sgmltest.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.util;

import com.aliasi.util.Sgml;

import com.aliasi.test.unit.BaseTestCase;

public class SgmlTest extends BaseTestCase {

    public void test() {
        assertNull(Sgml.entityToCharacter("foobar"));
        assertEquals(new Character('\u200C'), Sgml.entityToCharacter("zwnj"));

        assertSub("","");
        assertSub("foobar","foobar");
        assertSub("&Agr;","\u0391");
        assertSub("foo&Agr;","foo\u0391");
        assertSub("&Agr;foo","\u0391foo");
        assertSub("foo&Agr;bar","foo\u0391bar");
        assertSub("&Foobar;","?");
        assertSub("ℵ","\u2135");
        assertSub("ℵ´","\u2135\u00B4");
        assertSub("ℵfoo´","\u2135foo\u00B4");
        assertSub("fooℵbar´baz","foo\u2135bar\u00B4baz");
        assertSub("&Foobar;","baz","baz");
    }

    void assertSub(String in, String expected) {
        assertSub(in,expected,null);
    }

    void assertSub(String in, String expected, String unk) {
        assertEquals(expected,
                     unk != null
                     ? Sgml.replaceEntities(in,unk)
                     : Sgml.replaceEntities(in));


    }

}