修复店铺任务数据覆盖bug

2026-08-29 00:16:47 +08:00
parent 79596a970c
commit 322905607a
5 changed files with 120 additions and 88 deletions
@@ -14,7 +14,9 @@ import org.apache.poi.ss.usermodel.Drawing;
 import org.apache.poi.ss.usermodel.Row;
 import org.apache.poi.ss.usermodel.Sheet;
 import org.apache.poi.ss.usermodel.Workbook;
+import org.apache.poi.xssf.usermodel.XSSFDrawing;
 import org.apache.poi.xssf.usermodel.XSSFWorkbook;
+import org.openxmlformats.schemas.drawingml.x2006.spreadsheetDrawing.CTTwoCellAnchor;
 import org.springframework.core.io.ClassPathResource;
 import org.springframework.stereotype.Service;

@@ -66,10 +68,12 @@ public class ShopDataCrawlExcelAssemblyService {
    }

    /**
-     * Appends the supplied task rows to an already assembled daily workbook.
-     * Existing rows, styles and drawings are deliberately left untouched.
+     * Replaces the supplied countries' rows inside an already assembled daily
+     * workbook. Sheets for countries that carry no new rows are left untouched
+     * (rows, styles and drawings preserved). Returns the total data-row count
+     * across all sheets after replacement.
     */
-    public void appendWorkbook(File baseXlsx, File outputXlsx, List<ShopDataCrawlResultItemVo> items) {
+    public int replaceCountriesWorkbook(File baseXlsx, File outputXlsx, List<ShopDataCrawlResultItemVo> items) {
        if (baseXlsx == null || !baseXlsx.isFile()) {
            throw new BusinessException("当天累计文件不存在");
        }
@@ -84,14 +88,17 @@ public class ShopDataCrawlExcelAssemblyService {
            for (int i = 0; i < COUNTRIES.size(); i++) {
                List<ShopDataCrawlRowDto> rows = rowsByCountry.get(COUNTRIES.get(i));
                if (rows != null && !rows.isEmpty()) {
-                    appendSheet(workbook, workbook.getSheetAt(i), rows, imageCache, pictureIndexes);
+                    Sheet sheet = workbook.getSheetAt(i);
+                    clearSheetPictures(sheet, pictureIndexes);
+                    writeSheet(workbook, sheet, rows, imageCache, pictureIndexes);
                }
            }
            workbook.write(output);
+            return totalDataRows(workbook);
        } catch (BusinessException ex) {
            throw ex;
        } catch (Exception ex) {
-            throw new BusinessException("店铺数据抓取累计 Excel 追加失败: " + ex.getMessage());
+            throw new BusinessException("店铺数据抓取累计 Excel 按国家替换失败: " + ex.getMessage());
        }
    }

@@ -155,54 +162,21 @@ public class ShopDataCrawlExcelAssemblyService {
                sheet.removeRow(row);
            }
        }
+        clearSheetPictures(sheet, pictureIndexes);
        int rowIndex = 1;
        for (ShopDataCrawlRowDto value : rows == null ? List.<ShopDataCrawlRowDto>of() : rows) {
            Row row = sheet.createRow(rowIndex++);
-            String[] values = {value.getDate(), value.getAsin(), "", value.getInventorySales(), value.getSalesRank(),
-                    value.getPageViews(), value.getUnitsSold(), value.getPrice(), value.getRecommendedOffer(), value.getBrand()};
-            for (int column = 0; column < values.length; column++) {
-                Cell cell = row.createCell(column);
-                if (styles[column] != null) cell.setCellStyle(styles[column]);
-                cell.setCellValue(values[column] == null ? "" : values[column]);
-            }
-            if (!blank(value.getCommodityImage())) {
-                row.setHeightInPoints(IMAGE_ROW_HEIGHT_POINTS);
-                embedImage(workbook, sheet, row, value.getCommodityImage(), imageCache, pictureIndexes);
-            }
+            writeDataRow(workbook, sheet, row, value, styles, imageCache, pictureIndexes);
        }
    }

-    private void appendSheet(XSSFWorkbook workbook,
-                             Sheet sheet,
-                             List<ShopDataCrawlRowDto> rows,
-                             Map<String, SimilarAsinImageEmbedder.ResizedImage> imageCache,
-                             Map<String, Integer> pictureIndexes) {
-        Row header = sheet.getRow(0);
-        Row styleRow = sheet.getRow(1);
-        boolean currentTemplate = header != null && "商品图片".equals(cellText(header, IMAGE_COLUMN));
-        boolean templateHasBrand = header != null && "品牌".equals(cellText(header, BRAND_COLUMN));
-        CellStyle[] styles = new CellStyle[HEADERS.size()];
-        for (int column = 0; column < styles.length; column++) {
-            int sourceColumn = templateColumnForOutput(column, currentTemplate, templateHasBrand);
-            Cell cell = styleRow == null ? null : styleRow.getCell(sourceColumn);
-            styles[column] = cell == null ? null : cell.getCellStyle();
-        }
-        writeHeaders(sheet, currentTemplate, templateHasBrand);
-        sheet.setColumnWidth(IMAGE_COLUMN, IMAGE_COLUMN_WIDTH);
-        int rowIndex = Math.max(1, sheet.getLastRowNum() + 1);
-        for (ShopDataCrawlRowDto value : rows) {
-            Row row = sheet.createRow(rowIndex++);
-            writeRow(workbook, sheet, row, value, styles, imageCache, pictureIndexes);
-        }
-    }
-
-    private void writeRow(XSSFWorkbook workbook,
-                          Sheet sheet,
-                          Row row,
-                          ShopDataCrawlRowDto value,
-                          CellStyle[] styles,
-                          Map<String, SimilarAsinImageEmbedder.ResizedImage> imageCache,
-                          Map<String, Integer> pictureIndexes) {
+    private void writeDataRow(XSSFWorkbook workbook,
+                              Sheet sheet,
+                              Row row,
+                              ShopDataCrawlRowDto value,
+                              CellStyle[] styles,
+                              Map<String, SimilarAsinImageEmbedder.ResizedImage> imageCache,
+                              Map<String, Integer> pictureIndexes) {
        String[] values = {value.getDate(), value.getAsin(), "", value.getInventorySales(), value.getSalesRank(),
                value.getPageViews(), value.getUnitsSold(), value.getPrice(), value.getRecommendedOffer(), value.getBrand()};
        for (int column = 0; column < values.length; column++) {
@@ -306,4 +280,25 @@ public class ShopDataCrawlExcelAssemblyService {
        }
        return result;
    }
+
+    private void clearSheetPictures(Sheet sheet, Map<String, Integer> pictureIndexes) {
+        Drawing<?> patriarch = sheet.getDrawingPatriarch();
+        if (!(patriarch instanceof XSSFDrawing drawing)) {
+            return;
+        }
+        List<CTTwoCellAnchor> anchors = drawing.getCTDrawing().getTwoCellAnchorList();
+        for (int i = anchors.size() - 1; i >= 0; i--) {
+            if (anchors.get(i).getPic() != null) {
+                drawing.getCTDrawing().removeTwoCellAnchor(i);
+            }
+        }
+    }
+
+    private int totalDataRows(XSSFWorkbook workbook) {
+        int total = 0;
+        for (int i = 0; i < workbook.getNumberOfSheets(); i++) {
+            total += Math.max(0, workbook.getSheetAt(i).getLastRowNum());
+        }
+        return total;
+    }
 }
@@ -1704,12 +1704,14 @@ public class ShopDataCrawlTaskService {
            }

            int addedRowCount = excelAssemblyService.countRows(List.of(snapshot));
+            ShopDataCrawlDailyFileEntity baseDailyFile = resolveBaseDailyFile(
+                    preparation.dailyFile(), userId, shopKeyHash, businessDate);
            DailyWorkbookArtifact artifact = assembleDailyWorkbook(
-                    task, snapshot, preparation.dailyFile(), addedRowCount);
+                    task, snapshot, baseDailyFile, addedRowCount);
            try {
                persistedResult = executeShortTransaction(() -> persistDailyAggregation(
                        task, row, userId, shopKey, shopKeyHash, businessDate,
-                        preparation, artifact, addedRowCount));
+                        preparation, artifact));
                if (persistedResult.discardUploadedObject() && artifact.uploaded()) {
                    deleteObjectQuietly(artifact.objectKey());
                }
@@ -1743,23 +1745,35 @@ public class ShopDataCrawlTaskService {
        return new DailyAggregationPreparation(dailyFile, false);
    }

+    private ShopDataCrawlDailyFileEntity resolveBaseDailyFile(ShopDataCrawlDailyFileEntity currentDailyFile,
+                                                              Long userId,
+                                                              String shopKeyHash,
+                                                              LocalDate businessDate) {
+        if (currentDailyFile != null) {
+            return currentDailyFile;
+        }
+        List<ShopDataCrawlDailyFileEntity> older = dailyFileService.findOlder(userId, shopKeyHash, businessDate);
+        return older == null || older.isEmpty() ? null : older.get(0);
+    }
+
    private DailyWorkbookArtifact assembleDailyWorkbook(FileTaskEntity task,
                                                        ShopDataCrawlResultItemVo snapshot,
-                                                        ShopDataCrawlDailyFileEntity dailyFile,
+                                                        ShopDataCrawlDailyFileEntity baseDailyFile,
                                                        int addedRowCount) {
-        String filename = dailyFile != null && !blank(dailyFile.getResultFilename())
-                ? dailyFile.getResultFilename()
+        String filename = baseDailyFile != null && !blank(baseDailyFile.getResultFilename())
+                ? baseDailyFile.getResultFilename()
                : buildTaskWorkbookFilename(task);
-        String existingObjectKey = dailyFile == null ? null : dailyFile.getResultFileUrl();
+        String existingObjectKey = baseDailyFile == null ? null : baseDailyFile.getResultFileUrl();

        // A result with no new rows only needs a new database membership. Reusing
        // the canonical daily object avoids both a local copy and an OSS round trip.
-        if (dailyFile != null && addedRowCount == 0 && !blank(existingObjectKey)) {
+        if (baseDailyFile != null && addedRowCount == 0 && !blank(existingObjectKey)) {
            return new DailyWorkbookArtifact(
                    existingObjectKey,
-                    Math.max(0L, Objects.requireNonNullElse(dailyFile.getResultFileSize(), 0L)),
+                    Math.max(0L, Objects.requireNonNullElse(baseDailyFile.getResultFileSize(), 0L)),
                    false,
-                    filename);
+                    filename,
+                    Math.max(0, Objects.requireNonNullElse(baseDailyFile.getRowCount(), 0)));
        }

        File workRoot = FileUtil.mkdir(FileUtil.file(
@@ -1770,21 +1784,25 @@ public class ShopDataCrawlTaskService {
        File baseXlsx = FileUtil.file(workRoot, "base.xlsx");
        File outputXlsx = FileUtil.file(workRoot, filename);
        try {
-            if (dailyFile != null && !blank(existingObjectKey)) {
+            if (baseDailyFile != null && !blank(existingObjectKey)) {
                try {
                    Files.write(baseXlsx.toPath(), ossStorageService.readObjectBytes(existingObjectKey));
                } catch (Exception ex) {
-                    throw new BusinessException("读取当天累计文件失败: " + safeMessage(ex));
+                    throw new BusinessException("读取累计文件失败: " + safeMessage(ex));
                }
-                excelAssemblyService.appendWorkbook(baseXlsx, outputXlsx, List.of(snapshot));
-            } else {
-                excelAssemblyService.writeWorkbook(outputXlsx, List.of(snapshot));
+                int rowCount = excelAssemblyService.replaceCountriesWorkbook(baseXlsx, outputXlsx, List.of(snapshot));
+                String objectKey = ossStorageService.uploadResultFile(outputXlsx, MODULE_TYPE);
+                if (blank(objectKey)) {
+                    throw new BusinessException("累计文件上传后未返回文件地址");
+                }
+                return new DailyWorkbookArtifact(objectKey, outputXlsx.length(), true, filename, rowCount);
            }
+            excelAssemblyService.writeWorkbook(outputXlsx, List.of(snapshot));
            String objectKey = ossStorageService.uploadResultFile(outputXlsx, MODULE_TYPE);
            if (blank(objectKey)) {
-                throw new BusinessException("当天累计文件上传后未返回文件地址");
+                throw new BusinessException("累计文件上传后未返回文件地址");
            }
-            return new DailyWorkbookArtifact(objectKey, outputXlsx.length(), true, filename);
+            return new DailyWorkbookArtifact(objectKey, outputXlsx.length(), true, filename, addedRowCount);
        } finally {
            FileUtil.del(baseXlsx);
            FileUtil.del(outputXlsx);
@@ -1799,8 +1817,7 @@ public class ShopDataCrawlTaskService {
                                                            String shopKeyHash,
                                                            LocalDate businessDate,
                                                            DailyAggregationPreparation preparation,
-                                                            DailyWorkbookArtifact artifact,
-                                                            int addedRowCount) {
+                                                            DailyWorkbookArtifact artifact) {
        ShopDataCrawlDailyFileEntity dailyFile = dailyFileService.findForUpdate(
                userId, shopKeyHash, businessDate);
        if (handleExistingDailyMembership(row, dailyFile)) {
@@ -1826,9 +1843,7 @@ public class ShopDataCrawlTaskService {
        row.setResultFileUrl(objectKey);
        row.setResultFileSize(artifact.fileSize());
        row.setResultContentType(CONTENT_TYPE_XLSX);
-        row.setRowCount(dailyFile == null
-                ? addedRowCount
-                : Math.max(0, Objects.requireNonNullElse(dailyFile.getRowCount(), 0)) + addedRowCount);
+        row.setRowCount(artifact.rowCount());
        fileResultMapper.updateById(row);

        LocalDateTime now = dailyFileService.currentBusinessDateTime();
@@ -2140,7 +2155,12 @@ public class ShopDataCrawlTaskService {
    private record DailyWorkbookArtifact(String objectKey,
                                         long fileSize,
                                         boolean uploaded,
-                                         String filename) {
+                                         String filename,
+                                         int rowCount) {
+
+        DailyWorkbookArtifact(String objectKey, long fileSize, boolean uploaded, String filename) {
+            this(objectKey, fileSize, uploaded, filename, 0);
+        }
    }

    private record DailyAggregationResult(List<String> obsoleteObjectKeys,