feat:合同信息提取插件

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/LargeModelFunctionEnum.java

@@ -4,6 +4,7 @@ import cn.com.poc.common.utils.SpringUtils;
 import cn.com.poc.thirdparty.resource.demand.ai.function.calculator.CalculatorFunction;
 import cn.com.poc.thirdparty.resource.demand.ai.function.document_reader.DocumentReaderFunction;
 import cn.com.poc.thirdparty.resource.demand.ai.function.document_understanding.DocumentUnderstandIngFunction;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.ContractExtractionFunction;
 import cn.com.poc.thirdparty.resource.demand.ai.function.html_reader.HtmlReaderFunction;
 import cn.com.poc.thirdparty.resource.demand.ai.function.image_ocr.ImageOCRFunction;
 import cn.com.poc.thirdparty.resource.demand.ai.function.long_document_reader.LongDocumentReaderFunction;
@@ -46,6 +47,8 @@ public enum LargeModelFunctionEnum {
 
     long_document_reader(LongDocumentReaderFunction.class),
 
+    contract_extraction(ContractExtractionFunction.class),
+
     ;
 
     private Class<? extends AbstractLargeModelFunction> function;

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/extraction/ContractExtractionFunction.java

+package cn.com.poc.thirdparty.resource.demand.ai.function.extraction;
+
+import cn.com.poc.agent_application.entity.Variable;
+import cn.com.poc.common.utils.JsonUtils;
+import cn.com.poc.thirdparty.resource.demand.ai.function.AbstractFunctionResult;
+import cn.com.poc.thirdparty.resource.demand.ai.function.AbstractLargeModelFunction;
+import cn.com.poc.thirdparty.resource.demand.ai.function.entity.FunctionLLMConfig;
+import cn.com.poc.thirdparty.resource.demand.ai.function.entity.Parameters;
+import cn.com.poc.thirdparty.resource.demand.ai.function.entity.Properties;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity.KeyInfo;
+import cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api.TextInClient;
+import cn.hutool.core.collection.ListUtil;
+import com.alibaba.fastjson.JSONArray;
+import com.alibaba.fastjson.JSONObject;
+import org.springframework.stereotype.Component;
+
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * @author alex.yao
+ * @date 2025/5/12
+ */
+@Component
+public class ContractExtractionFunction extends AbstractLargeModelFunction {
+
+    private String DESC = "合同关键信息抽取";
+
+    private final FunctionLLMConfig functionLLMConfig = new FunctionLLMConfig.FunctionLLMConfigBuilder()
+            .name("contract_extraction")
+            .parameters(new Parameters("array")
+                    .addProperties("fileUrl", new Properties("string", "文件链接, 合同文件的在线地址"))
+                    .addProperties("key_info", new Properties("string", "关键信息名称, 长度限制20个字符"))
+                    .addProperties("paraphrase_names", new Properties("array", "相似名字段,字符串数组, 可根据相似名精准抽取关键信息, 最多填写3个,每个释义名称长度限制20个字符"))
+                    .addProperties("field_type", new Properties("string", "字段类型字段, 可选项有,时间:time, 金额:amount, 地址:address, 公司:company, 姓名:name, 描述(长文本):long_text_description, 其他:other, 印章:stamp, 分别对应产品段配置的字段类型"))
+                    .addProperties("keywords", new Properties("array", "关键字字段, 字符串数组, 可根据关键字信息，快速定位抽取信所在段落范围, 最多填写10个,且字符总长度不超过50"))
+            )
+            .description(DESC)
+            .build();
+
+    @Override
+    public AbstractFunctionResult<String> doFunction(String content, String identifier) {
+        AbstractFunctionResult<String> result = new AbstractFunctionResult<>();
+        JSONArray jsonArray = JSONArray.parseArray(content);
+        if (jsonArray.isEmpty()) {
+            return result;
+        }
+        String fileUrl = jsonArray.getJSONObject(0).getString("file_url");
+        List<KeyInfo> keyInfos = new ArrayList<>();
+        for (int i = 0; i < jsonArray.size(); i++) {
+            JSONObject jsonObject = jsonArray.getJSONObject(i);
+            KeyInfo keyInfo = new KeyInfo();
+            if (jsonObject.containsKey("field_type")) {
+                keyInfo.setField_type(jsonObject.getString("file_type"));
+            }
+            if (jsonObject.containsKey("key_info")) {
+                keyInfo.setKey_info(jsonObject.getString("key_info"));
+            }
+            if (jsonObject.containsKey("paraphrase_names")) {
+                keyInfo.setParaphrase_names(jsonObject.getJSONArray("paraphrase_names").toArray(new String[0]));
+            }
+            if (jsonObject.containsKey("keywords")) {
+                keyInfo.setKeywords(jsonObject.getJSONArray("keywords").toArray(new String[0]));
+            }
+            keyInfos.add(keyInfo);
+        }
+
+        TextInClient textInClient = new TextInClient();
+        String extraction = textInClient.extraction(fileUrl, keyInfos);
+        result.setFunctionResult(extraction);
+        result.setPromptContent(extraction);
+        return result;
+    }
+
+    @Override
+    public String getDesc() {
+        return DESC;
+    }
+
+    @Override
+    public List<String> getLLMConfig() {
+        return ListUtil.toList(JsonUtils.serialize(functionLLMConfig));
+    }
+
+    @Override
+    public List<String> getLLMConfig(List<Variable> variableStructure) {
+        return this.getLLMConfig();
+    }
+}

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/extraction/entity/Config.java

+package cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity;
+
+/**
+ * @author alex.yao
+ * @date 2025/5/12
+ */
+public class Config {
+
+    public String engine;
+    public String use_pdf_parser;
+    public String use_semantic_match;
+    public String remove_watermark;
+
+    public Config(String engine, String use_pdf_parser, String use_semantic_match, String remove_watermark) {
+        this.engine = engine;
+        this.use_pdf_parser = use_pdf_parser;
+        this.use_semantic_match = use_semantic_match;
+        this.remove_watermark = remove_watermark;
+    }
+}

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/extraction/entity/KeyInfo.java

+package cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity;
+
+/**
+ * @author alex.yao
+ * @date 2025/5/12
+ */
+public class KeyInfo {
+    public String key_info;
+    public String[] paraphrase_names;
+    public String field_type;
+    public boolean is_in_table;
+    public String[] keywords;
+
+    public KeyInfo() {
+    }
+
+    public KeyInfo(String key_info, String[] paraphrase_names, String field_type, boolean is_in_table, String[] keywords) {
+        this.key_info = key_info;
+        this.paraphrase_names = paraphrase_names;
+        this.field_type = field_type;
+        this.is_in_table = is_in_table;
+        this.keywords = keywords;
+    }
+
+    public String getKey_info() {
+        return key_info;
+    }
+
+    public void setKey_info(String key_info) {
+        this.key_info = key_info;
+    }
+
+    public String[] getParaphrase_names() {
+        return paraphrase_names;
+    }
+
+    public void setParaphrase_names(String[] paraphrase_names) {
+        this.paraphrase_names = paraphrase_names;
+    }
+
+    public String getField_type() {
+        return field_type;
+    }
+
+    public void setField_type(String field_type) {
+        this.field_type = field_type;
+    }
+
+    public boolean isIs_in_table() {
+        return is_in_table;
+    }
+
+    public void setIs_in_table(boolean is_in_table) {
+        this.is_in_table = is_in_table;
+    }
+
+    public String[] getKeywords() {
+        return keywords;
+    }
+
+    public void setKeywords(String[] keywords) {
+        this.keywords = keywords;
+    }
+}

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/extraction/entity/RequestData.java

+package cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity;
+
+/**
+ * @author alex.yao
+ * @date 2025/5/12
+ */
+public class RequestData {
+    public String creator;
+    public Config config;
+    public String filedata;
+    public String filename;
+    public KeyInfo[] key_info_list;
+
+    public RequestData(String creator, Config config, String filedata, String filename, KeyInfo[] key_info_list) {
+        this.creator = creator;
+        this.config = config;
+        this.filedata = filedata;
+        this.filename = filename;
+        this.key_info_list = key_info_list;
+    }
+}

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/text_in_pdf2md/PdfToMDFunction.java

@@ -7,7 +7,7 @@ import cn.com.poc.thirdparty.resource.demand.ai.function.AbstractLargeModelFunct
 import cn.com.poc.thirdparty.resource.demand.ai.function.entity.FunctionLLMConfig;
 import cn.com.poc.thirdparty.resource.demand.ai.function.entity.Parameters;
 import cn.com.poc.thirdparty.resource.demand.ai.function.entity.Properties;
-import cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api.OCRClient;
+import cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api.TextInClient;
 import cn.hutool.core.collection.ListUtil;
 import com.alibaba.fastjson.JSONObject;
 import com.fasterxml.jackson.databind.JsonNode;
@@ -63,7 +63,7 @@ public class PdfToMDFunction extends AbstractLargeModelFunction {
         options.put("paratext_mode", "annotation");
         options.put("parse_mode", "auto");
         options.put("table_flavor", "md");
-        OCRClient client = new OCRClient();
+        TextInClient client = new TextInClient();
         try {
             String response = client.recognize(fileContent, options);
             ObjectMapper mapper = new ObjectMapper();
@@ -72,6 +72,10 @@ public class PdfToMDFunction extends AbstractLargeModelFunction {
                 String markdown = jsonNode.get("result").get("markdown").asText();
                 result.setPromptContent(markdown);
                 result.setFunctionResult(markdown);
+            } else {
+                logger.warn("text in 文档信息提取异常:{}", response);
+                result.setFunctionResult(response);
+                result.setPromptContent("FAIL");
             }
             return result;
         } catch (Exception e) {

src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/text_in_pdf2md/api/OCRClient.java → src/main/java/cn/com/poc/thirdparty/resource/demand/ai/function/text_in_pdf2md/api/TextInClient.java

@@ -5,26 +5,36 @@ package cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api;
  * @date 2025/5/7
  */
 
+import cn.com.poc.common.utils.DocumentLoad;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity.Config;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity.KeyInfo;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.entity.RequestData;
+import cn.com.yict.framemax.core.exception.BusinessException;
+import com.fasterxml.jackson.databind.ObjectMapper;
+import org.jetbrains.annotations.NotNull;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
-import java.io.BufferedReader;
-import java.io.IOException;
-import java.io.InputStreamReader;
-import java.io.OutputStream;
+import java.io.*;
 import java.net.HttpURLConnection;
 import java.net.URL;
 import java.net.URLEncoder;
+import java.nio.charset.StandardCharsets;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.Base64;
 import java.util.HashMap;
+import java.util.List;
 import java.util.Map;
 
-public class OCRClient {
+public class TextInClient {
 
-    private Logger logger = LoggerFactory.getLogger(OCRClient.class);
+    private Logger logger = LoggerFactory.getLogger(TextInClient.class);
     private final String appId = "dafd04a574230c00ccba61132160de0c";
     private final String secretCode = "3bc03c7e6f9402963e6e71d16d786a9c";
-    private final String baseUrl = "https://api.textin.com/ai/service/v1/pdf_to_markdown";
-    public OCRClient() {
+
+    public TextInClient() {
     }
 
     public String recognize(byte[] fileContent, HashMap<String, Object> options) throws IOException {
@@ -37,16 +47,7 @@ public class OCRClient {
                     .append("=")
                     .append(URLEncoder.encode(entry.getValue().toString(), "UTF-8"));
         }
-
-        String fullUrl = baseUrl + (queryParams.length() > 0 ? "?" + queryParams : "");
-        URL url = new URL(fullUrl);
-
-        HttpURLConnection connection = (HttpURLConnection) url.openConnection();
-        connection.setRequestMethod("POST");
-        connection.setRequestProperty("x-ti-app-id", appId);
-        connection.setRequestProperty("x-ti-secret-code", secretCode);
-        connection.setRequestProperty("Content-Type", "text/plain;charset=utf-8");
-        connection.setDoOutput(true);
+        HttpURLConnection connection = getRecoGinzeHttpURLConnection(queryParams);
         try (OutputStream os = connection.getOutputStream()) {
             os.write(fileContent);
             os.flush();
@@ -69,4 +70,73 @@ public class OCRClient {
         }
     }
 
+    public String extraction(String fileUrl, List<KeyInfo> keyInfoList) {
+        try {
+            // 读取文件并将其转换为Base64编码
+            File file = DocumentLoad.downloadURLDocument(fileUrl);
+            byte[] fileData = Files.readAllBytes(file.toPath());
+            String base64FileData = Base64.getEncoder().encodeToString(fileData);
+
+            // 获取文件名
+            String fileName = file.getName();
+
+            // 构建请求数据
+            Config config = new Config("table", "true", "true", "false");
+            RequestData requestData = new RequestData("", config, base64FileData, fileName, keyInfoList.toArray(new KeyInfo[0]));
+
+            // 创建ObjectMapper对象，序列化Java对象为JSON
+            ObjectMapper objectMapper = new ObjectMapper();
+            String requestDataJson = objectMapper.writeValueAsString(requestData);
+
+            // 创建URL对象
+            URL url = new URL("https://doc-compare.intsig.com/api/contracts/v3/extraction/external/create");
+
+            // 打开HTTP连接
+            HttpURLConnection connection = (HttpURLConnection) url.openConnection();
+            connection.setRequestMethod("POST");
+            connection.setRequestProperty("x-ti-app-id", appId);
+            connection.setRequestProperty("x-ti-secret-code", secretCode);
+            connection.setRequestProperty("Content-Type", "application/json");
+            connection.setDoOutput(true); // 开启输出流
+
+            // 发送请求数据
+            try (OutputStream os = connection.getOutputStream()) {
+                byte[] input = requestDataJson.getBytes(StandardCharsets.UTF_8);
+                os.write(input, 0, input.length);
+            }
+
+            // 获取响应代码
+            int status = connection.getResponseCode();
+            logger.info("Response Code: {}", status);
+
+            // 读取响应内容
+            try (BufferedReader in = new BufferedReader(new InputStreamReader(connection.getInputStream()))) {
+                String inputLine;
+                StringBuilder response = new StringBuilder();
+                while ((inputLine = in.readLine()) != null) {
+                    response.append(inputLine);
+                }
+                // 输出响应内容
+                return response.toString();
+            }
+
+        } catch (IOException e) {
+            throw new BusinessException(e);
+        }
+    }
+
+    private HttpURLConnection getRecoGinzeHttpURLConnection(StringBuilder queryParams) throws IOException {
+        String baseUrl = "https://api.textin.com/ai/service/v1/pdf_to_markdown";
+        String fullUrl = baseUrl + (queryParams.length() > 0 ? "?" + queryParams : "");
+        URL url = new URL(fullUrl);
+
+        HttpURLConnection connection = (HttpURLConnection) url.openConnection();
+        connection.setRequestMethod("POST");
+        connection.setRequestProperty("x-ti-app-id", appId);
+        connection.setRequestProperty("x-ti-secret-code", secretCode);
+        connection.setRequestProperty("Content-Type", "text/plain;charset=utf-8");
+        connection.setDoOutput(true);
+        return connection;
+    }
+
 }
\ No newline at end of file

src/test/java/cn/com/poc/thirdparty/resource/demand/ai/function/PdfToMdFunctionTest.java

 package cn.com.poc.thirdparty.resource.demand.ai.function;
 
-import cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api.OCRClient;
+import cn.com.poc.common.utils.JsonUtils;
+import cn.com.poc.thirdparty.resource.demand.ai.function.extraction.ContractExtractionFunction;
+import cn.com.poc.thirdparty.resource.demand.ai.function.text_in_pdf2md.api.TextInClient;
 import cn.com.yict.framemax.core.spring.SingleContextInitializer;
 import com.fasterxml.jackson.databind.JsonNode;
 import com.fasterxml.jackson.databind.ObjectMapper;
@@ -10,6 +12,7 @@ import org.springframework.test.context.ContextConfiguration;
 import org.springframework.test.context.junit4.SpringJUnit4ClassRunner;
 import org.springframework.test.context.web.WebAppConfiguration;
 
+import javax.annotation.Resource;
 import java.nio.charset.StandardCharsets;
 import java.util.HashMap;
 
@@ -40,7 +43,7 @@ public class PdfToMdFunctionTest {
         options.put("paratext_mode", "annotation");
         options.put("parse_mode", "auto");
         options.put("table_flavor", "md");
-        OCRClient client = new OCRClient();
+        TextInClient client = new TextInClient();
         try {
             String response = client.recognize(fileContent, options);
             ObjectMapper mapper = new ObjectMapper();
@@ -48,9 +51,20 @@ public class PdfToMdFunctionTest {
             if (jsonNode.has("result") && jsonNode.get("result").has("markdown")) {
                 String markdown = jsonNode.get("result").get("markdown").asText();
                 System.out.println(markdown);
+            }else{
+                System.out.println(response);
             }
         } catch (Exception e) {
             System.out.println("1111111");
         }
     }
+
+    @Resource
+    private ContractExtractionFunction contractExtractionFunction;
+
+    @Test
+    public void test_cefunction() {
+        System.out.println(contractExtractionFunction.getLLMConfig());
+
+    }
 }