字符串 图片包括content和images
```
@Async //使用异步执行当前方法
public void auditWmNews(WmNews wmNews) {
log.info("开始审核文章");
// 判断状态是否待审核
if(wmNews.getStatus()!=1){
log.info("不是待审核状态，结束");
return ;
}
Integer wmNewsId = wmNews.getId();
// 获取文章中的文本和图片 ,key - text ,key - image
Map<String,Object> map = getTextAndImage(wmNews);
map就有值了----
String text = map.get("text").toString();
//只能强转
List<byte[]> imageList = (List<byte[]>)map.get("image");
@Autowired
private MinioService minioService;
/**
* 获取文章中的文本和图片 ,key - text ,key - image
* @param wmNews
* @return
private Map<String, Object> getTextAndImage(WmNews wmNews) {
// 文本包括 title、content、labels，获取的是字符串
StringBuilder stringBuilder = new StringBuilder();
//// 图片包括 content、images，获取的是List<byte[]>
List<byte[]> imageList = new ArrayList<>();
stringBuilder.append(wmNews.getTitle());
if(StringUtils.isNotBlank(wmNews.getLabels())){
//append分割
stringBuilder.append(",").append(wmNews.getLabels());
}
content是json所以解析content
try {
// 解析content
List<Map<String, Object>> list = JsonUtils.nativeRead(wmNews.getContent(), new TypeReference 表示类型<List<Map<String, Object>>>() {
});
for (Map<String, Object> map : list) {
String type = map.get("type").toString();
String value = map.get("value").toString();
if (type.equals("text")) {
如果是文本
stringBuilder.append(",").append(value);
} else {
// 获取图片url
if (StringUtils.isNotBlank(value)) {
// 根据图片的url，到minio下载图片，得到byte[]
byte[] bytes = minioService.downLoadFile(value);
imageList.add(bytes);
}
}
}
if (StringUtils.isNotBlank(wmNews.getImages())) {
String[] urls = wmNews.getImages().split(",");
for (String url : urls) {
// 从minio下载封面图片的字节
byte[] bytes = minioService.downLoadFile(url);
imageList.add(bytes);
}
}
}catch (Exception e){
log.error("图片从minio下载失败");
throw new LeadException(AppHttpCodeEnum.SERVER_ERROR);
}
// 返回map结果
Map<String,Object> map = new HashMap<>();
map.put("text",stringBuilder.toString());
map.put("image",imageList);
return map;
}
/ 阿里云文本审核
boolean b = checkText(text,wmNewsId);
if(!b){
log.error("阿里云审核文本不通过");
return ;
//throw new LeadException(AppHttpCodeEnum.TEXT_ILLEGAL);
}
// 阿里云图片审核
boolean b1 = checkImage(imageList,wmNewsId);
if(!b1){
log.error("阿里云审核图片不通过");
return ;
//throw new LeadException(AppHttpCodeEnum.PARAM_IMAGE_ILLEGAL);
}
// 敏感词审核
boolean b2 = checkSensitive(text,wmNewsId);
if(!b2){
log.error("敏感词检测不通过");
return ;
//throw new LeadException(AppHttpCodeEnum.PARAM_IMAGE_ILLEGAL);
}
if(!CollectionUtils.isEmpty(imageList)){
boolean b3 = true;
for (byte[] bytes : imageList) {
b3 = handleImageSensitiveScan(bytes,wmNewsId);
if(!b3){
log.error("图片敏感词检测不通过");
break ;
}
}
if(!b3){
log.error("图片敏感词检测不通过");
return ;
}
}
@Autowired
private WmSensitiveService wmSensitiveService;
/**
* 敏感词审核
* @param text
* @param wmNewsId
* @return
*/
private boolean checkSensitive(String text, Integer wmNewsId) {
// 查询所有的敏感词
List<WmSensitive> wmSensitiveList = wmSensitiveService.list();
// 只需要敏感词
List<String> sensitiveWord = wmSensitiveList.stream().
map(WmSensitive::getSensitives).
collect(Collectors.toList());
// 调用工具类检测敏感词
Map<String, Integer> map = SensitiveWordUtil.matchWords(sensitiveWord, text);
if(CollectionUtils.isEmpty(map)){
log.info("敏感词检测成功");
return true;
}
只要key
String reason = map.keySet().stream().collect(Collectors.joining(","));
log.info("reason=={}",reason);
// 修改文章状态
updateWmNewsStaus(wmNewsId,2,reason,null);
return false;
}
@Autowired
private GreenImageUploadScan imageUploadScan;
/**
* 阿里云检测图片
* @param imageList
* @param wmNewsId
* @return
*/
private boolean checkImage(List<byte[]> imageList, Integer wmNewsId) {
if(CollectionUtils.isEmpty(imageList)){
log.info("imagelist为空，停止检测");
return true;
}
// 调用阿里云图片审核接口
try {
Map<String, String> map = imageUploadScan.imageScan(imageList);
String suggestion = map.get("suggestion");
String label = map.get("label");
if("pass".equals(suggestion)){
log.info("阿里云图片审核通过");
return true;
}else if("review".equals(suggestion)){
log.error("阿里云图片审核不确定，需要人工");
this.updateWmNewsStaus(wmNewsId,3,label,null);
return false;
}else {
log.error("阿里云图片审核不通过！！");
this.updateWmNewsStaus(wmNewsId,2,label,null);
return false;
}
}catch (Exception e){
log.error("阿里云图片审核失败");
e.printStackTrace();
throw new LeadException(AppHttpCodeEnum.SERVER_ERROR);
}
}
@Autowired
private GreenTextScan textScan;
/**
* 阿里云审核文本
* @param text
* @return
*/
private boolean checkText(String text,Integer wmNewsId) {
if(StringUtils.isBlank(text)){
log.error("text 内容 为空，停止检测");
throw new LeadException(AppHttpCodeEnum.TEXT_ILLEGAL);
}
// 调用阿里云文本审核接口
try {
Map<String, String> map = textScan.greenTextScan(text);
String suggestion = map.get("suggestion");
String reason = map.get("reason");
if("pass".equals(suggestion)){
log.info("阿里云文本审核通过");
return true;
}else if("review".equals(suggestion)){
log.error("阿里云审核不确定，需要人工");
this.updateWmNewsStaus(wmNewsId,3,reason,null);
return false;
}else {
log.error("阿里云审核不通过！！");
this.updateWmNewsStaus(wmNewsId,2,reason,null);
return false;
}
}catch (Exception e){
log.error("阿里云文本审核失败");
e.printStackTrace();
throw new LeadException(AppHttpCodeEnum.SERVER_ERROR);
}
}
用正则把"去掉
@Autowired
private Tess4jClient tess4jClient;
/**
* 从图片中获取文字进行敏感词检测放redis中
*/
private boolean handleImageSensitiveScan(byte[] bytes,Integer wmNewsId){
try {
//图片识别文字审核---begin-----
//从byte[]转换为butteredImage
ByteArrayInputStream in = new ByteArrayInputStream(bytes);
BufferedImage imageFile = ImageIO.read(in);
//识别图片的文字
String result = tess4jClient.doOCR(imageFile);
//审核是否包含自管理的敏感词
boolean isSensitive = checkSensitive(result, wmNewsId);
return isSensitive;
}catch (Exception e){
e.printStackTrace();
return false;
}
}
