Parse Australian Immunisation Register Immunisation History Statement from PDF to CSV

parse-airihs.groovy

/*

"C:\Program Files\gs\gs10.03.1\bin\gswin64c.exe" -dBATCH -dNOPAUSE -sDEVICE=txtwrite -r300 -dUseCropBox -sOutputFile="{3}\{2}.txt" "{1}\{2}"
	
{1}	Process Property 'PDFs Location' from 'Vaccinations'
{2}	Document Property - Disk - File Name
{3}	Process Property 'Text Location' from 'Vaccinations'

*/

// Read "{3}\{2}.txt" into outData

def contents = outData.toString();
def matchesEmployee = contents =~ /(?m)For:\r\n(.+)/;
def matchesVaccs = contents =~ /([0-9][0-9]\s[a-zA-Z]{3}\s[0-9]{4})\t([a-zA-Z0-9\- ]+)\t([a-zA-Z \-]+)/;
def matchesAsAt = contents =~ /(?m)As at:\r\n([0-9][0-9]\s[a-zA-Z]+\s[0-9]{4})/;
def matchesDOB = contents =~ /(?m)Date of birth:\r\n([0-9][0-9]\s[a-zA-Z]+\s[0-9]{4})/;
def lastDate;
def dateGiven;

// DEBUG
//    contents += matchesEmployee.size()+LINE_SEPARATOR;
//    contents += matchesAsAt.size()+LINE_SEPARATOR;
//    contents += matchesDOB.size()+LINE_SEPARATOR;

def output = [["AsAt", "Employee", "DOB", "Date", "Immunisation", "Brand"]];
for (matchesVacc in matchesVaccs) { 
    if (matchesEmployee.size()
      && matchesVacc.size()
      && matchesAsAt.size()
      && matchesDOB.size()
    ) {
        if (matchesVacc[1] == '00 MMM 0000') {
            dateGiven = lastDate;
        } else {
            dateGiven = matchesVacc[1];
        }
        output.push([
            matchesAsAt[0][1],
            matchesEmployee[0][1],
            matchesDOB[0][1],
            dateGiven,
            matchesVacc[2],
            matchesVacc[3].trim()
        ]);
        lastDate = dateGiven;
    } else {}
}

contents = "";

for (line in output) {
    contents += "\"" + line.join("\",\"") + "\"" + LINE_SEPARATOR;
}