feat:年度报告的财务指标解析完成

service-base/src/main/java/com/simuwang/base/pojo/dto/report/ReportFinancialIndicatorsDTO.java

@@ -1,16 +1,20 @@
 package com.simuwang.base.pojo.dto.report;
 
+import cn.hutool.core.util.StrUtil;
 import com.simuwang.base.pojo.dos.report.ReportFinancialIndicatorsDO;
 import lombok.Getter;
 import lombok.Setter;
 
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
 @Setter
 @Getter
 public class ReportFinancialIndicatorsDTO extends BaseReportLevelDTO<ReportFinancialIndicatorsDO> {
     /**
      * 年度
      */
-    private Integer yearly;
+    private String yearly;
     /**
      * 期末基金净资产
      */
@@ -31,6 +35,14 @@ public class ReportFinancialIndicatorsDTO extends BaseReportLevelDTO<ReportFinan
      * 期末可供分配利润
      */
     private String undistributedProfit;
+    /**
+     * 期末可供分配基金份额利润
+     */
+    private String undistributedShareProfit;
+    /**
+     * 基金份额累计净值增长率
+     */
+    private String shareNavRet;
 
     public ReportFinancialIndicatorsDTO() {
         super();
@@ -49,12 +61,17 @@ public class ReportFinancialIndicatorsDTO extends BaseReportLevelDTO<ReportFinan
         ReportFinancialIndicatorsDO entity = new ReportFinancialIndicatorsDO();
         entity.setFileId(this.getFileId());
         entity.setLevel(this.getLevel());
-        entity.setEndDate(this.yearly);
         entity.setFundAssetSize(this.toBigDecimal(this.assetNet));
         entity.setNav(this.toBigDecimal(this.nav));
         entity.setProfit(this.toBigDecimal(this.profit));
         entity.setRealizedIncome(this.toBigDecimal(this.realizedIncome));
         entity.setUndistributedProfit(this.toBigDecimal(this.undistributedProfit));
+        if (StrUtil.isNotBlank(this.yearly)) {
+            Matcher matcher = Pattern.compile("\\d").matcher(this.yearly);
+            if (matcher.find()) {
+                entity.setEndDate(Integer.parseInt(matcher.group()));
+            }
+        }
         return entity;
     }
 
@@ -68,6 +85,8 @@ public class ReportFinancialIndicatorsDTO extends BaseReportLevelDTO<ReportFinan
                 ", profit=" + profit +
                 ", undistributedProfit=" + undistributedProfit +
                 ", realizedIncome=" + realizedIncome +
+                ", undistributedShareProfit=" + undistributedShareProfit +
+                ", shareNavRet=" + shareNavRet +
                 '}';
     }
 }

service-daq/src/main/java/com/simuwang/daq/components/report/parser/pdf/PDAnnuallyReportParser.java

@@ -3,16 +3,12 @@ package com.simuwang.daq.components.report.parser.pdf;
 import cn.hutool.core.collection.CollUtil;
 import cn.hutool.core.collection.ListUtil;
 import cn.hutool.core.map.MapUtil;
-import cn.hutool.core.util.StrUtil;
 import com.simuwang.base.mapper.EmailFieldMappingMapper;
 import com.simuwang.base.pojo.dto.report.*;
 import com.simuwang.daq.components.report.parser.ReportParserConstant;
 import org.springframework.stereotype.Component;
-import technology.tabula.RectangularTextContainer;
 import technology.tabula.Table;
 
-import java.awt.geom.Rectangle2D;
-import java.util.Comparator;
 import java.util.List;
 import java.util.Map;
 import java.util.function.Function;
@@ -23,12 +19,20 @@ import java.util.function.Function;
  * @description 年报解析逻辑:基本信息被拆分为多个表格，财务报表未解析
  */
 @Component(ReportParserConstant.PARSER_PDF_ANNUALLY)
-public class PDAnnuallyReportParser extends AbstractPDReportParser<AnnuallyReportData> {
+public class PDAnnuallyReportParser extends PDQuarterlyReportParser<AnnuallyReportData> {
+    private static final List<String> FINANCIAL_INDICATORS_COLUMN_NAMES = ListUtil.list(false);
+
+    static {
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("期末基金净资产");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("报告期期末单位净值");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("本期利润");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("本期已实现收益");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("期末可供分配利润");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("期末可供分配基金份额利润");
+        FINANCIAL_INDICATORS_COLUMN_NAMES.add("基金份额累计净值增长率");
+    }
+
     private final List<Table> fundInfoTables = ListUtil.list(true);
-    private final List<Table> shareChangeTables = ListUtil.list(true);
-    private final List<Table> assetAllocationTables = ListUtil.list(true);
-    private final List<Table> investmentIndustryTables = ListUtil.list(true);
-    private final List<Table> financialIndicatorsTables = ListUtil.list(true);
 
     public PDAnnuallyReportParser(EmailFieldMappingMapper fieldMappingMapper) {
         super(fieldMappingMapper);
@@ -49,41 +53,35 @@ public class PDAnnuallyReportParser extends AbstractPDReportParser<AnnuallyRepor
             }
             int colCount = table.getColCount();
             if (colCount == 2) {
-                // 用表格的第二行第一列的数据判断是否份额变动记录
-                String text = this.cleaningValue(table.getCell(1, 0).getText());
-                if (StrUtil.contains(text, "份额")) {
+                // 用表格的第一列的数据判断是否份额变动记录
+                List<String> texts = this.getTableColTexts(table, 0);
+                if (CollUtil.containsAny(texts, SHARE_CHANGE_COLUMN_NAMES)) {
                     this.shareChangeTables.add(table);
                 }
             } else if (colCount == 4) {
-                // 用表格的第一行第一列的数据判断是否主要财务指标数据
-                String text = this.cleaningValue(table.getCell(0, 0).getText());
-                if (StrUtil.contains(text, "期间数据和指标")) {
-                    this.financialIndicatorsTables.add(table);
-                    continue;
-                }
-                // 用表格的第一行第二列的数据判断是否行业配置数据
-                text = this.cleaningValue(table.getCell(0, 1).getText());
-                if (StrUtil.contains(text, "行业类别")) {
+                // 用表格的第二列的数据判断是否行业配置数据（内地）
+                List<String> texts = this.getTableColTexts(table, 1);
+                if (CollUtil.containsAny(texts, INDUSTRY_COLUMN_NAMES)) {
                     this.investmentIndustryTables.add(table);
                 }
             } else if (colCount == 3) {
-                // 用表格的第一行第一列的数据判断是否行业配置数据
-                String text = this.cleaningValue(table.getCell(0, 0).getText());
-                if (StrUtil.contains(text, "行业类别")) {
+                // 用表格的第一列的数据判断是否行业配置数据（港股通）
+                List<String> texts = this.getTableColTexts(table, 0);
+                if (CollUtil.containsAny(texts, INDUSTRY_COLUMN_NAMES)) {
                     this.investmentIndustryTables.add(table);
                     continue;
                 }
                 // 资产配置表格识别（兼容跨页的表格）获取表格中第二列的所有文字，判断所有文字中包含"股权投资"等字符串
-                List<String> details = ListUtil.list(false);
-                for (@SuppressWarnings("all") List<RectangularTextContainer> row : table.getRows()) {
-                    String detail = this.cleaningValue(row.get(1).getText(), false);
-                    if (StrUtil.isNotBlank(detail)) {
-                        details.add(detail);
-                    }
-                }
-                if (CollUtil.containsAny(details, ListUtil.of("股权投资", "股票投资", "债券投资"))) {
+                texts = this.getTableColTexts(table, 1);
+                if (CollUtil.containsAny(texts, ListUtil.of("股权投资", "股票投资", "债券投资", "另类投资", "其他资产", "其他融资总额"))) {
                     this.assetAllocationTables.add(table);
                 }
+            } else {
+                // 用表格的第一列的数据判断是否主要财务指标数据
+                List<String> texts = this.getTableColTexts(table, 0);
+                if (CollUtil.containsAny(texts, ListUtil.of("期间数据和指标", "期末数据和指标", "累计期末指标"))) {
+                    this.financialIndicatorsTables.add(table);
+                }
             }
         }
     }
@@ -101,46 +99,17 @@ public class PDAnnuallyReportParser extends AbstractPDReportParser<AnnuallyRepor
     }
 
     @Override
-    protected Map<String, Object> parseFundInfo(Table fundInfoTable) {
-        // 季报和年报的基金基本信息是两列的表格
-        Map<String, Object> baseInfoMap = MapUtil.newHashMap(32);
-        for (int i = 0; i < fundInfoTable.getRows().size(); i++) {
-            @SuppressWarnings("all")
-            List<RectangularTextContainer> cols = fundInfoTable.getRows().get(i);
-            for (int j = 0; j < 1; j++) {
-                baseInfoMap.put(cols.get(j).getText(), cols.get(j + 1).getText());
-            }
-        }
-        return baseInfoMap;
-    }
-
-    @Override
-    protected AnnuallyReportData parseExtInfoAndSetData(ReportBaseInfoDTO reportInfo, ReportFundInfoDTO fundInfo) {
-        Integer fileId = reportInfo.getFileId();
-        // 表格转换数据获取函数
-        Function<Table, Map<String, Object>> function = t -> {
-            Map<String, Object> extInfoMap = MapUtil.newHashMap(16);
-            for (int i = 0; i < t.getRowCount(); i++) {
-                String key = t.getCell(i, 0).getText();
-                String value = t.getCell(i, 1).getText();
-                extInfoMap.put(key, value);
-            }
-            return extInfoMap;
-        };
-        // 份额变动
-        List<ReportShareChangeDTO> shareChanges = this.buildLevelDto(fileId, this.shareChangeTables,
-                ReportShareChangeDTO.class, function);
-//        // 主要财务指标
-//        List<ReportFinancialIndicatorsDTO> financialIndicators = this.buildLevelDto(fileId, this.financialIndicatorsTables,
-//                ReportFinancialIndicatorsDTO.class, function);
-        // 资产配置
-        List<ReportAssetAllocationDTO> assetAllocations = this.buildAssetAllocationInfo(fileId);
-        // 行业配置
-        List<ReportInvestmentIndustryDTO> investmentIndustries = this.buildInvestmentIndustryInfo(fileId);
+    protected AnnuallyReportData buildExtData(ReportBaseInfoDTO reportInfo, ReportFundInfoDTO fundInfo,
+                                              List<ReportShareChangeDTO> shareChanges,
+                                              List<ReportAssetAllocationDTO> assetAllocations,
+                                              List<ReportInvestmentIndustryDTO> investmentIndustries,
+                                              Function<Table, Map<String, Object>> function) {
+        // 处理财务指标
+        List<ReportFinancialIndicatorsDTO> financialIndicators = this.buildFinancialIndicatorsInfo(reportInfo.getFileId());
         // 返回数据构建
         AnnuallyReportData reportData = new AnnuallyReportData(reportInfo, fundInfo);
         reportData.setShareChange(shareChanges);
-        reportData.setFinancialIndicators(null); // todo 财务指标
+        reportData.setFinancialIndicators(financialIndicators);
         reportData.setAssetAllocation(assetAllocations);
         reportData.setInvestmentIndustry(investmentIndustries);
         return reportData;
@@ -151,84 +120,28 @@ public class PDAnnuallyReportParser extends AbstractPDReportParser<AnnuallyRepor
         // todo 数据清洗
     }
 
-    /**
-     * 构建基金行业配置解析数据
-     *
-     * @return /
-     */
-    private List<ReportInvestmentIndustryDTO> buildInvestmentIndustryInfo(Integer fileId) {
-        List<ReportInvestmentIndustryDTO> dtos = ListUtil.list(false);
-        for (Table table : this.investmentIndustryTables) {
-            int colCount = table.getColCount();
-            // 投资地区: 1-境内, 2-港股通
-            int investType = colCount == 4 ? 1 : 2;
-            int j = colCount == 4 ? 1 : 0;
-            // 按行遍历
-            for (int i = 0; i < table.getRowCount(); i++) {
-                String text = this.cleaningValue(table.getCell(i, 0).getText());
-                if (StrUtil.containsAny(text, "序号", "行业类别")) {
-                    continue;
-                }
-                ReportInvestmentIndustryDTO dto = new ReportInvestmentIndustryDTO(fileId);
-                dto.setInvestType(investType);
-                dto.setIndustryName(this.cleaningValue(table.getCell(i, j).getText()));
-                dto.setMarketValue(this.cleaningValue(table.getCell(i, j + 1).getText()));
-                dto.setRatio(this.cleaningValue(table.getCell(i, j + 2).getText()));
-                dtos.add(dto);
-            }
-        }
-        return dtos;
-    }
-
-    /**
-     * 构建基金资产配置解析数据
-     *
-     * @param fileId 文件id
-     * @return /
-     */
-    private List<ReportAssetAllocationDTO> buildAssetAllocationInfo(Integer fileId) {
-        List<ReportAssetAllocationDTO> dtos = ListUtil.list(false);
-        String assetType = null;
-        for (Table table : this.assetAllocationTables) {
-            // 按行遍历
-            for (@SuppressWarnings("all") List<RectangularTextContainer> row : table.getRows()) {
-                // x坐标升序（防止部分行乱序问题）
-                row.sort(Comparator.comparing(Rectangle2D.Float::getX));
-                // 大类
-                String type = this.cleaningValue(row.get(0).getText());
-                if (StrUtil.isNotBlank(type)) {
-                    assetType = type;
-                }
-                // 金额、市值，有时是 “备注#金额”的格式
-                String marketValueAndRemark = this.cleaningValue(row.get(2).getText());
-                if (StrUtil.isBlank(marketValueAndRemark) || StrUtil.isBlank(assetType)) {
-                    continue;
-                }
-                // 资产明细
-                String detail = this.cleaningValue(row.get(1).getText(), false);
-                if (StrUtil.contains(marketValueAndRemark, "#")) {
-                    // 有#表示有备注，而且可能有多个，多个用分号分隔的.
-                    List<String> marketValueAndRemarks = StrUtil.split(marketValueAndRemark, ";");
-                    for (String mr : marketValueAndRemarks) {
-                        if (StrUtil.isBlank(mr)) {
-                            continue;
-                        }
-                        List<String> mrs = StrUtil.split(mr, "#");
-                        ReportAssetAllocationDTO dto = new ReportAssetAllocationDTO(fileId);
-                        dto.setAssetType(assetType);
-                        dto.setAssetDetails(detail);
-                        dto.setMarketValue(mrs.get(1));
-                        dto.setRemark(mrs.get(0));
-                        dtos.add(dto);
+    private List<ReportFinancialIndicatorsDTO> buildFinancialIndicatorsInfo(Integer fileId) {
+        List<ReportFinancialIndicatorsDTO> dtos = ListUtil.list(false);
+        // 这里不存在分级基金，可能会存在表格跨页
+        int colCount = this.financialIndicatorsTables.get(0).getColCount();
+        for (int j = 1; j < colCount; j++) {
+            Map<String, Object> infoMap = MapUtil.newHashMap(16);
+            for (Table table : this.financialIndicatorsTables) {
+                String year = this.cleaningValue(table.getCell(0, j).getText());
+                infoMap.put("年度", year);
+                for (int i = 0; i < table.getRowCount(); i++) {
+                    String columnName = this.cleaningValue(table.getCell(i, 0).getText());
+                    if (!CollUtil.contains(FINANCIAL_INDICATORS_COLUMN_NAMES, columnName)) {
+                        continue;
                     }
-                } else {
-                    ReportAssetAllocationDTO dto = new ReportAssetAllocationDTO(fileId);
-                    dto.setAssetType(assetType);
-                    dto.setAssetDetails(detail);
-                    dto.setMarketValue(marketValueAndRemark);
-                    dtos.add(dto);
+                    String value = this.cleaningValue(table.getCell(i, j).getText());
+                    infoMap.put(columnName, value);
                 }
             }
+            ReportFinancialIndicatorsDTO dto = new ReportFinancialIndicatorsDTO(fileId);
+            this.buildInfo(infoMap, dto);
+            dto.setLevel("母基金");
+            dtos.add(dto);
         }
         return dtos;
     }

service-daq/src/main/java/com/simuwang/daq/components/report/parser/pdf/PDQuarterlyReportParser.java

@@ -1,5 +1,6 @@
 package com.simuwang.daq.components.report.parser.pdf;
 
+import cn.hutool.core.collection.CollUtil;
 import cn.hutool.core.collection.ListUtil;
 import cn.hutool.core.map.MapUtil;
 import cn.hutool.core.util.StrUtil;
@@ -22,12 +23,60 @@ import java.util.function.Function;
  * @description pdf格式的季报解析逻辑
  */
 @Component(ReportParserConstant.PARSER_PDF_QUARTERLY)
-public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyReportData> {
+public class PDQuarterlyReportParser<T extends QuarterlyReportData> extends AbstractPDReportParser<T> {
+    protected static final List<String> INDUSTRY_COLUMN_NAMES = ListUtil.list(false);
+    protected static final List<String> SHARE_CHANGE_COLUMN_NAMES = ListUtil.list(false);
+
+    static {
+        // 中国证监会行业标准
+        INDUSTRY_COLUMN_NAMES.add("农、林、牧、渔业");
+        INDUSTRY_COLUMN_NAMES.add("采矿业");
+        INDUSTRY_COLUMN_NAMES.add("制造业");
+        INDUSTRY_COLUMN_NAMES.add("电力、热力、燃气及水生产和供应业");
+        INDUSTRY_COLUMN_NAMES.add("建筑业");
+        INDUSTRY_COLUMN_NAMES.add("批发和零售业");
+        INDUSTRY_COLUMN_NAMES.add("交通运输、仓储和邮政业");
+        INDUSTRY_COLUMN_NAMES.add("住宿和餐饮业");
+        INDUSTRY_COLUMN_NAMES.add("信息传输、软件和信息技术服务业");
+        INDUSTRY_COLUMN_NAMES.add("金融业");
+        INDUSTRY_COLUMN_NAMES.add("房地产业");
+        INDUSTRY_COLUMN_NAMES.add("租赁和商务服务业");
+        INDUSTRY_COLUMN_NAMES.add("科学研究和技术服务业");
+        INDUSTRY_COLUMN_NAMES.add("水利、环境和公共设施管理业");
+        INDUSTRY_COLUMN_NAMES.add("居民服务、修理和其他服务业");
+        INDUSTRY_COLUMN_NAMES.add("教育");
+        INDUSTRY_COLUMN_NAMES.add("卫生和社会工作");
+        INDUSTRY_COLUMN_NAMES.add("文化、体育和娱乐业");
+        INDUSTRY_COLUMN_NAMES.add("综合");
+
+        INDUSTRY_COLUMN_NAMES.add("港股通");
+
+        // 以下为国际标准
+        INDUSTRY_COLUMN_NAMES.add("能源");
+        INDUSTRY_COLUMN_NAMES.add("原材料");
+        INDUSTRY_COLUMN_NAMES.add("工业");
+        INDUSTRY_COLUMN_NAMES.add("非日常生活消费品");
+        INDUSTRY_COLUMN_NAMES.add("日常消费品");
+        INDUSTRY_COLUMN_NAMES.add("医疗保健");
+        INDUSTRY_COLUMN_NAMES.add("金融");
+        INDUSTRY_COLUMN_NAMES.add("信息技术");
+        INDUSTRY_COLUMN_NAMES.add("通讯服务");
+        INDUSTRY_COLUMN_NAMES.add("公用事业");
+        INDUSTRY_COLUMN_NAMES.add("房地产");
+
+        // 份额变动表格识别列
+        SHARE_CHANGE_COLUMN_NAMES.add("报告期期初基金份额总额");
+        SHARE_CHANGE_COLUMN_NAMES.add("减:报告期期间基金总赎回份额");
+        SHARE_CHANGE_COLUMN_NAMES.add("期末基金总份额/期末基金实缴总额");
+        SHARE_CHANGE_COLUMN_NAMES.add("报告期期间基金拆分变动份额");
+        SHARE_CHANGE_COLUMN_NAMES.add("报告期期间基金总申购份额");
+    }
+
+    protected final List<Table> financialIndicatorsTables = ListUtil.list(true);
+    protected final List<Table> shareChangeTables = ListUtil.list(true);
+    protected final List<Table> assetAllocationTables = ListUtil.list(true);
+    protected final List<Table> investmentIndustryTables = ListUtil.list(true);
     private final List<Table> navPerformanceTables = ListUtil.list(true);
-    private final List<Table> financialIndicatorsTables = ListUtil.list(true);
-    private final List<Table> shareChangeTables = ListUtil.list(true);
-    private final List<Table> assetAllocationTables = ListUtil.list(true);
-    private final List<Table> investmentIndustryTables = ListUtil.list(true);
 
     public PDQuarterlyReportParser(EmailFieldMappingMapper fieldMappingMapper) {
         super(fieldMappingMapper);
@@ -52,10 +101,10 @@ public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyRep
                 // 净值表现（未入库）
                 this.navPerformanceTables.add(table);
             } else if (colCount == 2) {
-                // 用表格的第二行第一列的数据判断是否份额变动记录
-                String text = this.cleaningValue(table.getCell(1, 0).getText());
+                // 用表格的第一列的数据判断是否份额变动记录
+                List<String> texts = this.getTableColTexts(table, 0);
                 // 主要财务指标或份额变动
-                if (StrUtil.contains(text, "份额")) {
+                if (CollUtil.containsAny(texts, SHARE_CHANGE_COLUMN_NAMES)) {
                     this.shareChangeTables.add(table);
                 } else {
                     this.financialIndicatorsTables.add(table);
@@ -64,9 +113,9 @@ public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyRep
                 // 行业配置
                 this.investmentIndustryTables.add(table);
             } else if (colCount == 3) {
-                // 用表格的第一个单元格判断是否资产配置表
-                String text = this.cleaningValue(table.getCell(0, 0).getText());
-                if (StrUtil.contains(text, "行业类别")) {
+                // 用表格的第一列单元格判断是否资产配置表
+                List<String> texts = this.getTableColTexts(table, 0);
+                if (CollUtil.containsAny(texts, INDUSTRY_COLUMN_NAMES)) {
                     this.investmentIndustryTables.add(table);
                 } else {
                     this.assetAllocationTables.add(table);
@@ -90,7 +139,7 @@ public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyRep
     }
 
     @Override
-    protected QuarterlyReportData parseExtInfoAndSetData(ReportBaseInfoDTO reportInfo, ReportFundInfoDTO fundInfo) {
+    protected T parseExtInfoAndSetData(ReportBaseInfoDTO reportInfo, ReportFundInfoDTO fundInfo) {
         Integer fileId = reportInfo.getFileId();
         // 表格转换数据获取函数
         Function<Table, Map<String, Object>> function = t -> {
@@ -118,11 +167,30 @@ public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyRep
         reportData.setFinancialIndicators(financialIndicators);
         reportData.setAssetAllocation(assetAllocations);
         reportData.setInvestmentIndustry(investmentIndustries);
-        return reportData;
+        return this.buildExtData(reportInfo, fundInfo, shareChanges, assetAllocations, investmentIndustries, function);
+    }
+
+    protected T buildExtData(ReportBaseInfoDTO reportInfo, ReportFundInfoDTO fundInfo,
+                             List<ReportShareChangeDTO> shareChanges,
+                             List<ReportAssetAllocationDTO> assetAllocations,
+                             List<ReportInvestmentIndustryDTO> investmentIndustries,
+                             Function<Table, Map<String, Object>> function) {
+        Integer fileId = reportInfo.getFileId();
+        // 主要财务指标
+        List<ReportFinancialIndicatorsDTO> financialIndicators = this.buildLevelDto(fileId, this.financialIndicatorsTables,
+                ReportFinancialIndicatorsDTO.class, function);
+        QuarterlyReportData reportData = new QuarterlyReportData(reportInfo, fundInfo);
+        reportData.setShareChange(shareChanges);
+        reportData.setFinancialIndicators(financialIndicators);
+        reportData.setAssetAllocation(assetAllocations);
+        reportData.setInvestmentIndustry(investmentIndustries);
+        @SuppressWarnings("unchecked")
+        T t = (T) reportData;
+        return t;
     }
 
     @Override
-    protected void cleaningReportData(QuarterlyReportData reportData) {
+    protected void cleaningReportData(T reportData) {
         // todo 数据清洗
     }
 
@@ -207,4 +275,22 @@ public class PDQuarterlyReportParser extends AbstractPDReportParser<QuarterlyRep
         }
         return dtos;
     }
+
+    /**
+     * 获取表格指定列的所有文字内容
+     *
+     * @param table 表格
+     * @param col   指定列
+     * @return /
+     */
+    protected List<String> getTableColTexts(Table table, Integer col) {
+        List<String> details = ListUtil.list(false);
+        for (@SuppressWarnings("all") List<RectangularTextContainer> row : table.getRows()) {
+            String detail = this.cleaningValue(row.get(col).getText(), false);
+            if (StrUtil.isNotBlank(detail)) {
+                details.add(detail);
+            }
+        }
+        return details;
+    }
 }

service-deploy/src/test/java/com/simuwang/ApplicationTest.java

@@ -45,8 +45,8 @@ public class ApplicationTest {
     @Test
     public void reportTest() {
         MailboxInfoDTO emailInfoDTO = this.buildMailbox();
-        Date startDate = DateUtil.parse("2024-10-10 16:40:30", DateConst.YYYY_MM_DD_HH_MM_SS);
-        Date endDate = DateUtil.parse("2024-10-10 19:59:30", DateConst.YYYY_MM_DD_HH_MM_SS);
+        Date startDate = DateUtil.parse("2024-10-11 08:30:30", DateConst.YYYY_MM_DD_HH_MM_SS);
+        Date endDate = DateUtil.parse("2024-10-11 19:59:30", DateConst.YYYY_MM_DD_HH_MM_SS);
         try {
             emailParseService.parseEmail(emailInfoDTO, startDate, endDate);
         } catch (Exception e) {