把字符串里中文大写数字转换成阿拉伯数字(通常在排序时使用)

在有的时候需要排序,但是字符串里又习惯使用"一二三。。。"这种的非阿拉伯数字的写法,所以我在网上找一段代码,根据自己的需求做了一些修改。代码如下:

import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class ConvertBigNum {

    private static String num = "零一二三四五六七八九十";
    private static String numF = "十百千万亿";
    private static String numReg = "[零一二三四五六七八九十百千万亿]{1,}";

    private static String internalFormat(final String s, final int fixDigitals) {
        String allStr = "";
        int[] mark = getIndexBigNum(s);
        int begin = mark[0];
        int end = mark[1];
        if (begin == -1) {// 没有大写数字
            return s;
        }
        if (begin > 0) {
            allStr += s.substring(0, begin);
        }
        int pos = 1;
        int allNum = 0;
        int last = 0;
        int lastPow = 0;
        if (begin >= 0) {
            while (end > begin) {
                // 当前数字是否可以算出数据
                pos = num.indexOf(s.charAt(begin));

                if (pos >= 0 && pos != 10) {
                    allNum += pos;
                    if (begin != 0 && num.indexOf(s.charAt(begin - 1)) != 10) {
                        last = pos;
                    } else {
                        last += pos;
                    }

                } else {
                    pos = numF.indexOf(s.charAt(begin)) + 1;
                    // 当前的为numF
                    if (pos > 0) {
                        if (pos >= 4) {
                            // 万以上特殊处理
                            last = 0;
                            allNum += (int) (Math.pow(10, pos) * (allNum - lastPow)) - (allNum - lastPow);
                            lastPow = allNum;

                        } else {
                            if (last > 0) {
                                // 前面是数字
                                allNum -= last;
                                last = (int) (last * Math.pow(10, pos));
                                allNum += last;
                            } else {
                                last = (int) (Math.pow(10, pos));
                                allNum += last;
                            }
                        }

                    }
                }

                begin++;

            }

        }

        if (allNum > 0) {
            if (fixDigitals > 0) {
                allStr += String.format("%0" + fixDigitals + "d", allNum);
            } else {
                allStr += allNum;
            }

        }

        if (begin < s.length()) {
            allStr += s.substring(begin);
        }
        return allStr;
    }

    private static int[] getIndexBigNum(String s) {
        if (s != null) {
            Matcher matcher = Pattern.compile(numReg).matcher(s);
            if (matcher.find()) {
                return new int[] { matcher.start(), matcher.end() };
            } else {
                return new int[] { -1, -1 };
            }
        } else {
            return new int[] { -1, -1 };
        }
    }

    public static String format(final String ori, final int fixDigitals) {
        String s = ori;
        int[] mark = getIndexBigNum(s);
        int begin = mark[0];
        if (begin > -1) {
            do {
                s = internalFormat(s, fixDigitals);
                mark = getIndexBigNum(s);
                begin = mark[0];
            } while (begin > -1);
        }
        return s;
    }

    public static String format(final String ori) {
        return format(ori, 0);
    }

    public static void main(String[] args) {
        System.out.println(format("第三年级第二十一A班", 3));
        System.out.println(format("五万二千三百年二班", 5));
    }
}

你可能感兴趣的:(字符串)