TextParse Support for GZip

gzip.jl

using GZip

MAX_WIDTH = 4

function Base.filesize(io::GZip.GZipStream)
    f = open(io.name, "r")
    seekend(f)
    skip(f, -4)
    n = Int64(first(reinterpret(Int32, read(f, 4))))
    close(f)
    n
end

type StringIO{T <: IO} <: AbstractString
    io::T
    size::Int64
end
StringIO(io::IO) = StringIO(io, filesize(io))
function Base.print(io::IO, s::StringIO)
    print(io, "StringIO(", s.io, ")")
end
Base.show(io::IO, x::StringIO) = print(io, x)
Base.start{T}(s::StringIO{T}) = 1
Base.endof{T}(s::StringIO{T}) = s.size
function Base.next{T}(s::StringIO{T}, i::Int)
    #p = position(s.io)
    seek(s.io, max(i - 1, 0))
    x = read(s.io, MAX_WIDTH)
    #seek(s.io, p)
    c, j = next(String(x), 1)
    (c, i + j - 1)
end
function Base.prevind{T}(s::StringIO{T}, i::Int)
    #p = position(s.io)
    seek(s.io, max(i - 1 - MAX_WIDTH, 0))
    x = read(s.io, 8)
    #seek(s.io, p)
    j = prevind(String(x), min(i, MAX_WIDTH + 1))
    i + j - min(i, MAX_WIDTH + 1)
end
function Base.getindex{T}(s::StringIO{T}, r::UnitRange{Int})
    #p = position(s.io)
    i, j = prevind(s, first(r)), nextind(s, last(r))
    seek(s.io, max(i - 1, 0))
    x = read(s.io, j - i - 1)
    #seek(s.io, p)
    String(x)
end

using PooledArrays
using TextParse

@inline function TextParse.alloc_string(str::StringIO, r::TextParse.StrRange)
    str[(r.offset + 2):(r.length + r.offset + 1)]
end
@inline function TextParse.nonallocating_setindex!{T}(pa::PooledArrays.PooledArray{T}, i, rng::TextParse.StrRange, str::StringIO)
    wstr = str[(rng.offset + 2):(rng.length + rng.offset + 1)]
    pool_idx = findfirst(pa.pool, wstr)
    if pool_idx <= 0
        # allocate only here.
        val = convert(T, wstr)
        pool_idx = PooledArrays.unsafe_pool_push!(pa, val)
    end

    pa.refs[i] = pool_idx
end

function TextParse.csvread(file::StringIO, delim=','; kwargs...)
    TextParse._csvread(file, delim; kwargs...)
end

s = StringIO(Gzip.open("foo.txt.gz"))
csvread(s, '\t')