From 068fc7f2e81178e55fa191a13709af64b1a163f6 Mon Sep 17 00:00:00 2001
From: EricsHu <hrr145632>
Date: 星期一, 05 十二月 2022 14:27:43 +0800
Subject: [PATCH] 处理编码
---
src/main/java/com/qxueyou/scc/exercise/service/impl/Handler.java | 128 +++++++++++++++++++++---------------------
1 files changed, 64 insertions(+), 64 deletions(-)
diff --git a/src/main/java/com/qxueyou/scc/exercise/service/impl/Handler.java b/src/main/java/com/qxueyou/scc/exercise/service/impl/Handler.java
index 185d785..fb58b2e 100644
--- a/src/main/java/com/qxueyou/scc/exercise/service/impl/Handler.java
+++ b/src/main/java/com/qxueyou/scc/exercise/service/impl/Handler.java
@@ -15,67 +15,67 @@
import com.qxueyou.scc.exercise.service.impl.parser.OptionParser;
public class Handler {
-
+
private Doc doc;
-
+
private Node currentNode;
-
- /** 失败 */
+
+ /** 澶辫触 */
public static final Integer HANDLER_RESULT_FAIL = 1 ;
-
- /** 上一类型 */
+
+ /** 涓婁竴绫诲瀷 */
public static final Integer HANDLER_RESULT_CONTINUE = 2;
-
- /** 成功*/
+
+ /** 鎴愬姛*/
public static final Integer HANDLER_RESULT_SUCCESS = 3;
-
- /** 校验是否有中文 */
- private static String regEx = "[\u4e00-\u9fa5]";
- private static Pattern pat = Pattern.compile(regEx);
-
+
+ /** 鏍￠獙鏄惁鏈変腑鏂� */
+ private static String regEx = "[\u4e00-\u9fa5]";
+ private static Pattern pat = Pattern.compile(regEx);
+
public Handler(){
doc = new Doc();
currentNode = doc;
}
-
+
/**
- * 解析文本,每次读取一个段落
- * @param content 段落文本内容
+ * 瑙ｆ瀽鏂囨湰,姣忔璇诲彇涓�涓钀�
+ * @param content 娈佃惤鏂囨湰鍐呭
*/
public ExerciseParse parse(String content){
-
+
ParseResult result = null;
-
+
content = content.replace((char)12288, (char)32);
content = content.replace((char)160, (char)32);
-
- //将得到的文本全角转半角
+
+ //灏嗗緱鍒扮殑鏂囨湰鍏ㄨ杞崐瑙�
//content = formatFullToHalf(content);
- content = content.replace('（', '(');
- content = content.replace('）', ')');
-
- // 非全部英文题 非全部空格， 需要将空格替换掉
+ content = content.replace('锛�', '(');
+ content = content.replace('锛�', ')');
+
+ // 闈炲叏閮ㄨ嫳鏂囬 闈炲叏閮ㄧ┖鏍硷紝 闇�瑕佸皢绌烘牸鏇挎崲鎺�
if(isContentContainsChinese(content) || content.trim().isEmpty()){
content = content.replace(" ", "");
}
-
+
content = content.replace("\t", "");
-
+
if(StringUtils.isEmpty(content)||content.equals("\r")){
return new ExerciseParse(HANDLER_RESULT_SUCCESS,content);
}
-
- //是否需要人为去掉回车
+
+ //鏄惁闇�瑕佷汉涓哄幓鎺夊洖杞�
content = content.replace("\r", "");
int count = 0;
-
+
do{
result = currentNode.getParser().parse(currentNode, content);
-
+
if(result.isSuccess()){
if(ParseResult.STEP_CUR.equals(result.getNextStep())){
- //成功后，取得当前parser
+ //鎴愬姛鍚庯紝鍙栧緱褰撳墠parser
generateLastSuccessType(doc,currentNode,result.getNextNode());
return new ExerciseParse(HANDLER_RESULT_SUCCESS,content);
}
@@ -89,7 +89,7 @@
}
}else{
if(ParseResult.STEP_CUR.equals(result.getNextStep())){
- //20150907 修改;抛到最上层doc解析器仍然无法定位，直接将文本内容添加到上一个解析成功的位置
+ //20150907 淇敼;鎶涘埌鏈�涓婂眰doc瑙ｆ瀽鍣ㄤ粛鐒舵棤娉曞畾浣嶏紝鐩存帴灏嗘枃鏈唴瀹规坊鍔犲埌涓婁竴涓В鏋愭垚鍔熺殑浣嶇疆
appendTextToLastPosition(doc,content);
return new ExerciseParse(HANDLER_RESULT_CONTINUE,doc.getLastSuccessType(),content);
}
@@ -106,24 +106,24 @@
currentNode = result.getNextNode();
}
}
-
+
}while( count++ < 15);
-
+
return new ExerciseParse(HANDLER_RESULT_FAIL,content);
-
+
}
-
+
public Doc result(){
return doc;
}
-
-
+
+
/**
- * 得到最新成功的解析类型
- *
+ * 寰楀埌鏈�鏂版垚鍔熺殑瑙ｆ瀽绫诲瀷
+ *
* @param doc doc
- * @param currNode 当前节点
- * @param analysisNode 如果是解析 部分的解析器，返回非空数据，用于获取解析器类型
+ * @param currNode 褰撳墠鑺傜偣
+ * @param analysisNode 濡傛灉鏄В鏋� 閮ㄥ垎鐨勮В鏋愬櫒锛岃繑鍥為潪绌烘暟鎹紝鐢ㄤ簬鑾峰彇瑙ｆ瀽鍣ㄧ被鍨�
*/
private void generateLastSuccessType(Doc doc,Node currNode,Node analysisNode){
Parser currParser = currNode.getParser();
@@ -140,39 +140,39 @@
return ;
}
}
-
+
/**
- * 得到当前解析成功的最新位置：如果是题目的：题干、选项或者解析，将他们组装对应位置
- *
+ * 寰楀埌褰撳墠瑙ｆ瀽鎴愬姛鐨勬渶鏂颁綅缃細濡傛灉鏄鐩殑锛氶骞层�侀�夐」鎴栬�呰В鏋愶紝灏嗕粬浠粍瑁呭搴斾綅缃�
+ *
* @param doc
* @param content
*/
private void appendTextToLastPosition(Doc doc,String content){
String currSuccessType = doc.getLastSuccessType();
-
- //第一层 ItemType
+
+ //绗竴灞� ItemType
List<Node> lstItemType = doc.getChildren();
if(null == lstItemType || lstItemType.isEmpty()){
return ;
}
-
- //第二层Item
+
+ //绗簩灞侷tem
ItemType objItemType = (ItemType)lstItemType.get(lstItemType.size() - 1);
List<Node> lstItem = objItemType.getChildren();
if(null == lstItem || lstItem.isEmpty()){
return ;
}
-
- //取到最新位置的Item
+
+ //鍙栧埌鏈�鏂颁綅缃殑Item
Item objItem = (Item)lstItem.get(lstItem.size() - 1);
if(null != objItem){
appendTextToCurrPosition(objItem,currSuccessType,content);
}
}
-
+
/**
- * 将当前没有归档的内容加入到对应的位置
- *
+ * 灏嗗綋鍓嶆病鏈夊綊妗ｇ殑鍐呭鍔犲叆鍒板搴旂殑浣嶇疆
+ *
* @param objItem
* @param currSuccessType
* @param content
@@ -181,25 +181,25 @@
if(Doc.CURR_SUCCESS_CONTENT == currSuccessType){
objItem.setTitle(objItem.getTitle()+ content);
currentNode = objItem;
-
+
}else if(Doc.CURR_SUCCESS_ANALYSIS == currSuccessType){
objItem.setAnalysis(objItem.getAnalysis() + content);
currentNode = objItem;
-
+
}else if(Doc.CURR_SUCCESS_OPTION == currSuccessType){
-
+
List<Option> lstOptions = objItem.getOptions();
if(null == lstOptions || lstOptions.isEmpty()){
return ;
}
-
+
Option opt = null ;
for(int i = lstOptions.size() -1 ; i>=0 ; i --){
opt = lstOptions.get(i);
if(null == opt){
return ;
}
- //将选项添加到最新位置
+ //灏嗛�夐」娣诲姞鍒版渶鏂颁綅缃�
if(StringUtils.isNotBlank(opt.getNo())){
opt.setContent(opt.getContent() + content);
currentNode = objItem;
@@ -208,20 +208,20 @@
}
}
}
-
+
/**
- * 字符串是否包含中文
+ * 瀛楃涓叉槸鍚﹀寘鍚腑鏂�
* @param content
* @return
*/
private boolean isContentContainsChinese( String content ){
-
+
Matcher matcher = pat.matcher(content);
boolean flg = false;
if (matcher.find()) {
flg = true;
}
- return flg;
+ return flg;
}
-
+
}
--
Gitblit v1.8.0