D-Net

package eu.dnetlib.data.mapreduce.hbase.dataexport;

import eu.dnetlib.miscutils.functional.xml.DnetXsltFunctions;

import org.apache.commons.lang3.StringUtils;

import org.dom4j.Document;

import org.dom4j.tree.DefaultText;

import java.text.ParseException;

import java.util.Map;

import java.util.Objects;

public class RecordFilter {

    private Map<String, String> criteria;

    private String yearXpath;

    private int fromYear;

    private int toYear;

    public RecordFilter(Map<String, String> criteria, String yearXpath, int fromYear, int toYear) {

        this.criteria = criteria;

        this.yearXpath = yearXpath;

        this.fromYear = fromYear;

        this.toYear = toYear;

    }

    public boolean matches(final Document record, final boolean strict) throws ParseException {

        final String date = record.valueOf(yearXpath);

        if (StringUtils.isBlank(date)) {

            return false;

        }

        final String yyyy = DnetXsltFunctions.extractYear(date);

        if (StringUtils.isBlank(yyyy)) {

            return false;

        }

        final Integer year = Integer.valueOf(yyyy);

        if (year < fromYear | year > toYear) {

            return false;

        }

        boolean matched = false;

        for(final Map.Entry<String, String> c : criteria.entrySet()) {

            boolean matches = matched = record.selectNodes(c.getKey()).stream()

                    .filter(Objects::nonNull)

                    .map(o -> textOf(o))

                    .map(s -> ((String) s).toLowerCase())

                    .map(s -> ((String) s).trim())

                    .anyMatch(s -> {

                        return ((String) s).matches(c.getValue().toLowerCase());

                    });

            if (matches && !strict) {

                return true;

            }

        }

        return matched;

    }

    private String textOf(final Object o) {

        if (o instanceof org.dom4j.tree.DefaultText) {

            return ((DefaultText) o).getText();

        }

        return o.toString();

    }

}