Takes a tab-delimited csv input file (tsv) produced by ner.ps1, and outputs a text file for each category found. A single category may be named, in which case a single output text file is created. If no output file is provided, results are written to the console instead.

parsetsv.ps1

<#
    parse tsv
    Categories: person, location, organization, misc, money, percent, date, time (depending on classifier used to produce the tsv)
    Outfile: Results written to console if outfile not provided. If all categories, outfile is used as a filename template.

#>

param(
    [Parameter(Mandatory=$true,Position=0)]
    [string]$file,
    [Parameter(Mandatory=$false,Position=1)]
    [string]$outfile,
    [Parameter(Mandatory=$false,Position=2)]
    [string]$category

)

$tsv = import-csv $file -delimiter "`t" -header "Entity","Category"
$outinfo = new-object System.IO.FileInfo($ExecutionContext.SessionState.Path.GetUnresolvedProviderPathFromPSPath($outfile))

# named category
if ($category -ne '')  {
    $entities = $tsv | where-object {$_.Category -eq $category} | sort -unique -property Entity
    if ($outfile) {
        if (!(test-path $outinfo.DirectoryName)) {
            mkdir $outinfo.DirectoryName
        }
        $stream = [System.IO.StreamWriter] $outinfo.FullName
        $entities | foreach-object {
            $stream.WriteLine($_.Entity)
        }
        $stream.close()
    }
    else {
        echo $entities
    }

    exit
}

# all categories
$categories = $tsv | where-object {$_.Category -ne ''} | sort -unique -property Category | select-object -expandproperty Category -unique
foreach ($cat in $categories) {
    $entities = $tsv | where-object {$_.Category -eq $cat} | sort -unique -property Entity
    if ($outfile) {
        if (!(test-path $outinfo.DirectoryName)) {
            mkdir $outinfo.DirectoryName
        }
        $stream = [System.IO.StreamWriter] ('{0}\{1}.{2}{3}' -f $outinfo.DirectoryName, $outinfo.BaseName, $cat, $outinfo.Extension)
        $entities | foreach-object {
            $stream.WriteLine($_.Entity)
        }
        $stream.close()
    }
    else {
        echo $entities
    }
}