From 50cb02efdeabbec4696c424a9bd4e1a3b8e26382 Mon Sep 17 00:00:00 2001
From: Jakob Nybo Nissen <jakobnybonissen@gmail.com>
Date: Thu, 16 Jun 2022 16:27:55 +0200
Subject: [PATCH] Invert parsing and construct from string

---
 src/BioSequences.jl               |  3 +++
 src/longsequences/constructors.jl | 33 +++++++++++++++++--------------
 src/longsequences/copying.jl      |  2 --
 3 files changed, 21 insertions(+), 17 deletions(-)

diff --git a/src/BioSequences.jl b/src/BioSequences.jl
index aef223a9..f4e3c1fe 100644
--- a/src/BioSequences.jl
+++ b/src/BioSequences.jl
@@ -204,6 +204,9 @@ using Random
 
 BioSymbols.gap(::Type{Char}) = '-'
 
+const SeqLike = Union{AbstractVector, AbstractString}
+const ASCIILike = Union{String, SubString{String}}
+
 include("alphabet.jl")
 
 # Load the bit-twiddling internals that optimised BioSequences methods depend on.
diff --git a/src/longsequences/constructors.jl b/src/longsequences/constructors.jl
index abdd2f94..2a69327e 100644
--- a/src/longsequences/constructors.jl
+++ b/src/longsequences/constructors.jl
@@ -60,20 +60,8 @@ function (::Type{T})(seq::LongSequence{<:NucleicAcidAlphabet{N}}) where
 end
 
 # Constructors from strings
-function LongSequence{A}(s::Union{String, SubString{String}}) where {A<:Alphabet}
-    return LongSequence{A}(s, codetype(A()))
-end
-
-# Generic method for String/Substring.
-function LongSequence{A}(s::Union{String, SubString{String}}, ::AlphabetCode) where {A<:Alphabet}
-    len = length(s)
-    seq = LongSequence{A}(undef, len)
-    return copyto!(seq, 1, s, 1, len)
-end
-
-function LongSequence{A}(s::Union{String, SubString{String}}, ::AsciiAlphabet) where {A<:Alphabet}
-    seq = LongSequence{A}(undef, ncodeunits(s))
-    return encode_chunks!(seq, 1, codeunits(s), 1, ncodeunits(s))
+function LongSequence{A}(s::AbstractString) where {A <: Alphabet}
+    return parse(LongSequence{A}, s)
 end
 
 function LongSequence{A}(
@@ -85,4 +73,19 @@ function LongSequence{A}(
     return copyto!(seq, 1, src, first(part), len)
 end
 
-Base.parse(::Type{LongSequence{A}}, seq::AbstractString) where A = LongSequence{A}(seq)
\ No newline at end of file
+Base.parse(::Type{T}, s::AbstractString) where {T <: LongSequence} = parse(T, String(s))
+
+function Base.parse(::Type{LongSequence{A}}, seq::ASCIILike) where {A<:Alphabet}
+    _parse(LongSequence{A}, seq, codetype(A()))
+end
+
+function _parse(::Type{LongSequence{A}}, s::ASCIILike, ::AlphabetCode) where {A<:Alphabet}
+    len = length(s)
+    seq = LongSequence{A}(undef, len)
+    return copyto!(seq, 1, s, 1, len)
+end
+
+function _parse(::Type{LongSequence{A}}, s::ASCIILike, ::AsciiAlphabet) where {A<:Alphabet}
+    seq = LongSequence{A}(undef, ncodeunits(s))
+    return encode_chunks!(seq, 1, codeunits(s), 1, ncodeunits(s))
+end
diff --git a/src/longsequences/copying.jl b/src/longsequences/copying.jl
index f939a811..1a37304e 100644
--- a/src/longsequences/copying.jl
+++ b/src/longsequences/copying.jl
@@ -107,8 +107,6 @@ function _copyto!(dst::SeqOrView{A}, doff::Integer,
 end
 
 #########
-const SeqLike = Union{AbstractVector, AbstractString}
-const ASCIILike = Union{String, SubString{String}}
 
 """
     copy!(dst::LongSequence, src)