Fix equation splitting for unicode variables

src/HallOfFame.jl

@@ -1,6 +1,7 @@
 module HallOfFameModule
 
 import DynamicExpressions: Node, string_tree
+import ..UtilsModule: split_string
 import ..CoreModule: MAX_DEGREE, Options, Dataset, DATA_TYPE, LOSS_TYPE
 import ..ComplexityModule: compute_complexity
 import ..PopMemberModule: PopMember, copy_pop_member
@@ -160,20 +161,4 @@ function string_dominating_pareto_curve(
     return output
 end
 
-"""
-    split_string(s::String, n::Integer)
-
-```jldoctest
-split_string("abcdefgh", 3)
-
-# output
-
-["abc", "def", "gh"]
-```
-"""
-function split_string(s::String, n::Integer)
-    length(s) <= n && return [s]
-    return [s[i:min(i + n - 1, end)] for i in 1:n:length(s)]
-end
-
 end

src/Utils.jl

@@ -48,6 +48,24 @@ function recursive_merge(x...)
     return x[end]
 end
 
+"""
+    split_string(s::String, n::Integer)
+
+```jldoctest
+split_string("abcdefgh", 3)
+
+# output
+
+["abc", "def", "gh"]
+```
+"""
+function split_string(s::String, n::Integer)
+    length(s) <= n && return [s]
+    # Due to unicode characters, need to split only at valid indices:
+    I = eachindex(s) |> collect
+    return [s[I[i]:I[min(i + n - 1, end)]] for i in 1:n:length(s)]
+end
+
 """
 Tiny equivalent to StaticArrays.MVector
 

test/test_print.jl

@@ -1,5 +1,6 @@
 using Test
 using SymbolicRegression
+using SymbolicRegression.UtilsModule: split_string
 
 include("test_params.jl")
 
@@ -45,3 +46,12 @@ for binop in [safe_pow, ^]
     minitree = Node(5, Node("x1"), Node("x2"))
     @test string_tree(minitree, opts) == "(x1 ^ x2)"
 end
+
+@testset "Test splitting of strings" begin
+    split_string("abcdefgh", 3) == ["abc", "def", "gh"]
+    split_string("abcdefgh", 100) == ["abcdefgh"]
+    split_string("⋅", 1) == ["⋅"]
+    split_string("⋅⋅", 1) == ["⋅", "⋅"]
+    split_string("⋅⋅⋅⋅", 2) == ["⋅⋅", "⋅⋅"]
+    split_string("ραβγ", 2) == ["ρα", "βγ"]
+end