find-replace-404.ps1

# based on https://dba.meta.stackexchange.com/revisions/3e578461-07e9-46f2-b81d-a5f3e582171a/view-source

Set-Location ~/repos/git/sqlserver-kit
# https://stackoverflow.com/a/43737097/4709762
# $regexUrl='(http[s]?)(:\/\/)([^\s,]+)(?=")'

# https://stackoverflow.com/questions/3809401/what-is-a-good-regular-expression-to-match-a-url
$regexUrl = 'https?:\/\/(www\.)?[-a-zA-Z0-9@:%._\+~#=]{2,256}\.[a-z]{2,6}\b([-a-zA-Z0-9@:%_\+.~#?&//=]*)'

$allUrls = (Get-ChildItem -Recurse . | Select-String -Pattern $regexUrl).Matches.Value | Select-Object -Unique 

$counter=1
$urlStatusCsv = '~/Desktop/url-status.csv'

if(-not (Test-Path $urlStatusCsv)){
    Write-Host '"URL,StatusCode"' | Export-Csv $urlStatusCsv
}
$status = Import-Csv $urlStatusCsv
# $status | Export-Csv $urlStatusCsv

$allUrls = $allUrls | Where-Object {$status.URL -notcontains $PSItem} 

$status += foreach($url in $allUrls){
    Write-Verbose "Testing URL $counter/$($allUrls.Count): $url" -Verbose
    $statusCode = try {
        (Invoke-WebRequest $url -UseBasicParsing -Method Head).StatusCode
    } catch {
        $_.Exception.Response.StatusCode.value__ 
    }
    $counter++
    $row = [PSCustomObject]@{
        URL     = $url
        Status  = $statusCode
    }
    $row | Export-Csv $urlStatusCsv -Append
    $row
}

($status | where-object Status -ne 200).Count # 33
# $status | where-object Status -ne 200 | Export-csv ~/Desktop/bad-urls.csv
# Get-ChildItem . -Recurse | Select-String "https://StorageAccount.blob.core.windows.net/Container/FileName.xel" # 'example link'
<#
$updates =@'
[{
    "oldLink":"https://github.com/ktaranov/sqlserver-kit/blob/master/Stored_Procedure/sp_BlitzInMemoryOLTP.sql",
    "newLink":"https://github.com/ktaranov/sqlserver-kit/blob/master/Stored_Procedure/dbo.sp_BlitzInMemoryOLTP.sql",
    "commitMsg":"renamed at 79b1d55 \n https://github.com/ktaranov/sqlserver-kit/commit/79b1d551780aae28117e4bca835ea4d4c180f630"
},
{
    "oldLink":"",
    "newLink":"",
    "commitMsg":""
}]
'@ | ConvertFrom-Json 
$updates=$updates| Where-Object {-not [string]::IsNullOrWhiteSpace($_.commitMsg)}
# $updates.Count


function Repair-Link {
    param (
        [string]$oldLink,
        [string]$newLink,
        [string]$commitMsg
    )
    
    Get-ChildItem -Recurse | 
      Select-String $oldLink | 
      Select-Object -Unique -Property Filename | 
      ForEach-Object { 
          (Get-Content $_.FileName).replace($oldLink,$newLink) | 
          Set-Content -LiteralPath $_.FileName
        }

    git stage . 
    git commit -m "$commitMsg"
}

$updates | ForEach-Object {
    $o = $_.oldLink
    $n = $_.newLink
    $c = $_.commitMsg
    Repair-Link -oldLink $o -newLink $n -commitMsg $c -Verbose
}

git log --pretty=oneline --abbrev-commit

$status | where-object Status -ne 200 | Export-Csv ~/Desktop/los-links-con-carne.csv
#>

follow-up.ps1

Set-Location '~/Desktop/'

$counter=1
$urlStatusCsv = 'sqlserver-kit-bad-urls.csv'
$exportCsv = 'export.csv'

$status = Import-Csv $exportCsv
$testUrls = (Import-Csv $urlStatusCsv).URL | Where-Object {$status.URL -notcontains $PSItem } 

$testUrls.Count

$newStatus=foreach($url in $testUrls){
    Write-Verbose "Testing URL $counter/$($testUrls.Count): $url" -Verbose
    try {
        $statusCode = (Invoke-WebRequest $url -UseBasicParsing -TimeoutSec 10).StatusCode
    } catch {
        $statusCode = $_.Exception.Response.StatusCode.value__ 
    }
    $counter++
    $row = [PSCustomObject]@{
        URL     = $url
        Status  = $statusCode
    }
    $row | Export-Csv $exportCsv -Append
    $row
}
# $newStatus=Import-Csv 'sqlserver-kit-bad-urls.csv'
$newStatus | Where-Object Status -NotIn ("200","301") | Export-Csv $urlStatusCsv

SQL-Server-First-Responder-Kit-bad-urls.csv

sqlserver-kit-bad-urls.csv

BrentOzarULTD/SQL-Server-First-Responder-Kit#2098

examples like "http://FirstResponderKit.org." are bad hits on the regex but obvious enough to ignore