git-all-find.ps1

[CmdletBinding()]
param(
  [Parameter(Mandatory=$true)]
  [ValidateScript({
    if (-Not ($_ | Test-Path) ) {
        throw "repos root path does not exist"
    }
    if (-Not ($_ | Test-Path -PathType Container) ) {
        throw "Repos root must be a folder"
    }
    return $true
  })]
  [System.IO.DirectoryInfo]$reposRoot,


  [Parameter(Mandatory=$true)]
  [ValidateScript({
    if (-Not ($_ | Test-Path) ) {
        throw "search strings path does not exist"
    }
    if (-Not ($_ | Test-Path -PathType Leaf) ) {
        throw "Search strings path must be a filename"
    }
    return $true
  })]
  [System.IO.FileInfo]$searchStringsFile,

  [Parameter(Mandatory=$false)]
  [String]$pathSpec,

  [Parameter(Mandatory=$false)]
  [String[]]$excludedRepos = @(),

  [Parameter(Mandatory=$false)]
  [int]$threads = 10
)

$pathSpec = $pathSpec.Replace("`"", "")
$searchFileInfo = (Get-Item $searchStringsFile)
$searchStringName = $searchFileInfo.Name.Replace($searchFileInfo.Extension, "")
$outputDir = (Join-Path $PSScriptRoot "output\$searchStringName")

if (!(Test-Path $outputDir)) {
    New-Item -Path $outputDir -ItemType Directory | Out-Null
}
else {
    Push-Location $outputDir
    Get-ChildItem | ForEach-Object { Remove-Item $_ -Force -ErrorAction SilentlyContinue }
    Pop-Location
}

$transcriptFile = (Join-Path $outputDir "$searchStringName.log")

try {
    Start-transcript $transcriptFile -ErrorAction Stop}
catch {
    Start-Transcript $transcriptFile
}

Write-Progress -Activity "Preprocessing" -Status "Reading search strings in $searchStringsFile"
$searchStrings = @(Get-Content -Path $searchStringsFile)

Write-Progress -Activity "Preprocessing" -Status "Scanning repos"

Push-Location $reposRoot
$repoDirs = Get-ChildItem -Recurse -Depth 2 -Force |
    Where-Object { $_.Mode -match "h" -and $_.FullName -like "*\.git"} |
    ForEach-Object { $_.Parent.FullName } |
    Where-Object { $excludedRepos -notcontains (Get-Item $_).Name }

Write-Progress -Activity "Preprocessing" -Completed

Write-Progress -Id 1 -Activity "Searching $($repoDirs.Length) repos for $($searchStrings.Length) terms" `
    -Status "Walking repo directories and running git grep"
$sw = [Diagnostics.Stopwatch]::StartNew()

$found = @{}
$repoIndex = 0
$pathSpecFinal = ""

if ($pathSpec) {
    $pathSpecFinal = " -- $pathSpec"
}

$repoDirs | ForEach-Object {
    $repoDir = $_
    $repoShortName = (Get-Item $repoDir).Name
    $repoIndex = $repoIndex + 1
    Write-Progress -Id 2 -ParentId 1 -Activity "Searching repo $repoIndex of $($repoDirs.Length) $repoDir" `
        -Status "Searching $repoDir" `
        -PercentComplete ($repoIndex/$repoDirs.Length*100)

    Push-Location $repoDir

    $repoMatchesFilename = (Join-Path $outputDir "$repoShortName.txt")
    "`n$repoDir ".PadRight(100, '*')
    $searchStringIndex = 0

    $searchStrings | Foreach-Object {
        $searchFor = $_
        Write-Progress -Id 3 -ParentId 2 -Activity "Searching $($searchStrings.Length) terms" `
            -Status "Searching for $searchFor" `
            -PercentComplete ($searchStringIndex/$searchStrings.Length*100)

        # Seem to have to use Invoke-Expression to get the git pathspec right here.
        # It'll work inline fine with -- '*.ext' '*.ext2' inline/hardcoded but as variable value isn't passed correctly
        $output = $null
        $expression = "`$output = git --no-pager grep -n -i --break --heading -A 0 -B 0 --threads $threads $searchFor $pathSpecFinal"
        Invoke-Expression $expression

        if ($output) {
            if (!$found.ContainsKey($searchFor)) {
                $found.Add($searchFor, 1)
            }
            else {
                $found[$searchFor] = $found[$searchFor] + 1
            }

            Add-Content $repoMatchesFilename $output
            "$($output.Trim())`n"
        }
        else {
            $notFound += $searchFor
        }

        $searchStringIndex = $searchStringIndex + 1
    }

    Pop-Location
}

$sw.Stop()
"Total search time: $($sw.Elapsed)"

$found.GetEnumerator() | sort-object Name | Format-Table -Autosize

$csvFilename = (Join-Path $outputDir "$searchStringName-found.csv")
$found.GetEnumerator() | sort-object Name | Export-Csv $csvFilename -NoTypeInformation
"Found results written to $csvFilename"

$notFoundStrings = $searchStrings `
    | Where-Object { $found.ContainsKey($_) -eq $false } `
    | Sort-Object `
    | ForEach-Object { [pscustomobject]@{ SearchTerm = $_ } }

$csvFilename = (Join-Path $outputDir "$searchStringName-not-found.csv")
$notFoundStrings | Export-Csv $csvFilename -NoTypeInformation
"Not found results written to $csvFilename"

Pop-Location