public class MailBodyParser
{
    //Two patterns / languages will be supported in this first Version (EN/DE)
    public static final String[] FROM = {"Von", "From"};
    public static final String[] TO = {"An", "To"};
    public static final String[] SENT = {"Gesendet", "Sent"};
    public static final String[] SUBJECT = {"Betreff", "Subject"};
    public static final String[] CC = {"Cc", "Cc"};
    public static final String[] IMPORTANCE = {"Wichtigkeit","Importance"};

    public static MailItemBody parseBody( MailBody body )
    {
        String plainText = body.getBodyPlain();
        String regex = buildPlainRegEx();
        String[] bodyParts = plainText.split(regex);

        //RegEx for fixing image tags
        String regexImage = "IMAGE (.*?)>";
        String regexLink = "HYPERLINK \"(.*?)\"([^\\s]*)";
        String regexContentId = "cid:(.*)";
        Pattern pattern = Pattern.compile(regexImage);
        Pattern contentIdPattern = Pattern.compile(regexContentId);
        Pattern hyperlinkPattern = Pattern.compile(regexLink);
        Matcher matcher;

        //Trim the parts and replace things
        for (int i = 0; i < bodyParts.length; i++) {

            //Check if outlook mail or not ==> apply with outlook mails, else
            if( body.getBodyHtml().contains("urn:schemas-microsoft-com:office:word") )
            {
                //Paragraph
                bodyParts[i] = bodyParts[i].replaceAll("\r\n\r\n \r\n\r\n", "\n\n");
                //New Lines
                bodyParts[i] = bodyParts[i].replaceAll("\r\n\r\n", "\n");

                bodyParts[i] = bodyParts[i].replaceAll("\r\n", "\n");
                bodyParts[i] = bodyParts[i].trim();
            }
            else
            {
                bodyParts[i] = bodyParts[i].replaceAll("\r\n", "\n");
                bodyParts[i] = bodyParts[i].trim();
            }

            //Find images and fix indices
            matcher = pattern.matcher(bodyParts[i]);
            while( matcher.find() )
            {
                String replaceWhat = matcher.group(0);
                String replaceWith = null;
                String attributes = matcher.group(1);
                String[] singleAttributes = attributes.split(" ");

                for (int j = 0; j < singleAttributes.length; j++) {
                    if( singleAttributes[j].startsWith("src=") )
                    {
                        Matcher m = contentIdPattern.matcher(singleAttributes[j]);
                        if( m.find() )
                        {
                            String match = m.group(1);
                            //Can occur if src is surrounded by ""
                            if( match.endsWith("\"") )
                                match = match.replace("\"", "");

                            replaceWith = "IMAGE<!" + getAttachmentId( body, match ) + "!>";
                        }
                    }
                }

                if( replaceWith == null )
                    replaceWith = "--Image could not be restored from mail--";

                bodyParts[i] = bodyParts[i].replace(replaceWhat, replaceWith);
            }

            matcher = hyperlinkPattern.matcher(bodyParts[i]);
            while( matcher.find() )
            {
                String replaceWith = "HYPERLINK<!" + matcher.group(1) + "||" + matcher.group(2) + "!>";
                bodyParts[i] = bodyParts[i].replace(matcher.group(0), replaceWith);
            }
        }

        //anonymize the html body
        String htmlText = body.getBodyHtml();

        //Loop over text and find the parts to replace
        pattern = Pattern.compile(buildHtmlRegEx());
        matcher = pattern.matcher(htmlText);

        /*
         * TODO: ADMIN OPTION
        while(matcher.find())
        {
            String from = matcher.group(1);
            String to   = matcher.group(2);
            String cc   = null;
            if( matcher.groupCount() > 2 )
                cc = matcher.group(3);

            htmlText = htmlText.replaceAll(from, "n/a");
            htmlText = htmlText.replaceAll(to, "n/a");
            if( cc != null )
                htmlText = htmlText.replaceAll(cc, "n/a");
        }
         */
        MailItemBody ret = new MailItemBody();
        ret.setAttachmentsIds(toIntArray(body.getAttachmentIds()));
        ret.setBodyParts(bodyParts);
        ret.setAnonymizedHtmlBody(htmlText);
        return ret;
    }

    private static String getAttachmentId( MailBody body, String contentId )
    {
        String[] contentIds = body.getAttachmentContentIds();
        for( int i = 0; i < contentIds.length; i++ )
        {
            if( contentId.equals(contentIds[i]) )
                return body.getAttachmentIds()[i];
        }

        return null;
    }

    private static int[] toIntArray( String[] array )
    {
        if( array == null )
            return new int[0];

        int[] ret = new int[array.length];
        for (int i = 0; i < ret.length; i++) {
            ret[i] = Integer.parseInt(array[i]);
        }
        return ret;
    }

    private static String buildPlainRegEx()
    {
        //Example: (Von|From):.*?\n(Gesendet|Sent):.*?\n(An|To):.*?\n(Betreff|Subject):.*?\n
        String regex;
        String fromTerm = "(?:";
        String sentTerm = "(?:";
        String toTerm = "(?:";
        String ccTerm = "(?:(?:";
        String subjectTerm = "(";
        String importanceTerm = "(?:(?:";
        String sep = ":.*?(?:\r)?\n";

        for (int i = 0; i < FROM.length; i++)
        {
            fromTerm += FROM[i] + ( i < (FROM.length - 1) ? "|":")" );
            sentTerm += SENT[i] + ( i < (SENT.length - 1) ? "|":")" );
            toTerm += TO[i] + ( i < (TO.length - 1) ? "|":")" );
            ccTerm += CC[i] + ( i < (CC.length - 1) ? "|":"):.*?(?:\r)?\n)?" );
            subjectTerm += SUBJECT[i] + ( i < (SUBJECT.length - 1) ? "|":")" );
            importanceTerm += IMPORTANCE[i] + ( i < (IMPORTANCE.length - 1) ? "|":"):.*?(?:\r)?\n)?" );
        }

        regex = "(?s)"+ fromTerm + sep + sentTerm + sep + toTerm + sep + ccTerm + subjectTerm + sep + importanceTerm;
        return regex;
    }


    private static String buildHtmlRegEx()
    {
        //(?s)(Von|From):.*?(?:<.*?>)*(.*?)(?:<.*?>)*(Gesendet|Sent).*?(An|To):(?:<.*?>)*(.*?)<.*?>
        String regex;
        String fromTerm = "(?:";
        String sentTerm = "(?:";
        String toTerm = "(?:";
        String ccTerm = "(?:(?:";
        String subjectTerm = "(?:";
        String dotAllFlag = "(?s)";

        for (int i = 0; i < FROM.length; i++)
        {
            fromTerm += FROM[i] + ( i < (FROM.length - 1) ? "|":")" );
            sentTerm += SENT[i] + ( i < (SENT.length - 1) ? "|":")" );
            toTerm += TO[i] + ( i < (TO.length - 1) ? "|":")" );
            subjectTerm += SUBJECT[i] + ( i < (SUBJECT.length - 1) ? "|":")" );
            ccTerm += CC[i] + ( i < (CC.length - 1) ? "|":"):(?:<.*?>)*(.*?)<.*?>)?" );
        }

        regex = dotAllFlag + fromTerm + ":(?:<.*?>)+(.*?)(?:<.*?>)+" + sentTerm +
                ":.*?" + toTerm + ":(?:<.*?>)+(.*?)(?:<.*?>)+" + ccTerm + ".*?" + subjectTerm;
        return regex;
    }
}