Add Deserializer

src/ProteinChains.jl

@@ -1,7 +1,5 @@
 module ProteinChains
 
-using StaticArrays: SVector
-
 using Backboner
 
 include("ideal.jl")
@@ -38,4 +36,6 @@ export writecif, writepdb
 export PDBFormat, MMCIFFormat
 export pdbentry, @pdb_str
 
+include("serialization/serialization.jl")
+
 end

src/atom.jl

@@ -1,4 +1,5 @@
 using PeriodicTable: elements
+using StaticArrays: SVector
 
 const ELEMENT_SYMBOL_TO_NUMBER = Dict(uppercase(elements[number].symbol) => number for number in 1:118)
 const number_TO_ELEMENT_SYMBOL = Dict(n => s for (s, n) in ELEMENT_SYMBOL_TO_NUMBER)
@@ -22,10 +23,10 @@ end
 Atom{T}(atom::Atom) where T = Atom{T}(atom.name, atom.number, atom.x, atom.y, atom.z)
 
 @inline Atom(name::UInt32, number::Integer, x::T, y::T, z::T) where T = Atom{T}(name, number, x, y, z)
-@inline Atom(name::AbstractString, element_symbol::AbstractString, x::T, y::T, z::T) where T =
+@inline Atom(name::AbstractString, element_symbol::AbstractString, x, y, z) =
     Atom(encode_atom_name(name, element_symbol), element_symbol_to_number(element_symbol), x, y, z)
 
-@inline Atom(name, number, coords::AbstractVector{T}) where T = Atom(name, number, coords...)
+@inline Atom(name, number, coords::AbstractVector) = Atom(name, number, coords...)
 
 coords(atom::Atom) = SVector(atom.x, atom.y, atom.z)
 

src/io/io.jl

@@ -18,5 +18,3 @@ include("renumber.jl")
 include("read.jl")
 include("write.jl")
 include("download.jl")
-
-include("serialization.jl")

src/serialization/lazy.jl

@@ -0,0 +1,27 @@
+mutable struct Deserializer <: AbstractDict{String,ProteinStructure}
+    filename::String
+    file::HDF5.File
+    keys::Set{String}
+
+    function Deserializer(filename::AbstractString)
+        file = HDF5.h5open(filename, "r")
+        deserializer = new(filename, file, Set(keys(file)))
+        finalizer(close, deserializer)
+        return deserializer
+    end
+end
+
+Base.close(deserializer::Deserializer) = close(deserializer.file)
+
+Base.keys(deserializer::Deserializer) = deserializer.keys
+Base.length(deserializer::Deserializer) = length(keys(deserializer))
+Base.getindex(deserializer::Deserializer, key::AbstractString) = read(deserializer.file[key], ProteinStructure)
+
+function Base.iterate(deserializer::Deserializer, state...)
+    handler(::Nothing) = nothing
+    handler((key,state)::Tuple{String,Int}) = (key => deserializer[key], state)
+    return handler(iterate(deserializer.keys, state...))
+end
+
+Base.show(io::IO, deserializer::Deserializer) = print(io, "$(Deserializer)(\"$(deserializer.filename)\")")
+Base.show(io::IO, ::MIME"text/plain", deserializer::Deserializer) = print(io, summary(deserializer))

src/io/serialization.jl → src/serialization/serialization.jl

@@ -79,7 +79,9 @@ end
 
 function serialize(path::AbstractString, dataset::ProteinDataset; mode="w")
     HDF5.h5open(path, mode) do f
-        foreach(((key, structure),) -> write(f, structure, key), dataset)
+        for (key, structure) in dataset
+            write(f, structure, key)
+        end
     end
     return path
 end
@@ -96,3 +98,5 @@ function deserialize(path::AbstractString)
     end
     return ProteinDataset(structures)
 end
+
+include("lazy.jl")