Julia: arrow，一种革命性的数据格式

准备了一个csv文件，大约约59万行，14列，大小约61M，格式如下：

https://blog.csdn.net/wowotuo/article/details/109828399

上代码。

一、写入arrow文件

1、相关库

using DataFrames
using CSV
using Arrow;

2、csv => dataframe

csv_path = s"C:\Users\songroom\Desktop\test.csv"
println("csv => DataFrame: ")
@time df =  CSV.File(csv_path) |> DataFrame;

3、dataframe => arrow文件

println("df => arrow file ")
arrow_path = "C:\\Users\\songroom\\Desktop\\test.arrow"
io = open(arrow_path, "r+"); # 不能是w,或w+，否则会有内存泄露
try
    @time Arrow.write(io, df)
finally
    close(io)
end

上面，完成arrow文件的生成。

二、读出arrow文件

println("read arrow file ")
arrow_path = "C:\\Users\\songroom\\Desktop\\test.arrow"
@time df = DataFrame(Arrow.Table(arrow_path))

三、整体代码和输出

using DataFrames
using CSV
using Arrow;

csv_path = "C:\\Users\\songroom\\Desktop\\test.csv"
arrow_path = "C:\\Users\\songroom\\Desktop\\test.arrow"
@time df = get_dataframe_from_csv(csv_path);
@time write_arrow_file(df,arrow_path)
@time df = read_arrow_file(arrow_path)

function get_dataframe_from_csv(csv_path:: String)
   df =  CSV.File(csv_path) |> DataFrame;
   return df
end

function write_arrow_file(df:: DataFrame,arrow_path::String)
    println("df => arrow file ")
    open(arrow_path, "r+") do io # 不能是w,或w+，否则会有内存泄露
        Arrow.write(io, df)
    end
end

function read_arrow_file(arrow_path::String)
    println("read arrow file ")
    df = DataFrame(Arrow.Table(arrow_path))
    return df
end

输出：

julia> @run test
  0.313795 seconds (590.24 k allocations: 163.234 MiB, 12.18% gc time)
df => arrow file
  0.132979 seconds (594.55 k allocations: 35.252 MiB)
read arrow file
  0.000664 seconds (900 allocations: 57.250 KiB)
read_arrow_file (generic function with 1 method)

四、其它

arrow文件和CSV文件比较：
速度绝对快，但arrow文件是一个吃硬盘的。