Improve parsing logic more

src/FASTX.jl

@@ -96,6 +96,14 @@ const UTF8 = Union{AbstractVector{UInt8}, String, SubString{String}}
     error(String(take!(buf)))
 end
 
+function truncate(s::AbstractString, len::Integer)
+    if length(s) > len
+        return string(String(collect(Iterators.take(s, len - 1))), '…')
+    else
+        return s
+    end
+end
+
 include("fasta/fasta.jl")
 include("fastq/fastq.jl")
 
@@ -132,6 +140,7 @@ function FASTA.Record!(record::FASTQ.Record)
 end
 
 Base.parse(::Type{T}, s::AbstractString) where {T <: Record} = parse(T, String(s))
+Base.parse(::Type{T}, s::Union{String, SubString{String}}) where {T <: Record} = parse(T, codeunits(s))
 
 "Get the indices of `data` that correspond to sequence indices `part`"
 function seq_data_part(record::Record, part::AbstractUnitRange)

src/fasta/fasta.jl

@@ -20,7 +20,7 @@ import TranscodingStreams: TranscodingStreams, TranscodingStream, NoopStream
 # Trivial use, I only use it here because it's a dep of Automa anyway.
 # Can be removed with no big problems
 using ScanByte: memchr, ByteSet
-import ..FASTX: identifier, description, sequence, UTF8, seqlen, throw_parser_error
+import ..FASTX: identifier, description, sequence, UTF8, seqlen, throw_parser_error, truncate
 
 include("record.jl")
 include("readrecord.jl")

src/fasta/record.jl

@@ -68,19 +68,27 @@ function Base.empty!(record::Record)
     return record
 end
 
-function Base.parse(::Type{Record}, data::UTF8)
+function Base.parse(::Type{Record}, data::AbstractVector{UInt8})
+    # Error early on empty data to not construct buffers
+    isempty(data) && throw(ArgumentError("Cannot parse empty string as FASTA record"))
+
     record = Record(Vector{UInt8}(undef, sizeof(data)), 0, 0, 0)
-    stream = NoopStream(IOBuffer(data))
-    _, _, found = readrecord!(stream, record, (1, 1))
-    found || throw(ArgumentError("invalid FASTA record"))
-    next = false
-    try
-        _, _, found = readrecord!(stream, Record(), (1, 1))
-        next = found
-    catch e
-        next = true
+    stream = NoopStream(IOBuffer(data), bufsize=sizeof(data))
+    cs, _, found = readrecord!(stream, record, (1, 1))
+
+    # If found is not set, then the data terminated early
+    found || throw(ArgumentError("Incomplete FASTA record"))
+
+    # In this case, the machine ran out of data exactly after one record
+    p = stream.state.buffer1.bufferpos
+    p > sizeof(data) && iszero(cs) && return record
+
+    # Else, we check all trailing data to see it contains only \r\n
+    for i in p-1:sizeof(data)
+        if !in(data[i], (UInt8('\r'), UInt8('\n')))
+            throw(ArgumentError("Invalid trailing data after FASTA record"))
+        end
     end
-    next && throw(ArgumentError("Multiple FASTA records"))
     return record
 end
 
@@ -137,15 +145,7 @@ function Base.show(io::IO, record::Record)
     print(io, summary(record), ':')
     println(io)
     println(io, "description: \"", description(record), '"')
-    print(io,   "   sequence: \"", truncate(sequence(String, record), 40), '"')
-end
-
-function truncate(s::String, len::Integer)
-    if length(s) > len
-        return string(String(collect(Iterators.take(s, len - 1))), '…')
-    else
-        return s
-    end
+    print(io,   "   sequence: \"", truncate(sequence(record), 40), '"')
 end
 
 function memcmp(p1::Ptr, p2::Ptr, n::Integer)

src/fastq/fastq.jl

@@ -13,7 +13,7 @@ import BioGenerics: BioGenerics
 import BioGenerics.Automa: State
 import StringViews: StringView
 import TranscodingStreams: TranscodingStreams, TranscodingStream, NoopStream
-import ..FASTX: identifier, description, sequence, UTF8, seqlen, throw_parser_error
+import ..FASTX: identifier, description, sequence, UTF8, seqlen, throw_parser_error, truncate
 
 include("quality.jl")
 include("record.jl")

src/fastq/record.jl

@@ -96,19 +96,27 @@ function Base.empty!(record::Record)
     return record
 end
 
-function Base.parse(::Type{Record}, data::UTF8)
+function Base.parse(::Type{Record}, data::AbstractVector{UInt8})
+    # Error early on empty data to not construct buffers
+    isempty(data) && throw(ArgumentError("Cannot parse empty string as FASTQ record"))
+
     record = Record(Vector{UInt8}(undef, sizeof(data)), 0, 0, 0)
-    stream = NoopStream(IOBuffer(data))
-    _, _, found = readrecord!(stream, record, (1, 1))
-    found || throw(ArgumentError("invalid FASTQ record"))
-    next = false
-    try
-        _, _, found = readrecord!(stream, Record(), (1, 1))
-        next = found
-    catch e
-        next = true
+    stream = NoopStream(IOBuffer(data), bufsize=sizeof(data))
+    cs, _, found = readrecord!(stream, record, (1, 1))
+
+    # If found is not set, then the data terminated early
+    found || throw(ArgumentError("Incomplete FASTQ record"))
+
+    # In this case, the machine ran out of data exactly after one record
+    p = stream.state.buffer1.bufferpos
+    p > sizeof(data) && iszero(cs) && return record
+
+    # Else, we check all trailing data to see it contains only \r\n
+    for i in p-1:sizeof(data)
+        if !in(data[i], (UInt8('\r'), UInt8('\n')))
+            throw(ArgumentError("Invalid trailing data after FASTQ record"))
+        end
     end
-    next && throw(ArgumentError("Multiple FASTQ records"))
     return record
 end
 
@@ -199,9 +207,9 @@ end
 
 function Base.show(io::IO, record::Record)
     println(io, "FASTQ.Record:")
-    println(io, "  description: ", description(record))
-    println(io, "     sequence: ", sequence(String, record))
-    print(io,   "      quality: ", quality(String, record))
+    println(io, "  description: \"", description(record))
+    println(io, "     sequence: \"", truncate(sequence(record), 40), '"')
+    print(io,   "      quality: \"", truncate(quality(record), 40), '"')
 end
 
 function memcmp(p1::Ptr, p2::Ptr, n::Integer)

test/fasta/record.jl

@@ -100,6 +100,7 @@ end
 
     # Trailing extra record
     @test_throws Exception parse(Record, ">A\nT\n>G\nA\n")
+    @test_throws Exception parse(Record, ">A\nT\n>")
     @test parse(Record, ">A\nT\n\r\n\r\n") isa Record
 end
 

test/fastq/record.jl

@@ -117,6 +117,7 @@ end
 
     # Trailing extra record
     @test_throws Exception parse(Record, "@A\nT\n+\nJ\n@B\nU\n+\nK")
+    @test_throws Exception parse(Record, "@A\nT\n+\nJ\n@B\nU\n")
     @test parse(Record, "@A\nT\n+\nJ\n\r\n") isa Record
 end