DeedleSaveCsvExample.fsx

open System
open System.IO
open System.Text.RegularExpressions
open System.Globalization

(*
For this example let's say we have data describing the processing of a batch of steps eg something like this:

BatchId
Start DateTime, End DateTime, Step Description
Start DateTime, End DateTime, Step Description
Start DateTime, End DateTime, Step Description
Start DateTime, End DateTime, Step Description

And repeating eg
1
2015-10-01 00:01:22, 2015-10-01 00:01:25, Step1
2015-10-01 00:01:29, 2015-10-01 00:03:12, Step2
2015-10-01 00:03:13, 2015-10-01 01:23:25, Step3

2
2015-10-01 00:02:22,2015-10-01 00:02:56,Step1
2015-10-01 00:03:29,2015-10-01 00:04:05,Step2
2015-10-01 00:04:13,2015-10-01 02:01:05,Step3
*)

let txt = File.ReadAllText("c:\\temp\\somedata.txt")
let rawTimes = Regex.Split(txt, "\n\r\n")

type RunTime = {
    Id: string;
    BatchDuration: TimeSpan;
    MaxStepDuration: TimeSpan;
    Step: string}

let stepParser (s:string) =
    let csvs = Regex.Split(s, ",")
    let start = DateTime.ParseExact(csvs.[0], "yyyy-MM-dd HH:mm:ss", CultureInfo.CurrentCulture)
    let ``end`` = DateTime.ParseExact(csvs.[1], "yyyy-MM-dd HH:mm:ss", CultureInfo.CurrentCulture)
    start, ``end``, ``end``.Subtract(start), csvs.[2]

let batchDurations = rawTimes
                    |> Array.map (fun rt -> 
                                            let splitLines = Regex.Split(rt, "\n",RegexOptions.Multiline)
                                            let id = splitLines.[0].Trim()
                                            let stepDurations = splitLines.[1..] |> Array.map stepParser
                                            let maxStepDurationTime, maxStepDurationStep = stepDurations 
                                                                                            |> Array.maxBy (fun (_,_,ts,_) -> ts) 
                                                                                            |> fun (_,_,ts,st) -> ts,st
                                            let batchStart = stepDurations |> Array.map (fun (s,_,_,_) -> s) |> Array.min
                                            let batchEnd = stepDurations |> Array.map (fun (_,e,_,_) -> e) |> Array.max
                                            {Id = id; BatchDuration = batchEnd.Subtract(batchStart); MaxStepDuration = maxStepDurationTime; Step = maxStepDurationStep.Trim() }
                                    )


#r @"packages\Deedle.1.2.4\lib\net40\Deedle.dll"
open Deedle

let frame = Frame.ofRecords batchDurations

frame.SaveCsv(@"c:\temp\data.csv")