Merge pull request #44 from jakobnissen/noautoma

Removed Automa parsing

Project.toml

@@ -1,13 +1,11 @@
 name = "BioSymbols"
 uuid = "3c28c6f8-a34d-59c4-9654-267d177fcfa9"
 authors = ["Ben J. Ward <[email protected]"]
-version = "4.0.4"
+version = "4.0.5"
 
 [deps]
-Automa = "67c07d97-cdcb-5c2c-af73-a7f9c32a568b"
 
 [compat]
-Automa = "0.8"
 julia = "1"
 
 [extras]

src/BioSymbols.jl

@@ -99,8 +99,6 @@ export
     encoded_data,
     encode
 
-import Automa
-import Automa.RegExp: @re_str
 
 """
 The BioSymbol type is an abstract type that represents

src/aminoacid.jl

@@ -142,73 +142,6 @@ const threeletter_to_aa = Dict(
     "PYL" => AA_O, "SEC" => AA_U,
 )
 
-# Generate an amino acid parser.
-let
-    re = Automa.RegExp
-    function aapat(three, aa)
-        one = convert(Char, aa)
-        pat = re.alt(
-            re.cat([re.alt(lowercase(x), uppercase(x)) for x in three]...),
-            lowercase(one),
-            uppercase(one))
-        pat.actions[:exit] = [Symbol(three)]
-        return pat
-    end
-    aminoacids = [
-        ("ALA", AA_A),
-        ("ARG", AA_R),
-        ("ASN", AA_N),
-        ("ASP", AA_D),
-        ("CYS", AA_C),
-        ("GLN", AA_Q),
-        ("GLU", AA_E),
-        ("GLY", AA_G),
-        ("HIS", AA_H),
-        ("ILE", AA_I),
-        ("LEU", AA_L),
-        ("LYS", AA_K),
-        ("MET", AA_M),
-        ("PHE", AA_F),
-        ("PRO", AA_P),
-        ("SER", AA_S),
-        ("THR", AA_T),
-        ("TRP", AA_W),
-        ("TYR", AA_Y),
-        ("VAL", AA_V),
-        ("ASX", AA_B),
-        ("XLE", AA_J),
-        ("GLX", AA_Z),
-        ("XAA", AA_X),
-        ("PYL", AA_O),
-        ("SEC", AA_U),
-    ]
-    aa_term = re"\*"
-    aa_term.actions[:exit] = [:Term]
-    aa_gap = re"-"
-    aa_gap.actions[:exit] = [:Gap]
-    whitespaces = re"[ \t\r\n]*"
-    aminoacid = re.cat(
-        whitespaces,
-        re.alt(vcat((aapat(three, aa) for (three, aa) in aminoacids)..., aa_term, aa_gap)...),
-        whitespaces)
-    machine = Automa.compile(aminoacid)
-    actions = Dict(Symbol(three) => :(aa = $(aa)) for (three, aa) in aminoacids)
-    actions[:Term] = :(aa = AA_Term)
-    actions[:Gap]  = :(aa = AA_Gap)
-
-    ctx = Automa.CodeGenContext(checkbounds=false)
-    @eval function Base.tryparse(::Type{AminoAcid}, data::AbstractString)
-        $(Automa.generate_init_code(ctx, machine))
-        p_end = p_eof = sizeof(data)
-        aa = AA_INVALID
-        $(Automa.generate_exec_code(ctx, machine, actions))
-        if cs != 0
-            return nothing
-        end
-        return aa
-    end
-end
-
 function Base.tryparse(::Type{AminoAcid}, c::Char)
     @inbounds aa = c <= '\x7f' ? char_to_aa[Int(c)+1] : AA_INVALID
     if aa == AA_INVALID
@@ -218,6 +151,17 @@ function Base.tryparse(::Type{AminoAcid}, c::Char)
     end
 end
 
+function Base.tryparse(::Type{AminoAcid}, s::AbstractString)
+    if sizeof(s) == 1
+        tryparse(AminoAcid, first(s))
+    else
+        stripped = strip(s)
+        sizeof(stripped) == 1 && return tryparse(AminoAcid, first(stripped))
+        sizeof(stripped) == 3 || return nothing
+        return get(threeletter_to_aa, uppercase(stripped), nothing)
+    end
+end
+
 function Base.parse(::Type{AminoAcid}, c::Union{AbstractString,Char})
     aa = tryparse(AminoAcid, c)
     if aa === nothing