package com.walker.support.es.util;
|
|
import com.walker.infrastructure.utils.StringUtils;
|
import com.walker.support.es.Constants;
|
import com.walker.support.es.SaveData;
|
import com.walker.support.es.SaveDataFile;
|
import org.apache.commons.codec.binary.Base64;
|
import org.elasticsearch.common.text.Text;
|
import org.elasticsearch.search.fetch.subphase.highlight.HighlightField;
|
import org.elasticsearch.xcontent.XContentBuilder;
|
import org.elasticsearch.xcontent.XContentFactory;
|
import org.springframework.util.FileCopyUtils;
|
|
import java.io.File;
|
import java.util.HashMap;
|
import java.util.Map;
|
|
/**
|
* 描述：
|
* @author 时克英
|
* @date 2020年7月13日 下午2:19:24
|
*/
|
|
public class FullTextUtils {
|
|
/**
|
* 返回高亮对象中的文本
|
* @param hl
|
* @return
|
*/
|
public static final String getOneFieldHighlightText(Map<String, HighlightField> hl){
|
StringBuilder sb = new StringBuilder();
|
if(hl != null){
|
Text[] texts = null;
|
for(HighlightField hf : hl.values()){
|
texts = hf.getFragments();
|
if(texts != null && texts.length > 0){
|
// sb.append(texts[0].toString()).append("\n");
|
sb.append(texts[0].toString());
|
}
|
}
|
}
|
return sb.toString();
|
}
|
|
/**
|
* 返回检索需要的json字符串
|
* @param queryText 查询字符串
|
* @param dataSecurity 权限类型：0无限制，1个人，2部门，3单位
|
* @param securityInfo 用户id | 部门id | 单位id
|
* @return
|
*/
|
public static final String acquireSearchQueryJson(String queryText, int dataSecurity, String securityInfo){
|
StringBuilder text = new StringBuilder();
|
text.append("{\"bool\":{");
|
text.append("\"must\":[");
|
// 如果存在空格，说明要多个条件同时匹配
|
String[] findKeys = queryText.split(" ");
|
if(findKeys.length > 1){
|
// 通过设置多个multi_match + must来实现and效果
|
int i = 0;
|
for(String k : findKeys){
|
if(i > 0){
|
text.append(",");
|
}
|
text.append("{\"multi_match\":{\"query\":\"").append(k).append("\",");
|
// 这里设置查询字符串按照语意分词
|
text.append("\"analyzer\":\"ik_smart\",");
|
text.append("\"fields\":[\"title\", \"text_db\", \"attachment.content\", \"xmname\"]}}");
|
i++;
|
}
|
} else {
|
text.append("{\"multi_match\":{\"query\":\"").append(queryText).append("\",");
|
text.append("\"fields\":[\"title\", \"text_db\", \"attachment.content\"]}}");
|
}
|
if(dataSecurity == Constants.DATA_SECURITY_PERSON){
|
text.append(",{\"match\":{\"user_id\":\"").append(securityInfo).append("\"}}");
|
} else if(dataSecurity == Constants.DATA_SECURITY_DEPT){
|
text.append(",{\"match\":{\"dept_id\":\"").append(securityInfo).append("\"}}");
|
} else if(dataSecurity == Constants.DATA_SECURITY_ORG){
|
text.append(",{\"match\":{\"org_id\":\"").append(securityInfo).append("\"}}");
|
} else {
|
// 无限制
|
}
|
text.append("]}}");
|
return text.toString();
|
}
|
|
/**
|
* query 里面的内容，不带query本身
|
* @param _id
|
* @return
|
*/
|
public static String acquireSearchOneById(String _id){
|
StringBuilder text = new StringBuilder();
|
text.append("{");
|
text.append("\"term\":{\"_id\":\"").append(_id).append("\"}");
|
text.append("}");
|
return text.toString();
|
}
|
|
public static final Map<String, String> translate(SaveData d){
|
Map<String, String> map = new HashMap<>();
|
if(d instanceof SaveDataFile){
|
SaveDataFile sdf = (SaveDataFile)d;
|
map.put("id", sdf.getId());
|
map.put("create_time", sdf.getCreateTime());
|
map.put("data", sdf.getData());
|
map.put("dept_id", sdf.getDeptId());
|
map.put("org_id", sdf.getOrgId());
|
map.put("path", sdf.getPath());
|
map.put("text_db", sdf.getTextDb());
|
map.put("title", sdf.getTitle());
|
map.put("update_time", sdf.getUpdateTime());
|
map.put("user_id", sdf.getUserId());
|
// map.put("text_html", d.getTextHtml());
|
map.put("xmid", sdf.getXmid());
|
map.put("xmname", sdf.getXmname());
|
map.put("user_name", sdf.getUserName());
|
|
} else {
|
map.put("id", d.getId());
|
map.put("create_time", d.getCreateTime());
|
map.put("data", d.getData());
|
map.put("dept_id", d.getDeptId());
|
map.put("org_id", d.getOrgId());
|
map.put("path", d.getPath());
|
map.put("text_db", d.getTextDb());
|
map.put("title", d.getTitle());
|
map.put("update_time", d.getUpdateTime());
|
map.put("user_id", d.getUserId());
|
map.put("text_html", d.getTextHtml());
|
map.put("user_name", d.getUserName());
|
}
|
return map;
|
}
|
|
/**
|
* 创建简单的索引表，为测试返回分词用
|
* @param tokenType
|
* @return
|
*/
|
public static final String createSimpleIndex(String tokenType, boolean attachment){
|
StringBuilder json2 = new StringBuilder();
|
json2.append("{\n");
|
json2.append("\"properties\":{\n");
|
// 项目id
|
json2.append("\"id\":{\"type\":\"keyword\"},");
|
json2.append("\"title\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"},");
|
if(attachment){
|
json2.append("\"create_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"},");
|
// 附件信息
|
json2.append("\"attachment\":{");
|
json2.append("\"properties\":{\"content\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"}}");
|
json2.append("}");
|
} else {
|
json2.append("\"create_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"}");
|
}
|
|
json2.append("}");
|
json2.append("}");
|
return json2.toString();
|
}
|
|
/**
|
* 创建索引库
|
* @param type 表类型（暂未使用）
|
* @return
|
*/
|
public static final String createIndexJson(String type, String tokenType){
|
// if(StringUtils.isEmpty(type)){
|
// type = "table";
|
// }
|
StringBuilder json2 = new StringBuilder();
|
json2.append("{\n");
|
if(!StringUtils.isEmpty(type)){
|
json2.append("\"").append(type).append("\":{\n");
|
}
|
json2.append("\"properties\":{\n");
|
System.out.println("+++++++++++++++++");
|
// 项目id
|
json2.append("\"id\":{\"type\":\"keyword\"},");
|
// json.append("\"title\":{\"type\":\"text\",\"analyzer\":\"ik_max_word\"},");
|
json2.append("\"title\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"},");
|
json2.append("\"path\":{\"type\":\"keyword\"},");
|
// 原始html内容
|
json2.append("\"text_html\":{\"type\":\"text\"},");
|
json2.append("\"text_db\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"},");
|
json2.append("\"create_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"},");
|
json2.append("\"update_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"},");
|
json2.append("\"user_name\":{\"type\":\"keyword\"},");
|
// 用户数据权限
|
json2.append("\"user_id\":{\"type\":\"keyword\"},");
|
json2.append("\"dept_id\":{\"type\":\"keyword\"},");
|
json2.append("\"org_id\":{\"type\":\"keyword\"},");
|
// 附件信息
|
/** */
|
json2.append("\"attachment\":{");
|
json2.append("\"properties\":{\"content\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"}}");
|
json2.append("}");
|
|
json2.append("}");
|
if(!StringUtils.isEmpty(type)){
|
json2.append("}");
|
}
|
json2.append("}");
|
return json2.toString();
|
}
|
|
public static final String createXmFileJson(String type, String tokenType){
|
StringBuilder json = new StringBuilder();
|
json.append("{\n");
|
if(!StringUtils.isEmpty(type)){
|
json.append("\"").append(type).append("\":{\n");
|
}
|
json.append("\"properties\":{\n");
|
json.append("\"id\":{\"type\":\"keyword\"},");
|
// json.append("\"title\":{\"type\":\"text\",\"analyzer\":\"ik_max_word\"},");
|
json.append("\"title\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"},");
|
// 附件绝对路径
|
json.append("\"path\":{\"type\":\"keyword\"},");
|
// 项目id、名称
|
json.append("\"xmid\":{\"type\":\"keyword\"},");
|
json.append("\"xmname\":{\"type\":\"keyword\"},");
|
// 原始html内容
|
// json.append("\"text_html\":{\"type\":\"keyword\"},");
|
// json.append("\"text_db\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"},");
|
json.append("\"create_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"},");
|
json.append("\"update_time\":{\"type\":\"date\",\"format\":\"yyyy-MM-dd HH:mm:ss\"},");
|
json.append("\"user_name\":{\"type\":\"keyword\"},");
|
// 用户数据权限
|
json.append("\"user_id\":{\"type\":\"keyword\"},");
|
json.append("\"dept_id\":{\"type\":\"keyword\"},");
|
json.append("\"org_id\":{\"type\":\"keyword\"},");
|
// 附件信息
|
json.append("\"attachment\":{");
|
// json.append("\"properties\":{\"content\":{\"type\":\"text\",\"analyzer\":\"ik_max_word\"}}");
|
json.append("\"properties\":{\"content\":{\"type\":\"text\",\"analyzer\":\"").append(tokenType).append("\"}}");
|
json.append("}");
|
json.append("}");
|
if(!StringUtils.isEmpty(type)){
|
json.append("}");
|
}
|
json.append("}");
|
return json.toString();
|
}
|
|
public static String getFileBase64(String fileContent){
|
return Base64.encodeBase64String(fileContent.getBytes());
|
}
|
|
public static String getFileBase64Value(String path) throws Exception{
|
File file = null;
|
try{
|
file = new File(path);
|
byte[] fileBase64 = Base64.encodeBase64(FileCopyUtils.copyToByteArray(file));
|
return new String(fileBase64);
|
} catch (Exception e){
|
e.printStackTrace();
|
return null;
|
} finally {
|
if(file != null){
|
}
|
}
|
}
|
|
public static String getFileBase64ValueByContent(String content){
|
byte[] fileBase64 = Base64.encodeBase64(content.getBytes());
|
return new String(fileBase64);
|
}
|
|
public static byte[] getFileContent(String base64){
|
return Base64.decodeBase64(base64);
|
}
|
|
/**
|
* 创建表结构的结构对象
|
* @param type 表类型（暂未使用）
|
* @return
|
* @throws Exception
|
*/
|
@Deprecated
|
public static final XContentBuilder createFullTextIndexJson(String type) throws Exception{
|
if(StringUtils.isEmpty(type)){
|
type = "table";
|
}
|
XContentBuilder builder = XContentFactory.jsonBuilder();
|
builder.startObject();
|
{
|
builder.startObject(type);
|
{
|
// builder.startObject("_id");
|
// {
|
// builder.field("path", "mainkey");
|
// }
|
// builder.endObject();
|
builder.startObject("properties");
|
{
|
/**
|
* 设置默认id
|
"_id":{"path":"mainkey"},
|
"_source" : { "enabled" : false },
|
"properties" : {
|
"mainkey" : { "type" : "string", "index" : "not_analyzed" }
|
}
|
*/
|
// builder.startObject("mainkey");
|
// {
|
// builder.field("type", "keyword");
|
// builder.field("index", "not_analyzed");
|
// }
|
// builder.endObject();
|
// 业务主键
|
builder.startObject("id");
|
{
|
builder.field("type", "keyword");
|
}
|
builder.endObject();
|
//
|
builder.startObject("title");
|
{
|
builder.field("type", "text");
|
builder.field("analyzer", "ik_max_word");
|
}
|
builder.endObject();
|
//
|
builder.startObject("path");
|
{
|
builder.field("type", "keyword");
|
}
|
builder.endObject();
|
// 数据库中的内容
|
builder.startObject("text_db");
|
{
|
builder.field("type", "text");
|
builder.field("analyzer", "ik_max_word");
|
}
|
builder.endObject();
|
//
|
builder.startObject("create_time");
|
{
|
builder.field("type", "date");
|
builder.field("format", "yyyy-MM-dd HH:mm:ss");
|
}
|
builder.endObject();
|
//
|
builder.startObject("update_time");
|
{
|
builder.field("type", "date");
|
builder.field("format", "yyyy-MM-dd HH:mm:ss");
|
}
|
builder.endObject();
|
}
|
builder.endObject();
|
|
// 附件
|
builder.startObject("attachment");
|
{
|
builder.startObject("properties");
|
{
|
builder.startObject("content");
|
{
|
builder.field("type", "text");
|
builder.field("analyzer", "ik_max_word");
|
}
|
builder.endObject();
|
}
|
builder.endObject();
|
}
|
builder.endObject();
|
}
|
builder.endObject();
|
}
|
builder.endObject();
|
return builder;
|
}
|
}
|
