Remove N from def of Cell to make it more type stable

Project.toml

@@ -6,18 +6,16 @@ version = "0.5.3"
 [deps]
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 StructHelpers = "4093c41a-2008-41fd-82b8-e3f9d02b504f"
-UnPack = "3a884ed6-31ef-47d7-9d2a-63182c4928ed"
 spglib_jll = "ac4a9f1e-bdb2-5204-990c-47c8b2f70d4e"
 
-[extras]
-Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
-
 [compat]
 StaticArrays = "0.8.3, 0.9, 0.10, 0.11, 0.12, 1.0"
 StructHelpers = "0.1"
 julia = "1.3"
-UnPack = "1.0"
 spglib_jll = "1.14"
 
+[extras]
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
 [targets]
 test = ["Test"]

src/Spglib.jl

@@ -1,7 +1,6 @@
 module Spglib
 
 using spglib_jll: libsymspg
-using UnPack: @unpack
 
 # All public methods
 export get_symmetry,

src/model.jl

@@ -1,7 +1,7 @@
 using StaticArrays: MMatrix, MVector
 using StructHelpers: @batteries
 
-export Cell, Dataset, SpacegroupType, basis_vectors
+export Cell, Dataset, SpacegroupType, basis_vectors, natoms
 
 """
     Cell(lattice, positions, types, magmoms=zeros(length(types)))
@@ -11,31 +11,45 @@ The basic input data type of `Spglib`.
 Lattice parameters `lattice` are given by a ``3×3`` matrix with floating point values,
 where ``𝐚``, ``𝐛``, and ``𝐜`` are given as columns.
 Fractional atomic positions `positions` are given
-by a ``3×N`` matrix with floating point values, where ``N`` is the number of atoms.
+by a vector of ``N`` vectors with floating point values, where ``N`` is the number of atoms.
 Numbers to distinguish atomic species `types` are given by a list of ``N`` integers.
 The collinear polarizations `magmoms` only work with `get_symmetry` and are given
 as a list of ``N`` floating point values.
 """
-struct Cell{N,L,P,T,M}
+struct Cell{L,P,T,M}
     lattice::MMatrix{3,3,L,9}
-    positions::MMatrix{3,N,P}
-    types::MVector{N,T}
-    magmoms::MVector{N,M}
+    positions::Vector{MVector{3,P}}
+    types::Vector{T}
+    magmoms::Vector{M}
 end
 function Cell(lattice, positions, types, magmoms = zeros(length(types)))
-    if lattice isa AbstractVector
-        lattice = hcat(lattice...)
+    if !(lattice isa AbstractMatrix)
+        lattice = reduce(hcat, lattice)  # Use `reduce` can make it type stable
     end
-    if positions isa AbstractVector
-        positions = hcat(positions...)
+    N = length(types)
+    if positions isa AbstractMatrix
+        P = eltype(positions)
+        if size(positions) == (3, 3)
+            error("ambiguous `positions` size 3×3! Use a vector of `Vector`s instead!")
+        elseif size(positions) == (3, N)
+            positions = collect(eachcol(positions))
+        elseif size(positions) == (N, 3)
+            positions = collect(eachrow(positions))
+        else
+            throw(DimensionMismatch( "the `positions` has a different number of atoms from the `types`!"))
+        end
+    else  # positions isa AbstractVector or a Tuple
+        P = eltype(Base.promote_typeof(positions...))
+        positions = collect(map(MVector{3,P}, positions))
     end
-    N, L, P, T, M =
-        length(types), eltype(lattice), eltype(positions), eltype(types), eltype(magmoms)
-    return Cell{N,L,P,T,M}(lattice, positions, types, magmoms)
+    L, T, M = eltype(lattice), eltype(types), eltype(magmoms)
+    return Cell{L,P,T,M}(lattice, positions, types, magmoms)
 end
 
 @batteries Cell eq = true hash = true
 
+natoms(cell::Cell) = length(cell.types)
+
 """
     basis_vectors(cell::Cell)
 
@@ -48,15 +62,16 @@ end
 
 # This is an internal function, do not export!
 function _expand_cell(cell::Cell)
-    @unpack lattice, positions, types, magmoms = cell
+    lattice, positions, types, magmoms =
+        cell.lattice, cell.positions, cell.types, cell.magmoms
     # Reference: https://github.com/mdavezac/spglib.jl/blob/master/src/spglib.jl#L32-L35 and https://github.com/spglib/spglib/blob/444e061/python/spglib/spglib.py#L953-L975
-    clattice = Base.cconvert(Matrix{Cdouble}, lattice) |> transpose
-    cpositions = Base.cconvert(Matrix{Cdouble}, positions)
+    clattice = Base.cconvert(Matrix{Cdouble}, transpose(lattice))
+    cpositions = Base.cconvert(Matrix{Cdouble}, reduce(hcat, positions))
     ctypes = Cint[findfirst(isequal(u), unique(types)) for u in types]
     if magmoms !== nothing
         magmoms = Base.cconvert(Vector{Cdouble}, magmoms)
     end
-    return Cell(clattice, cpositions, ctypes, magmoms)
+    return clattice, cpositions, ctypes, magmoms
 end
 
 # This is an internal type, do not export!

src/reciprocal.jl

@@ -37,7 +37,7 @@ function get_ir_reciprocal_mesh(
     end
     @assert all(isone(x) || iszero(x) for x in is_shift)
     # Prepare for input
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     mesh = Base.cconvert(Vector{Cint}, mesh)
     is_shift = Base.cconvert(Vector{Cint}, is_shift)
     is_time_reversal = Base.cconvert(Cint, is_time_reversal)

src/standardize.jl

@@ -14,7 +14,7 @@ function standardize_cell(
     no_idealize = false,
     symprec = 1e-5,
 )
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     to_primitive = Base.cconvert(Cint, to_primitive)
     no_idealize = Base.cconvert(Cint, no_idealize)
     num_atom = Base.cconvert(Cint, length(types))

src/symmetry.jl

@@ -41,7 +41,7 @@ function get_symmetry!(
     if !(size(rotation, 1) == size(rotation, 2) == size(translation, 1) == 3)
         throw(DimensionMismatch("`rotation` & `translation` don't have the right size!"))
     end
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     rotation = Base.cconvert(Array{Cint,3}, rotation)
     translation = Base.cconvert(Matrix{Cdouble}, translation)
     max_size = Base.cconvert(Cint, size(rotation, 3))
@@ -82,7 +82,7 @@ function get_symmetry_with_collinear_spin!(
     cell::Cell,
     symprec = 1e-5,
 ) where {T}
-    @unpack lattice, positions, types, magmoms = _expand_cell(cell)
+    lattice, positions, types, magmoms = _expand_cell(cell)
     rotation = Base.cconvert(Array{Cint,3}, rotation)
     translation = Base.cconvert(Matrix{Cdouble}, translation)
     equivalent_atoms = Base.cconvert(Vector{Cint}, equivalent_atoms)
@@ -220,7 +220,7 @@ end
 Return the exact number of symmetry operations. An error is thrown when it fails.
 """
 function get_multiplicity(cell::Cell, symprec = 1e-5)
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     num_atom = Base.cconvert(Cint, length(types))
     num_sym = ccall(
         (:spg_get_multiplicity, libsymspg),
@@ -244,7 +244,7 @@ end
 Search symmetry operations of an input unit cell structure.
 """
 function get_dataset(cell::Cell, symprec = 1e-5)
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     num_atom = Base.cconvert(Cint, length(types))
     ptr = ccall(
         (:spg_get_dataset, libsymspg),
@@ -270,7 +270,7 @@ end
 Search symmetry operations of an input unit cell structure, using a given Hall number.
 """
 function get_dataset_with_hall_number(cell::Cell, hall_number::Integer, symprec = 1e-5)
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     num_atom = Base.cconvert(Cint, length(types))
     hall_number = Base.cconvert(Cint, hall_number)
     ptr = ccall(
@@ -332,7 +332,7 @@ end
 Return the space group type in Hermann–Mauguin (international) notation.
 """
 function get_international(cell::Cell, symprec = 1e-5)
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     symbol = Vector{Cchar}(undef, 11)
     exitcode = ccall(
         (:spg_get_international, libsymspg),
@@ -357,7 +357,7 @@ end
 Return the space group type in Schoenflies notation.
 """
 function get_schoenflies(cell::Cell, symprec = 1e-5)
-    @unpack lattice, positions, types = _expand_cell(cell)
+    lattice, positions, types = _expand_cell(cell)
     symbol = Vector{Cchar}(undef, 7)
     exitcode = ccall(
         (:spg_get_schoenflies, libsymspg),

test/standardize.jl

@@ -26,7 +26,7 @@
               -1.999719735 1.999719735 0.0
               0.0 0.0 8.57154746
           ]
-    @test primitive_cell.positions ≈ [  # Python results
+    @test reduce(hcat, primitive_cell.positions) ≈ [  # Python results
         0.15311561 0.34688439 0.65311561 0.84688439
         0.84688439 0.65311561 0.34688439 0.15311561
         0.1203133 0.6203133 0.3796867 0.8796867
@@ -69,7 +69,7 @@ end
         -1.99971973 1.99971973 0.0
         0.0 0.0 8.57154746
     ]
-    @test primitive_cell.positions ≈ [  # Python results
+    @test reduce(hcat, primitive_cell.positions) ≈ [  # Python results
         0.15311561 0.34688439 0.65311561 0.84688439
         0.84688439 0.65311561 0.34688439 0.15311561
         0.1203133 0.6203133 0.3796867 0.8796867
@@ -105,7 +105,7 @@ end
             2.0 -2.0 2.0
             2.0 2.0 -2.0
         ]
-        @test new_cell.positions ≈ [0.0 0.0 0.0]'
+        @test new_cell.positions ≈ [[0.0, 0.0, 0.0]]
         @test new_cell.types == [1]
     end
     @testset "Test `refine_cell`" begin
@@ -123,7 +123,7 @@ end
             0.0 4.0 0.0
             0.0 0.0 4.0
         ]
-        @test new_cell.positions ≈ [
+        @test reduce(hcat, new_cell.positions) ≈ [
             0.0 0.5
             0.0 0.5
             0.0 0.5