Export-WhisperCsvToTxt.ps1

[CmdletBinding()]
param (
    [string]$CsvPath = "Z:\BossmanJack\Kick\Transcripts",
    [string]$TextOutput = "Z:\BossmanJack\Kick\TranscriptsTxt",
    [switch]$Overwrite = $false
)

# This script exports the CSV files generated by Whispercpp to a nice, legible
# format similar to what's printed to the console while Whisper is running as
# Whispercpp's own txt format excludes timestamps for whatever reason

function Get-TsText([TimeSpan]$TS)
{
    # e.g. 01:15:07.100
    return "$($TS.Hours.ToString('00')):$($TS.Minutes.ToString('00')):$($TS.Seconds.ToString('00')).$($TS.Milliseconds.ToString('000'))"
}

# The cast to object[] is to ensure it's always an array,
# even if only a single item is returned
[object[]]$CsvFiles = Get-ChildItem -File -Recurse -Include "*.csv" -Path $CsvPath
Write-Host "$($CsvFiles.Count) files to process" -ForegroundColor Magenta

$i = 0
$Count = $CsvFiles.Count
foreach ($CsvFile in $CsvFiles)
{
    $i++
    Write-Progress -Activity "Processing CSVs" -Status "Processing $($CsvFile.FullName) ($i of $Count)" -PercentComplete ($i / $Count * 100)
    #Write-Host "Processing $($CsvFile.Name) ($i of $Count)" -ForegroundColor Cyan
    $CsvData = Import-Csv -LiteralPath $CsvFile.FullName
    $Lines = @("VOD at https://mines.expert/BossmanJack/Kick/$($CsvFile.BaseName).mp4")
    $TsTextPath = Join-Path $TextOutput "$($CsvFile.BaseName).txt"
    if (-not $Overwrite -and [System.IO.File]::Exists($TsTextPath))
    {
        #Write-Host "Skipping $TsTextPath as it already exists" -ForegroundColor Yellow
        continue
    }
    foreach ($Row in $CsvData)
    {
        $Start = [TimeSpan]::FromMilliseconds($Row.start)
        $End = [TimeSpan]::FromMilliseconds($Row.end)
        $Lines += "[$(Get-TsText $Start) -> $(Get-TsText $End)]: $($Row.text)"
    }

    # Out-File will write a BOM to the file, which seemingly serves to make the
    # file an unreadable mess 99% of the time.
    # Out-File is also buggy when dealing with very weird file names
    [System.IO.File]::WriteAllLines($TsTextPath, $Lines)
    Write-Host "$($CsvFile.Name) done" -ForegroundColor Green
}

Write-Host "Done" -ForegroundColor Green