lineparsertest.java

一个自然语言处理的Java开源工具包。LingPipe目前已有很丰富的功能

package com.aliasi.test.unit.corpus;

import com.aliasi.corpus.ObjectHandler;
import com.aliasi.corpus.LineParser;
import com.aliasi.corpus.Parser;

import com.aliasi.test.unit.BaseTestCase;

import java.io.ByteArrayInputStream;
import java.io.CharArrayReader;
import java.io.IOException;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

import org.xml.sax.InputSource;

public class LineParserTest extends BaseTestCase {

    static final String[] DOCS = new String[] {
	"",
	"\n",
	"\n\n",
	"abc",
	"abc\n",
	"abc\rdef",
	"\nabc\rdef\r\n\n",
    };

    static final String[][] EXPECTED_LINES = new String[][] {
	{ },
	{ "" },
	{ "", "" },
	{ "abc" },
	{ "abc" },
	{ "abc", "def" },
	{ "", "abc", "def", "" }
    };

    // public void test0() {
    // assertEquals(Arrays.asList(new String[] { "ab", "", "c", "", "" }),
    // Arrays.asList("ab::c:::::::".split(":")));
    // }

    public void test1() throws IOException {
	IdParser parser = new IdParser();
	Collector<String> collector = new Collector<String>();
	parser.setHandler(collector);

	for (int i = 0; i < DOCS.length; ++i) {
	    String input = DOCS[i];
	    char[] cs = input.toCharArray();
	    List<String> lineList = Arrays.<String>asList(EXPECTED_LINES[i]);

	    collector.clear();
	    CharArrayReader reader = new CharArrayReader(cs);
	    InputSource in = new InputSource(reader);
	    parser.parse(in);
	    assertEquals(lineList, collector.result());
	    
	    collector.clear();
	    parser.parseString(cs,0,cs.length);
	    assertEquals(lineList, collector.result());
	}
    }

    public class IdParser extends LineParser<ObjectHandler<String>> {
	public void parseLine(String line, int number) {
	    getHandler().handle(line);
	}
    }
    

    public class Collector<E> implements ObjectHandler<E> {
	List<E> mCollector = new ArrayList<E>();
	public void handle(E e) {
	    mCollector.add(e);
	}
	public List<E> result() {
	    return mCollector;
	}
	public void clear() {
	    mCollector.clear();
	}
    }

}