An attempt at faster Julia IO

dlm2dict.jl

# Chris DuBois
# September 2012

# Inspired by: http://vijayinterviewquestions.blogspot.com/2007/07/write-your-own-c-program-to-implement.html
function atoi(s)
    i = 0
    L = strlen(s)

    # skip preceding whitespaece
    a = 1
    while s[a] == ' '
        a+=1
    end

    j = a
    while j <= L
        # skip trailing non digits
        if s[j] < '0' || s[j] > '9'
            j = L + 1
        else
            i = (i << 3) + (i << 1) + s[j] - '0'
            j += 1
        end
    end
    return i
end
@assert atoi("390709") == 390709
@assert atoi(" 390709") == 390709
@assert atoi(" 390709 ") == 390709
@assert atoi(" 390709 lxkjd") == 390709

function atof(s)
    out = Array(Float64,1)
    ccall(:jl_strtod, Int32, (Ptr{Uint8},Ptr{Float64}), s, out)
    out[1]
end
@assert atof("1.234") == 1.234
@assert atof(" 1.234") == 1.234
@assert atof(" 1.234 ") == 1.234
@assert atof(" 1 ") == 1.0

# Read a delimited file with known types in each column.
# Return a Dict with an element for every column.
# TODO: how to handle empty final lines
function dlm2dict(filename, types, dlm::Char, header)
    d = Dict()
    io = open(filename)
    if header
        cn = split(readline(io),dlm)
    else
        cn = [strcat("column_",x) for x in 1:length(types)]
    end
    for i in 1:length(cn)
        d[cn[i]] = Array(types[i], 0)
    end

    for line in each_line(io)
        pos = 1
        for i in 1:length(cn)
           if types[i] == Int64
               push(d[cn[i]], atoi(line[pos:end]))
               pos = strchr(line, dlm, pos) + 1
           elseif types[i] == Float64
               push(d[cn[i]], atof(line[pos:end]))
               pos = strchr(line, dlm, pos) + 1
           else
               e = strchr(line, dlm, pos)
               push(d[cn[i]], line[pos:e])
               pos = e + 1
           end
        end
    end
    return d
end