diff --git a/backend-java/src/main/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookService.java b/backend-java/src/main/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookService.java index d1b990948..b171fbe3e 100644 --- a/backend-java/src/main/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookService.java +++ b/backend-java/src/main/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookService.java @@ -170,7 +170,7 @@ public class PublishWorkbookService { row.setSourceId(values.get(0)); row.setAsin(values.get(1)); row.setCountry(values.get(2)); - row.setBrand(values.get(3)); + row.setBrand(stripBrandLabel(values.get(3))); row.setPrice(values.get(4)); return row; } @@ -293,6 +293,22 @@ public class PublishWorkbookService { return value.isBlank() ? "publish" : value; } + /** + * 剥掉亚马逊 byline 残留的「Brand:」标签前缀。 + * + * 这类值送进 WIPO 品牌查询会被当成 Solr 字段限定符,返回 + * isError / status 500 / undefined field Brand(2026-10-08 主机 A 实测 19,247 次, + * 不带前缀的查询 0 次)。只剥开头一次:值中间的冒号属品牌自身内容, + * "Brands:" 也不是标签。 + */ + private String stripBrandLabel(String value) { + if (value == null || value.isEmpty()) { + return value; + } + return value.replaceFirst( + "(?i)^\\s*(?:brand|品牌|marke|marca|marque|marka)\\s*[::]\\s*", "").trim(); + } + private String normalize(String value) { if (value == null) { return ""; diff --git a/backend-java/src/test/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookServiceTest.java b/backend-java/src/test/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookServiceTest.java index bf8e6f14d..49d62048a 100644 --- a/backend-java/src/test/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookServiceTest.java +++ b/backend-java/src/test/java/com/nanri/aiimage/modules/publish/service/PublishWorkbookServiceTest.java @@ -171,6 +171,51 @@ class PublishWorkbookServiceTest { } } + @Test + void stripsAmazonBylineLabelFromBrandButKeepsLookalikes() throws Exception { + // 2026-10-08:品牌列带 "Brand: " 前缀的行送进 WIPO 会被当 Solr 字段限定符, + // 回 undefined field Brand(主机 A 实测 19,247 次;不带前缀的查询 0 次)。 + // 上架 Excel 是这类脏值的入口,所以在这里拦;后两条是排他断言, + // 打死「无条件砍掉开头若干字符」的写法。 + Path directory = Files.createTempDirectory("publish-brand-"); + try { + String[] brands = { + "Brand: TOWN & COUNTRY BASICS", + "\u54c1\u724c\uff1aNIKE", + "Brands: NIKE", + "Brand: NIKE: X", + }; + File file = directory.resolve("brand.xlsx").toFile(); + try (Workbook workbook = new XSSFWorkbook(); + FileOutputStream output = new FileOutputStream(file)) { + Sheet sheet = workbook.createSheet("UK"); + Row header = sheet.createRow(0); + for (int index = 0; index < PublishWorkbookService.SOURCE_HEADERS.size(); index++) { + header.createCell(index) + .setCellValue(PublishWorkbookService.SOURCE_HEADERS.get(index)); + } + for (int index = 0; index < brands.length; index++) { + Row data = sheet.createRow(index + 1); + data.createCell(0).setCellValue(String.valueOf(index + 1)); + data.createCell(1).setCellValue("B00" + index); + data.createCell(2).setCellValue("UK"); + data.createCell(3).setCellValue(brands[index]); + data.createCell(4).setCellValue(50); + } + workbook.write(output); + } + + List rows = service.parse(file).rows(); + assertEquals(brands.length, rows.size()); + assertEquals("TOWN & COUNTRY BASICS", rows.get(0).getBrand()); + assertEquals("NIKE", rows.get(1).getBrand()); + assertEquals("Brands: NIKE", rows.get(2).getBrand()); + assertEquals("NIKE: X", rows.get(3).getBrand()); + } finally { + FileUtil.del(directory.toFile()); + } + } + private void writeSourceSheet(Sheet sheet, String country, String asin) { Row header = sheet.createRow(0); for (int index = 0; index < PublishWorkbookService.SOURCE_HEADERS.size(); index++) {