$tesseractPath = "tesseract.exe"
$inputImage = "input.jpg"

$tsvData = & $tesseractPath $inputImage stdout -l eng --psm 6 `
    -c tessedit_char_whitelist=" abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789[]{}()=+-*/\|&$%^#@!~`';:,<>.?_" `
    tsv

# Parse as tab separated values
$lines = $tsvData -split "`r?`n"
$headers = $lines[0] -split "`t"
$records = $lines[1..($lines.Length - 1)] | Where-Object { $_.Trim() -ne "" } | ForEach-Object {
    $values = $_ -split "`t"
    if ($values.Length -eq $headers.Length) {
        $obj = @{}
        for ($i = 0; $i -lt $headers.Length; $i++) {
            $obj[$headers[$i]] = $values[$i]
        }
        [PSCustomObject]$obj
    }
}

# Level 5 is "word". Would have been nice to use level 4 "line" but that has no text
# So, instead, we grab level 5 "line"
$words = $records | Where-Object {
    $_.level -eq '5' -and $_.text -match '\S'
}

# Now group them by line number to get the text by line
$groupedLines = $words | Group-Object line_num

# Assume the first left position > 0 is the size of the indents
$minLeft = ($words | Where-Object { [int]$_.left -gt 0 } | Measure-Object -Property left -Minimum).Minimum

# Set tab width to that minimum value
$tabWidth = [int]$minLeft



# Now create your indented text
foreach ($line in $groupedLines) {
    $sortedWords = $line.Group | Sort-Object { [int]$_.left }
    $firstLeft = [int]$sortedWords[0].left
    $indent = [math]::Round($firstLeft / $tabWidth)
    $textLine = ($sortedWords | ForEach-Object { $_.text }) -join ' '
    Write-Output (' ' * $indent + $textLine)
}