From 12b4f2bb9421bbbd89601cbf8d12641392bf53c8 Mon Sep 17 00:00:00 2001 From: HuangHai <10402852@qq.com> Date: Mon, 16 Jun 2025 09:02:59 +0800 Subject: [PATCH] 'commit' --- .../java/com/dsideal/base/AI/TestMax32K.java | 259 ++++++------------ 1 file changed, 79 insertions(+), 180 deletions(-) diff --git a/src/main/java/com/dsideal/base/AI/TestMax32K.java b/src/main/java/com/dsideal/base/AI/TestMax32K.java index 6a65c62e..2cb1241f 100644 --- a/src/main/java/com/dsideal/base/AI/TestMax32K.java +++ b/src/main/java/com/dsideal/base/AI/TestMax32K.java @@ -18,15 +18,16 @@ 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; + import java.util.Scanner; + import com.dsideal.base.AI.Generator.WordGenerator; public class TestMax32K { private static final int MAX_CHUNK_SIZE = 8000; // 8000字符限制 private static boolean dataSubmitted = false; // 标记数据是否已提交 - + private static final String[] regions = {"楚雄州", "文山州"}; // 或者根据需要设置具体的地区 + public static void main(String[] args) { Scanner scanner = new Scanner(System.in); LocalMysqlConnectUtil.Init(); @@ -37,7 +38,7 @@ public class TestMax32K { submitAllDataChunks(); dataSubmitted = true; System.out.println("\n数据提交完成！"); - + // 第二步：询问用户想生成哪种格式 while (true) { System.out.println("\n=== 报告生成选项 ==="); @@ -47,9 +48,9 @@ public class TestMax32K { System.out.println("3. PPT演示文稿 (.pptx)"); System.out.println("4. 退出程序"); System.out.print("\n请选择您想生成的报告格式 (1-4): "); - + String choice = scanner.nextLine().trim(); - + switch (choice) { case "1": generateWordReport(); @@ -67,7 +68,7 @@ public class TestMax32K { System.out.println("无效选择，请输入 1-4 之间的数字。"); } } - + } catch (Exception e) { System.err.println("程序执行出错: " + e.getMessage()); e.printStackTrace(); @@ -75,7 +76,7 @@ public class TestMax32K { scanner.close(); } } - + /** * 提交所有数据块到大模型 */ @@ -84,22 +85,21 @@ public class TestMax32K { System.out.println("数据已经提交过了，无需重复提交。"); return; } - + // 添加获取参数的代码 - String[] regions = { "楚雄州","文山州"}; // 或者根据需要设置具体的地区 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 tableList = Db.use(DataEaseModel.DB_NAME).find(sql); - + // 修改这一行，传递正确的参数 String[] dataChunks = getDataChunks(regions, tableList); System.out.println("总共需要提交 " + dataChunks.length + " 个数据块"); - + CountDownLatch latch = new CountDownLatch(dataChunks.length); - + for (int i = 0; i < dataChunks.length; i++) { final int chunkIndex = i; final boolean isLastChunk = (i == dataChunks.length - 1); - + String prompt; if (isLastChunk) { prompt = "这是最后一个数据块（第" + (chunkIndex + 1) + "/" + dataChunks.length + "个）。\n" + @@ -110,7 +110,7 @@ public class TestMax32K { "请接收此数据块，无需分析，等待所有数据提交完成。\n\n" + "数据内容：\n" + dataChunks[chunkIndex]; } - + CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() { @Override public void onData(String data) { @@ -118,27 +118,27 @@ public class TestMax32K { System.out.println("\n大模型确认: " + data); } } - + @Override public void onComplete(String fullResponse) { System.out.println("数据块 " + (chunkIndex + 1) + "/" + dataChunks.length + " 提交完成"); latch.countDown(); } - + @Override public void onError(String error) { System.err.println("数据块 " + (chunkIndex + 1) + " 提交失败: " + error); latch.countDown(); } }); - + // 避免请求过于频繁 Thread.sleep(1000); } - + latch.await(); } - + /** * 生成Word报告 */ @@ -147,9 +147,9 @@ public class TestMax32K { System.out.println("请先提交数据！"); return; } - + System.out.println("\n开始生成Word报告..."); - + try { String prompt = "基于之前提交的所有数据，请生成一份详细的Word格式分析报告。\n" + "要求：\n" + @@ -157,52 +157,50 @@ public class TestMax32K { "2. 包含完整的数据分析和洞察\n" + "3. 结构清晰，包含标题、摘要、详细分析、结论等部分\n" + "4. 适合转换为Word文档格式"; - + StringBuilder wordContent = new StringBuilder(); CountDownLatch latch = new CountDownLatch(1); - + CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() { @Override public void onData(String data) { wordContent.append(data); System.out.print("."); // 显示进度 } - + @Override public void onComplete(String fullResponse) { System.out.println("\nWord报告生成完成！"); - + // 保存Word文件 try { String timestamp = new SimpleDateFormat("yyyyMMdd_HHmmss").format(new Date()); String fileName = "analysis_report_" + timestamp + ".docx"; - // 使用WordGenerator生成Word文档 - String[] regions = {"云南省"}; // 或者根据实际情况设置地区 WordGenerator.generateWordDocument(wordContent.toString(), fileName, regions); - + System.out.println("Word报告已保存至: WebRoot/upload/" + fileName); } catch (Exception e) { System.err.println("保存Word文档失败: " + e.getMessage()); } - + latch.countDown(); } - + @Override public void onError(String error) { System.err.println("\n生成Word报告失败: " + error); latch.countDown(); } }); - + latch.await(); - + } catch (Exception e) { System.err.println("生成Word报告时出错: " + e.getMessage()); } } - + /** * 生成HTML报告 */ @@ -211,9 +209,9 @@ public class TestMax32K { System.out.println("请先提交数据！"); return; } - + System.out.println("\n开始生成HTML报告..."); - + try { String prompt = "基于之前提交的所有数据，请生成一份详细的HTML格式分析报告。\n" + "要求：\n" + @@ -232,51 +230,51 @@ public class TestMax32K { "4. 使用ECharts库实现图表\n" + "5. 页面结构清晰，包含标题、摘要、图表展示、数据分析、结论等部分\n" + "6. 确保所有图表都有真实的数据和完整的配置"; - + StringBuilder htmlContent = new StringBuilder(); CountDownLatch latch = new CountDownLatch(1); - + CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() { @Override public void onData(String data) { htmlContent.append(data); System.out.print("."); // 显示进度 } - + @Override public void onComplete(String fullResponse) { System.out.println("\nHTML报告生成完成！"); - + // 保存HTML文件 try { String timestamp = new SimpleDateFormat("yyyyMMdd_HHmmss").format(new Date()); String fileName = "analysis_report_" + timestamp + ".html"; String filePath = "WebRoot/upload/" + fileName; - + Files.write(Paths.get(filePath), htmlContent.toString().getBytes("UTF-8")); - + System.out.println("HTML报告已保存至: " + filePath); } catch (Exception e) { System.err.println("保存HTML文件失败: " + e.getMessage()); } - + latch.countDown(); } - + @Override public void onError(String error) { System.err.println("\n生成HTML报告失败: " + error); latch.countDown(); } }); - + latch.await(); - + } catch (Exception e) { System.err.println("生成HTML报告时出错: " + e.getMessage()); } } - + /** * 生成PPT报告 */ @@ -285,9 +283,9 @@ public class TestMax32K { System.out.println("请先提交数据！"); return; } - + System.out.println("\n开始生成PPT报告..."); - + try { String prompt = "基于之前提交的所有数据，请生成一份PPT演示文稿的详细内容大纲。\n" + "要求：\n" + @@ -296,30 +294,30 @@ public class TestMax32K { "3. 每页PPT都要有明确的标题和要点\n" + "4. 适合制作成专业的演示文稿\n" + "5. 内容要简洁明了，重点突出"; - + StringBuilder pptContent = new StringBuilder(); CountDownLatch latch = new CountDownLatch(1); - + CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() { @Override public void onData(String data) { pptContent.append(data); System.out.print("."); // 显示进度 } - + @Override public void onComplete(String fullResponse) { System.out.println("\nPPT内容生成完成！"); - + // 调用PptGenerator生成PPT try { String timestamp = new SimpleDateFormat("yyyyMMdd_HHmmss").format(new Date()); String fileName = "analysis_report_" + timestamp + ".pptx"; - + // 使用PptGenerator生成PPT String token = PptAIKit.createApiToken("dsideal", 1000); String pptInfo = PptGenerator.generatePptFromMarkdown(pptContent.toString(), token); - + // 保存PPT信息到文本文件 String infoFilePath = "WebRoot/upload/ppt_info_" + timestamp + ".txt"; if (pptInfo != null) { @@ -329,28 +327,28 @@ public class TestMax32K { Files.write(Paths.get(infoFilePath), "PPT生成失败".getBytes("UTF-8")); System.out.println("PPT生成失败，错误信息已保存至: " + infoFilePath); } - + } catch (Exception e) { System.err.println("生成PPT失败: " + e.getMessage()); } - + latch.countDown(); } - + @Override public void onError(String error) { System.err.println("\n生成PPT报告失败: " + error); latch.countDown(); } }); - + latch.await(); - + } catch (Exception e) { System.err.println("生成PPT报告时出错: " + e.getMessage()); } } - + /** * 分割过大的单表数据 */ @@ -358,12 +356,12 @@ public class TestMax32K { List allTableData, int maxSize) { List chunks = new ArrayList<>(); StringBuilder currentTableChunk = new StringBuilder(); - + for (Record dataRecord : allTableData) { Map columns = dataRecord.getColumns(); StringBuilder rowData = new StringBuilder(); rowData.append("["); - + boolean first = true; for (String fieldName : fieldNames) { if (!first) rowData.append(","); @@ -376,7 +374,7 @@ public class TestMax32K { first = false; } rowData.append("]\n"); - + // 检查是否超过限制 if (currentTableChunk.length() + rowData.length() > maxSize) { if (!currentTableChunk.isEmpty()) { @@ -386,138 +384,39 @@ public class TestMax32K { } currentTableChunk.append(rowData); } - + if (!currentTableChunk.isEmpty()) { chunks.add(currentTableChunk.toString()); } - + return chunks; } - public static String generateComprehensiveReport(String[] regions) { - - 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 tableList = Db.use(DataEaseModel.DB_NAME).find(sql); - - // 获取分块数据 - String[] dataChunks = getDataChunks(regions, tableList); - - System.out.println("开始逐步提交 " + dataChunks.length + " 个数据块..."); - - final StringBuilder finalReport = new StringBuilder(); - final CountDownLatch finalLatch = new CountDownLatch(1); - - // 逐步提交数据块 - for (int i = 0; i < dataChunks.length; i++) { - final int chunkIndex = i; - final boolean isLastChunk = (i == dataChunks.length - 1); - - String prompt; - if (isLastChunk) { - // 最后一个数据块：要求返回完整分析报告 - prompt = "这是最后一部分教育数据，请基于之前提交的所有数据生成一份完整的综合分析报告（3000字以内）：\n" + dataChunks[i]; - } else { - // 中间数据块：只提交数据，不要求返回分析 - prompt = "这是第" + (i + 1) + "部分教育数据，共" + dataChunks.length + "部分，请接收并记录，暂不需要分析：\n" + dataChunks[i]; - } - - System.out.println("\n=== 提交第 " + (i + 1) + "/" + dataChunks.length + " 个数据块 ==="); - - final CountDownLatch chunkLatch = new CountDownLatch(1); - - CallDeepSeek.callDeepSeekStream(prompt, new CallDeepSeek.SSEListener() { - @Override - public void onData(String data) { - if (isLastChunk) { - // 只有最后一个数据块才显示和保存返回内容 - System.out.print(data); - finalReport.append(data); - } else { - // 中间数据块的响应不显示（或只显示确认信息） - // System.out.print("."); // 可选：显示进度点 - } - } - - @Override - public void onComplete(String fullResponse) { - if (isLastChunk) { - 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(); - } else { - System.out.println("第 " + (chunkIndex + 1) + " 个数据块已提交"); - } - chunkLatch.countDown(); - } - - @Override - public void onError(String error) { - System.err.println("提交第 " + (chunkIndex + 1) + " 个数据块时出错: " + error); - if (isLastChunk) { - finalReport.append("生成失败: ").append(error); - finalLatch.countDown(); - } - chunkLatch.countDown(); - } - }); - - try { - // 等待当前块处理完成 - chunkLatch.await(); - if (!isLastChunk) { - Thread.sleep(1000); // 中间块之间稍微延迟 - } - } catch (InterruptedException e) { - System.err.println("等待数据块处理时被中断: " + e.getMessage()); - } - } - - 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 tableList) { List dataChunks = new ArrayList<>(); StringBuilder currentChunk = new StringBuilder(); - + 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"; + "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"; currentChunk.append(header); - + // 遍历所有相关数据表 for (Record record : tableList) { String tableName = record.getStr("TABLE_NAME"); - + // 为当前表收集所有数据 List allTableData = new ArrayList<>(); Set fieldNames = new LinkedHashSet<>(); - + // 为每个地区收集数据 for (String region : regions) { String sql = "select * from `" + tableName + "` where `行政区划`=?"; List listContent = Db.use(DataEaseModel.DB_NAME).find(sql, region); - + if (!listContent.isEmpty()) { allTableData.addAll(listContent); // 收集字段名（使用第一条记录的字段结构） @@ -526,18 +425,18 @@ public class TestMax32K { } } } - + if (!allTableData.isEmpty()) { // 构建当前表的完整数据块 StringBuilder tableData = new StringBuilder(); tableData.append("\n表: ").append(tableName).append("\n"); tableData.append("字段: ").append(String.join(",", fieldNames)).append("\n"); - + // 输出压缩格式的数据 for (Record dataRecord : allTableData) { Map columns = dataRecord.getColumns(); tableData.append("["); - + boolean first = true; for (String fieldName : fieldNames) { if (!first) tableData.append(","); @@ -549,10 +448,10 @@ public class TestMax32K { } first = false; } - + tableData.append("]\n"); } - + // 检查是否需要分块 String tableDataStr = tableData.toString(); if (currentChunk.length() + tableDataStr.length() > MAX_CHUNK_SIZE) { @@ -562,7 +461,7 @@ public class TestMax32K { currentChunk = new StringBuilder(); currentChunk.append(header); } - + // 如果单个表数据超过限制，需要进一步分割 if (tableDataStr.length() > MAX_CHUNK_SIZE - header.length()) { List tableChunks = splitLargeTable(fieldNames, allTableData, MAX_CHUNK_SIZE - header.length()); @@ -582,12 +481,12 @@ public class TestMax32K { } } } - + // 添加最后一个块 if (currentChunk.length() > header.length()) { dataChunks.add(currentChunk.toString()); } - + return dataChunks.toArray(new String[0]); } }