Converts PDFs to JPGs and OCRed text with imagemagick and tesseract.

raw-ocr.ps1

<#
 Processes raw source pdfs, producing per page: 1 txt, 1 hocr, 1 jpg.
 Requires imagemagick w/ ghostscript, tesseract.
 
 Subscripts: pdf2png.ps1, ocr.ps1, hocr.ps1, png2jpg.ps1
#>

param(
    [string]$indir = ".",
    [string]$outbase = $indir
)

function WebSafe([string]$s) {
    return $s.ToLowerInvariant().Replace(" ", "-").Replace("&", "-")
}

$files = ls "$indir\*.*" -include *.pdf

foreach ($file in $files) {

    # get metadata from filename
    $split = $file.BaseName.Split("_")
    $uni = $split[0]
    $col = $split[1]
    $title = $split[2]

    # convert pdf to png per page
    & .\pdf2png.ps1 -in $file.FullName

    # convert png to txt
    & .\ocr.ps1 -ext "png" -indir "$indir\png" -outdir "$indir\txt"

    # convert png to xml (hocr)
    & .\hocr.ps1 -ext "png" -indir "$indir\png" -outdir "$indir\xml"
    # rename hocr extension .html => .xml
    ls "$indir\xml\*.html" | foreach-object { ren $_.FullName  ((join-path $_.DirectoryName $_.BaseName) + ".xml")}

    # create path for jpgs
    $uniSafe = WebSafe($uni)
    $colSafe = WebSafe($col)
    $titleSafe = WebSafe($title)
    $jpgdir = "$outbase\$uniSafe\$colSafe\$titleSafe"
    if (!(test-path $jpgdir)) {
        mkdir $jpgdir
    }

    # convert .png to .jpg
    & .\png2jpg.ps1 -size 1000 -indir "$indir\png" -outdir $jpgdir

    # get page count
    $pagecount = (ls "$jpgdir\*.*" -include *.jpg).Count

    # write metadata to manifest.xml
    $manifest = get-content ("$indir\manifest.xml")
    $manifest | foreach-object {
        $_  -replace '{UNIVERSE}', $uni `
            -replace '{COLLECTION}', $col `
            -replace '{TITLE}', $title `
            -replace '{PAGECOUNT}', $pagecount `
            -replace '{OCRTYPE}', 'hocr'
    } | set-content "$indir\manifest.xml"

    # clean up
    rm "$indir\png" -recurse

}