Add the dynamic size limit fitness for regexp-search

config/regexp_search/dynamic_size_limit.json

@@ -0,0 +1,100 @@
+{
+    "pop_size": 1000,
+    "growth_factor": 2.0,
+    "crossover_ratio": 0.5,
+    "mutation_ratio": 0.45,
+    "remove_duplicates": true,
+
+    "plugin_dir":"_build/plugins/regexp_search/",
+    "plugins":
+    [
+        "../selection/Tournament.cmxs",
+        "../parent_chooser/UniformParentChooser.cmxs",
+        "creation/RandomCreation.cmxs",
+        "creation/CreationFromExample.cmxs",
+        "mutation/RebuildSubtreeMutation.cmxs",
+        "mutation/ReduceMutation.cmxs",
+        "crossover/TakeSubtreeCrossover.cmxs",
+        "crossover/CombineCrossover.cmxs",
+        "fitness/MultiobjectiveFitness.cmxs",
+        "fitness/DynamicSizeLimitFitness.cmxs",
+        "simplification/RepeatSimplification.cmxs",
+        "simplification/GroupChars.cmxs"
+    ],
+
+    "creation":
+    [
+        {
+            "proba":0.1, "method":"from_example",
+            "params":
+            {
+                "replacement_patterns":
+                [
+                    { "proba":0.5, "regexp":"[A-Z]" },
+                    { "proba":0.5, "regexp":"[a-z]" },
+                    { "proba":0.7, "regexp":"[0-9]" },
+                    { "proba":0.5, "regexp":"[ \t]" },
+                    { "proba":0.8, "regexp":"[A-Za-z]" },
+                    { "proba":0.2, "regexp":"." }
+                ]
+            }
+        },
+        { "proba":0.4, "method":"fill", "params":"&random_gen" },
+        { "proba":0.5, "method":"grow", "params":"&random_gen" }
+    ],
+    "mutation":
+    [
+        { "proba":0.7, "method":"rebuild_subtree", "params":"&random_gen" },
+        { "proba":0.3, "method":"reduce" }
+    ],
+    "crossover":
+    [
+        { "proba":0.9, "method":"take_subtree" },
+        { "proba":0.1, "method":"combine", "params":{"alt_proba":0.1} }
+    ],
+    "fitness":
+    {
+        "method":"dynamic_size_limit",
+        "params":
+        {
+            "full_example_score":10.0,
+            "full_counter_example_score":50.0,
+            "partial_example_score":5.0,
+            "partial_counter_example_score":0.0,
+            "size_score":1.0,
+            "elim_size":150,
+            "initial_size_limit":25
+        }
+    },
+    "simplifications":
+    [
+        { "schedule":25, "method":"repeat_simplification" },
+        { "schedule":50, "method":"group_chars" }
+    ],
+    "selection" : { "method":"tournament" },
+    "parent_choice": { "method":"uniform" },
+
+    "&random_gen":
+    {
+        "min_init_depth": 3,
+        "max_init_depth": 5,
+        "max_depth": 10,
+
+        "concat_proba": 0.5,
+        "alt_proba": 0.05,
+        "opt_proba": 0.05,
+        "plus_proba": 0.05,
+        "star_proba": 0.05,
+        "rand_char_proba": 0.25,
+        "predefined_proba": 0.05,
+        "predefined_list":
+        [
+            { "proba":0.2, "regexp":"." },
+            { "proba":0.3, "regexp":"[A-Za-z]" },
+            { "proba":0.1, "regexp":"[A-Z]" },
+            { "proba":0.1, "regexp":"[a-z]" },
+            { "proba":0.2, "regexp":"[0-9]" },
+            { "proba":0.1, "regexp":"[ \t]" }
+        ]
+    }
+}

plugins/regexp_search/fitness/DynamicSizeLimitFitness.ml

@@ -0,0 +1,52 @@
+(** This fitness function is the same as MultiobjectiveFitness except it gives a null fitness value to all individuals bigger and not better than previous encountered ones with a dynamic size limit.
+    It uses internally the MultiobjectiveFitness plugin and accept the same parameters.
+    It is compatible with a static size threshold and is intended to prevent bloat. *)
+let build_dynamic_size_limit_fitness json =
+    let open Yojson.Basic.Util in
+    let init_size_limit = json |> member "initial_size_limit" |> to_int in
+
+    let module MultiobjectiveFitness = (val RegexpSearchHooks.Fitness.get "multiobjective" json) in
+
+    (module struct
+        (** The fitness contains Evaluated with the real fitness value if the individual is small enough and Rejected if it is not. *)
+        type t =
+            | Rejected
+            | Evaluated of MultiobjectiveFitness.t
+        ;;
+
+        type individual = RegexpTree.t
+        type target_data = ExampleList.t
+
+        let to_float = function
+            | Rejected -> 0.
+            | Evaluated fit -> MultiobjectiveFitness.to_float fit
+        ;;
+
+        let to_string = function
+            | Rejected -> "Above dynamic size limit"
+            | Evaluated fit -> MultiobjectiveFitness.to_string fit
+        ;;
+
+        let compare x y = Pervasives.compare (to_float x) (to_float y);;
+
+        let best_fitness = ref 0.;;
+        let size_limit = ref init_size_limit;;
+
+        let compute examples regexp =
+            let fitness = MultiobjectiveFitness.compute examples regexp in
+            let float_fitness = MultiobjectiveFitness.to_float fitness in
+            let size = RegexpTree.size regexp in
+            if float_fitness > !best_fitness then
+            (
+                best_fitness := float_fitness;
+                size_limit := max !size_limit size
+            );
+            if size > !size_limit then Rejected
+            else Evaluated fitness
+        ;;
+    end : EvolParams.Fitness with type individual = RegexpTree.t and type target_data = ExampleList.t)
+;;
+
+let () =
+    RegexpSearchHooks.Fitness.register "dynamic_size_limit" build_dynamic_size_limit_fitness
+;;

plugins/regexp_search/regexp_search.itarget

@@ -7,5 +7,6 @@ mutation/ReduceMutation.cmxs
 fitness/FullMatchFitness.cmxs
 fitness/PartialMatchFitness.cmxs
 fitness/MultiobjectiveFitness.cmxs
+fitness/DynamicSizeLimitFitness.cmxs
 simplification/RepeatSimplification.cmxs
 simplification/GroupChars.cmxs