vgaltes

let numberOfAuthors =
    totalCommits
    |> Array.groupBy(fun c -> c.CommitInfo.Author)
    |> Array.length
    
----------------------------------------------------
val numberOfAuthors : int = 22

vgaltes

let numberOfEntitiesChanged =
    totalCommits
    |> Array.collect(fun c -> c.Files)
    |> Array.length
    
-----------------------------------------------
val numberOfEntitiesChanged : int = 37765

vgaltes

let numberOfEntities = 
    totalCommits
    |> Array.collect(fun c -> c.Files)
    |> Array.groupBy(fun f -> f.FileName)
    |> Array.length
    
--------------------------------------------
val numberOfEntities : int = 8119

vgaltes

let numberOfCommits = totalCommits |> Array.length

-------------------------------------------
val numberOfCommits : int = 5061

vgaltes

type CommitInfo = {Hash : string; Author : string; TimeStamp : DateTime; Message : string}
type CommittedFile = {LinesAdded: int option; LinesDeleted: int option; FileName: string}
type Commit = {CommitInfo: CommitInfo; Files: CommittedFile[]}

val totalCommits : Commit []

vgaltes

#r "../packages/FSharp.Data.2.2.5/lib/net40/FSharp.Data.dll"
open System
open System.IO
open FSharp.Data

[<Literal>]
let lineBreak = "\r\n"

type CommitInfo = {Hash : string; Author : string; TimeStamp : DateTime; Message : string}
type CommittedFile = {LinesAdded: int option; LinesDeleted: int option; FileName: string}

vgaltes

let totalCommits = 
    commits
    |> Array.map extractCommitInfo

vgaltes

let extractCommitInfo (commit:string) =
    let isCommitInfoLine (line: string) =
        line.StartsWith("[")

    let extractCommitedFilesInfo c = 
        let commitFileLine = CommitLineCsv.Parse(c).Rows |> Seq.head
        {LinesAdded = commitFileLine.LinesAdded; LinesDeleted = commitFileLine.LinesDeleted; FileName = commitFileLine.FileName}

    let commitLines = commit.Split([|lineBreak|], StringSplitOptions.RemoveEmptyEntries)    
    let commitInfoLine = commitLines |> Array.takeWhile(isCommitInfoLine) |> Array.last

vgaltes

{CommitInfo = commitInfo; Files = fileRows}

vgaltes

type CommitLineCsv = CsvProvider<"LinesAdded\tLinesDeleted\tFile", HasHeaders = false, Schema = "LinesAdded(int option),LinesDeleted(int option),FileName">

let extractCommitedFilesInfo c = 
  let commitFileLine = CommitLineCsv.Parse(c).Rows |> Seq.head
  {LinesAdded = commitFileLine.LinesAdded; LinesDeleted = commitFileLine.LinesDeleted; FileName = commitFileLine.FileName}
  
let fileRows = fileLines |> Array.map extractCommitedFilesInfo