Support Unicode subscripts when parsing spin-{orbitals,configurations} (#41)

Author: jagot

src/AtomicLevels.jl

@@ -9,6 +9,7 @@ using HalfIntegers
 using Combinatorics
 
 include("common.jl")
+include("unicode.jl")
 include("parity.jl")
 include("orbitals.jl")
 include("relativistic_orbitals.jl")

src/spin_orbitals.jl

@@ -132,9 +132,19 @@ end
 
 function Base.parse(::Type{O}, orb_str) where {OO<:AbstractOrbital,O<:SpinOrbital{OO}}
     m = match(r"^(.*)\((.*)\)$", orb_str)
-    m === nothing && throw(ArgumentError("Invalid spin-orbital string: $(orb_str)"))
-    o = parse(OO, m[1])
-    SpinOrbital(o, (split(m[2], ",")...,))
+    # For non-relativistic spin-orbitals, we also support specifying
+    # the m_ℓ quantum number using Unicode subscripts and the spin
+    # label using α/β
+    m2 = match(r"^(.*?)([₋]{0,1}[₁₂₃₄₅₆₇₈₉₀]+)([αβ])$", orb_str)
+    if !isnothing(m)
+        o = parse(OO, m[1])
+        SpinOrbital(o, (split(m[2], ",")...,))
+    elseif !isnothing(m2) && OO <: Orbital
+        o = parse(OO, m2[1])
+        SpinOrbital(o, from_subscript(m2[2]), m2[3])
+    else
+        throw(ArgumentError("Invalid spin-orbital string: $(orb_str)"))
+    end
 end
 
 """

src/unicode.jl

@@ -0,0 +1,17 @@
+function from_subscript(s)
+    inverse_subscript_map = Dict('₋' => '-', '₊' => '+', '₀' => '0',
+                                 '₁' => '1', '₂' => '2', '₃' => '3',
+                                 '₄' => '4', '₅' => '5', '₆' => '6',
+                                 '₇' => '7', '₈' => '8', '₉' => '9',
+                                 '₀' => '0')
+    map(Base.Fix1(getindex, inverse_subscript_map), s)
+end
+
+function from_superscript(s)
+    inverse_superscript_map = Dict('⁻' => '-', '⁺' => '+', '⁰' => '0',
+                                   '¹' => '1', '²' => '2', '³' => '3',
+                                   '⁴' => '4', '⁵' => '5', '⁶' => '6',
+                                   '⁷' => '7', '⁸' => '8', '⁹' => '9',
+                                   '⁰' => '0')
+    map(Base.Fix1(getindex, inverse_superscript_map), s)
+end

test/orbitals.jl

@@ -311,6 +311,26 @@ using Random
             @test parse(SpinOrbital{Orbital}, "1s(0,α)") == SpinOrbital(Orbital(1, 0), (0, up))
             @test parse(SpinOrbital{Orbital{Int}}, "1s(0,α)") == SpinOrbital(Orbital(1, 0), (0, up))
             @test parse(SpinOrbital{Orbital{Symbol}}, "ks(0,α)") == SpinOrbital(Orbital(:k, 0), (0, up))
+
+            @test so"1s₀β" == so"1s(0,β)"
+            @test so"2p₋₁α" == so"2p(-1,α)"
+            @test so"k[31]₋₁₃α" == so"k[31](-13,α)"
+
+            for o in [SpinOrbital(o"1s", (0,-1/2)),
+                      SpinOrbital(o"1s", (0,1/2)),
+                      SpinOrbital(o"2p", (1,-1/2)),
+                      SpinOrbital(ro"1s", (1/2)),
+                      SpinOrbital(ro"2p", (3/2)),
+                      SpinOrbital(ro"3d", (5/2)),
+                      SpinOrbital(ro"3d-", (-3/2)),
+                      SpinOrbital(ro"3p", (-1/2)),
+                      SpinOrbital(ro"3p", (-3/2)),
+                      SpinOrbital(ro"3p", (1/2)),
+                      SpinOrbital(ro"3p-", (-1/2)),
+                      SpinOrbital(ro"3p-", (1/2))]
+                O = typeof(o.orb)
+                @test parse(SpinOrbital{O}, string(o)) == o
+            end
         end
     end
 

test/runtests.jl

@@ -3,6 +3,11 @@ using WignerSymbols
 using HalfIntegers
 using Test
 
+@testset "Unicode super-/subscripts" begin
+    @test AtomicLevels.from_subscript("₋₊₁₂₃₄₅₆₇₈₉₀") == "-+1234567890"
+    @test AtomicLevels.from_superscript("⁻⁺¹²³⁴⁵⁶⁷⁸⁹⁰") == "-+1234567890"
+end
+
 include("parity.jl")
 include("orbitals.jl")
 include("configurations.jl")