1. 程式人生 > >JAVA 無BOM utf8檔案編碼判斷

JAVA 無BOM utf8檔案編碼判斷

/**
	 * @Comments :獲取檔案編碼格式
	 * @param fileName
	 * @return
	 */
	private static String getCharset(File fileName) {
		BufferedInputStream bin;
		int bom = 0;
		String str = " ";
		String str2 = "";
		try {
			bin = new BufferedInputStream(new FileInputStream(fileName));
			bom = (bin.read() << 8) + bin.read();
			// 獲取兩個位元組內容,如果檔案無BOM資訊,則通過判斷字的位元組長度區分編碼格式
			byte bs[] = new byte[10];
			while(str.matches("\\s+\\w*")){
				bin.read(bs);
				str = new String(bs, "UTF-8");
			}
			str2 = new String(bs, "GBK");
		} catch (FileNotFoundException e) {
			e.printStackTrace();
		} catch (IOException e) {
			e.printStackTrace();
		}


		String code = null;
		// 有BOM
		switch (bom) {
		case 0xefbb:
			code = "UTF-8";
			break;
		case 0xfffe:
			code = "Unicode";
			break;
		case 0xfeff:
			code = "UTF-16BE";
			break;
		default:
			// 無BOM
			if (str.length() <=str2.length()) {
				code = "UTF-8";
			} else {
				code = "GBK";
			}
		}


		return code;
	}