Java IOUtil.saveTxt方法代码示例

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
public static boolean fix(String path)
{
    StringBuilder sbOut = new StringBuilder();
    for (String line : IOUtil.readLineListWithLessMemory(path))
    {
        if (line.trim().length() == 0)
        {
            sbOut.append(line);
            sbOut.append('\n');
            continue;
        }
        String[] args = line.split("\t");
        for (int i = 10 - args.length; i > 0; --i)
        {
            line += "\t_";
        }
        sbOut.append(line);
        sbOut.append('\n');
    }
    return IOUtil.saveTxt(path + ".fixed.txt", sbOut.toString());
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
public void testSegmentCorpus() throws Exception
{
    File root = new File("D:\\Doc\\语料库\\搜狗文本分类语料库精简版");
    for (File folder : root.listFiles())
    {
        if (folder.isDirectory())
        {
            for (File file : folder.listFiles())
            {
                System.out.println(file.getAbsolutePath());
                List<Term> termList = HanLP.segment(IOUtil.readTxt(file.getAbsolutePath()));
                StringBuilder sbOut = new StringBuilder();
                for (Term term : termList)
                {
                    if (CoreStopWordDictionary.shouldInclude(term))
                    {
                        sbOut.append(term.word).append(" ");
                    }
                }
                IOUtil.saveTxt("D:\\Doc\\语料库\\segmented\\" + folder.getName() + "_" + file.getName(), sbOut.toString());
            }
        }
    }
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
/**
 * 将NGram词典重新写回去
 *
 * @param map
 * @param path
 * @return
 */
private static boolean reSaveDictionary(TreeMap<String, Integer> map, String path)
{
    StringBuilder sbOut = new StringBuilder();
    for (Map.Entry<String, Integer> entry : map.entrySet())
    {
        sbOut.append(entry.getKey());
        sbOut.append(' ');
        sbOut.append(entry.getValue());
        sbOut.append('\n');
    }

    return IOUtil.saveTxt(path, sbOut.toString());
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
private static boolean dumpEnum(Set<String> set, String path)
{
    StringBuilder sb = new StringBuilder();
    for (String s : set)
    {
        sb.append(s);
        sb.append(",\n");
    }
    return IOUtil.saveTxt(path, sb.toString());
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
public void testGeneratePatternJavaCode() throws Exception
{
    CommonStringDictionary commonStringDictionary = new CommonStringDictionary();
    commonStringDictionary.load("data/dictionary/organization/nt.pattern.txt");
    StringBuilder sb = new StringBuilder();
    Set<String> keySet = commonStringDictionary.keySet();
    CommonStringDictionary secondDictionary = new CommonStringDictionary();
    secondDictionary.load("data/dictionary/organization/outerNT.pattern.txt");
    keySet.addAll(secondDictionary.keySet());
    for (String pattern : keySet)
    {
        sb.append("trie.addKeyword(\"" + pattern + "\");\n");
    }
    IOUtil.saveTxt("data/dictionary/organization/code.txt", sb.toString());
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
public void testMakePinyinJavaCode() throws Exception
{
    StringBuilder sb = new StringBuilder();
    for (Pinyin pinyin : PinyinDictionary.pinyins)
    {
        // 0声母 1韵母 2音调 3带音标
        sb.append(pinyin + "(" + Shengmu.class.getSimpleName() + "." + pinyin.getShengmu() + ", " + Yunmu.class.getSimpleName() + "." + pinyin.getYunmu() + ", " + pinyin.getTone() + ", \"" + pinyin.getPinyinWithToneMark() + "\", \"" + pinyin.getPinyinWithoutTone() + "\"" + ", " + Head.class.getSimpleName() + "." + pinyin.getHeadString() + ", '" + pinyin.getFirstChar() + "'" + "),\n");
    }
    IOUtil.saveTxt("data/dictionary/pinyin/py.txt", sb.toString());
}
 

import com.hankcs.hanlp.corpus.io.IOUtil; //导入方法依赖的package包/类
public void testMakeSimpleCRFTemplate() throws Exception
{
    Set<String> templateList = new LinkedHashSet<String>();
    int maxDistance = 4;
    // 字特征
    for (int i = -maxDistance; i <= maxDistance; ++i)
    {
        templateList.add("%x[" + i + ",0]");
    }
    // 细词性特征
    for (int i = -maxDistance; i <= maxDistance; ++i)
    {
        templateList.add("%x[" + i + ",1]");
    }
    // 粗词性特征
    for (int i = -maxDistance; i <= maxDistance; ++i)
    {
        templateList.add("%x[" + i + ",2]");
    }
    // 组合特征
    for (int i = 1; i <= maxDistance; ++i)
    {
        templateList.add("%x[-" + i + ",0]/" + "%x[0,0]");
        templateList.add("%x[0,0]/" + "%x[" + i + ",0]");

        templateList.add("%x[-" + i + ",1]/" + "%x[0,1]");
        templateList.add("%x[0,1]/" + "%x[" + i + ",1]");

        templateList.add("%x[-" + i + ",2]/" + "%x[0,2]");
        templateList.add("%x[0,2]/" + "%x[" + i + ",2]");
    }

    int id = 0;
    StringBuilder sb = new StringBuilder();
    for (String template : templateList)
    {
        sb.append(String.format("U%d:%s\n", id, template));
        ++id;
    }
    System.out.println(sb.toString());
    IOUtil.saveTxt("D:\\Tools\\CRF++-0.58\\example\\dependency\\template.txt", sb);
}