil?
         values.each do |v|
-          @idx[cache_key] << [v, m]
+          new_pairs << [v, m]
         end
       end
-      @idx[cache_key].sort_by! { |pair| pair[0] }
+      unless new_pairs.empty?
+        entry[:sorted].concat(new_pairs)
+        entry[:sorted].sort_by! { |pair| pair[0] }
+      end
+      entry[:indexed_count] = maps_array.size
     end
+
     threshold = @term.operands[1].is_a?(Symbol) ? params[@term.operands[1].to_s]&.first : @term.operands[1]
     return nil if threshold.nil?
-    i = @idx[cache_key].bsearch_index { |pair| pair[0] >= threshold } || @idx[cache_key].size
-    result = @idx[cache_key][0...i].map { |pair| pair[1] }.uniq
+    i = entry[:sorted].bsearch_index { |pair| pair[0] >= threshold } || entry[:sorted].size
+    result = entry[:sorted][0...i].map { |pair| pair[1] }.uniq
     (maps & []) | result
   end
 

data/lib/factbase/indexed/indexed_not.rb

@@ -13,16 +13,25 @@ class Factbase::IndexedNot
   def predict(maps, fb, params)
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
-    if @idx[key].nil?
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { facts: nil, indexed_count: 0, yes_set: nil }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       yes = @term.operands.first.predict(maps, fb, params)
       if yes.nil?
-        @idx[key] = { r: nil }
+        entry[:facts] = nil
+        entry[:yes_set] = nil
       else
-        yes = yes.to_a.to_set
-        @idx[key] = { r: maps.to_a.reject { |m| yes.include?(m) } }
+        yes_set = yes.to_a.to_set
+        entry[:yes_set] = yes_set
+        entry[:facts] = maps_array.reject { |m| yes_set.include?(m) }
       end
+      entry[:indexed_count] = maps_array.size
     end
-    r = @idx[key][:r]
+    r = entry[:facts]
     if r.nil?
       nil
     else

data/lib/factbase/indexed/indexed_one.rb

@@ -13,13 +13,19 @@ class Factbase::IndexedOne
   def predict(maps, _fb, _params)
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
-    if @idx[key].nil?
-      @idx[key] = []
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { facts: [], indexed_count: 0 }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       prop = @term.operands.first.to_s
-      maps.to_a.each do |m|
-        @idx[key].append(m) if !m[prop].nil? && m[prop].size == 1
+      maps_array[entry[:indexed_count]..].each do |m|
+        entry[:facts] << m if !m[prop].nil? && m[prop].size == 1
       end
+      entry[:indexed_count] = maps_array.size
     end
-    (maps & []) | @idx[key]
+    (maps & []) | entry[:facts]
   end
 end

data/lib/factbase/indexed/indexed_unique.rb

@@ -16,10 +16,19 @@ class Factbase::IndexedUnique
   def predict(maps, _fb, _params)
     return nil if @idx.nil?
     key = [maps.object_id, @term.operands.first, @term.op]
-    if @idx[key].nil?
+    entry = @idx[key]
+    maps_array = maps.to_a
+    if entry.nil?
+      entry = { facts: [], indexed_count: 0 }
+      @idx[key] = entry
+    end
+    if entry[:indexed_count] < maps_array.size
       props = @term.operands.map(&:to_s)
-      @idx[key] = maps.to_a.select { |m| props.all? { |p| !m[p].nil? } }
+      maps_array[entry[:indexed_count]..].each do |m|
+        entry[:facts] << m if props.all? { |p| !m[p].nil? }
+      end
+      entry[:indexed_count] = maps_array.size
     end
-    (maps & []) | @idx[key]
+    (maps & []) | entry[:facts]
   end
 end

data/lib/factbase/version.rb

@@ -9,5 +9,5 @@
 # License:: MIT
 class Factbase
   # Current version of the gem (changed by .rultor.yml on every release)
-  VERSION = '0.18.0' unless const_defined?(:VERSION)
+  VERSION = '0.19.0' unless const_defined?(:VERSION)
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: factbase
 version: !ruby/object:Gem::Version
-  version: 0.18.0
+  version: 0.19.0
 platform: ruby
 authors:
 - Yegor Bugayenko