huanghai
/
YunNanDsBase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

'commit'

Browse Source
main
HuangHai 1 month ago
parent 729c812330
commit 035eed1882
2 changed files with 194 additions and 86 deletions
Show Stats Download Patch File Download Diff File

252
src/main/java/com/dsideal/base/AI/TestMax32K.java
Unescape View File

@ -4,12 +4,17 @@ import com.dsideal.base.DataEase.Model.DataEaseModel;
import com.dsideal.base.Util.LocalMysqlConnectUtil;
import com.jfinal.plugin.activerecord.Db;
import com.jfinal.plugin.activerecord.Record;
import com.dsideal.base.Util.CallDeepSeek;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.LinkedHashSet;
import java.util.ArrayList;
import java.util.concurrent.CountDownLatch;
import java.text.SimpleDateFormat;
import java.util.Date;
import cn.hutool.core.io.FileUtil;
import java.io.File;
public class TestMax32K {
private static final int MAX_CHUNK_SIZE = 30000; // 30K字符限制
@ -17,15 +22,182 @@ public class TestMax32K {
public static void main(String[] args) {
LocalMysqlConnectUtil.Init();
// 直接调用生成综合报告
String report = generateComprehensiveReport();
System.out.println("\n=== 最终报告 ===");
System.out.println(report);
}
/**
* 分割过大的单表数据
*/
private static List<String> splitLargeTable(String tableName, Set<String> fieldNames,
List<Record> allTableData, int maxSize) {
List<String> chunks = new ArrayList<>();
StringBuilder currentTableChunk = new StringBuilder();
for (Record dataRecord : allTableData) {
Map<String, Object> columns = dataRecord.getColumns();
StringBuilder rowData = new StringBuilder();
rowData.append("[");
boolean first = true;
for (String fieldName : fieldNames) {
if (!first) rowData.append(",");
Object value = columns.get(fieldName);
if (value instanceof String) {
rowData.append("\"").append(value).append("\"");
} else {
rowData.append(value);
}
first = false;
}
rowData.append("]\n");
// 检查是否超过限制
if (currentTableChunk.length() + rowData.length() > maxSize) {
if (currentTableChunk.length() > 0) {
chunks.add(currentTableChunk.toString());
currentTableChunk = new StringBuilder();
}
}
currentTableChunk.append(rowData);
}
if (currentTableChunk.length() > 0) {
chunks.add(currentTableChunk.toString());
}
return chunks;
}
/**
* 获取分块数据的方法（可供其他类调用）
*/
public static String[] getDataChunks() {
// 这里可以将main方法中的逻辑提取出来，返回分块数据
// 为了简化，这里只是示例
return new String[]{"示例数据块1", "示例数据块2"};
}
public static String generateComprehensiveReport() {
String[] regions = {"文山州", "楚雄州"};
String sql = "select table_name as TABLE_NAME from core_dataset_table where dataset_group_id in (select id from core_dataset_group where pid='1036317909951057920')";
List<Record> tableList = Db.use(DataEaseModel.DB_NAME).find(sql);
// 使用字符串数组存储分块数据
// 获取分块数据
String[] dataChunks = getDataChunks(regions, tableList);
List<String> chunkAnalyses = new ArrayList<>();
System.out.println("开始分析 " + dataChunks.length + " 个数据块...");
// 第一阶段：流式分析各个数据块
for (int i = 0; i < dataChunks.length; i++) {
final int chunkIndex = i;
final StringBuilder chunkResult = new StringBuilder();
final CountDownLatch latch = new CountDownLatch(1);
String prompt = "请对以下教育数据进行简要分析，重点关注关键指标和趋势，控制在500字以内：\n" + dataChunks[i];
System.out.println("\n=== 正在分析第 " + (i + 1) + " 个数据块 ===");
CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() {
@Override
public void onData(String data) {
System.out.print(data);
chunkResult.append(data);
}
@Override
public void onComplete(String fullResponse) {
System.out.println("\n--- 第 " + (chunkIndex + 1) + " 个数据块分析完成 ---\n");
chunkAnalyses.add(chunkResult.toString());
latch.countDown();
}
@Override
public void onError(String error) {
System.err.println("分析第 " + (chunkIndex + 1) + " 个数据块时出错: " + error);
chunkAnalyses.add("分析失败: " + error);
latch.countDown();
}
});
try {
// 等待当前块分析完成
latch.await();
Thread.sleep(1000); // 稍微延迟，避免API调用过于频繁
} catch (InterruptedException e) {
System.err.println("等待分析结果时被中断: " + e.getMessage());
}
}
// 第二阶段：流式生成综合报告
System.out.println("\n=== 开始生成综合分析报告 ===");
StringBuilder combinedAnalysis = new StringBuilder();
combinedAnalysis.append("基于以下分块分析结果，请生成一份完整的教育数据综合分析报告（3000字以内）：\n\n");
for (int i = 0; i < chunkAnalyses.size(); i++) {
combinedAnalysis.append("数据块").append(i + 1).append("分析：\n");
combinedAnalysis.append(chunkAnalyses.get(i)).append("\n\n");
}
final StringBuilder finalReport = new StringBuilder();
final CountDownLatch finalLatch = new CountDownLatch(1);
CallDeepSeek.callDeepSeekStream(combinedAnalysis.toString(), new CallDeepSeek.SSEListener() {
@Override
public void onData(String data) {
System.out.print(data);
finalReport.append(data);
}
@Override
public void onComplete(String fullResponse) {
System.out.println("\n\n=== 综合分析报告生成完成 ===");
// 保存报告到文件
try {
String timestamp = new SimpleDateFormat("yyyyMMdd_HHmmss").format(new Date());
String fileName = "教育数据综合分析报告_" + timestamp + ".txt";
String filePath = "WebRoot/upload/" + fileName;
FileUtil.writeString(finalReport.toString(), new File(filePath), "UTF-8");
System.out.println("报告已保存到: " + filePath);
} catch (Exception e) {
System.err.println("保存报告时出错: " + e.getMessage());
}
finalLatch.countDown();
}
@Override
public void onError(String error) {
System.err.println("生成综合报告时出错: " + error);
finalReport.append("生成失败: ").append(error);
finalLatch.countDown();
}
});
try {
finalLatch.await();
} catch (InterruptedException e) {
System.err.println("等待最终报告时被中断: " + e.getMessage());
}
return finalReport.toString();
}
/**
* 提取数据分块逻辑为独立方法
*/
private static String[] getDataChunks(String[] regions, List<Record> tableList) {
List<String> dataChunks = new ArrayList<>();
StringBuilder currentChunk = new StringBuilder();
// 添加数据说明头部
String header = "数据说明: 以下是云南省教育数据的压缩格式\n" +
"格式: 表名 -> 字段列表 -> 数据行(数组格式)\n" +
"地区范围: " + String.join(",", regions) + "\n\n";
@ -41,7 +213,7 @@ public class TestMax32K {
// 为每个地区收集数据
for (String region : regions) {
sql = "select * from `" + tableName + "` where `行政区划`=?";
String sql = "select * from `" + tableName + "` where `行政区划`=?";
List<Record> listContent = Db.use(DataEaseModel.DB_NAME).find(sql, region);
if (!listContent.isEmpty()) {
@ -114,76 +286,6 @@ public class TestMax32K {
dataChunks.add(currentChunk.toString());
}
// 输出分块结果统计
System.out.println("总共分成 " + dataChunks.size() + " 个数据块:");
for (int i = 0; i < dataChunks.size(); i++) {
String chunk = dataChunks.get(i);
System.out.println("数据块 " + (i + 1) + " 长度: " + chunk.length() + " 字符");
}
// 返回分块数据数组供后续使用
String[] chunksArray = dataChunks.toArray(new String[0]);
// 示例：如何使用分块数据
System.out.println("\n=== 可以这样使用分块数据 ===");
for (int i = 0; i < chunksArray.length; i++) {
System.out.println("处理第 " + (i + 1) + " 个数据块...");
// 这里可以调用DeepSeek API处理每个块
// String result = CallDeepSeek.callDeepSeek(chunksArray[i]);
System.out.println("块 " + (i + 1) + " 内容预览: " +
chunksArray[i].substring(0, Math.min(200, chunksArray[i].length())) + "...");
}
}
/**
* 分割过大的单表数据
*/
private static List<String> splitLargeTable(String tableName, Set<String> fieldNames,
List<Record> allTableData, int maxSize) {
List<String> chunks = new ArrayList<>();
StringBuilder currentTableChunk = new StringBuilder();
for (Record dataRecord : allTableData) {
Map<String, Object> columns = dataRecord.getColumns();
StringBuilder rowData = new StringBuilder();
rowData.append("[");
boolean first = true;
for (String fieldName : fieldNames) {
if (!first) rowData.append(",");
Object value = columns.get(fieldName);
if (value instanceof String) {
rowData.append("\"").append(value).append("\"");
} else {
rowData.append(value);
}
first = false;
}
rowData.append("]\n");
// 检查是否超过限制
if (currentTableChunk.length() + rowData.length() > maxSize) {
if (currentTableChunk.length() > 0) {
chunks.add(currentTableChunk.toString());
currentTableChunk = new StringBuilder();
}
}
currentTableChunk.append(rowData);
}
if (currentTableChunk.length() > 0) {
chunks.add(currentTableChunk.toString());
}
return chunks;
}
/**
* 获取分块数据的方法（可供其他类调用）
*/
public static String[] getDataChunks() {
// 这里可以将main方法中的逻辑提取出来，返回分块数据
// 为了简化，这里只是示例
return new String[]{"示例数据块1", "示例数据块2"};
return dataChunks.toArray(new String[0]);
}
}

24
src/main/java/com/dsideal/base/Util/CallDeepSeek.java
Unescape View File

@ -200,18 +200,23 @@ public class CallDeepSeek {
/**
* 同步调用DeepSeek API（非流式响应）
*
* @param prompt 用户提示词
* @return DeepSeek的完整响应内容
* @return 完整的响应内容
*/
public static String callDeepSeek(String prompt) {
try {
System.out.println("开始同步调用DeepSeek API...");
JSONObject jsonPayload = createRequestPayload(prompt);
// 设置为非流式响应
jsonPayload.set("stream", false);
jsonPayload.put("stream", false);
HttpClient client = createHttpClient();
java.net.http.HttpRequest request = createHttpRequest(jsonPayload);
java.net.http.HttpRequest request = java.net.http.HttpRequest.newBuilder()
.uri(URI.create(API_URL))
.header("Content-Type", "application/json")
.header("Authorization", "Bearer " + API_KEY)
.POST(java.net.http.HttpRequest.BodyPublishers.ofString(jsonPayload.toString(), StandardCharsets.UTF_8))
.build();
HttpResponse<String> response = client.send(request, HttpResponse.BodyHandlers.ofString());
@ -224,15 +229,16 @@ public class CallDeepSeek {
.getStr("content", "");
}
} else {
System.err.println("DeepSeek API调用失败: " + response.statusCode() + " Body: " + response.body());
return null;
System.err.println("API请求失败，状态码: " + response.statusCode());
System.err.println("响应内容: " + response.body());
return "API调用失败: " + response.statusCode();
}
} catch (Exception e) {
System.err.println("调用DeepSeek API时出错: " + e.getMessage());
System.err.println("调用DeepSeek API时发生错误: " + e.getMessage());
e.printStackTrace();
return null;
return "调用失败: " + e.getMessage();
}
return null;
return "";
}
}

Write Preview
Loading…
Cancel
Save