feat: support contains and occursin method (#37)

- Added codeunit and codeunit(s, i) methods for accessing code units
- Implemented isvalid and thisind for proper UTF-8 character boundary detection

Author: Beforerr

src/staticstring.jl

@@ -1,5 +1,6 @@
 # https://github.com/mkitti/StaticStrings.jl
 # https://github.com/JuliaPy/PythonCall.jl/blob/main/src/Utils/Utils.jl
+using Base: between
 
 struct StaticString{N, T} <: AbstractString
     codeunits::NTuple{N, T}
@@ -60,10 +61,33 @@ function Base.String(x::StaticString{N, T}) where {N, T}
     return String(b .= x.codeunits)
 end
 
-@inline Base.ncodeunits(::StaticString{N}) where N = N
+@inline Base.ncodeunits(::StaticString{N}) where {N} = N
+Base.codeunit(::StaticString{N, T}) where {N, T} = T
+Base.@propagate_inbounds Base.codeunit(s::StaticString, i::Int) = s.codeunits[i]
 
-function StaticString(cu::Base.CodeUnits{T}) where T
+function StaticString(cu::Base.CodeUnits{T}) where {T}
     N = length(cu)
     return StaticString{N, T}(NTuple{N, T}(cu))
 end
-StaticString(s::AbstractString) = StaticString(codeunits(s))
+StaticString(s::AbstractString) = StaticString(codeunits(s))
+
+Base.isvalid(s::StaticString, i::Int) = checkbounds(Bool, s, i) && thisind(s, i) == i
+Base.thisind(s::StaticString, i::Int) = _thisind_str(s, i)
+
+@inline function _thisind_str(s, i::Int)
+    i == 0 && return 0
+    n = ncodeunits(s)
+    i == n + 1 && return i
+    @boundscheck between(i, 1, n) || throw(BoundsError(s, i))
+    @inbounds b = codeunit(s, i)
+    (b & 0xc0 == 0x80) & (i - 1 > 0) || return i
+    @inbounds b = codeunit(s, i - 1)
+    between(b, 0b11000000, 0b11110111) && return i - 1
+    (b & 0xc0 == 0x80) & (i - 2 > 0) || return i
+    @inbounds b = codeunit(s, i - 2)
+    between(b, 0b11100000, 0b11110111) && return i - 2
+    (b & 0xc0 == 0x80) & (i - 3 > 0) || return i
+    @inbounds b = codeunit(s, i - 3)
+    between(b, 0b11110000, 0b11110111) && return i - 3
+    return i
+end

test/staticstring.jl

@@ -1,4 +1,5 @@
 using CommonDataFormat: StaticString
+using Test
 
 @testset "StaticString" begin
     s = "Hello, World!"
@@ -8,4 +9,9 @@ using CommonDataFormat: StaticString
     @test ss == s
     @test String(ss) == s
     @test !isempty(ss)
-end
+    @test contains(ss, "World")
+    @test occursin("World", ss)
+    @test replace(ss, "World" => "Julia") == "Hello, Julia!"
+
+    @test codeunit(ss) == UInt8
+end